Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
'-:-;!'
. -
.:;;
t. ''
Economa
Urbana
QUINTA
EDICIN
''
1
,>
LIB OS&
COMPANIA
Calle 19 No. 3-16 Local 104
librosycomp@hotmail.com .
Telefax: 341 71 00 Bogot, D.C. ..
McGraw-Hill
Lewi
n
Bastan Burr Ridge, IL Dubuque, lA Madison, Wl New York Sar Francisco St. Louis
Bangkok Bogot Caracas Kuala Lumpur Lisbon London Madrid Mexico City
Milan Montreal New Delhi Santiago Seoul Singapore Sydney Taipei Toronto
Sorne auxiliares, incluyendo componentes electrnicos e impresos, pueden no estar disponibles para clientes fuera de los Estados Unidos.
3 4 5 6 7 8 9 O DOCIDOC O 9 8 7 6 5 4 3
2 3 4 5 6 7 8 9 O DOC!DOC O 9 8 7 6 5 4 3
ISBN 0-07-248784-4
Publisher: Gary Burke
Executive sponsoring editor: Paul Shensa
Editorial assistant: Jennifer Leon
Marketing manager: Martin D. Quinn
Producer, Media technology: Melissa Kansa
Project manager: Natalie J. Ruffatto
Senior production supervisor: Michael R. McCormick
Senior designer: Pam Verros
Supplement producer: Matthew Perry
Cover design: John Isaacs Design
Cover art: Leger, Fernand, The City, 1919, Philadelphia Museum of Art: A.E. Gallatin Collection.
"Leger captures the cold steel surfaces of the urban landscape... cubes and cylinders, evoke the metallic
textures of industry. The jumbled girders, poles, high walls, and steps, together with the human silhouettes,
create a sense of the anonymous, mechanical movement associated with the fast pace of city life."
Quoted from Adams, Laurie Schneider. Art Across Time, 2nd ed. New York: McGraw-Hill, 2002.
p. 897. The material is reproduced with permission of The McGraw-Hill Companies.
Typeface: 10112 Times Roman
Compositor: Interactive Composition Corporation
Printer: R.R. Donnelley & Sons Company
Preface
Este libro utiliza anlisis econmico para explicar por qu existen las ciudades,
donde se desarrollan, crecen, y cmo se arreglan las diferentes actividades dentro de las
ciudades. Tambin se estudia la economa de la pobreza urbana, vivienda inadecuada, la
segregacin, problemas de congestin; la contaminacin y el crimen.El texto est
diseado para su uso en cursos de pregrado en economa urbana y asuntos
urbanos. Tambin utilizarlo para cursos de posgrado en
planificacin urbana, las polticas pblicas y administracin
pblica. Todos los conceptos econmicos usados en el libro estn
cubiertos en el curso tpico de Microeconoma intermedia, As que
los estudiantes que han completado un curso ser capaces de moverse a travs del libro a
un ritmo acelerado. Para los estudiantes cuya exposicin a Microeconoma islimited a un
curso introductorio o que podran beneficiarse de una revisin de los conceptos cubiertos
en un curso de Microeconoma intermedia-me han proporcionado un apndice
(herramientas de la microeconoma) que cubre los conceptos esenciales.
Qu es nuevo en la 5 edicin?
Hay muchos cambios en el libro para la quinta edicin. Las tablas y grficos han sido completamente
actualizados con los datos ms recientes. A travs del libro, he incorporando las ideas y los hechos de
recientes investigaciones tericas y empricas. El anlisis de la poltica ha sido updatedto refiect
recientes cambios en las polticas pblicas y refinamientos en el anlisis econmico de alternativas
polticas. Aqu estn los aspectos ms destacados de los cambios ms importantes en el libro, organizado
por las seis partes del libro.
vili
Pref
ace
promover el aprendizaje entre los trabajadores y (e) por qu las ciudades generan
beneficios de consumo. Adems de explicar por qu existen ciudades, captulo 2 aborda
los lmites del crecimiento urbano, centrndose en crecientes distancias de transporte.
Un nuevo captulo 3, grandes y pequeas ciudades, explica las fuerzas del mercado que
generan las grandes ciudades, con trabajos tericos y empricos recientes en economas
externas de escala en la produccin. El captulo tambin explica por qu las ciudades
difieren en tamao y alcance.
La ltima parte del captulo 8 explica las interacciones entre el mercado laboral urbano y los mercados de tierra urbana, mediante la introduccin
de un sistema de tranvas para ilustrar las interacciones del mercado. Esta presentacin es una alternativa para el anlisis de equilibrio general ms
compleja en la edicin anterior. En captulos posteriores, utilizamos el marco desarrollado en este captulo para discutir los efectos de diferentes
polticas (un servicio urbano de lmites y congestin tax.es) en el mercado laboral y el mercado de tierras.
Estos captulos fueron trasladados hacia adelante en el libro para proporcionar una transicin suave desde el anlisis
.de patrones de uso de la tierra a la poltica de transporte urbano.
ix
Preface
PARTE V: LA
VIVIENDA
Los captulos de vivienda estn ahora en su propia seccin del libro, con la discusin de la falta de vivienda incluido en el
primer captulo. La discusin de la segregacin racial ha sido trasladada al captulo de la clasificacin.
Los captulos sobre los gobiernos locales se han consolidado en dos captulos, uno sobre gasto local y un segundo sobre
ingresos locales.
Agradecimientos
Estoy en deuda con muchos personas que lean el libro y sugiere formas de mejorar la cobertura y la exposicin. En
particular, me gustara dar las gracias a los profesores que participaron en las encuestas y comentarios que son
indispensables en el desarrollo de la quinta edicin de Urban Economics.The aparicin de sus nombres no necesariamente
constituye su endorsernent el texto o su rnethodology.
( '
l.
Ed Coulson
Stuart S. Rosenthal Virginia Polytechnic Institute y la Universidad Estatal
/
David Figlio
University of Oregon
Mary Stevenson
. University of Massachusetts, Boston
Andrew Gold
Trinity College
1
John Yinger
Syracuse University
Kenneth Lipner
Florida Intemational University
Arthur O'Sullivan
1
1
i
1
Contents in Brief
/
Chapter 1
Introduction
Parti
MARKET FORCES IN THE DEVELOPMENT OF CITIES
Chapter 2
19
Chapter 3
39
65
Chapter 5
92
Chapter 6
119
Part II
LAND RENT AND LAND USE PATTERNS
Chapter 7
Chapter 8
167
Chapter 9
202
153
228
xi
xiv
Contents in Brief
PartiD
URBAN TRANSPORTATION
Chapter 11 Autos and Highways
257
291
PartN URBAN
PROBLEMS
Chapter 13 Household Sorting
323
351
372
397
PartV
HOUSING
Chapter 17 Why Is Housing Different?
433
467
PartVI
LOCAL GOVERNMENT
Chapter 19 Local Government Spending
503
524
Contents
Captulo 1 Introduccin 1 Qu es economa urbana? 1 el papel de las ciudades en la economa 2 creacin e innovacin en
ciudades 2 aprendizaje en ciudades 3 comercio y la produccin en el consumo de las ciudades 4 en ciudades 5 Qu es
urbano? 5 censo De_JJ.nitions 6 suma, mary: lengua de urbano economa 13 Coming'Attractions 13 las fuerzas del
mercado en el desarrollo de ciudades 14 uso de la tierra dentro de metropolitana reas 14 urbano transporte 14 los
problemas urbanos y las polticas pblicas 15 vivienda y gastos del Gobierno Local polticas pblicas 15 y 15 impuestos
Partl mercado fuerzas en las DEVELOP1\1ENT ciudades
Captulo 2 por qu existen las ciudades? 19 una regin sin ventaja comparativa
ciudades comercio 19 ciudades 20 genera comercio 20
XV
xvi
J,
'
Content
xv
Captulo 4 donde ubicar las empresas? 65 transferible entradas y salidas 66 orientada a recursos firmas 67 empresas orienta das al mercado 70 escalan las economas en el
transporte 71 el principio de mediana ubicacin 72 puntos de transbordo y ciudades portuarias 73 energa orientacin 75 local izacin y urbanizacin economas 76 mercados y
opciones de ubicacin 76 transporte coste versus costo laboral 76 laborales a largo plazo 78 laboral y amenidades naturales aprendizaje 78 y 79 de ubicacin laboral
80 impuestos servicios pblicos e impuestos locales y ubicacin DecisioBs 80 servicios pblicos y las decisiones de ubicacin 81 subsidios y deportes
profesionales programas de incentivos 81, estadios y trabajos de 82 casos y resultados empricos 83 el. Semiconductor 84 la industria japonesa del
automvil firmas 84 Mex.ican prendas de vestir industria 85 auto: animal domstico fabricacin y localizacin economas 85 GM; s Saturn planta 86
efectos de salarios y los sindicatos 87 captulo 5 reas de mercado y Central lugar teora 92 equilibrio con 92 competencia monopolstica los precios con un
monopolista 93 entrada y competencia 94 eficiencia Trade-Offs 95
Determinantes del mercado reas 97 An Algebraic modelo de mercado reas 98 cambios en poblacin densidad
98 Area de mercado y la demanda y S cale economas 99 Market Area y viajes costos 99 Parcel Post y pulperas
100 Market Area y renta 100 la desaparicin de pequeas tiendas 101: mercado de reas de diferentes
industrias 102
xvili
Contents
Teora del lugar central 103 un lugar Central Simple modelo 103 relajante supuestos 107 otros tipos de industria 108
Teora del lugar central y los estudios empricos del mundo Real 108 de lugar Central tamao 109 sistemas
distribucin de ciudades: la regla Rank-Size 109 Urban Giants: The Puzzle ofthe Large Primary City 112 The Role of Trade 112 The Effect
ofTrade Restrictions 113 The Role of Infrastructure and Politics 114
xix
Contents
Part II
LAND RENT AND LAND USE PATTERNS
Captulo 7 Introduccin a tierra Alquiler tierra 153 Alquiler Alquiler Alquiler terreno y fertilidad competencia y las interacciones
polticas pblicas mercado 159 155 157 157 el Coro leyes Debate 159 precio de la vivienda y tierra precios impuestos 161 161 tierra
alternativas al impuesto nico. 163 Tax.ation de terrenos urbanos en Pittsburgh 163 captulo 8 uso de la tierra en el. Uso de tierra de la
ciudad de monocentric en el Distrito Central de negocios: Faetones y oficinas 168 el oferta alquiler de Factor fijo fabricantes
proporciones 168 entrada de sustitucin y la funcin de oferta-alquiler convexa 170 utilizar la funcin de oferta-alquiler de oficina
empresas 173 tierras en el Distrito Central de negocios 174 terrenos residenciales utilizan 177 funcin lineal de la vivienda-precio:
ningn consumidor Substittition 177 167 C. ? sumer sustitucin y una convexa vivienda-precio funcin The.Residential oferta-alquiler
funcin 181 densidad 182 tierra uso residencial en el Monocentric ciudad 183 relajante los supuestos 185 desplazamientos supuestos
185 variacin espacial en otros atributos Locational 186. ...XX
Contents
Captulo 9 uso del suelo en modernas ciudades el espacial distribucin de empleo y poblacin la suburbanizacin de
poblacin 205 suburbanizacin de mayores ingresos y menor transporte costo 206 los problemas urbanos y suburbanizacin
207 suburbanizacin de fabricacin 208 el Intracity camin 208 el Intercity camin 209 otros factores: automvil, un piso
pants y aeropuertos 21O suburbanizacin de la oficina de empleo 212 suburbanizacin y Clustering de oficinas 212
tecnologa de las comunicaciones y oficina suburbanizacin 214 personas seguir empleos y puestos de trabajo siga gente
215 suburbano Subcentros 216 subcentros en Los ngeles 217 subcentros en Houston 219 subcentros en Chicago borde
219 ciudades 219 subcentros y centros urbanos 221 202 202 captulo JO tierra utilizan controles y zonificacin control
poblacin crecimiento 228 crecimiento urbano lmite 228 edificio permisos de zonificacin de uso de la tierra 231: tipos y
efectos en el mercado de zonificacin para molestias Industrial 233. Rendimiento zonificacin Retail 236
Nuisa...Contents
xxi
PartiD
URBAN TRANSPORTATION
Captulo 11 Autos y congestin de las carreteras: equilibrio frente a un ptimo volumen 258 de
trfico la demanda de transporte urbano 259 el privado y Social los costos del viaje 260 equilibrio
frente a volumen de trfico ptima 261 la respuesta poltica: congestin 262 beneficios fiscales y los
costes de congestin Tax.es 262 pico contra pico viajan 264 estimaciones de congestin impuestos
265 aplicando el impuesto de congestin. 266 experiencias con congestin precios 266 Altematives
a una congestin 267 gasolina impuestos. 268 aparcamiento impuestos 268 expansin de
capacidad y trfico diseo 269 subsidios para trnsito 271 carretera precios y volumen de trfico en
el largo plazo 273 Costo Total medio de dos carriles y carreteras de cuatro carriles 273 largo plazo
costo promedio y costo Marginal curvas 275 ptimo volumen y camino ancho 275 Cong stion pagar
peajes por el ptimo camino 275 quin paga por las carreteras? 277 257 congestin y uso de la
tierra Pattems congestin impuesto. y residencial tierra uso 277 277 congestin Tax.e,s y las Autos
de mercado laboral urbano andA...:xxii
Contents
Un ejemplo de eleccin Modal 295 instalaciones servicio y eleccin Modal vehculos de alta ocupacin (HOV) 297 298
elegir un sistema de trnsito: Problem del planificador Costo 300 del coste del autobs sistema 301 costo del BART 302
sistema eleccin 302 luz Auto sistema 300 303 subsidios del carril para trnsito masivo 305 justificacin para trnsito
subsidios 305 razones para la reforma de la poltica de trnsito dficits 307: contratacin y Paratransit 308 contratacin
por servicios de trnsito 309 desregulacin y servicios de paratrnsito suscripcin 309 cercanas -Precios 311 Resumen
de paratrnsito 311 experiencia britnica con la desregulacin 312 transporte y uso de la tierra Pattems gasolina 312 313
sistemas fijos-Rail y uso de la tierra 313
PartiV
URBAN PROBLEMS
Captulo 13 hogar clasificacin clasificacin por las caractersticas de bienes pblicos 323
exigeuna Local de bienes pblicos locales 324 la cantidad eficiente de bienes pblicos
locales 325 el votante mediano recoge el Gobierno Local equilibrio cantidad 327: votando
con los pies 328 pujando por tierra y clasificacin 330 pruebas de clasificacin para los
bienes pblicos locales 333 el impuesto predial y clasificacin por demanda de vivienda
333 los impuestos sobre la propiedad y exclusin Zbning 334 el impuesto predial aumenta
el nmero de jurisdicciones 335 clasificacin para Peer grupos 335 licitacin para Peer
clasificacin 336 mezcla en lugar de la clasificacin y la tierra? 338323
xxiii
Contents
37
2
xxiv
Contents
Vales de educacin ciudades centrales 389 aumento de gasto en educacin ciudades centrales
390 390 vales dirigidos para familias de bajos ingresos 391
Captulo 16 delincuencia urbana
crimen hechos 397 las vctimas
del delito de crimen 398 y el
precio de la vivienda 399 los
costos del crimen 401
l:
:
1:
;.
1
397
417
'
i
l
'
XXV
Contents
PartV
HOUSING
Captulo 17 por qu es diferente la vivienda?
Heterogeneidad y lmmobility 433 el precio de la
vivienda: enfoque hednico el barrio efectos 435
comodidades y Disamenities 435 elegir una vivienda
437 segmentado pero relacionados con los mercados
438
xxvi
Contents
Condiciones de vivienda pblica y vivienda consumo 472 vida del 474 viviendas
pblicas y bienestar destinatario en vivienda pblica 476 los efectos en el mercado de la
vivienda pblica 476 oferta poltica: subsidios para privado 479 demanda poltica de
vivienda: subvenciones a los consumidores 479 Rent Certificates: Section 8 480 Housing Vouchers
481 Market Effects of Consumer Subsidies 483 The Experimental Housing Allowance Program 485 Supply versus
Demand Policies 486 Community Development and Urban Renewal 487 Urban Renewal 487 Recent Community
Development Programs 488
Control de alquileres 489 efectos en el mercado de alquiler controlan 490 efectos distributivos de Control de alquileres
492 deterioro y mantenimiento 493 vivienda submercados 495 controlando el suministro ante Alquiler controlan 495
alternativas para alquilar Control 496
Part VI
LOCAL GOVERNMENT
Captulo 19 Local G wemment gasto Local gasto hechos 503 el papel de Local estado
505 monopolio Natural. 506 externalidades 507 nivel ptimo de un bien pblico
Local. 509 beneficio tributario 510 federalismo y el ptimo nivel de gobierno 512
economas de escala frente a la demanda diversidad 512 externalidades versus
demanda diversidad 514 Trade-Offs de consolidacin 514 alternativa enfoques a los
derrames de 516 A Closer Look en el saldo de mediana votantes resultado 517 A serie
de B elecciones 517503
'!
Contents
xxvii
CHAPTER 1
introduction
Chapter 1 Introduction
Este enfoque en las decisiones de ubicacin permite examinar por qu ciudades existen, por
eso crecen, cmo se arreglan las diferentes actividades en las ciudades y los aspectos
espaciales de los problemas urbanos como la decadencia urbana, la contaminacin, la
congestin, el crimen y la pobreza. En este libro, exploraremos las razones de la creciente
urbanizacin en los ltimos dos siglos. Durante este perodo, la proporcin de la poblacin
mundial vive en ciudades aumentadas de alrededor del 9 por ciento a 46 por ciento y el
porcentaje de la poblacin en las ciudades de Estados Unidos aument de alrededor del 6
por ciento a 75 por ciento (Barioch, 1988). Como veremos, el aumento de la urbanizacin
fue causado por cambios fundamentales en las tecnologas de produccin, transporte y
comunicacin. Debido a estos cambios tecnolgicos, hay algunos beneficios sutiles de la
urbanizacin que ms que compensan los costos evidentes. Tambin exploraremos algunos
cambios fundamentales en el interna. estructura de las ciudades. En el siglo XIX, la ciudad
typicallarge era monocentric, la mayor parte de su empleo en el rea del ncleo central. En
el rea metropolitana tpica modero, slo 22 por ciento de puestos de trabajo estn a tres
kilmetros del centro de la ciudad, con la mayora de los-re restante trabajos dispersadas en
zonas suburbanas (Glaeser, Kahn y Chu, 2001). La transformacin espacial de ciudades fue
causada por muchos factores, incluyendo el cambio lgico techno, aumento de los ingresos
y la poltica del gobierno. Mirando hacia el futuro, muchos futuristas han predicho que las
innovaciones en telecomunicaciones (por ejemplo, los Internet y mo bilis, telfonos)
causar ciudades encoger o incluso desaparecer. Hace varias dcadas, la introduccin del
telfono condujo a las predicciones calamitosas similares para las ciudades. Vamos a ver
por qu los futuristas parecen estar mal otra vez.
Tambin exploraremos los problemas urbanos de economa ofvarious, incluyendo la pobreza,: decadencia urbana, congestin, la
contaminacin y el crimen. La ubicacin de las decisiones ofhouseholds y las empresas contribuyen a estos problemas. Por ejemplo,
cuando las empresas trasladar desde la ciudad central a the & uburbs, habr menos empleos para los residentes del centro de la
ciudad. Si el centro de la ciudad residents\don ' t siga empleos a los suburbios, las tasas de pobreza y criminalidad en las ciudades ser
incre.ase. A su vez, estos problemas afectan las decisiones de ubicacin de los hogares y las empresas. Por ejemplo, una ciudad central
que es responsable de los programas de transferencia como bienestar y vivienda con subsidio tendr relativamente altas tasas de
impuestos, causando muchos contribuyentes de ingresos medianos y altos ingresos a trasladarse a los suburbios. La pelea oftaxpayers
reducir la base impositiva de la ciudad, obligando a la ciudad para recortar el gasto en programas vitales como la educacin y la
delincuencia .control. Asimismo, ser el crimen urbano. ' causar muchos hogares a fiee suburbios de seguro.
Chapter 1 Introduction
Evidencia adicional para los derrames de conocimiento proviene de datos sobre producto en novations. En los Estados Unidos,
el 70 por ciento de la poblacin vive en ciudades, pero 96 por ciento de todas las innovaciones de producto provienen de las
ciudades. Las 10 ciudades ms innovadores producen dos tercios de las innovaciones de la nacin con slo una cuarta parte de
la poblacin de la na cin. Las cuatro ciudades ms innovadoras (Nueva York, San Francisco, Boston y Los ngeles) producen el
48 por ciento de las innovaciones de la nacin, pero tienen slo 17 por ciento de la poblacin del pas (Feldman y Audretsch,
1999). Por qu ciudades aumentar la creatividad y la innovacin? Feldman y Florida (1994) explican el papel de las ciudades en
la creatividad como sigue:
Las ciudades tambin ofrecen oportunidades para los trabajadores a aprender. Capital humano se define como el conocimiento y habilidades adquiridas por los
trabajadores de la educacin formal, experiencia laboral e interacciones sociales. Una forma de aumentar su capital humano es aprender por ing por
imitacin: Si ves a alguien ms realice una tarea eficientemente, entonces puedes aplicar lo aprendido a tu propio trabajo. Las ciudades ofrecen ms contactos
con tareas similares, facilitando el aprendizaje que aumenta la productividad de los trabajadores.1 i
4
Chapter
1 Introduction
,'
,:
Areas
!:
NewYork,
C cago,
Los Angeles
College graduates as percent of
Other
Metropolitan Nonmetropolitan
Areas
Areas
23.6%,
22.5%
9.1
7.9
13.4%
nonsenior
population
Young (ages 18-34) college graduates as
3.9
percent of nonsenior
p
o
p
u
l
a
t
i
o
n
M
i
d
d
l
e
a
g
e
d
(
a
g
e
s
3
5
4
9
)
c
o
l
l
e
g
e
g
r
a
d
u
a
t
e
s
9
.
8
1
0
.
0
6
.
4
a
s
p
e
r
c
e
n
t of nonsenior population
Old (ages 50-65) college graduates as percent
4.7
4.5
3.1
of nonsenior
population
Source: Calculations based on Glaeser, "Learning in Cities," Joumal of Economic Geography 1 (2001), pp.
27-50.
Chapter 1 Introduction
el tercer milenio A.c. equipado con ayuno veleros; los fenicios eran los
intermediarios para toda la cuenca mediterrnea, comercio de materias primas,
productos alimenticios, textiles y joyera. Atenas prosperaron como un sitio para el
comercio regional en su apogeo alrededor del ao 500 A.C., intercambio de oficios
domsticos y productos olecolas de alimentos y materias primas procedentes del
campo. Durante los siglos XI y XII, ciudades-estado italianas forjaron acuerdos
con los gobernantes bizantinas e islmicas Asia-Amrica con frica del norte y el
este. Estos europeos negocian madera, irongrain, vino y pao de lana para
medicamentos, colorantes, lino, algodn, cuero y metales preciosos. Este comercio
fue la fuerza principal detrs del crecimiento ofVenice, Gnova y Pisa. Ms
adelante en el libro, analizaremos las ciudades comerciales ms detalladamente. Las
ciudades tambin sirven como centros de produccin porque la produccin de muchos bienes y
servces es ms eficiente en un entorno urbano de alta densidad. Ms adelante en el libro 11
tomar una mirada cercana a las ventajas de la produccin de las ciudades. Las ciudades fenicias
mejoraron las tcnicas empleadas para producir productos de vidrio y tela teida y exportaron
bienes por ambos. Durante la poca medieval, las ciudades inglesas produjeron pao de lana para
la exportacin. Ms recientemente, la revolucin industrial condujo a una explosin de actividad
manufacturera y el rpido crecimiento de las ciudades; la produccin masiva de bienes atrajo a
personas de zonas rurales en ciudad faetones. En las ltimas dcadas, la produccin de bienes
basados en conocimiento ha aumentado en importancia, causando cambios profundos en las
ciudades.
El consumo en las
ciudades
Las ciudades tambin proporcionan a los consumidores con una amplia variedad de
bienes y servicios. Considere un producto con una demanda baja per cpita (por
ejemplo, teatro en vivo). Supongamos que 2 de cada 100 personas asisten a un
espectculo de teatro en vivo en cualquier ao dado. Adems, supongamos que
romper ni siquiera, el teatro debe vender 20.000 boletos anuales. Para sobrevivir, la
ciudad sU:rrounding el teatro debe tener una poblacin de al menos 1 milln
(1.000.000 veces 0.02 = 20.000). En general, cuanto mayor sea la ciudad, cuanto
mayor sea el nmero de productos que puede soportar. En ciudades ms grandes,
usted puede comprar todos los productos que estn disponibles en las ciudades ms
pequeas, as como otros bienes que tienen una demanda per cpita relativamente
baja. Tiene la mayor variedad de consumo disponible en materia de ciudades? Hay
alguna evidencia que ha aumentado el valor consumidores lugar en variedad. En las
ltimas dcadas, las ciudades con el mayor .variety de ciertos bienes de consumo han
crecido ms rpidamente. En los Estados Unidos, el ms rpido crecimiento...
QU ES
URBANA?
Para un economista urbano, una zona geogrfica se considera urbana si contiene un gran nmero de
personas en un rea relativamente pequea: en otras palabras, se define un rea urbana
-1
Chapter 1 Introduction
como uno con una densidad de poblacin relativamente alta. Para exarnple, suponga que la densidad de poblacin de
av escritas de un condado particular es de 20 personas por kilmetro cuadrado. Si parte del Condado contiene 50.000
personas en un rea de 20 kilmetros cuadrados (es decir, la densidad de poblacin es 2.500 personas por milla
cuadrada), es considerado un rea urbana porque tiene una densidad de poblacin relativamente alta. Esta definicin
acomoda las zonas urbanas de muy diferentes tamaos, desde una pequea ciudad a una gran rea metropolitana. Los
economistas definen un rea urbano en trminos de densidad poblacional porque una economa urbana se basa en
contacto frecuente entre las distintas actividades econmicas, y tales contactos es posible slo si las empresas y los
hogares se embalan en un rea relativamente pequea.
Defini
ciones
del
Censo
La oficina del censo de Estados Unidos define un nmero de
reas geogrficas. Puesto que la mayora empir ical trabajo
en economa urbana se basa en datos del censo, un claro
entendimiento de estas definiciones es importante. Las tres
primeras definiciones lidiar con la poblacin urbana.l. Urban
Place. An urban place is defined as a geographical area con menos de 2.500
habitantes en un rea relativamente pequea. 2. rea urbanizada. Una zona
urbanizada incluye al menos una gran ciudad central (un municipio) y el rea
circundante con densidad de poblacin superior a 1.000 personas por milla
cuadrada. Para tener una zona urbanizada, la poblacin total del rea debe ser por
lo menos 50.000. Figura 1-1 muestra una zona urbanizada como el gran crculo
centrado en un municipio. La zona urbanizada fuera de la ciudad central se llama
la franja urbana. En 1990, haba 396 reas urbanizadas en los Estados Unidos,
que contienen 63.6 por ciento de la poblacin del pas.
3. urbana poblacin. La oficina del censo define la poblacin urbana del pas como todas
personas que viven en las reas urbanizadas ms gente fuera de las reas urbanizadas que
viven en lugares urbanos. Segn esta definicin, en 1990 aproximadamente tres cuartas
partes de la poblacin estadounidense viva en zonas urbanas.
FIGURE 1-1
Centro de la ciudad
(Municipio)
rea urbanizada del Condado
Chapter 1 Introduction
Figura 1-2 por ciento ofU.S. Poblacin que vive en zonas urbanas, 1800-199080
75
70
65
60
55
-ae
::::1
;::
50
45
-40
;::
] 35
30
25
20
15
10
5
Figura 1-2 muestra el porcentaje de la poblacin vive en zonas urbanas (aieas urbanizada y
lugares urbanos). En la mayor parte de este perodo, la proporcin de la popu lation en las
zonas urbanas aument, con la nica excepcin siendo la dcada de 1930 y 1940, durante la
poca de la gran depresin y la segunda guerra mundial. Entre 1800 y 1990, la tasa de
urbanizacin se increment de 6 por ciento a 75 por ciento. Figura 1.:...-3 muestra las tasas
de urbanizacin para las diferentes regiones del mundo, con proyecciones para el ao 2025.
En 1970, las tasas de urbanizacin fueron aproximadamente 25 por ciento en Asia y Mrica
entre 50 y 60 por ciento en Amrica Central y Sudamrica y por encima del 60 por ciento en
Europa y Amrica del norte. Para el ao 2025, las tasas de urbanizacin se esperan que
aumenten en todas partes, con los aumentos ms grandes en Asia y Mrica. Para el mundo en
su conjunto, la tasa de urbanizacin fue 30 por ciento en 1950 y se espera que aumente a 61
por ciento para el ao 2025. El segundo conjunto de definiciones de lidiar con los municipios
y reas metropolitanas del censo.
l. Un municipio se define como un rea sobre la cual una corporacin municipal exer cises
autoridad poltica, proporcionar al gobierno local servicios como alcantarillado, proteccin
del crimen y fue. Desafortunadamente, el censo utiliza las palabras Municipio y ciudad
indistintamente. Cuando un documento del censo se refiere a una ciudad se refiere a la
poltica ciudad o municipio, no la econmica de la ciudad. Para definer 8
Chapter 1 Introduction
100
90
en
d.)
80
D
D
1970
2000
- 2025
70
-eo
.S
b.O
. i.l
J!l
'
:
en
d.)
60
50
40
'E 30
'1
d.)
[S
1:
p..
20
10
o
1
1
Africa
Asia
Centr2l
America
Europe
North
America
South
America
Source: United Nations (U.N.) Population Division, World Urbanization Prospects (1996
revision), on diskette (New York: United Nations, 1996).
;i .
1
"'i
1
' '
'
1
1,
1
1
the economic city, we would draw boundaries to include all the activities that are
an integral part of the area's economy. When using census data, it's important
to rernember the distinction between the political city and the economic city.
2. A metropolitan area is defined as an area containing a large population nucleus
and the nearby cornmunities that are integrated, in an economic sense, with the
nucleus. The-,nucleus is either a central city with at least 50,000 people or
an urbanized area. The integrated cornmunities are the surrounding counties
from which a relatively large number of people cornrnute to the nucleus. To
be designated a metropolitan area, the area in and around an urbanized area
must have either a central city with a population greater than 50,000 or a total
population (including the integrated cornrnunities) of at least 100,000 (75,000 in
New England). In 1999, there were 284 metropolitan areas in the United States,
ranging in population from just over 50,000 to just over 20 million people.
Figure 1-1 shows the simplest possible arrangement of a metropolitan area. The
metropolitan area is the county containing the urbanized area.
The third set of census definitions deal with different types of rnetropolitan
areas. If the population of a metropolitan area exceeds one million, the metropolitan area may be divided into two or more primary metropolitan statistical areas
(PMSAs). A PMSA consists of a large urbanized county (ora cluster of counties) that
is integrated, in an economic sense, with other portions of the larger rnetropolitan
area. When a metropolitan area is divided into two or more PMSAs, the rnetropolitan area is called a consolidated metropolitan statistical area (CMSA). In 1999,
there were 20 of these large metropolitan complexes, ranging in population from
just over one million (Hartford-New Britain-Middletown) to just over 20 million
(New York-Northern New Jersey-Long Island). Figure 1-4 shows the maps oftwo
Chapter 1 Introduction
Chicago PMSA
Joliet
PMSA
Kankakee
PMSA
Oakland PMSA
10
1
Chapter 1 Introduction
',
1:1
..: i
,:!
i
'
1
.
'
';j
Cherokee
i:
1
1'
1 '
'i
'1,
1:
d1:
:1
! .
1
1.'
I
I
Chapter 1 Introduction
1999
,-
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Metropolitan Area:
CMSAorMSA
Population
1999
Population
1980
% Change,
1980-99
20,033
18,906
5.96
15,781
8,810
7,285
6,816
5,988
5,867
11,498
8,115
5,791
5,368
5,649
5,122
37.25
8.56
25.80
26.97
6.00
14.55
5,458
4,802
4,408
3,746
3,656
3,424
2,931
2,912
2,831
2,781
2,564
2,365
2,346
2,257
2,149
1,948
1,737
1,686
1,646
1,542
1,538
1,519
1,505
5,293
3,046
3,118
2,233
2,644
2,409
1,600
2,938
2,198
1,862
2,414
1,742
2,571
1,614
1,584
2,938
1,449
1,100
1,570
1,201
1,098
1,306
805
3.12
57.65
41.37
67.76
38.28
42.13
83.19
-0.88
28.80
49.36
6.21
35.76
-8.75
39.84
35.67
-33.70
19.88
53.27
4.84
28.39
40.07
16.31
86.96
Chapter 1 Introduction
12
;,
1
City
1995
2015
Tokyo, Japan
Mexico City, Mexico
Sao Paulo, Brazil
Bombay, India
Shanghai, China
Calcutta, India
Buenos Aires, Argentina
Seoul, Korea
Beijing, China
Osaka, Japan
Lagos, Nigeria
Rio de Janeiro, Brazil
Delhi, India
Karachi, Pakistan
Cairo, Egypt
Paris, France
Tianjin, China
Manila, Philippines
Moscow, Russian Fed.
Jakarta, Indonesia
Dhaka, Bangladesh
26.96
16.56
16.53
15.14
13.58
11.92
11.80
11.61
11.30
10.61
10.29
10.18
9.95
9.73
9.69
9.52
9.42
9.29
9.27
8.62
8.55
28.89
19.18
20.32
26.22
17.97
17.31
13.86
12.98
15.57
10.61
24.61
11.86
16.86
19.38
14.42
9.69
13.53
14.66
9.30
13.92
19.49
Percentage Increase,
1995-2015
7%
16
23
73
32
45
17
12
38
139
17
69
99
49
2
44
58
61
128
1'
Source: United Nations Population Division, Urban Agglomerations, 1950-2015 (1996 revision),
on diskette (New York: United Nations, 1996).
13
Chapter 1 Introduction
FIGURE Cambio de 1-6 por ciento en la poblacin metropolitana dentro y fuera de las
ciudades centrales, por regin y divisin de 1990 a 1999
, - (Pacific)
WestRegion
Northeast Region
Midwest Region
4.1
35.8
-J
15.0
14.9
4.2
20.3
WestNorth
Central
East North
Central
Middle At[an:tic
20.2
18.0
23.8
1.9
9.5
East South
Centr31
2.5
South
Atlantic
11.2
South Region
4.2
U.S.
West
CO:MING ATTRACTIONS
Economa urbana puede dividirse en seis reas relacionadas que corresponden a los seis partes de este libro.
Estas son las fuerzas del mercado (1) en el desarrollo de las ciudades, la tierra (2)
1'
1\
1
14
Chapter 1 Introduction
Utilice dentro de ciudades, transporte urbano (3), (4) problemas urbanos y polticas pblicas, vivienda (5) y las polticas pblicas y (6)
gastos del gobierno local y los impuestos.
Fuerzas del mercado en el desarrollo de ciudades
,,!:
15
Chapter 1 lntroduction
16
Chapter 1 lntroduction
Referencias
'1
PRIMERA
PARTE
Parees de mercado en el desarrollo de ciudades
/n to rnarket ec.onorny, individuals exchange their labor for wage incorne, which they use to buy
consurner goods and services. How do these rnarket transactions affect cities? Chapter 2 explains
that. cities exist because there are economies of scale in the production of sorne goods and services.
Chapter 3 explains why sorne cities are big and others are srnall, focusing on the rationale for firms to
cluster in cities. Chapter 4 shows how the location decisions of firms and households to determine
where cities develop. Chapter 5 explores the rnarket forces behind the developrnent of a hierarchical
systern of cities in region, with cities differing in size and scope. Chapter 6 examines the determinants
of urban economic growth and explains sorne of the techniques used by economists to predict growth
ernployment. \
17
t.
CHAPTER 2
Por qu existen las ciudades?
El hombre es el animal _Slo que hace gangas; un perro no cambia los huesos con otro
perro. -ADAMSMITH
ciudades existen porque es e ciente para producir algunos bienes a gran escala. Una
actividad est sujeta a las economas de escala si el costo promedio de produccin
disminuye cuando la escala la operacin y producir una salida ms. La forma de
aprovechar las economas de escala es que concentran la produccin ata pequeo
nmero de sitios. Porque los trabajadores pueden economizar en el tiempo de viaje,
las ciudades se desarrollan alrededor de estos sitios de produccin. Una de las teoras
populares del econornics urbano es que si no hubiera ninguna econornies de escala,
no habra ningn ciudades. En este caso, cada hogar sera autosuficiente, produciendo
_La propias mercancas y servicios en su propia casa. No habra ninguna razn para
que las ciudades porque el costo de la vida urbana (por ejemplo, un alto precio de la
tierra) no se compensara por algn beneficio. Como veremos en este captulo, las
ciudades existen porque las bene:fits asociadas con la produccin concentrada ms
que compensar los costos de vida en una alta densidad de poblacin.
19
20
,./
;_
',
'
1
1
,.
Part 1
'1
',
1
En nuestro modelo de la regin sin ciudades, suponemos que todos los hogares son igualmente
productivos en la produccin de camisas y pan. Si esta hiptesis es relajada, uno
. '-JI.'::J
--... .
ti
!.'\ ,-
C-1,-
- . '-':::.'1- 1 O T
..
21
i
.f..:..
- --,
parte de la regin puede tener una produccin de advantagfilrs1irt
comparativo, y la otra parte tenga una ventaja comparativa en la produccin de pan.
Es posible pero no seguro-que ventaja comparativa causar la especializacin y el
comercio. La nocin de ventaja comparativa puede explicarse con un ejemplo
numrico simple. Supongamos que el ern no$ la mitad de la regin es ms
productiva que la mitad sur. En la tabla 2-1, Corea del norte tiene una ventaja
absoluta en la produccin tanto pan y camisas; los residentes del norte producen dos
veces como mucho pan por hora y camisas seis veces ms por hora. Las diferencias
de productividad podran deberse a diferencias en habilidades laborales, el tiempo o
los recursos naturales.
La nocin de ventaja comparativa se basa en el principio de costo de
oportunidad. En un perodo de una hora, un trabajador del Norte puede producir seis
camisas o dos Hogazas de pan. Por lo tanto, el costo de oportunidad de una camisa
es pan de un tercio, y el costo de oportunidad del pan es tres camisas. Un trabajador
del sur puede producir un pan o una camisa, as que el coste de oportunidad del pan
es una camiseta y viceversa. El norte tiene una ventaja comparativa en la produccin
de camisetas porque el costo de la oportunidad de camisas es una tercera barra de
pan, comparado con un pan en el sur. Sirnilarly, el sur tiene una ventaja comparativa
en el pan, porque su costo de oportunidad de una hogaza de pan es una camisa,
comparada con tres camisetas en el norte.
-
'
Ventaja comparativa puede conducir al comercio entre Nm; th y sur. Para explicar las posibles ventajas del comercio,
supongo que todos househlds en la regin son inicialmente autosuficientes. Supongamos que la tasa de cambio es dos
camisas por hogaza de pan; es decir, el precio del pan es dos camisas. Si ambas regiones se mueven en la direccin de la
especializacin, ambos pueden obtener ms bienes de consumo. Como se muestra en la primera fila de la tabla 2-2, si un
hogar Norte cambia una hora de trabajo de produccin del pan a la produccin de camisa, produce seis camisas adicionales
pero dos menos hogazas de pan. RR: un h9usehold sur cambia a tres horas de produccin camisa al pan, que produce tres
ms '-panes pero tres camisas menos. Como se muestra en la segunda fila de nmeros, dada la tasa de cambio de dos
camisas por pan, puede intercambiar la casa northrn
South
North
1 shirt
1loaf
3 shirts
1/3loaf
South
Bread
Shirts
North
Change in production
Exchange 2 shirts per loaf
Net effect of change in production
and exchange
Bread
Shirts
Bread
Shirts
-2loaves
+3loaves
+lloaf
+6 shirts
-6 shirts
+3loaves
-3loaves
-3 shirts
+6 shirts
+3 shirts
22
sus seis camisas extras para tres hogazas de pan de la casa de cono. En el lado
de fiip, el hogar cono intercambia sus tres panes extras para seis camisas. Como
se muestra en la ltima .row, el hogar del norte tiene una ganancia de una
hogaza de pan, mientras que la casa de cono tiene una ganancia de tres camisas.
En este caso, cada hogar se beneficia de un movimiento hacia la especializacin
en el bien para el cual tiene una ventaja comparativa.
What about transportation costs? Trade between the two households will be
benefits! if the transportation costs involved in trading aren't too large. Suppose
that it takes an hour to execute a trade. For the northem household, the
opportunity cost of the hour spent is the amount the household could ofbread
trading produces for itself during .this tiri: j. e. The opportunity cost is two
loaves so it would n.ot be sensitive to trade. If, however, the travel time is less
than a half hour, the household would have a net gain from trade. For example,
it the trade takes only one-fifth of an hour, the opportunity cost is four-tenths of
a loaf, leaving a net gain of six-tenths of a loaf. Similarly, the southern
household would have a net gain. Therefore, households in the two regions will
specialize and trade, exploiting their comparative advantages to increase
consumption.
Aunque ahora hay especializacin y el comercio, no habr ningn ciudades. Dado el supuesto cuarto (no
hay economas de escala en el transporte), los hogares en las dos regiones participarn en comercio
directo. Cada hogar northem unir para arriba con un hogar sur para intercambiar camisetas y pan. No
hay razn para un n intermediario porque las personas pueden transportar mercancas tan
eficientemente como alguien que transporta mercancas a granel.
23
CIUDADES
DE FBRICA
La tercera hiptesis subyacentes a la regin sin ciudades es que la constante vuelve a escala en la
produccin. Concretamente, cada trabajador puede producir una camisa o una hogaza de pan por hora, sin
importar cunto el trabajador produce. No hay ningn adva. permiten de centralizar. d produccin en
fbricas, as que todas las mercancas se producirn en los hogares. Si dejamos la asuncin de retornos
constantes a escala e introducir las economas de escala en la produccin de camisas, produccin de la
fbrica podr sustituir a la produccin nacional, provocando el desarrollo de las ciudades de fbrica.
La escala de las economas y tbe Sbirt Fa: ctory suponga que hay ventajas de la produccin a gran escala de las
camisas. En otras palabras, a medida que aumenta el volumen de produccin, disminuye la mano de obra necesaria
para producir una camisa. Figura 2-1 muestra los tiempos de trabajo promedio para diferentes cantidades de camisas
1 fbrica. Disminuye el tiempo laboral promedio de una hora por una camisa (la t a la misma hora como h <? me
produccin) a la hora de una cuarta parte para una produccin de 400 camisetas. stos. las economas de escala
surgen por dos razones.
24
0.50
-::;-
....
:a
."..'.
Cl)
o.
"...'.
o
;:J
e
Cl)
Se 0.25
....
o
.D
.....:
200
.400
600
Number of shirts
1
1
0.75
t;
" 0.50
CJ
1
1
Cost of
homemade shirt
zo
0.25
Travel cost
_._
1
1
--+
Factory cost
-L--
--
--4--
--- MWes
from factory
Market area of
shirt factory
El rea de la fbrica de camisas de mercado es el rea sobre la cual el costo neto de una camisa de
fbrica (tiempo de produccin ms tiempo de viaje) es menor que el tiempo de produccin de una
camisa casera (1 hora).
26
mide el precio neto de camisas, la suma del precio de fbrica (0.25loaf) y tiempo de viaje a la fbrica. El eje
horizontal mide la distancia a la fbrica en millas. Si tiempo es ocho millas por hora (ida y vuelta cuatro millas por
hora), la fbrica ser precios camisas hechas en casa para los residentes de tres kilmetros de la fbrica (en O.75
hora caminando a tiempo). Por lo tanto, rea del mercado de la fbrica es un crculo con un radio de tres
kilmetros.
Se desarrollar una fbrica ciudad una pequea urbana zona alrededor de la fbrica de camisas. La camisa
trabajadores willlive cerca de la fbrica para ahorrar en costos de comrnuting y har una oferta por el tierra de
precio cerca de la fbrica. A medida que aumenta el precio de la tierra, los trabajadores se economizan en tierra
ocupando lotes ms pequeos. En otras palabras, la densidad de poblacin alrededor de la camisa factory.will ser
mayor que la densidad de poblacin en el resto de la regin. Una ciudad se define como un lugar con una
relativamente alta densidad de poblacin, para que la fbrica produce el desarrollo de una pequea fbrica de la
ciudad. La ciudad de fbrica se desarrolla porque se cumplen dos condiciones. Primero, agricultura. la
productividad es lo suficientemente alta como para que los trabajadores fuera de la ciudad pueden generar
suficiente comida para alimentarse y tener lo suficiente para alimentar a los trabajadores de camisa en la ciudad.
En segundo lugar, las economas de escala son grandes en relacin con los gastos de viaje, lo que la fbrica de
camisas puede precios camisas hechas en casa, al menos para personas que viven cerca de la fbrica.
Lmites de tamao de ciudad: coste y el ferrocarril de carga
27
de los Estados Unidos, una gran parte de los alimentos consumidos en Nueva
Inglaterra fue producida por agricultores de la regin. Asimismo, una gran parte de
los productos consumidos en el oeste (por ejemplo, zapatos) fue producida por los
artesanos locales. El ferrocarril corta los costos de flete, fbricas orientales podran
precios a occidentales artesanos y agricultores occidentales podran precios a
agricultores orientales. Como resultado, ambas regiones especializan en mayor
medida y comercio aumentado. El aumento en la produccin de las fbricas en
Nueva Inglaterra aument el tamao de sus ciudades de fbrica.
Lmites de
Sewing machine, which was developed in the middle of th... The sewing machine also
LMITES de tamao de ciudad: Costes de transporte hasta este punto, implcitamente asumimos que los
salarios pagados a obreros es independiente del tamao de la ciudad. Cada trabajador se paga una hogaza de
pan por hora de trabajo de fbrica, sin importar el tamao de la ciudad. De hecho, los salarios pagados a
obreros se incrementar a medida que la ciudad crece porque have'longer de las ciudades ms grandes
desplazamientos veces. La fbrica debe compensar a los trabajadores para trayectos ms largos, as el salario
aumentar a medida que crece la ciudad. A medida que aumenta el salario, igual que el costo de la fbrica
camisas relativa el costo de las camisas hechas en casa.
28
Commute Time
WorkTime
HourlyWage
100
20,000
O hours
4 hours
8 hours
4 hours
Incremente los salarios con el tamao de la ciudad para ilustrar los efectos de
desplazamientos los costos de los salarios y el precio de las camisas de fbrica,
considere una ciudad fbrica donde cada trabajador tiene ocho horas por da para
dividir entre desplazamientos y fbrica de trabajo. Cada hora tiempo
desplazamientos disminuye trabajo por una hora. Para ser indiferente entre
produccin de pan en casa y ser empleados en la fbrica (que implica viajar y
trabajar a veces), el trabajador deber ser pagado ocho rebanadas de pan de cada
da de trabajo y desplazamientos. Podemos utilizar el sencillo ejemplo que se
muestra en el cuadro 2-3 para mostrar por qu los salarios aumentan con el
tamao de la ciudad. Supongamos que en una ciudad pequea de 100
trabajadores, tiempo de trayecto es cero: cada trabajador vive a pocos pasos de la
fbrica de camisas. Cada trabajador pasa ocho horas en la fbrica y gana ocho
panes al da, que se traduce en un pan por hora de trabajo de fbrica. En el otro
extremo, supongo que en una ciudad con 20.000 trabajadores, tiempo de viaje es
de cuatro horas al da, dejando slo cuatro horas de trabajo de fbrica. Para...
Figura 2-3 muestra la relacin entre el tamao de la ciudad y los salarios,
utilizando los dos extremos que se muestra en el cuadro 2-3. El salario se eleva de un
pan por hora en la pequea ciudad a dos panes por hora en la ciudad con 20.000
trabajadores. Cules son las implicaciones de salarios ms altos para el productor de
la camisa y la ciudad de fbrica? Cuanto mayor sea el salario, mayor ser el costo de
produccin para camisas de fbrica y la ms alto el precio de las camisas de fbrica. Recordar
que la zona del mercado de la fbrica se define como el rea sobre la cual el precio neto de fbrica
camisetas (el precio de fbrica ms recorrido los costos incurridos por la camisa del comprador) es
menor que el costo de las camisas hechas en casa. Como una ciudad crece, aumento de los salarios
aumentar el factoryprice de en relacin con el costo de camisetas camisas hechas en casa. En otras
palabras, ms desplazamientos costos willlimit la zona del mercado de la fbrica y el tamao de la
ciudad.
Disminucin en los costos de transporte y la experiencia de Boston porque desplazamientos costos lmite ciudad
tamao, una disminucin en el costo unitario de transporte provocar una ciudad crecer. Supongo que la ciudad
de desplazamientos velocidad fueron dobles. En el cuadro 2-3, el tiempo de viaje asociados con una ciudad
of20, 000 trabajadores se reducira a la mitad, reduciendo tiempo desplazamientos trabajadores, los salarios, el
costo por camiseta y el precio de las camisas de fbrica. Como la fbrica precio disminuye en relacin con el
costo de camisetas caseras; la zona del mercado de la fbrica se incrementar, aumentando el nmero de
trabajadores
de
la
fbrica
y
el
tamao
de
la
ciudad.
29
t
j
"'....
..e
... ....
.,,
.9
'
">'
.S
en
20
Nlll"! lber trabajadores (miles)
30
Lasprimeras ciudades
Las primeras ciudades se desarrollaron en los valles del ro frtiles en
el Cercano Oriente alrededor de 3000 A.C. Las ciudades-Estados que
se desarroll en el valle de Mesopotamia incluyen Eiridu, your, .
hectreas y una poblacin de unos 25.000. En el 2000 A.C., la ciudad
de Babilonia tena una poblacin de unos 50.000. Estas ciudades se
desarrollaron poco despus de una serie de agri
. hectreas y una poblacin de unos 25.000. En el 2000 A.C., la ciudad
de Babilonia tena una poblacin de unos 50.000. Estas ciudades se
desarrollaron poco despus de una serie de agri cultura!break.throughs
genera un excedente de alimentos, permitiendo que la gente sorne
perseguir las actividades no agrcolas en las ciudades. Entre las
innovaciones fueron la domesticacin del grano, el desarrollo de
sistemas de riego y el arado (reemplazando la azada). Segn
Davis (1976), entre 50 y 150 agricultores deban apoyar el habitante de una sola
ciudad.
Por qu se desarroll las primeras ciudades? Hay alguna
incertidumbre acerca de las caractersticas sociales y econmicas de
las sociedades tempranas, as que los historiadores slo pueden
especular sobre los orgenes de las ciudades. La mayora est de
acuerdo que las primeras ciudades sirvieron tanto religiosos y
defensiva
propsitos. Se explica el desarrollo de las ciudades defensivas y religiosas por
los mismos factores que explican el desarrollo de ciudades en el mercado?
Consideremos primero la ciudad defensiva. Un agricultor que genera
una agricultura} supervit eventualmente usar el excedente, lo
consumiendo mismo o cambiarlo por otro
mercancas. En cualquier caso, debe almacenar la comida sobrante por sorne
perodo de tiempo. El alimento almacenado proporciona un objetivo lucrativo
para los ladrones. Si hay economas de escala en proteccin y
almacenamiento de alimentos, los agricultores pueden ser mejores guardar sus
excedentes en un almacn central en lugar de un conjunto de servicios, uno
para cada agricultor. La gente trabajando en el willlive almacenamiento
fortificada instalacin (gerentes y guardias) cerca la facil ity, generando un
lugar con una densidad de poblacin relativamente alta, una pequea ciudad.
Esta es la teora de la ciudad defensiva: las primeras ciudades desarrollaron
debido a las economas de escala en el almacenaje de la lechera supervit.
Apoyo para la teora viene de ologists Archaea que descubri los restos de
instalaciones de almacenamiento fortificado en las primeras ciudades.
Al considerar la ciudad religiosa. Segn Murnford (1961), el desarrollo de
las primeras ciudades coincidi con el desarrollo de la religin en gran
escala. Antes el desarrollo de las ciudades, la mayora de la gente adorada en
pequeos grupos, ya sea en el hogar o en la aldea local. En la poca que
desarrollaron las primeras ciudades, la tierra local los dioses fueron
substituidos por los dioses celestiales, que pueden tener d emanded cult o
en una escala ms gr ande. Si adoracin y splica er an
consider ados ms eficaces en gr an escala, una duplicacin del
tamao del Santuar io sera ms del doble la salida del religiosa. Los
pequeos santuarios en los hogares y aldeas fueron reemplazados por
31
Griegos
y
romanos
Ciudades
La prxima etapa de la urbanizacin occidental ocurri en Grecia. En
el ao 500 A.C. haba cientos de ciudades-estado independientes, que
se extienden en la poblacin de unos pocos cientos a d e c e n a s d e
miles. Atenas fue la ms grande, con una
poblacin de unos 150.000, y Sparta tena una
poblacin de u nos 40.000. A diferencia de
a n t e r i o r e s las ciudades, que fueron controladas por los
sacerdotes cacique, casi todas las ciudades griegas fueron
funcionadas por sus ciudadanos, mediante asambleas pblicas para
hacer poltica. Atenas era una ciudad de mercado, con actividad
centrada en el gora comercial. Temprano en su
la historia, la ciudad estableci colonias independientes y
comercializan sus oficios domsticos y olivo produ ts para alimentos y
materias primas de las colonias. Comercio fue facilitado por el la
estampacin de monedas de oro y plata en el siglo VII A.C. En este
perodo temprano, Atenas se alimentan a travs del comercio
voluntario con otras reas. El Imperio ateniense se convirti en las secuelas de la exitosa
guerra contra Persia en el siglo v A.C. Despus de las ciudadesEstados repelieron el persa invasin, formaron la Liga de delos para
llevar la guerra a Minar de Asia. Al final de la exitosa campaa, Atenas
haban asumido el control completo de la liga, dictat ing poltica las
.32
The Roman cities were eventually overrun by marauders from rural areas. According to Hohenberg and Lees (1985), the Roman economy emphasized the collection of the agricultura! surplus and neglected production activity. Instead of exchanging urban goods for agricultura! products, Rome used conquest and tribute to
feed its population. In the fourth and fifth centuries, German tribes invaded from
the north, disrupting the Roman collection system. It appears that there was little
interest outside of Rome in restoring the "trade" routes, s6 the losses from successive invasions were cumulative. If Rome had relied to a greater extent on voluntary
exchange, the colonies rnight have been more interested in maintaining the exchange network, and the Westem empire rnight have recovered from the Germanic
raids.
What are the lessons from the rise and fall of Athens and Rome? Early in
its history, Athens engaged in voluntary trade with other areas, exchanging urban
goods for food from the countryside. The city thrived under this system of voluntary
exchange: The Athenians eventually switched to a system of conquest and tribute,
resulting in war and the decline of the city. Mumford (1961) suggests that the city
of Rome should have been called "Parasitopolis" to indicate the extent to which its
population lived off the labors of outsiders. The decline of Rome was caused in part
by the disruption of its collection system by the Germanic raids. Perhaps the lesson
is that cities based on voluntary exchange are viable, but cities based on coercive
transfer payments are not.
Feudal Cities
In the first few centuries following the fall of the Roman Empire, cities in the West
declined. The Islamic conquest disrupted trading on the Mediterranean, causing the
decline of many port cities. Waves of marauding barbarians continued to sweep
across Europe, mking travel and trade dangerous. People sought safety inside city
walls, where, once again there was safety in numbers.
The feudal economy of the eleventh through the fourteenth centuries was based
on manorial estates and small walled cities. On the manor, the lord inherited his
serfs, who worked the lord's land and served in the rnilitary in exchange for the use
of a small plot of land. In the small towns, workers produced handicrafts (cloth,
leather goods, metal goods), which they exchanged for the agricultura! surplus of
the manor. Cities in the feudal era were small arid numerous. In the eleventh century,
London was the largest city in England, with a population of only 16,000. There
were severa!other English cities with populations of around 5,000. The largest of
the many German cities had a population of no more than 40,000.
The largest cities served the Byzantine Empire and the Muslim areas of Spain.
Venice was one of the few large cities beyond the Byzantine and Muslim areas, but it
depended heavily on trade with the Eastem empire. During the eleventh and twelfth
centuries, the Italian city-states forged agreements with the Byiantine and Islarnic
rulers for trade with North Africa and the East. The Europeans traded wood, iron,
grain, wine, and wool cloth for medicines, dyes, linen, cotton, leather, and precious
metals. The increased tradecontributed to the growth of Venice, Genoa, and Pisa.
i1
i
1
33
Mercantiles
Ciudades
A partir del siglo XV, las grandes ciudades mercantiles desarrollaron en Europa. Uno
factor en este desarrollo fue la transferencia de poder de un gran nmero de seores
feudales a un nmero relativamente pequeo de Reyes, reinas y prncipes. El cambio
en el poder fue causada en gran parte por los avances militares que las maniobras de
defensa feudal tradicional obsoleta.
34
Part 1
35
.'
Figura 2-4 muestra la tendencia en la urbanizacin de EEUU desde 1800 hasta 1990. En 1800,
slo 6 por ciento de la poblacin viva en ciudades. El porcentaje de poblacin en las
ciudades aument a 15 por ciento por el 1850, 40 por ciento por el 1900, 64 por ciento en
1950 y el 75 por ciento en 1990. La revolucin industrial provoc dramtico crecimiento en
ciudades de Estados Unidos. Entre 1850 y 1890, las poblaciones de Nueva York y Boston ms
que triplicaron; ms que duplicado la poblacin cin de Baltimore; y ms que multiplic la
poblacin de Filadelfia. Oeste, en promedio, la poblacin de Chicago, Detroit, Cleveland,
Pittsburgh y Los ngeles incrementado /por entre 1850 y 1890. Al acecho detrs de estos
cambios fueron grandes aumentos en la actividad manufacturera urbano. La proporcin del
empleo en la fabricacin se duplic con creces en la mayora de estos ciudades, desde
aproximadamente 10 por ciento emplyment comprometes a alrededor del 20 por ciento.
r
36
FIGURE 2-4- OfU.S por ciento. Poblacin que vive en zonas urbanas, 1800-1990
80
75
70
65
60
55
50
-e
45
:::::1
.S 40
0.0
:::
&
35
"E
o 30
p..
25
20
15
10
5
0 -- -- ---J----
-- -- --
-- -- --
Oportunidades de empleo l. Resumen se concentran en las ciudades porque algunas actividades estn sujetos a las economas
de escala. 2. una ciudad comercial se desarrollar si [a] las clifferences en productividad subyacente ventaja comparativa son
grandes en relacin con los costos de transporte y [b] hay economas de escala en el transporte. 3. una ciudad fbrica
desarrollar si las economas de escala en la produccin son grandes en relacin con los costos de transporte. 4. ciudad
tamao es limitado por el costo de transporte de mercancas (coste de la carga) y los trabajadores de oftransporting costo
dentro de la ciudad (costos de transporte). 5. las primeras ciudades desarrollaron por razones defensivas y religiosas. 6. la
centralizacin del poder durante el perodo mercantil caus el desarrollo de ciudades administrativas y reduce las barreras
comerciales. Combinado con viajes ocano eficiente, esto result en el desarrollo del comercio en las ciudades. 7. la rpida
urbanizacin en los siglos XIX y XX fue causado por la revolucin industrial y el asociado i...
37
1 ";
:t
East
West
1
3
5
6
West
5. Region F has a single factory city (a city that developed as a result of scale
economies in production). Suppose that a matter transmitter is introduced into .
the regan. The transmitter can costlessly transpot goods, but cannot be used
38
REFERENCES
l. D vis, Kingsley. "Urbanizacin".En la economa urbana, ed. Harold Hochman. Nueva York: W. W. Norton, 1976. 2.
Hohenberg, Paul M. y Lynn H. las. La realizacin de la Europa urbana 1000-1950. Cambridge, Mass.: Harvard University
Press, 1985. Analiza las razones de la urbanizacin de Europa. 3. Mumford, Lewis. La ciudad en la historia. Nueva York:
Harcourt Brace Jovanovich, 1961.
.....
CAPITULO 3
Grandes y Pequeas
Ciudades
39
40
poblacin entre 3 millones y 6 millones y 22 reas con poblaciones de entre 1,5 millones y 3 millones. En otras palabras, la
srnaller la zona urbanizada, el nurnerous ms las reas de tamao similar. Grupo B de la figura 3-1 muestra la distribucin
de tamao de los 51 a travs de las reas urbanizadas 396th.
12
_......
":::':
:$
]_
::::
o
.=
co
'"o5.
:..,
11
17
23
29
Rank of urbanized area
35
600
1
; 200
o
100
200
Rank of urbanized area
300
396
41
ECONOMAS localizacin y
grupos de la industria
..
' .i
.
1
!
_(
Compartiendo
Proveedores de Insumos
Racimos de industria Sorne ocurren debido a comprar las empresas en una industria en
particular que un intermedi comi entrada del mismo proveedor. Las empresas se agrupan
alrededor de un proveedor de entrada comn si se cumplen dos condiciones:
l. La demanda de inpv.t de una empresa individual no es suficientemente grande como para
explotar las economas de escala en la produccin de la entrada intermedia. 2. transporte
costos son relativamente altos. Si los demandantes y proveedor de interactuan en la
orfabrication de diseo de la entrada intermedia, contacto cara a cara entre comprador y
vendedor.es necesario, y proximidad con el proveedor de entrada es importante.
Asimismo, ifthe entrada intermedia es voluminosa, frgil, o debe entregarse rpidamente,
proximidad
es
importante.
Vernon (1972) utiliza la industria de la confeccin de Manhattan para ilustrar la localizacin '.
economas de compartir el proveedor de insumo intermedio. Debido a la demanda para los vestidos de
alta costura es impredecible, una firma de confeccin no puede comprometerse t a la produccin a gran
escala de un determinado tipo de vestido, pero debe estar preparada para cambiar rpidamente a otro
diseo si el primer vestido es infructuoso. La firma debe ser pequeo y gil, con su encargado
supervisar cuidadosamente el mercado del vestido, el proceso de diseo y los procesos de produccin.
Para la firma pequea y gil, mano de obra especializada y maquinaria-diseado para producir una
salida en un gran escala-son no es factible. El ' modista compra alguna de sus insumos intermedios de
las empresas especializadas en esas entradas.42 Part 1 Market Forces in the Development of Cities
Una de las entradas ntermediate de confeccin es botones, un bien que est sujeta a
grandes economas de escala. Un buttonmaker grande puede utilizar mano de obra
especializada y grandes mquinas para producir botones ata un costo relativamente
bajo. Supongamos que tarda 10 pequeos modistas para generar demanda suficiente
botn para ex.haust las economas de escala en la produccin de botn. Al compartir
una buttonmaker, los modistas 1O pueden obtener sus botones a un menor costo. Es
eficiente para compartir una buttonmaker porque hay una escala. desajuste de la
economa: las economas de escala en confeccin son pequeas en comparacin con
las economas de escala en inflable. Debe racimo modistas alrededor de la firma de
botn, o pudieran compartir el buttonmaker ordenando a los botones de un catlogo?
Modistas ubicar cerca del productor botn porque cada vestido de diseo requiere un
diseo nico botn, y modistas deben supervisar el diseo y la fabricacin de botones
para un vestido en particular con el contacto cara a cara. En contraste con muchos otro
intermedi...
l. sede corporativa de de produce una gran variedad de salidas. Un ejecutivo
puede supervisar el desarrollo de una nueva campaa publicitaria de una semana,
elija una ubicacin para una nueva planta de la prxima semana y desarrollar una
estrategia para defenderse de una demanda la semana ing siga. La mayora de las
empresas contratan empresas de publicidad porque las economas sce en la
produccin de campaas de publicidad son grandes en relacin con la demanda
de publicidad de una empresa individual. Los ejecutivos corporativos necesitan
estar cerca de la firma de publicidad porque los ejecutivos de asistir en el diseo
de la campaa publicitaria; cara a cara es necesaria. En general, el agrupamiento
de sede corporativa permite corporaciones para explotar las economas de escala
en la produccin de bienes intermedios (publicidad, eco nmica consultores,
servicios legales) que tiempo es una parte importante del proceso de produccin.
2. las empresas que producen bienes de alta tecnologa nuevas cara incierta demanda de
sus productos. En vez de producir sus propios componentes, que adquieren los
componentes electrnicos de las empresas que pueden aprovechar las economas de
escala en la produccin de estas piezas. Las firmas de interactuarcon con sus proveedores
en el diseo y la fabricacin de los componentes y deben ubicar lo suficientemente cerca
para facilitar frecuente cara a cara. Aunque una empresa de alta tecnologa no necesita
localizar cerca de su proveedor de tuercas y tornillos (insumos estandarizados que pueden
solicitarse un catlogo), existen beneficios sustanciales de localizar a su distribuidor de
componentes electrnicos no normalizados. Adems, las empresas de alta tecnologa
explotan las economas de escala en prueba por compartir las empresas que proporcionan
las instalaciones de prueba, localizando lo suficientemente cerca para aprovechar las
instalaciones de las prueba a la mxima velocidad del producto. . 3. revistas y apers noticias
producen salidas (texto e ilustraciones) que cambian m unpred1ctable formas da a da. En
la
produccin
de
sus
productos,
editores...
43
.;
,,,,
,1
,i
44
Part 1
$20
16
Switch
Cost
Probability
of Switch
$8
1/2
1/2
(B)
Isolated Finn
Finn in Cluster
$
--
30
20
10
D ad
120
Number of workers
80 120 160
Number of workers
Un aislado centraremos paga salarios diferentes en tiempos buenos y malos momentos, pero emplea el mismo nmero de
trabajadores en ambos casos. En contraste, un centraremos en un clster de industria paga el mismo sueldo en ambos
casos, pero contrata a ms trabajadores en los buenos tiempos. El beneficio de la agrupacin y agrupacin de mano de obra
es que durante los buenos tiempos la empresa paga un salario inferior y contrata a ms trabajadores. El costo es que en los
malos tiempos, la empresa paga un salario ms alto. El beneficio supera el costo porque la empresa contrata a ms
trabajadores durante los buenos tiempos que en los malos tiempos.
ha marcado una relativamente alta demanda de mano de obra, segn lo indicado por la
curva de demanda Dgood En b'ad veces que el precio de su producto es bajo, por lo que
la firma tiene una relativamente baja demanda de mano de obra, segn lo indicado por la
curva de demanda con la etiqueta Dbad El panel izquierdo de la figura 3-2 muestra qu
sucede si la empresa elige un sitio aislado, lejos de otras firmas de software. El suministro
de mano de obra disponible para la firma de software en el sitio aislado est fijado en 120
trabajadores. En los buenos tiempos del: empresa paga un salario de $30, butin malas
pocas el fiirn paga un salario de slo $1O. En otras palabras, una empresa aislada
contrata a la same nmero de trabajadores en tiempos buenos y malos, pero paga un
salario ms alto en los buenos tiempos. El panel de la derecha ofFigure 3-2 muestra qu
sucede ifthe empresa se localiza en un racimo de otras empresas de software y se basa en
una piscina comn de los trabajadores de software. Supongo que en cualquier ao, pasarlo
bien experimentados por una sola firma es equilibrada por un mal momento
experimentado por otra firma, as la demanda de formacin para los trabajadores de
software
es
estable.
T
...
46
47
Type 1
Type 2
$4
$5
Type 3
Type 4
Type5
$5
$4
$6
Type 1
Actual types
Output produced
Type 2
$6
Type 3
Type 4
Type 5
X
$5
$6
X
$5
$6
'
1
48
es ms frecuente en las industrias con la volatilidad de los mercados. Rosenthal y Strange (2000a) proporcionan evidencia
adicional para clustering basado en agrupacin de mano de obra. Compartir informacin: Tercer tipo de conocimiento A los
derrames de economa localizacin surge desde el intercambio de informacin entre las empresas en la misma industria.
Como vimos en el captulo 1, se produce una cantidad desproporcionada de la innovacin en las ciudades, especialmente en
las ciudades ms grandes. Una ciudad proporciona oportunidades para las interacciones entre personas con intereses
comunes y por lo tanto promueve el pensamiento creativo. Un grupo de empresas en la misma industria promueve la
innovacin al reunir gente produciendo mercancas similares con tecnologa de produccin similares. En palabras de Alfred
Marshall (1920), cuando una industria ha elegido un lugar para s mismo, es probable que all quedarme mucho tiempo; tan
grandes son las ventajas que la gente sigue el mismo comercio experto por barrio cerca uno al otro. Los misterios del
comercio se convierten sin misterios; Pero son como wer...
Chapter 3
49
URBANIZATION ECONOMIES
Un segundo tipo de extemal economa de escala se
produce si el costo de produccin de un finn individual
disminuye a medida que la produccin total de la
areaincreases urbana. Las economas de urbanizacin se
diferencian de las economas de localizacin de dos
maneras. Primero, las economas de urbanizacin,
resultado de la escala de toda la economa urbana, no
simplemente la escala de una industria en particular. En
50
51
por 11,7 puestos de trabajo. Las ganancias de trabajo son rnuch srnaller para el empleo de software ms distante de
la zona del cdigo postal. Si sumamos 1.000 puestos de trabajo de software en un anillo de 1 a 5 millas escandalizan
al rea postal, software ernployment en la zona aumenta slo 0,8 empleos. Cmo locales son economas de
localizacin? El estudio de Rosenthal y Strange (2000b) muestra que los beneficios asociados con las economas de
localizacin (compartiendo entrada sup alicates, una mano de obra e informacin), rpidamente con la distancia.
Computa la entrega del porcentaje del efecto de localizacin para un moverse una milla lejos escandalizan a ter
clus. En promedio, el efecto de localizacin agota ata tasa de 50 por ciento por milla. La atenuacin rpida de los
resultados de econornies localizacin explica al local de "economas de localizacin". Evidencia de las economas
de urbanizacin all es sorne mezclado pruebas relativas a las economas de urbanizacin. El consenso entre los
economistas urbanos es que el doble del tamao de una ciudad aumentar laboral agrcola pro por 5 a...
J
f
l'
52
es mayor en las ciudades con una amplia variedad de industrias, lo que sugiere que la ecologa urbana nornic
diversidad promueve el crecimiento. Tenga en cuenta que estos resultados confl.ict con los estudios que encuentran
especializacin-grandes concentraciones de empleo en una industria en particular promueve el crecimiento del
empleo. Parece que el tema es mejor para emplear ment crecimiento-diversidad o especializacin-no ha sido resuelta.
No hay duda que la diversidad econornic en ciudades promueve la innovacin en prod uct diseo y produccin.
Glaeser, Kallal, Shein.krnan y Shleifer (1992) pro vide evidencia que los derrames de conocimiento contribuyen al
crecimiento de las ciudades, con los derrames que se producen principalmente entre las industrias en lugar dentro de
un determinado in dustria. Harrison, Kelley y Gant (1996) y Kelley y ayudante (1999) muestran que la diversidad en el
empleo local aumenta la tasa de adopcin de nueva produccin pro cesos. Feldman y Audretsch (1999) muestran que
la diversidad local tiene un fuerte efecto positivo sobre el desarrollo...
.:
A
r
_,\
1
,.
'
53
las tcnicas se estaban desarrollando. Las empresas pequeas computadoras agrupan en el.Silicon
Valley en California y a lo largo de inBastan Rou te 128 a explotar las economas de localizacin. La
estandarizacin de productos y procesos de produccin permiti que las empresas para establecer instalaciones
de gran produccin fuera de los racimos. Por qu sobrevivieron las instalaciones exo dus extensinde los
clusters? Nuevos productos seguirn desarrollndose, y las empresas con continuar en clster para explotar las
economas de localizacin. Mayora de las empresas mantienen sus instalaciones de investigacin y desarrollo en
los grupos y establece plantas de rama para fabricacin y montaje. Las diferencias en la ciudad de tamao
como se muestra en la figura 3-1, all es una variacin sustancial l.n ciudad tamaos. En este captulo, hemos
visto por qu sorne tipos offirms se agrupan en las ciudades. Cmo lleva los agrupamiento offirms el tipo de
distribucin de tamaos de la ciudad que se muestra en la Figur3-1? Vamos a pregunta contestes en tres pasos,
mirando los papeles de las economas de localizacin, las economas de urbanizacin y localmente c...
54
FIGURE 3-3 Variation in City Size
600
500
400
400
2
8
s 300
-a
8
200
80
120
120
100
60
CityL
City S-1
City S-2
City S-3
City S-4
Industrias sujetas a fuertes economas Iocalization forman racimos grandes, causando grandes concentraciones de puestos de trabajo.
Industrias sujetas a las economas ms dbiles de localizacin forman pequeos racimos y las concentraciones de empleo. Las economas de
urbanizacin amplifican diferencias en el empleo debido a las grandes ciudades proporcionan mayores economas de urbanizacin. Locales de
empleo amplifica las diferencias en el tamao de la ciudad porque las ciudades ms grandes tienen una mayor variedad de bienes de consumo.
Podemos usar.Figura 3-3 para mostrar el efecto combinado de las economas urbanas izacin y localizacin en nuestros hipotticos siete
ciudades. Las reas oscuras inferiores muestran empleo base en industrias sujetas a las economas localizatin y las reas ms claras muestran
empleo resultantes de la urbanizacin ec6nomies. La ciudad grande (L), que tiene 120 puestos de trabajo en una industria sujeta a economas
de localizacin, gana 80 puestos de trabajo de las economas de urbanizacin, mientras que cada una de las ciudades de mediano (M) gana
slo 20 puestos de trabajo. Cada ciudad mediana tiene menos que ofrecer que la ciudad grande en trminos de...
55
Porque cuanto mayor sea la ciudad, mayor ser el atractivo de la ciudad a firmas sujetas a las economas de
urbanizacin. El papel de los bienes de consumo. Hasta ahora nuestra discusin de la distribucin de tamao
de las ciudades ha descuidado al lado de los consumidores del mundo urbano. Hemos asumido implcitamente
que todos los trabajadores en la ciudad de producen mercancas para la exportacin de la ciudad, pero no
consuman ningn producto localmente. De hecho, los trabajadores en una ciudad pasan algunos de sus
ingresos en productos de consumo pro provistas dentro de la ciudad: comprar comida, haz cortes de cabello,
alquiler de vivienda, utilizar medica! ser vicios e ir a eventos deportivos. Este gasto en productos de consumo
local sup puertos empleos en tiendas de abarrotes, barberas, hospitales, construccin de viviendas y equipos
deportivos. Los trabajos apoyados por el gasto de los consumidores locales a menudo se denomina empleo
local. Algunos productos estn disponibles en todas las ciudades, grandes, medianas y pequeas. Si un
producto tiene gran demanda per cpita en relacin con econornies de escala en la produccin, incluso una
pequea ciudad generar...
J
'
56
Podemos usar el simple modelo representado en la figura 3-3 para mostrar los efectos del empleo
local sobre las diferencias en el tamao de la ciudad. Supongamos que empezamos con el empleo de exportacin:
ciudad L inicialmente tiene 200 exportjobs (120 de las economas de localizacin) y 80 de las economas de
urbanizacin, mientras que cada una de las ciudades M tiene 80 puestos de trabajo y cada una de las ciudades de S
tiene 30 puestos de trabajo de exportacin. Pequea ciudad: un trabajo local por trabajo de exportacin. Para
proporcionar los productos que pueden ser apoyados incluso en una ciudad pequea, agregamos un trabajador
local (barberos y empleados) por cada trabajador de exportacin. Gran ciudad: dos puestos de trabajo locales
por exportacin de trabajo. Agregamos bastante barberos y empleados para atender a la poblacin ms grande y
tambin aadir que aadimos los cantantes de pera y neurocirujanos. Ciudad media: 1,5 empleos locales por
exportacin de trabajo. Figura 3-3 muestra cmo las economas de urbanizacin y empleo local soy plify cualquier
inicial diferencias en el empleo. La ciudad grande comenz con cuatro veces ms puestos de trabajo en industrias
sujetas a las economas de localizacin como el pequeo...
60
Part 1
of Cities
una mayor demanda de telecomunicaciones porque las dos formas de comunicacin son complementos. Una
cuestin ms sutil refiere al efecto de las telecomunicaciones en la compleja realidad de las interacciones en las ciudades.
Comunicacin ms fcil puede permitir una mayor especializacin en diseo y produccin, aumentando la necesidad de
comunicacin cara a cara coordinar los esfuerzos especializados de las diferentes personas involucradas. Adems, la
comunicacin ms fcil puede aumentar la complejidad de la informacin y las ideas trans enguantada entre personas,
que requieren ms contactos cara a cara para comunicar estas ideas. Gottman (1977) analiza los efectos del telfono (el
primer Reglamento de l tion dispositivo) en las ciudades: proporciona el telfono, cuando sea necesario, cuasi-inmediata
comunicacin verbal entre todas las unidades interdependientes a un costo mnimo. Habra sido muy difcil para todas
estas complejas e integradas las redes (en las ciudades) a trabajar en unsono sin el telfono, que hizo posible la constante
y efic...
La misma lgica se aplica a las innovaciones ms recientes tales como faxes, correo electrnico e Internet.
Cules son los hechos sobre las innovaciones en las telecomunicaciones y la urbanizacin? Durante el
reciente perodo de rpido desarrollo de las telecomunicaciones, urbanizacin aument. Basado en las
tendencias recientes, la proporcin de la poblacin mundial vive en ciudades se espera que el aumento de su
nivel actual de 45 por ciento al 61 por ciento para el ao 2025. Durante este perodo, la proporcin de
poblacin viviendo en ciudades se espera que aumente de alrededor del 74 por ciento a 84 por ciento en
Amrica del norte, Amrica del sur y Europa; de alrededor de 34 por ciento a 54 por ciento en frica y Asia; y
del 70 por ciento a 75 por ciento en la regin del Pacfico. Parece las innovaciones en tecnologa de
telecomunicaciones no causar ciudades a desaparecer. Sorne interacciones entre los productores requieren
contacto cara a cara, as que siempre habr una necesidad para las ciudades y la proximidad fsica que
proporcionan. El nmero offace...
SUMM
ARY
l. las economas de localizacin (menores costos de produccin como salida formacin de pliegues) ocurren
porque las empresas en un clster de la industria se benefician de compartir (1) los proveedores de insumos
intermedios, (2) una mano de obra y (3) la informacin. 2. las economas urbanizacin ocurren Si cuesta la
produccin de un especial de firma / pliegues como la salida total de los aumentos del rea urbana. ) )
61
62
..
t:
J
J
l
l
I
l.
!
l
.,. "
'
10
3
8
4
7
5
6
8
3
63
sell the same quantity of cars." Critically appraise the Honda dealer's statement.
If the statement is incorrect, what's wrong with the reasoning? illustrate your
answer with a graph.
.1
REFERENCES
l. Carlina, Gerald A. "Productivity in Cities: Does City Size Matter?" Business Review ofthe Federal Reserve Bank of Philadelphia (November/December 1987),
pp. 3-12.
2. Carlton, D.W. "The Location and Employment Choices ofNew Firms." Review
of Economics and Statistics 65 (1983), pp. 440-49.
3. Dumais, Guy, Glen Ellison, and Edward Glaeser. "Geographic Concentration
as a Dynamic Process." National Bureau ofEconomic Research Working Paper
6270, November 1997.
4. Duranton, G., and D. Puga. "Diversity and Specialization in Cities: Why, Where
and When Does It Matter?" Urban Studies (2000).
5. Duranton, G., and D. Puga." "Nursery Cities: Urban Diversity, Process Innovation, and the Life Cycle of Products." CEPR Discussion Paper No. 2376
(1999).
6. Feldman, Maryann P., and David B. Audretsch. "Innovation in Cities: ScienceBased Diversity, Specialization and Localized Competition." European Economic Review 43 (1999), pp. 409-29.
7. Fujita, Masahisa, and Ryoichi Ishii. Global Location Behavior and Organizational Dynamics of Japarrese Electronics Firms and Their Impact on Regional conomies. InAlfredD. Chandler, Jr., Peter Hagstrom, andrjan Solvell
(eds.), The Dynamic Firm: The Role of Technology, Strategy, Organization and
Regions. Oxford: Oxford University Press, 1998, pp. 343-83.
8. Gaspar, Jess, and Edward L. Glaeser. "Information Technology and the Future
of Cities." Joumal of Urban Economics 43 (1998), pp. 136-56.
9. Glaeser, Edward L.; Hedi D. Kallal; Jose A. Scheinkman; and Andrei Shleifer.
"Gowth in Cities." Journal of Political Economy 100 (1992), pp. 1126-52.
1O. Gottman, J. "The Telephone and the Structure of the City." In The Social Impact
ofthe Telephone, ed. l. De Sola Pool. Cambridge, Mass.: MIT Press, 1977.
11. Graham, Stephen. "Telecommunications and the Future of Cities: Debunking
the Myths." Cities 14 (1997), pp. 21-29.
12. Hanson, Gordon. "Scale Economies and the Geographic Concentration of Industry." Joumal ofEconomic Geography 1: 255-276 (2001).
13. Harrison, Bennett; Maryellen R. Kelley; and Jon Gant. Specialization versus
Diversity in Local Econornies: The Implications for Innovative Private-Sector
Behavior. Cityscape 2 (1996), pp. 61-93.
14. Head, K.; J. Ries; aiid D. Swenson. ''Agglomeration Benefits and Location
Choice." Joumal of Intemational Economics 38 (1995), pp. 223-48.
15. Henderson,J.V. "Effic;iency ()f Resource Usage and City Size." Joumal of U rban
Economics 19 (1986), pp. 47-90.
,------,
1
64
16. Henderson, J.V. Urban Development: Theory, Fact, and Jllusion. New York:
Oxford University Press, 1988.
17. Henderson, J.V.; Kuncoro, A.; and M. Turner. "Industrial Development and
Cities." Joumal of Political Economy 103 (1995), pp. 1067-81.
18. Jacobs, Jane, The Economy of Cities New York: Random House, 1969.
19. Kelley, Maryellen R., and Susan Helper. Firm Size and Capabilities, Regional
Agglomeration, and the Adoption ofNew Technology. Economics oflnnovation
and New Technology 8 (1999), pp. 79-103.
20. Marshall, Alfred. Principies of Economics. London: Macmillan, 1920, p. 352.
21. McLuhan, H. Understanding Media-The Extension of Man. London: Sphere,
1964.
22. Milis, Edwin. "A Thematic History of Urban Economics." Brookings-Wharton
Papers on Urban Affairs (2000), pp. 1-52.
23. Mun, Seil, and Bruce G. Hutchinson. "Empirical Analysis of Office Rent and
Agglomeration Economies: A Case Study of Toronto." Joumal of Regional
Science 35 (1995), pp. 437-55.
24. Nakamura, R. "Agglomeration eco omies in urban manufacturing industries: a
case of Japanese cities," Journal ofUrban Economics 17 (1985), pp. 108-24.
25. Negroponte, N. Being Digital: The Roadmap to the Information Superhighway.
London: Hodder and Stoughton, 1995.
26. Rosenthal, Stuart, and William Strange. "The Determinants of Agglomeration,"
Working Paper (2000a).
27. Rosenthal, Stuart, and William Strange. "Geography, Industrial Organization,
and Agglomeration," Working Paper (2000b).
28. Saxenian, Annalee. Regional Advantage: Culture and Competition in Silicon
Valley and Route 128. Cambridge, Mass.: Harvard University Press, 1994.
29. Toffier, A. The Third Wave. New York: Morrow, 1980.
30. Vernon, Raymond. "External Economies." In Readings in Urban Economics,
ed. M. Edel and J. Rothenberg. New York: Macrnillan, 1972.
CHAPTER 4
In
earlier chapters we explained why cities exist and why sorne cities are so big. In
this chapter, we'll explore the where of cities. To address the where question, we'll
examine the location decisions of firms. When a firm chooses a particular location
for its production facility, the resulting concentration of ernployrnent either generates
a new city, Or, more often, Ca"\].SeS an existing city tO grow. Although location theory
is usually cast in terms of where new firms locate, it applies as well to the expansion
of existing firms. The economic conditions that attract new firms to a city also rnake
it profitable for firms aiready in the city toexpand their operations. In other words,
location theory helps us explain both why cities arise in particular locations and why
sorne cities grow more rapidly than others.
Figure.4-1 provides a useful backdrop for our discussion of location decisions
and urban g!owth. A city will grow if the nurnber of new jobs gained frorn the birth
of new firms and the expansion of existing firms exceeds the nurnber of old jobs
lost from firm deaths and contractions. The question is whether growing cities have
relatively large job gains, or relatively small job losses. The clear rnessage from
Figure 4-1 is that cities differ in their job gains, not their job losses. Although most
cities have roughly the same percentage of job losses, growing cities have larger
gains from births and expansions.
The location decisions of firms are based on profit rnaximization. A firm's potenti81 profit vares across space for several reasons. First, it is costly to transport
inputs and outputs, and locations with relatively low transport costs will generate
higher. profits, ceteris paribus (everything else being equal).. Second, sorne inputs
cannot be transported at all, and locatons with inexpensive local (nontransferable)
inputs will generate higher profits, ceteris paribus. Third, sorne firms benefit from
proximity to other firms in the sarne industry (localization economies) and other
firms benefit from being in a large diverse city (urbanization economies). Fourth,
the public sector levies taxes and provides public goods and services, and locations with a relatively efficient public sector will generate higher profits, ceteris
paribus.
-
-
65
..
;&_
-.
r
66
FIGURE 4-1 Gross Gains and Losses in Employment in Selected Cities, 1984-1986
45 - .-----------,
Openings and expansions
40 -
1
O Closings and contractions
r-
r--
r-- r-
Q)
OJl
c:l
..0
25-r-
r----------------
20-r-
1---
--
15-
--
105-
---
---
- --- -
--
.--
--
--
'-
'--
--
--
--
::
Q)
:l
r-
---
--r-
r-
'
--r-
::,..
..0
OJl
.....
::::1
><
..0
....
s
.J:j
Q)
o
Q)
o
.......
::::1
:Q
OJl
"'::
"o:!'
:E
<:<:
c:l
Q)
"'
:Eo.
"o'
......
;.=
c:l
"O
::
<:<:
IZl
:.a
.S
:pE.,
o.
Q)
'E
c:l
Q)
IZl
<:<:
.....
::
OJl
o
Q)
::
c:l
IZl
Source: Table 2.3 in Rr+_dall W. Eberts and Joe A. Stone, Wage and Adjustm.ent in Local Labor Markets
(Kalamazoo, MI: Upjohn'J:nstitute, 1992).
'--
....
67
Fixed-factor proportions. The firm produces its fixed quantity with fixed
arnounts of each input. In other words, the firm uses a single recipe to produce
its good, regardless of the prices of its inputs. There is no factor substitution.
Fixed prices. The firm is so small that it does not affect the prices of its inputs
or its product.
Under these four assumptions, the firm maximizes its profit by minimizing its
transportation costs. The firm's profit equals total revenue (price times the quantity
of output) less input costs and transport costs. Total revenue is the same at all
locations because the firm sells a fixed quantity of output at a fixed price. Input costs
are the same at all locations because the firm buys a fixed arnount of each input
at fixed prices. The only costs that vary across space are procurement costs (the
costs of transporting the firm's transferable input) and distribution costs (the costs
of transporting the firm's output). Therefore, the firm will choose the location that
minimizes its total transport costs.
The firm's location choice is determined by the outcome of a tug-of-war. The
firm is pulled toward its input source because the closer to the input source, the lower
the firm's procurement costs. On the other side, the firm is pulled toward the market
because proximity to the market reduces the firm's distribution costs.
Resource-Oriented Firms
1'
.
A resource-oriented firm is defined as a firm that has relatively high costs for transporting its input. Table 4-1 shows the transport characteristics for such a firm. The
firm produces baseball bats, using 10 tons of wood to produce 3 tons of bats. The
firm is involved in a weight-losing activity in the sense that its output is lighter than
its transferable input.
The key factors in the tug-of-war are the monetary weights of the firm's input
and output. The monetary weight of the input is equal to the physical weight of
the input (10 tons) times the transportation rate ($1 per ton per mile), or $10 per
mile. Similarly, the onetary weight of the output is 3 tons times $2, or $6 per mile.
The firm is considered a resource-oriented firm because the monetary weight of its
transferable input exceeds the monetary weight of its output. Although the unit cost
of transporting output is higher (because finished bats must be packed carefully,
but logs can be tossed onto a truck),' the loss of weight in the production process
generates a lower monetary weight for the output.
Figure 4-2 shows the firm's transportation costs. If we use x as the distance
from the input source (the forest) to .the production site (the factory), the firm's
TABLE 4-1 Monetary Weights for a Resource-Oriented Firrn
Input (wood)
Physical weight (tons)
Transportrate (cost per ton per mile)
Monetary weight (physical weight times rate)
10
$1
$10
Output (bats)
3
$2
$6
68
100
1
1
1
80
1
1
1
1
60
1
1
1
40
1
1
1
. 1
20
: Market (M)
1
6
4
x (Distance from forest)
Forest (F)
Total transport cost (the sum ofprocurement cost and distribution cost) is minimized at the forest
because the monetary weight of the input ($10) exceeds the monetary weight of the output ($6). The
weight-gaining activity locates at its source of raw materials.
procurement cost is
PC = W. t.
that is, the monetary weight of the input (the physical weight W times the transport cost rate t) times the distance between forest and factory. The slope of the
procurement-cost curve is the monetary weight of the input, so PC rises by $1O per
mile, from zero at the forest to $100 at the market 10 miles away.
The distribution costs are computed in an analogous way. If XM is the distance
between the forest and the market, the firm's distribution cost is
DC = W 0
t0
(xM- x)
that is, the monetary weight ofthe output (weight w0 times the transport cost rate t0 )
times the distance from the factory to the market. The slope of the distribution-cost
curve is the monetary weight of the output, so as we move from the forest toward
the market, DC decreases by $6 per mile, from $60 at the forest (lO miles from the
market) to zero at the market.
Total transport cost is the sum of procurement and distribution costs. In Figure 4-2, total transport cost is minimized at the forest site. at $60. To see why
transport cost is minimized here, suppose the firm started at the forest site and
then moved one mile toward the market. Its distribution cost would decrease by $6
(the monetary weight of bats) but its procurement cost would increase by $10 (the
'
'
..
..
r;.
..,
.'
69
C318 L 1 O TE CA- y---monetary weight of the wood), so its total transport cost would increases by $4. The
firm's total transport cost is minimized at the forest because the monetary weight
of the input exceeds the monetary weight of the output. The resource-oriented firm
locates near its input source.
The bat firm is resource-oriented because it is a weight-losing activity, using
10 tons of wood to produce only three tons of bats. The cost of transporting wood
is large relative to the cost of trarisporting the finished output, so the firm saves on
transport costs by locating near the forest. In this case, the tug-of-war is won by the
input source because there is more physical weight on the input side. Here are sorne
other examples of weight-losing firms.
.,
-.
l. Beet-sugar factories locate near sugar-beet farros because one pound of sugar
beets generates only about 2.7 ounces of sugar.
2. Onion dehydratorslocate near oniqn fields because one pound of fresh onions
becomes less than one pound of dried onions.
3. Ore processors locate near mines because they use only a fraction of the materials
extracted from the ground.
Sorne firms are resource oriented because their inputs are relatively expensive
to transport. Consider a firm that cans fruit, producing one ton of canned fruit with
roughly a ton of rawfruit. The firm's input is perishable, and must be transported in
refrigerated trucks, while its output can be transported less expensively on regular
trucks. Because the cost of shipping a ton of raw fruit exceeds the cost of shipping a
ton of canned fruit, the monetary weight of the input exceeds the monetary weight
of the output, and the firm willlocate near its input source, a fruit farro.
In general, a firm's input will be more expensive to ship if it is more bulky,
perishable, fragile, or hazardous than the output. Hoover (1975) provides sorne
examples of such activities:
.
1
l
1
..al.
l. Cotton baling. The input (raw cotton) is more bulky than the output (baled
cotton). The cost of shipping a tonof fl.uffy cotton exceeds the cost of shipping
a ton of.compacted cotton, so the monetary weight of the input is higher and the
resource-oriented cotton baler willlocate near the cotton field.
2. Skunk deodorizing. The input (fully arroed skunks) is more fragile and hazardous than the output (disarmed skunks). The cost of shipping a ton of armed
skunks exceeds the cost of shipping a ton of disarroed ones, so the skunk deodorizer willlocate near the skunk farro.
In general, when a firm's input is relatively bulky, perishable, fragile, or hazardous,
the tug-of-war will be won by the input source, not because the input is heavier, but
because it is more expensive to transport.
There are many examples of industries that locate close to their transportable
inputs (Ellison and Glaeser, 1999). The producers of soybean and vegetable oil
are concentrated in Nebraska, North Dakota, and South Dakota, close to the farros
that supply soybeans and corn. Mi1'1( and cheese producersoare concentrated in South
70
Dakota, Nebraska, and Montana, elose to dairy fanns. Sawmills and other wood processors are concentrated in Arkansas, Montana, and Idaho, elose to vast tirnberlands.
Market-Oriented Firms
A rnarket-oriented firrn is defined as a firrn that has relatively high costs for transporting its output to the rnarket. Table 4-2 shows the transport characteristics for
such a firrn. The bottling firrn uses one ton of sugar and three tons of water (a ubiquitous input) to produce four tons of bottled beverages. The firrn is involved in a
weight-gaining activity in the sense that its output is heavier than its transferable
input. The rnonetary weight of the output exceeds the rnonetary weight of the input,
so this rnarket-oriented firrn willlocate near its rnarket.
As shown in Figure 4-3, the firrn's transport cost is rninirnized at the market.
Because the monetary weight of the output exceeds the monetary weight of the input,
a one-rnile rnove away frorn the rnarket increases the distribution cost by more than
TABLE 4-2 Monetary Weights for a Market-Oriented Firrn
Input (sugar)
Physical weight (tons)
Transport rate (cost per ton per mile)
Monetary weight (physical weight times rate)
Output (beverage)
$1
$1
$1
$4
40
30
20
6
x (Distance from input source)
".t--,(":'",;' :'
,l:,""
.-
71
Fixed cost. There are fixed costs associated with processing paperwork, and
l
!
I
l
72
Number of consumers
Monetary weight
$4
$16
Distance from W
10
$2
$20
Ann locates her pizza parlor at point Y because it is the median 1ocation: she delivers 1O pizzas to
consumers to the west of Y, and 1O to consumers to the east of Y. A move from Y to S would
decrease delivery cost for the 10 consumers at]X>int Z (savings of$20), but iricrease delivery costs
for the 11 consumers at points W, X, and Y (increase in cost of $22). A move in the opposite
clirection would also increase total transport costs.
73
Demand
L
4
17
The median location is in the large city (L). Any move to the left Ofpint L will increase travel costs
for the majority of consumers, increasing total transportation costs.
To show that the median location rninirnizes total transport costs, suppose that
Ann starts at the median location, and then moves to point S, one mile east of Y. As
she moves to the east, the good news is that she spends less on delivery to point Z:
she saves $2 per trip to Z, saving a total of $10 in eastward delivery costs. The bad
news is that westward costs increase: she pays $2 more per trip to points W, X, and
Y; since there are 11 customers to the west of S, her westward delivery costs increase
by a total of$22 ($4 for W, $16 for X, and $2 for Y). Since the increase in westward
costs exceeds the decrease in eastward costs, the move from Y to S increases total
delivery costs. The same is true for a move in the opposite direction; if Ann moves
from Y toward W, her delivery costs will increase.
The median location rninimizes total transport costs because it splits pizza consumers into two equal parts. As Ann moves eastward away from Y, she moves
further away from 11 customers, but moves closer to only 10 customers. Similarly,
a westward move will cause her to move eloser to 1O customers, but further from
11 customers. In general, any move away from the median location will increase
delivery costs for the majority of consumers, so total delivery cost increases. It is
important .to note that the distance between the consumers is irrelevant to the firm's
location choice. Por example, if the Z consumers were located 100 miles from W
instead of 9 io.p.es from W, the median location would still be point Y. Total delivery
costs would still be minirnized (ata higher level, of course) at point Y.
The principie of median location provides another explanation of why large
cities become larger. Considera firm that delivers its product to consumers in five
different cities. In Figu,re 4-5, there is a large city at location L, and four small cities
at locations S1, Sz, S3, and S4. The firm sells 4 units in each small city, and 17 units
in the large city. The median locatin is in the large city, even though the large city is
at the end of the line. A one-mile move westward from L would decrease transport
costs by $16 (as the firm moved closerto consumers in the small cities), but would
increase transport costs by $17 (as the firm moved away from consumers in the large
city). The lesson from this example is that the concentration of demand in large cities
causes large cities to grow.
74
M
Monetary weight = $10
Input source B
Monetary weight = $15
The firrn locates its sawrnill at the port (P) because it is the median transport location. A move
from .P toward either A or B would increase output transport costs by $10 without affecting
input transport costs. A move from P toward M would increase input transport costs by $30
but decrease output transport costs by only $10.
Figure 4-6 shows the location options for a sawmill. The firm harvests logs
from locations A and B, processes the logs into lumber, and then sells the lumber in
an overseas market at point M. Highways connect points A and B to the port, and
ships travel from the port to point M. The sawmill is a weight-losing activity: The
monetary weights of the inputs are $15 for point A and $15 for point B, and the
monetary weight ofthe output is $10.
Where will the firm locate its sawmill? Although there is no true median location,
the port is the closest to a median location. If the firm starts at the port (P), it could
move either toward one of its input sources or to its market.
1)
l. Toward input source A. A one-mile move from P toward point A will cause
offsetting changes in the costs of transporting logs from the two input sources:
the cost of logs from A would decrease, but the cost of logs from B would
increase. At the same time, the cost of transporting output would increase by
$10. Given the offsetting changes in input transport costs and the increase in
output costs, the port location is superior to locations between P and A. The
same argument applies for a move from P, toward B.
2. To market (M). Unless the firm wants to operate a floating sawmill, it would
not move to points between the port and the overseas market at M. It could,
however, move all the way to the market."A move from P to M would decrease
output transport cost by $10 (the monetary 'weight of output) times the distance
between M and P, and increase input transport cost by $30 (the monetary weight
of the inputs) times the distance. Therefore, the port loction is superior to the
market location.
Although the sawmill is a weight-losing activity, it willlocate at the port, not at
one of its input sources. The port location is efficient because it provides a central
collection point for the firm's inputs.
75
There are many examples of port cities that developed as a result of the location decisions of industrial firms. Seattle started in 1880 as a sawrnill town: Firms
harvested trees in western Washington, processed the logs in Seattle sawrnills, and
then shipped the wood products to other states and countries. Baltimore was the
nation's first boomtown: Flour rnills processed wheat from the surrounding agricultura! areas for export to the West Indies. Buffalo was the midwestem center for fl.our
rnills, providing consumers in eastem cities with fl.our produced from rnidwestem
wheat. Wheat was shipped from rnidwestern states across the Great Lakes to Buffalo,
where it was processed into fl.our for shipment, by ral, to cities in the eastem United
States. In contrast with Baltimore, which exported its output (fl.our) by ship, Buffalo
imported its input (wheat) by ship.
ENERGY ORIENTATION
What is the role of energy in the location decisions of firms? In the first half of the
nineteenth century, energy was a _local input, defined asan input that could not be
transported from one location to another. The waterwheel was the first device used
to generate nonanimal mechanical energy. The waterwheel was tumed by waterfalls
and fast-moving streams, providing power for production facilities located along
rivers and streams. Textile manufacturers built factories along small backcountry
streams in New England, with waterwheels powering their machinery. Among the
cities that developed as a result of the waterwheel are Lowell, Lawrence, Holyoke,
and Lewiston.
The development of the steam engine in the second half of the nineteenth century
made energy.. a transportable input. The steam engine could be operated anywhere,
with the only constraint being the availability of coal to fuel the engine. Sorne energyintensive manufacturers located near the coal mines in Pennsylvania. Others located
along navigable waterways and shipped coal from the mines to their factories. In
New England, textile firms shifted from backcountry waterfall sites to sites along
navigable waterways. Production shifted to the Fall River-New Bedford area along
the south coast of New England. The development of the railroad gave coal users
another transport option, causing the deVelopment of production sites along raillines.
The development of electricity affected the location patterns of manufacturers
Because electricity can be transmitted over distances of several hundred miles, a firm
can use the energy.generated from water power without locating along a backcountry
stream. Similarly, a firm can use coal resources without shipping the bulky fuel from
the mine to its factory. In general, the development of electricity decreased the
importance of energy considerations in location decisions.
For sorne activities, the availability of cheap energy is still an importan!locati"on
factor (Ellison and Glaeser, 1999). Sorne intensive users of electricity (the production
ofprimary aluminum and chlorine), locate in the Pacific Northwest to exploit cheap
electricity from hydroelectric power. The producers of clay tiles and fertilizer are
intensive users of natural :gas, and locatein Arkansas, Louisiana, and Texas, where
natural gas is cheap. The production of cement and lime requires large amounts
76
Part 1
of coal, and firrns in these industries locate in Montana, Nevada, and Wyoming to
exploit cheap coal.
How will the deregulation of electricity markets affect the location decisions
of energy-oriented industries? In the regulated market of the 1990s, there was substantial variation in the price of electricity across states, a result of regulations that
restricted the fiow of electricity across state lines. Deregulation is expected to free
producers to "wheel" power from state to state, and the free fiow of electricity is
expected to reduce the spatial variation in the price of electricity. As a result, fewer
firrns will choose locations based on the price of electricity.
)
)
)
)
)
)
We can modify ourmodel of a firm's location choice to incorporate the cost oflabor.
To simplify matters, suppose the firm gets its inputs and sells its output at the same
location (T). The productivity of labor is the same at alllocations, but the price of
labor (the wage) decreases as the firm moves away from locatin T. In Figure 4-7
the firm's total cost is the sum oflabor costs and transport costs, which is minimized
)
)
_)
)
77
4
6
Distance om market and inputs
10
Wben inputs and outputs are relatively heavy and costly to transport, the firm
minimizes the sum of transport and lal;>or costs by locating elose to the market and
inputs (Panel A, with cost = $30 as shown by point i). A decrease in the physical
weight or unit freight cost causes the firm to.locate far from the market and inputs,
where labor costs are lower (Panel B, with total cost = $10 as shown by point j).
at point T. In this case, transport costs are high relative to the variation in labor
costs, meaning that the transport-cost curve is relatively steep. As a result, the forces
pulling the finntoward location Tare strongertt1.an the forces pulling the firm toward
locations with lower wages and labor costs, so the tug-of-war is won by location T.
78
In the last several decades, transport costs have become less important in firms'
location decisions. In many industries that have traditionally been resource-oriented
or market-oriented, firms now locate far from their input sources and markets. The
changes in locational orientation resulted from innovations in transportation and
production that have decreased transport costs.
Transportation technology. The development of fast ocean ships and container technology decreased shipping costs, while improvements in railroads
and trucks lowered the cost of overland travel. Faster and more efficient aircraft
have decreased the cost of air travel.
Production technology. Improvements in production techniques decreas"ed the
physical weight of inputs. For example, the amount of coal and ore required to
produce one ton of steel has decreased steadily, a result of improved production methods and the us of scrap metal (a local input) instead of iron ore (a
transportable input).
Panel B of Figure 4-7 shows the effects of a decrease in the unit transport costs
and the physical weights of inputs. These changes flatten the transport-cost curve,
shifting the cost-minimizing location to a location 1O miles from the market and
inputs, where labor costs are lower. The decrease in transport costs causes the firm
to switch from transfer orientation to labor orientation.
As unit transport costs and input weights decrease, firms are more likely to base
their location decisions on access to inexpensive local inputs rather than access to
transportable inputs. A recent example is the movement of the assembly operations
of many U.S. manufacturers to sites along the Mexican border. The steel industry
has moved from the eastern United States, with its rich coal and ore deposits, to
Brazil, Korea, and Mexico-far from both raw materials and steel markets. Similar!y,
manufacturers have moved from the United States to Asia and Mexico, far from U.S.
markets, because the savings in labor costs are greater than the increase in transport
costs.
Labor in the Long Run
In the long run, workers can move between cities. In the typical year, about 6 percent
of the U.S. population move across county lines, and about 3 percent move across
state borders {Black, 2000). A firm that locates in one city could hire workers currently living in another city with the expectation that the workers will transfer to the
firm's city. According to Bartik (1991), the vast majority of new jobs in cities are
filled with people who relocate from other cities: When the number of jobs in a city
increases by 1,000, on average, only 230 jobs go to current residents; the rest go to
workers who transfer from other cities.
Labor and Natural Amenities
Considera nation that has two regions-with ariD. weather in the North and 'Y\/
weather in the South-and workers prefer wami weather. To achieve Iocational
,',
Cho.pter 4
79
equilibrium for workers, wages in the South must be lower than wages in the North;
otherwise, there would be an incentive for workers to move to the Solith, getting
the same wage and better weather as well. In equilibrium, the wage for a given skill
level will be lower in the South, and northern workers will be compensated for bad
weather by a higher wage.
The lower wage caused by weather will encourage finns to locate their production facilities in the South. In this case, the local nature of weather and workers'
preference for warm weather makes labor a local input. The firm's location depends
on the location choice of its workforce: instead of workers following firms, firms
follow workers. Graves (1979) and Porell (1982) provide empirical support for this
phenomenon. There is evidence that weather has played an important role in location
decisions and urban growth. In the past few decades, the most rapidly growing urban
areas are ones with warm, dry weather (Black and Henderson, 1999; Glaeser, Kolko,
and Saiz, 2001).
Natural amenities appear to be most important for firms that employ high-.
income workers. Since the demands for these amenities are income-elastio, highincome workers are attracted to locations with amenities, and the firms hiring these
people follow. For example, research and development finns employ engineers and
computer scientists, who place a high value on good weather anda clean environment.
One explanation for the shift of employment from northern states to the southern and
western states is that rising income has increased the demand for natural amenities,
causing workers to move to areas that provide these amenities.
.,
1l
'
What is the role of learning on firms' location choices? As we saw in Chapter 1, cities
promote Ie"arning by facilitating contact between workers of different skill levels,
causing learning by imitation. A city that provides better learning opportunities will
attract workers who are eager to learn as well as firms that are eager to hire such
workers. Productivity is higher in places that provide better Iearning opportunities,
attracting firms to th9se places.
Learning opportunities in cities ci.epend on tbe average education level of the
city's workers. Rauch (1 93) shows that an increase in a city's average education
level provides a better environment for Iearning, and increases the productivity of
individual workers. In other words, a worker learns more and is thus more productive
if he or she is surrounded by better educated people. One implication of this result
is that when a particular worker moves from a city with a low average education
level to one with a high education level, the worker becomes more productive and
earns a higher wage. Specifically, a one-year increase in the average education level
increases a worker's productivity by 2.8 percent.
Human capital is not distributed uniformly across space. As we saw in Chapter 1,
college graduates are overrepresented in cities, especially large ones. Black (2000)
computes an index of dissimilarity for college graduates in U.S. cities. The index
shows the percentage of college gra,duates that would haveto switch cities in order to
ensure an even distribution of coliege graduates across cities. For example, suppose
80
that for the nation as a whole, 25 percent of the metropolitan workers are college
graduates. A dissimilarity index of 12 would indicate that 12 percent of all graduates
would have to switch cities to ensure that in every city 25 percent of workers would
be college graduates. For a set of 318 metropolitan areas, the dissimilarity index
increased from 10.2 in 1940, to 11.6 in 1970, to 14.9 in 1990. In other words, the
concentration of college graduates in cities is substantial and growing.
The uneven distribution ofhuman capital has important implications for location
decisions and urban growth. As we saw earlier, cities with above-average education
levels will have larger productivity gains, and will attract more firms and thus grow
more rapidly.
-.
'f
81
it appears that rnanufacturers are more sensitive than other types of firms to tax
differentials within and between rnetropolitan areas. This is sensible because rnanufacturers are likely to be oriented toward the national rnarket and thus have a wider
range of location options. Second, metropolitan areas with relatively high taxes on
capital (in the form of taxes on business property) tend to repel capital-intensive
industries and attract labor-intensive industries.
1
-
:4.
.:t
,.
Many cities try to attract new firms by offering special subsidies. Levy (1981) describes severa! types df subsidy prograrns, including the following:
l. Tax abatement. In sorne cities, new :firms are exempt frorn local property taxes
for a fixed period (e.g., 10 years).. Sorne cities offer tax abatements to all new
developrnents, while others offer abatements only to :firms that are particularly
sensitive to tax differentials.
2. Industrial bonds. Sorne cities issue tax-free industrial bonds to finance property developrnents. The local government uses the revenue from the bonds to
purchase the land and leases the property toa private firm. Because the interest
incorne frorn industrial bonds is not subject to federal taxes, the bond buyer
accepts a relatively low interest rate (e.g., 8 percent instead of 12 percent).
Therefore;the lessee pays less than the market interest rate on the rnoney borrowed to finance the project. The use of industrial bonds was sharply curtailed
oythe Tax Reform Act of 1986.
82
3. Government loans and loan guarantees. Sorne cities loan money directly
to developers, and others guarantee loans from private lenders. In both cases,
developers borrow money ata relatively low interest rate: Either the city charges
an interest rate below the market rate, or the city decreases the risk associated
with a private loan, allowing the developer to borrow private money ata relatively
low interest rate.
4. Site development. Sorne cities subsidize the provision of land and public services for new development. The city purchases a site, clears the land, builds
roads and sewers, and then sells the site to a developer at a fraction of the cost
of acquiring and developing the site.
Wassmer (1994) explores the effects of local economic-development programs
on business activity in the Detroit metropolitan area. Municipalities in the area differ
in their use of industrial development bonds and their property-tax abatement programs for commercial and industrial property. Wassmer shows that these programs
have a positive effect on business activity in only 5 of 16 cases. For the other 11 cases,
the programs have either no effect on business activity, ora negative effect. Several
studies of enterprise zones (areas of a City where firms pay low tax rates, receive
subsidies for worker training, and are exempted from local regulations) suggest that
such zones are not very ef.fective in luring firms to their areas (Boamet and Bogart,
1996; Papke, 1993 and 1994; Dowall, 1996).
There are many anecdotes about firms shopping around for the best tax-abatement
deal. The classic example is a firm that secretly decides to locate in a particular city
. but then goes to an inferior city and asks for a tax-abatement package. With the
"bid" of the inferior city in hand, the firm then asks the preferred city to match
the tax package. The preferred city matches the tax package, and the firm locates
in that city, which it-had intended to do even in the absence of any tax incentives.
A recent study of property-tax abatement (Anderson and Wassmer, 1995) provides
evidence that this sort of tax shopping occurs frequently. The authors suggest that
state policymakers should develop policies to discourage this sort of competition
between cities.
Professional Sports, Stadiums, and Jobs
In San Francisco in 1997, billboards around the metropolitan area read, "Build the
Stadium-Create the Jobs." That was the campaign slogan for the proponents of a
plan to use $100 million of public money to help pay for a new football stadium
for the San Francisco 49ers (Noll and Zimbalist, 1997). San Francisco is one of
many cities that have subsidized the construction of facilities for professional sports.
Between 1989 and 1997, thirty-one new stadiums were built, at an average cost of
about $150 million. Are sports stadiums effective tools of economic development?
Do they create jobs?
83
American
84
industry. We'll also discuss empirical evidence concerning the effects oflabor costs
and unions on location decisions.
The Semiconductor lndustry
The semiconductor industry illustrates sorne of the complexities oflocational choices
in modero industry. The industry's output is light and compact relative to its value,
so transport costs are relatively unimportant in location decisions. What matters are
localization economies and access to different types of labor.
As explained by Castells (1988), the making of semiconductors involves severa!
distinct operations, requiring three different types of labor:
l. Research and development. Engineers and scientists design new circuits and
prepare the circuits for implantation into silicon chips.
2. Wafer fabrication. Skilled technicians and manual workers make the chips
holding the circuits.
3. Assembly into components. Unskilled workers assemble the chips into electronic components.
Many semiconductor firms split their operations into three parts. Research and
development occurs in the Silicon Valley to exploit localization economies generated by the large concentration of semiconductor firms. The Silicon Valley is also
considered a desirable location by engineers and scientists. Advanced manufacturing (wafer production) is typically located outside the Silicon Valley. For example,
National Semiconductor has manufacturing facilities in Utah, Arizona, and the state
of Washington; Intel has plants in Oregon, ATizona, and Texas; and Advanced Micro Devices has a plant in Texas. These facilities are located in areas that provide
(1) a plentiful supply of skilled manuallaborers, (2) an environment attractive to
engineers and technicians, and (3) easy access, by air transportation, to the Silicon
Valley. The firm's assembly facilities are typically overseas, in locations such as
Southeast Asia that provide a plentiful supply of low-skilled workers.
Japanese Automobile Firms
A recent study of Japanese firms in the automobile industry (Smith and Florida,
1994) explores the relative importance of various locational attributes. Under the
just-in-time inventory system, a firm links the delivery of intermediate inputs with
its production schedule, the idea being that the inputs should arrive just in time for
production. This inventory system requires geographic proximity to input suppliers
to both save on shipping time and facilitate communication between firms. The study
suggests that the major factor in the location of Japanese-affiliated automotive suppliers in the United States is proximity to Japanese-affiliated automotive assemblers.
In other words, agglomeration is the key feature in the location choices of such firms.
The study of Japanese firms provides sorne other insights into locational choices.
Japanese-affiliated automotive suppliers prefer locations with relatively large populations, a high density of manufacturing activity, an educated workforce, and better
85
transportation. Furthermore, these firms tend to pick locations with relatively high
wages, refl.ecting their willingness to pay higher wages in exchange for higher labor
productivity. Finally, Japanese-affiliated manufacturers tend to locate in areas with
relatively high concentrations of minority workers.
86
Part 1
with knowledge and experience in tufting. Support firms located nearby, supplying
dyes, backing, and other intermediate inputs. Sorne of the old firms that had produced
woven carpets went out ofbusiness because they were underpriced by tufting firms,
while others moved frorn the Northeast to the Dalton area and switched to the new
technique. By the rniddle of the 1950s, Dalton was the carpet capital of the nation.
Of the top 20 carpet rnanufacturers in the United States, 6 are located in Dalton, and
13 others are located nearby.
Nashville, TN
Lexington, KY
St. Louis, MO
Bloomington, rr..,
Kalamazoo, :MI
Terre Haute, IN"
Marysville, OH
Transport Costs
Total Measured
Cost
$426
423
419
417
430
413
427
$159
186
172
202
244
209
219
$118
106
134
162
116
168
169
$703
715
725
781
790
790
815
1
'\
Source.' Timothy J. Bartik, Charles Becker, Steve Lake, and John Bush, "Saturo and State Economic Development," Forum
for Applied Research and PubTic Policy (Spring 1987), pp. 29-40.
-
Chapter 4
'
87
would pay low wages and pass on the savings to GM. As shown in Table 4-3, the
variation in labor cost was relatively large. The difference between the lowest-cost
site (Nashville) and the highest-cost site (Kalamazoo) was $85 per car.
The third step was to estimate tax.es for the seven possible sites. Table 4-3
shows the tax. cost per car in the absence of special subsidies and exemptions. The
di:fference between the site with the highest tax.es (Marysville) and the site with the
lowest tax.es (Lexington) was $63 per car. Nashville was ranked third, with tax. costs
per car $12 higher than Lexington.
This case study suggests that in the absence of special tax. treatment, Nashville
had the lowest total cost. The most important factor was the lower labor costs in
Nashville. Although the city hada slight disadvantage to Terre Haute in terms of
market access, this disadvantage was more than offset by Nashville's lower labor
costs.
The state ofTennessee offered three types of inducements to GM. The first was a
$30 million highway project that connected the plant with Interstate 65. The second
was a subsidized job-training program for Saturn workers, worth about $4 per car.
The third was a program of property tax. subsidies, worth about $30 per car. As
shown by the numbers in Table 4-3, these subsidies would not have been necessary
in the absence of special subsidies from other states. Other states did offer special
subsidies, and Tennessee responded in kind to stay competitive in the bidding for
the plant.
How do wages affect firms location decisions? Carlton (1983) examines the relationship between the number of new :firms in a metropolitan area and several
variables, including local wages. He looks at firm births in three industries: plastics
products, electronic transmitting equipment, and electronic components. These industries have relatively low transport costs, so they are oriented toward local inputs,
not toward output markets or natural resources. His results suggest that for the electronic components industry the elasticity of the number of plant births with respect
to the metropolitan wage is -1.07. In other words, a 1O percent increase in the wage
decreases the number of births by 10.7 percent. Bartik (1991) examines and summarizes the results of dozens of studis and concludes that the long-run elasticity of
bus.iJiess activity with respect to the wage is between -1.0 and -2.0. This means
that a 1O percent decrease in the metropolitan wage will increase business activity
between 1O percent and .20 percent.
How do unions affect location decisions and the volume of business activity?
One ofthe effects ofunions is to increase wages, so ahighly unionized metropolitan
area is likely to have relatively high wages. Since the elasticity of business activity
with respect to the wage is relatively large, to the extent that unions increase wages,
they decrease business activity. Unions may have other (nonwage) effects that influence business location decisions. For example, unions may affect labor productivity.
According to Bartik (1991), empirical studies of the nonwage effects of unions
have generated mixed results. Although most studies fund that the presence of
r
88
unions decreases business activity, many studies suggested that the negative effects
are rather small.
SUMMARY
l. Cities develop around the concentrations of employment generated by firms, so
the location choices of firms determine the location of cities and urban_ growth.
2. A resource-oriented firm has relatively high transport costs for its inputs, so it
locates near its input source.
.
3. A market-oriented firm has relatively high transport costs for its output, so it
locates close to its market.
4. A market-oriented firm with multiple inputs and outputs willlocate at its median..
transport location. The median location is often in large cities, providing one
reason for the growth of big cities.
5. Sorne firms are oriented toward local inputs. Energy-intensive firms are drawn
toward locations with cheap energy, while firms that benefit from agglomeration
economies are pulled toward industry clusters or large cities.
6. In the last several decades, the relative importance of transport costs has decreased, reftecting decreases in unit transport costs and the physical weight of
inputs.
7. Natural amenities attract workers, and the resulting lower wages attract firms.
8. Large cities provide learning opportunities that increase productivity and attract
firms, causing urban growth.
don't use location theory to make location decisions. I chose this location for
my plywood mill because it is elose to my favorite fishing spot."
2. Depict graphically the effects of the following changes on the bat firm's cost
curves (shown in Figure 4-2). Explain any changes in the optimum location.
a. The cost of shipping bats increases froin $1 per tonto $4 per ton, while the
cost of shipping wood remains at $1 per ton.
b. The forest at point F burns down, forcing ihe firm to use wood from point
G which is 10 miles west of point F (20 miles from the market).
c. The firm starts producing bats with wood and cork, using three tons of wood
and two tons of cork to produce three tons of bats. Cork is ubiquitous (available at alllocations for the same price).
3. Why do breweries typically locate near their markets (far from their input
sources), while wineries typically locate near their input sources (far from their
markets)?
4. The building ofwooden ships was a weight-losing activity, as evidenced by the
piles of scrap wood generated by shipbuilders. Yet shipbuilders located in ports,
far from their input sources (inland forests). Why?
89
5. Considera firm that delivers video rentals to its customers. The spatial distribution of customers is as follows: 10 videos are delivered to location W, lO miles
due west of the city center; 50 videos are delivered to the city center; 25 videos
are delivered to E, 1 mile due east of the city center; and 45 videos are delivered
to point F, 2 mJJ.es east of the city center. Production costs are the same at all
locations.
a. Using a graph, show where the firm should locate. Explain your location
choice.
b. Suppose that point W is in a valley and point F is at the top of a mountain.
Therefore, the unit cost of easterly transport (shipments from west to east)
is twice the unit cost of westerly transport. If production costs are the same
at alllocations, where should the firm locate? Explain.
6. Figure 4-4 shows the location choice of Ann's pizza firm. Discuss the effects
of the following changes on Ann's location choice.
a. A tripling of the distance between Y and Z (from 7 miles to 21 miles).
b. A tripling of the number of customers at point W. Instead of two customers
at W there are six.
c. Ann stops delivery service, forcing consumers to travel to the pizza parlar.
7. In Figure 4-6, the weight-losing firm is located at point P (the port). If the
monetary weight of location B is $27 instead of $15, will the firm stilllocate at
point P?
8. There is sorne evidence that people have become more sensitive to air pollution.
In other words, people are willing to pay more for clean air. If this is true, what
infl.uence will it have on the location decision of firms?
9. Consider a firm that uses one transferable input to produce one output. The
monetary weight of the output is $4, and the monetary weight of the input is $3.
The dist nce between M (the market) and F (the input source) is 10 miles.
a. Suppose that production costs are the same at alllocations. Using a diagram
like the one in Figure 4-2, explain where the firm willlocate.
b. Suppose that the cost of land (a local input) increases as one approaches the
market. Specifically, suppose that the cost of land is zero at F but increases
ata rate of $2 per mile as the firm approaches M. Depct the location choice
of the firm graphically.
10. Chapter 2 discusses the incubator process. When industries mature, they move
from single-activity clusters to ateas with lower land and labor costs. Explain
this process in terms of changes in the orientation of firms as they mature.
11. Suppose that country L has a plentiful supply of labor (and low wages) but a
relatively low supply of raw materials. In contrast, H has a plentiful supply of
raw materials, but a relatively low supply of labor (and high wages). The two
countries are separated by a mountain range that makes travel between the two
countries prohibitively costly. Suppose that a weight-losing product is initially
produced in H (clase to the supply of raw materials). Suppose that a tunnel is
bored through the mountain, decreasing the costs of shipping raw materials and
output between the two countries. Assume that laborers do not migrate from
one country to th oi:b.er.
90
a. How will the tunnel affect the location choices of weight-losing firms?
b. How will the tunnel affect wages in the two countries?
c. How might this analysis be used to explain (1) the shift in manufacturing from the United States to East Asian countries and (2) the narrowing of the wage differential between the United States and East Asian
countries?
REFERENCES
l. Anderson, John E., and Robert W. Wassmer. "The Decision to 'Bid for Business':
Municipal Behavior in Granting Property Tax Abatements." Regional Science
and Urban Economics 25 (1995), pp. 739-57.
2. Baade, RobertA., andAllenR. Sanderson. ''TheEmploymentEffects ofTeams
and Sports Facilities." Sports, Jobs and Taxes, eds. Roger Noll and Andrew
Zimbalist. Washington: Brookings, 1997.
3. Bartik, Timothy J. Who Benefitsfrom State and Local Economic Development
Policies? Kalamazoo, MI: W. E. Upjbhn Institute, 1991.
4. Bartik, Timothy J.; Charles Becker; Steve Lake; and John Bush. "Saturo and
State Economic Development." Fontm for Applied Research and Public Policy,
Spring 1987, pp. 29-40.
5. Black, Duncan, and Vernon Henderson. "A Theory ofUrban Growth." Joumal
of Political Economy 107 (1999), pp. 252-84.
6. Black, Duncan. "Local Knowledge Spillovers and Inequality." Working Paper,
University of California, Irvine (2000).
7. Boarnet, Marlon, and William T. Bogart. "Enterprise Zones and Employment:
Evidence from New Jersey." Joumal ofUrban Economics 40 (1996), pp. 198215.
8. Carlton, D.W. "The Location and Employment Choices ofNew Firms." Review
of Economics and Statistics 65 (1983), pp. 440-49.
9. Castells, Manuel. "The New Industrial Space: Information Technology Manufacturing and Spatial Structure in the United States." In America's New Market
Geography, eds. George Sternlieb and James Hughes. New Brunswick, NJ:
Center for Urban Policy Research, 1988.
1O. Dowall, David. "An Evaluation of California's Enterprise Zone Programs." Econome Development Quarterly 10 (1996), pp. 352-68.
11. Eberts Randall W., and Joe A. Stone. Wage and Adjustment in Local Labor
Markets. Kalamazoo, MI: Upjohn Institute, 1992.
12. Ellison, Glen, and Edward Glaeser. "The Geographic Concentration of Industry:
Does Natural Advantage Explain Agglomeration?"American Economic Review
89 (1999), pp. 311-16.
13. Glaeser, Edward L., Jed Kolko, and Albert Saiz. "Consumer City." Joumal of
Economic Geography 1 (2001), pp. 27-50.
14. Graves, Phillip E. "A Life-Cycle Empirical Analysis ofMigration and Climate
by Race." Joumal of Urban Economics 6 (1979), pp. 135-47.
91
15. Greenwood, Michael. "Human Migration: Theory, Models, and Empirical Studies." Journal of Regional Science 25 (1985), pp. 521-44.
16. Hanson, GordonH. "Agglomeration, Dispersion, and the Pioneer Firm." Journal
ofUrban Economics 39 (1996), pp. 255-81.
17. Helms, L. Jay. "The Effect of State and Local Taxes on Economic Growth: A
Times Series-Cross Section Approach." Review of Economics and Statistics 68
(1985), pp. 574-82.
18. Hoover, Edgar M. Regional Economics. New York: Knoph, 1975.
19. Krugman, Paul. Geography and Trade. Cambridge, MA: MIT Press, 1991.
20. Levy, John M. Economic Development Programs for Cities, Counties, and
Towns. New York: Praeger Publishers, 1981.
21. Munnell, Alicia H. "How Does Public Infrastructure Affect Regional Economic
Performance?" New England Economic Review, 1990, pp. 11-33.
22. Noll, Roger G., and Andrew Zimbalist. "Build the Stadium-Create the Jobs!"
Sports, Jobs and Taxes, eds. Roger Noll and Andrew Zimbalist. Washington:
Brookings, 1997.
23. Papke, Leslie. "Tax Policy and Urban Development: Evidence from the Indiana
Enterprise Zone Program."Journal of Public Economics 54 (1994), pp. 37-49.
24. Papke, Leslie. "What Do We Know About Enterprise Zones?" In Tax Policy and
the Economy, ed. James Poterba. Cambridge, MA: MIT Press, 1993.
25. Porell, Frank W. "Intermetropolitan Migration and Quality of Life." Journal of
Regional Science 22 (1982), pp. 137-58.
CHAPTER 5
92\ }'
. r\ -:;
/
\:
\
93
.e
Number of Areas
1
2
4
14
19
33
52
99
172
;t
..
:.
.>f
18-------1 -------------------------1
1
1
1
.g
13 ------ ------
0..
-.,
f.
-ji
1
1
1
1
1
1
1
1
1
1
1
1.
1
:f
1
1
- --------------\
.1
1
1
30
20 '
1
1
10
10
o
Distance from center ofregion (miles)
,.
20
30
T
h
e
n
e
t
p
ri
c
e
o
f
C
D
s
i
s
t
h
e
s
u
m
o
f
t
h
e
st
o
r
e
p
ri
c
e
(
$
8
)
a
n
d
t
h
e
c
o
st
o
f
tr
a
v
e
li
n
g
t
o
t
h
e
m
u
si
c store
(50 cents per round-trip mile per CD).
Because music stores sell the same product at the same price, they differ only in
location.
94
.r
Average
production cost
Marginal
cost
Demand = Average
revenue
--Marginal revenue
qm
next to the music store, to $13 for a household living 10 miles from the music store,
to $23 for a household 30 miles from the store.
Figure 5-2 shows how the monopolist chooses the profit-maximizing output.
The monopolist's demand curve is negatively sloped; as the store price increases,
people buy fewer CDs. The negatively sloped demand curve generates a negatively
sloped marginal-revenue curve. Profit is maximized where marginal revenue equals
marginal cost, so the monopolist produces qm U!ts of output. If the monopolist sells
qm CDs, the market-clearing price is Pm (the point on the demand curve above qm).
Total profit (revenue less cost) is indicated by the shaded area.
_)
_/
1
.../
95
Average
production cost
Marginal
cost
Pe-----
Cm-------
"'&..J..._-+----Demand =Average
revenue
'&--------1-----p.._..---Marginal
revenue
qe
qm qc
Quantity of CDs per store
Entry shifts the demand curve of the individual store to the left, decreasing price and profit. Entry
continues until profit is zero. The equilibrium output is qe because marginal revenue = marginal
cost (firms are maximizing profit) and profit =average production cost (profit equals zero).
profit, producing the output at which marginal revenue equals marginal cost. Second,
economic profit is zero: The store price equals the average total cost of production.
Figure 5-3 shows the equilibrium situation. Atan output qe, marginal revenue
equals marginal cost (firms are maximizing profit), and the store price Pe equals the
average production cost (profits are zero). At the equilibrium output, the marginal
revenue curve intersects the marginal-cost curve, and the demand curve is tangent
to the average production-cost curve.
This is the theory of monopolistic competition (covered in most intermediate
microeconomics textbooks). Each music store is a monopolist within its own territory; but its monopoly power is limited by entry and competition. Each store faces
a negatively sloped demand curve, and entry shifts the demand curve to the point at
which profits are zero. Like a perfectly competitive firm, a firm with a local monopoly
earns zero economic profit.
Efficiency Trade-Offs
There are sorne efficiency trade-offs associa!ed with entry and competition. On the
one hand, entry decreases output per firm, so individual firms move upward along
their average cost curves. Scale economies are lost because output is divided among a
larger number of firms, each of which produces at a higher average cost. On the other
r
Part 1 Market Forces in the Development of Cities
96
FIGURE 5-4 Trade-Offs with Entry: Production Cost versus Travel Cost
$
Average total
cost
Average
production cost
Average travel
cost
qe
q,
qm
qc
Quantity of CDs per store
Entry increases the number of firrns and decreases output per firrn. As output falls, average
production cost increases and average travel cost decreases. In this case, entry goes "too far" in the
sense that average total cost is higher than the feasible minimum (e,).
hand, the increase in the number of stores decreases travel distances for consumers,
decreasing travel costs.
Figure 5-4 shows the trade-offs between production and travel costs. -=----The --------average
total cost of CDs is defined as the sum of average production cost (from Figure 5-3)
_anaver ge travel cost. Average travel cost is the.trvercosncurred by the typical
CD consumer. As the output of the music store increases, the market area increases
and the typical consumer travels a longer distance to buy CDs. Therefore, the average
travel cost increases with the output of the store. Average total cost (production cost
plus travel cost) reaches its minimum point at q, far below the output that minimizes
average production cost (qc).
Entry decreases output per firm, gener ting bad news and good news. The
bad news is that average production cost inct ases as firms move up the average
production-cost curv. The good news is that average travel costs decrease as firms
move down the average travel-cost curve. Starting from qm in Figure 5-4, the good
news dominates the bad news until output per firm reaches qt. From that point on,
further entry increases production costs by more than it decreases travel costs, so
average total cost increases. In equilibrium, the output per firm drops to qe, meaning
that the entry process goes "too far" in the sense that average total cost is higher than
the feasible minimum.
97
Figure 5-4 shows only one possible equilibrium outcome. For different sets of
cost and demand curves, the equilibrium output per firm might be greater than qt
or equal to qt. In other words, unless one knows something about the demand and
cost curves, it's impossible to predict exactly where on the average total-cost curve
the firm ends up. However, we do know that (1) entry decreases output per firm and
(2) in equilibrium, the firm does not produce at the minimum point of its average
production-cost curve, but at sorne point to the left of the minimum point.
11
1
1
1
1
1
1
1
1
1
1
1
1
30
20
20
Tammy's territory
Bob's territory
Dick's territory
Each store's market area is the area over which its n price is less than the net prices of other stores.
Each store has a circular territory with a 10-mile radius:
30
98
Bob and charge the same price for CDs ($6). Compared to Bob, Tammy has a lower
net price in the westem third of the region, and Dick has a lower net price in the
eastem third of the region. The three music stores split the region into three equal
market areas, so every store has a circular market area with a 10-mile radius.
The market arrangement in Figure 5-5 has two implications for CD consumers.
First, if the spaces between the circular market areas are ignored, the maximum net
price is $11, the sum of the price charged by music stores ($6) and the maximum
travel cost ($5). Second, every household patronizes the music store closest to its
home. All stores charge the same price for the same product, so a household will
buy from the store with the lowest travel costs.
The market area ofthe firm (in square miles) is the territory required for the firm to
sell its target quantity (q). In other words, the market area is
q
-M=d e
The denominator is the demand for CDs per square mile (also known as demand
density), equal to the quantity demanded per person (d) times the number of people
per square mile (e). liiTable 5-2, the demand density is 200 CDs, equal to 4 CDs
per person times 50 people per square mile. The market area equals the target output
(q) divided by the demand density. For example, if q = 1, 000 CDs, the firm needs
5 square miles (each of which has a demand of 200 CDs) to sell its target output.
Symbol
Numerical Example
d
e
4
50
200
1,000
5.0
d-e
q
M= qj(d- e)
99
example, if per capita demand doubles to 8 CDs per person, the market area shrinks
from 5 miles (1,000/200) to 2.5 miles (1,000/400).
. How does an increase in population density a:ffect the market area? Like an
increase in per capita demand, an increase in density increases the volume of CDs
sold per square mile, so each firm needs a smaller territory to exhaust its scale
economies. For example, if population density increases to 125 people per square
mile, the market area shrinks from 5 miles (1,000/200) to 2 miles (1,000/500).
100
area will increase from 5 square miles (1,000/200) to 7 square miles (1,400/200).
This is the output effect of a decrease in travel costs.
If consumers obey the law of demand, a decrease in travelcosts has an ambiguous
effect on the size of the market area. A decrease in travel costs decreases the net
price of CDs, increasing demand per capita (d). The increase in demand tends to
reduce the firm's market area because the firm needs a smaller territory to sell a given
quantity of output. The price effect at least partly offsets the increase in market area
caused by the increase in output per firm, so we cannot predict whether the market
area will increase or decrease.
Recalling the expression for market areas (M= q j(d e)), we expect a common
output per firm in the two cities (q is the same), so the city with the larger demand per
square rnile (d e) will have the smaller market area for music stores. This is sensible
because a larger demand per square mile means a firm needs a smaller territory to
exploit its scale economies.
Which city will have a larger demand per square mile? Since a CD is a normal
good, the wealthy city will have a higher Pt?r capita deiDand for CDs. Since land is
a normal good, population density will be lower in the wealthy city. For example,
the wealthy may live iri single-farnily homes and the poor may live in apartment
buildings. Because the wealthy city has a higher per capita demand and lower popu-
101
TABLE 5-3 Income and Market Area with Different Income Elasticities of Demand
forLand
Wealthy City
Highincome
Elasticity.
Lowincome
Elasticity
l.O
$1,200
4.4
50
200
1,000
5
176
1,000
5.68
0.25
$1,200
4.4
47.5
209
1,000
4.78
Poor City
Income elasticity for land
Per capita income
Per capita demand for CDs (d)
Population density (e)
Demand density (d e)
Output per store (q)
Market area in square miles (qf d e)
l.O
$1,000
40
lation density, we can't predict whether it will have larger or smaller market areas for
music stores. The relationship between income and market area is ambiguous because income affects percapita demand and population density in opposite directions.
Under what circumstances. will the wealthy community have a larger market
area? Suppose that the income elasticity of demand for land is large relative to the
income elasticity of demand for CDs. If so, the density effect (increased income
decreases population density) will domnate the demand effect (increased income
increases per capita demand), and the wealthy city will have a lower demand density and a larger market area. Music stores will need a larger market area because:
ialthough each household consumes more CDs, there are many fewer households per 1
\_square mile.
Table 5-3 shows a simple example of how differences in income affect market
areas. The income elasticity of demand for CDs is 0.50. In the wealthy city, where
per capita income is 20 percent higher, the per capita demand for CDs is 1O percent
higher. In the m.iddle column, the income elasticity of demand for land is assumed
to be 1.0. Therefore, the wealthy city has a 20 percent higher demand for 1and and
a 20 percent lower population density (40 instead of 50). Given these assumptions,
the wealthy city has a lower demand density, so music stores need a larger market
area to exploit their scale economies (5.68 square miles instead of 5 square miles).
This result depends, of course, on numerical assumptions. If the demand for
land were relatively income-inelastic, the difference in population density would
be relatively srnall. The demand effect would domnate the density effect, so the.
wealthy city would have a smaller market area. In the third column ofTable 5-3, the
income elasticity of demand for land is assumed to be 0.25: population density in
the wealthy city is 47.5 (a 5 percent difference) instead of 40; demand density is 209
instead of 176; and the rnarket area is 4.78 instead of 5.68. In general, if the income
elasticity for land is srnall relative to the income elasticity for CDs, the wealthier
city will have a smaller rnarket area.
'
Ij
102
United States were purchased in small stores. People bought food in small "mom
and pop" grocery stores and purchased other goods in small general stores. The
traditional merchant was involved in every transaction, playing two roles:
Market areas vary from indilstry to industry, refiecting diffrences in travel costs,
per capita demand, and scale economies. If scale economies are large relative to per
Chapter 5
103
capita demand, the industry will have a small number of firms, each of which has
a large market area. In contrast, if scale economies are small relative to per capita
demand, there will be a large number of firms with small market areas.
Consider the market areas of pizza parlors and Tibetan restaurants. Suppose that
both activities are subject to the same sort of scale economies, so the optimum output
for both types of restaurants is 200 meals per day. If the total demand for pizza is
10,000 meals per day, the region will have 50 pizza parlors (10,000/200). Ifthe total
demand for Tibetan food is 200 meals per day, a single Tibetan restaurant will serve
the entire region. The market area is not determined by scale economies per se, but
by scale economies relative to per capita demand.
CENTRALPLACETHEORY
Central place theory, which was developed by Christaller (translated in 1966) and
refined by Losch (translated in 1954), is used to predict the number, size, and scope of
cities in a region. The theory is based on a simple extension of market-area analysis.
Market areas vary from industry to industry, depending on scale economies and
per capita demand, so every industry has a different location pattem. Central place
theory shows how the location patterns of different industries are merged to form a
regional system of cities. The theory answers two questions about cities in a regional
economy:
l. How many cities will develop?
2. Why are sorne cities larger than others?
A Simple C tral Place Model
Consider a region with three consumer products: CDs, pizzas, and jewelry. The
region has the following characteristics:
l. Population density. The initial distribution of population is uniform. The total
population of the region is 80,000.
2. No shopping externalities. As discussed in Chapter 2, shopping extemalities
normally occur with complementary goods (one-stop shopping) and imperfect
substitutes (comparison shopping). The simple central place model assumes that
there are no shopping extemalities.
3. Ubiquitous inputs. All inputs are available at alllocations at the same prices.
4. Uniform demand. For each product, per capita demand is the same throughout
the region.
5. Number of stores. The three goods have different percapita demands and scale
economies:
a. Jewelry. Scale econornies are large relative to per capita demand. Every
jewelry store requires a population of 80,000, so a single jeweler will serve
the entire region.
.. b. Compact discs. Scale_ economies are moderate relative to per capita demand. Every riusic store requires a population of 20,000, so there
be
four music stores in the region.
will
104
Part 1
e. Pizza. Scale econornies are small relative to per capita demand. Every pizza
parlar requires a population of 5,000, so there will be 16 pizza parlors in the
region.
The central place model is a model of market-oriented firms, defined earlier in
the book as firms that base their location decisions exclusively on access to their
consumers. Because all inputs are ubiquitous, the firms ignore input costs in thir
location decisions.
The single jeweler willlocate at the center of the region. Because production
costs are the same at alllocations (allinputs are ubiquitous), the jeweler will minimize
its total costs by rninimizing its travel costs. According to the principie of median
location (discussed earlier in the book), travel costs are minimized at the median
location. Because population density is uniform, the median location is the center
of the region and the jeweler willlocate there.
A city will develop around the jewelry store. Jewelry workers willlocate near
the store to economize on commuting costs. The populatio_n density near the jeweler
will increase, generating a city (a place of relatively high density) at the center of
the region. In Figure 5-6, a city develops at point L.
The music stores will carve up the region into market areas, causing the development of additional cities. If the region's population density were uniform, music
firms would carve out four equal market areas. However, because there is a city
surrounding the jeweler in the center of the region,-there will be enough demand to
support more than one music store in city L.If city Lalong with the surrounding area
1
1
1
1
1
1
1
1
1
1
1
1
1
1
r-------
0
----
1
1
1
1
1
1
1
1
0
L
---8------0
1
1
1
1
1
1
1
CityL
M Cities
S Cities
1 Jeweler
2 Music stores
4 Pizza parlors
1 Music store
2 Pizza parlors
1 Pizza parlor
There are 11 cities.i.ri the region: one large city (L), two mediui:n sized cities (M), and eight
small cities (S). The larger the city, the greater the variety of goods sold.
Chapter 5
105
has enough people to support two music stores, the two other music stores will split
the rest of the region into two market areas. In Figure 5-6, two more cities develop
at the locations marked with an M.
The pizza parlors will also carve up the region into market areas, causing the
development of more cities. Because the population densityis higher in the cities
that develop around the jewelry store and the music stores, there will be more than
one pizza parlar in L and the two M cities. Suppose that L will support four pizza
parlors, and each of the M cities will support two pizza parlors. If so, a total of eight
pizza parlors willlocate in cities L and M. The remaining eight pizza parlors will
divide the rest of the region into eight market areas, causing the development of
eight additional cities (the places marked with an S in Figure 5-6).
The rectangular region has a total of 11 cities. The large city at the center of the
region sellsjewelry, CDs, and pizza. City L has a population of20,000, meaning that
it is large enough to support four pizza parlors (5,000 people per pizza parlar). The
city sells CDs to consumers from the four surrounding S cities, so the total number of
CD consumers is 40,000 (20,000 from L and 5,000 each from four S cities), enough
to support two music stores. The t\YO medium-sized cities sell CDs and pizza. Each
of the M cities has a population of 10,000, meaning that each city is large enough
to support two pizza parlors. Each city sells CDs to consumers from two nearby S
cities, so the total number of CD consumers in each M city is 20,000 (10,000 from
M and 5,000 each from two S cities), enough to support one music store per M city.
The eight small cities sell only pizza. Each ofthe S cities has a population of 5,000,
meaning that each city can support one pizza parlar.
Figure 5-7 shows the size distribution of cities in the region. The vertical axis
measures city size (population), and the horizontal axis measures the rank of the city.
The largest cty (L) has a population of 20,000; the second and third largest cities
t
(M cities) have populations of 10,000; and the fourth through the eleventh largest
t
cities have populations of 5,000.
l
The simple central place model generates a hierarchical system of cities. There
are three distinct types of cities: L (high arder), M (medium arder), and S (low arder).
The larger the city, the 'greater the varie ty of goods sold. Each city imports goods
l
from higher-order cities and exports goods to lower-order cities. Cities of the same
l
arder do not interact. For example, an M city imports jewelry from L and exports
l
CDs to S cities, but does not interaCt y;ith the other M city. Similarly, an S city
J
imports jewelry from L and CDs from either L oran M city, but does not trade with
other
S cities. The system of cities is hierarchical in the sense that there are distinct t
l
types of cities and distinct patterns of trade dominance.
The simple central place model provides sorne important insights into how the
l
market-area decisions of firms combine to generate the urban hierarchy. Here are
l
sorne lessons from central place theory.
l
t
l
J
l
l. Diversity and scale economies. The region's cities differ in size and scope.
This diversity occurs becase the tbree consumer products have different scale
economies relative to percapita demand, so they have di:fferent market areas. To
explain the importance of differences in relative scale economies, suppose that
r
'
106
FIGURE 5-7 Size Distribution of Cities with Simple Central Place Model
20,000
,.....,
::
o
"5
0 ..
.e 10,000
i:i
5,000
10
11
Rank
The simple central place model generates one large city (L) with a population of 20,000, two
medium-sized cities (M), each of which has a population of 10,000, and eight small cities (S),
each of which has a population of 5,000.
the three goods have the same scale econornies relative to per capita demand,
so the region has 16 jewelers, 16 music stores, and 16 pizza parlors. The market
areas ofthe three goods would coincide, so the region would have 16 identical
cities, each of which would provide all three goods. In other words, if there are
no differences in scale econornies relative to per capita demand, the region's
cities will be identical.
2. Large means few. The region has a sml number of large cities and a large
number of small cities. Why isn't there a large number oflarge cities anda small
number of small cities? A city is relatively large if it provides more goods than
a smaller city. The extra goods provided by a large city are those goods that are
subject to relatively large scale econornies. Since there are few stores selling
the goods subject to relatively large scale econornies, few cities can be large.
In the simple central place rnodel, the L city is larger than an M city because L
sells CDs, pizza, and jewelry. Since there is only one jewelry store in the region,
there is only mie city larger than the M cities.
.
3. Shopping paths. Consurners travel to bigger cities, not to srnaller cities or
-cities of the same size. For example, consumers from an M city travel to L to
,
.'
--
...
41(,
:1
:4
Il
l.
1
1
l
!
1
J
1
r
l
1
l
i
1
107
buy jewelry, but do not travel to the other M city oran S city to consume CDs or
pizza. Instead, they buy CDs and pizza in their own city. Similarly, consumers
in S cities travel to larger cities for jewelry and CDs, but do not shop in other S
cities.
108
109
Type of Place
._,
.
:
- --
'.
.-
Number of places
Average population
Average number of establishments per place
Average number offunctions per place
Average number of establishments per function
Town
Village
Harnlet
4
2,433
149
59.8
2.5
9
948
54.4
32.1
1.7
20
417
6.9
5.9
1.2
Source: Brian J. L. Berry and William Garrison, ''The Functional Bases of the Central Place Hierarchy," Economic
Geography 34, pp. 145-54, @1958, Clark University.
f
-
"l
._
:,f
l
l
(
{
i
t
j
t
f
'
'
l
-t
tt
l
J
::=
o
18,000,000
16,000,000
14,000,000
12,000,000
1=1
o
,:j
10,000,000
].
p..
8,000,000
6,000,000
4,000,000
2,000,000
lllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll!!llllll!!lll!!llllll!!l
10 13 16 19 22 25 28 31 34
37 40 43
46 49 52 55 58 61 64 67 70 73 76 79
Rank
" )
82 85 88
"'\
-----....
91 94
'
'
';
'.
111
40
35
30
10
100
200
300
396
Geographers and economists have estimated the relationship between city size
and rank. The relationship is approximated by the rank-size rule
Rank Size = Constant
This rule suggests that the product of rank and size is the same in all cities in a
region. For example, if the largest metropolitan area has a population of 16 million,
the rank-size rule suggests that the second largest city has a population of 8 million,
the third largest city has a population of 5.33 million, and the sixteenth largest city
has a population of 1 million. Figure 5-9 shows the rank-size relationship for U.S.
urbanized areas in 1990. The rank of the urbanized area is on the horizontal axis and
the product of rank and size (in millions) is on the vertical axis. The largest deviation
from the rank-size rule occurs in cities ranked 10th through 80th; these cities are
much larger than would be predicted by the rank-size rule. In general, it does not
appear that the size distribution of U.S. urbanized areas satisfies the rank-size rule.
Empirical evidence concerning the rank-size rule comes from statistical studies
of city-size distributions. The hypothesized relationship between rank and size is
Size = Constant/Ra.nk!l
Part 1
112
The rank-size rule is a good approximation of the relationship between rank and
size if B (the exponent of rank) is close to 1.0. Rosen and Resnick (1980) estimated
the rank-size relationship for 44 countries. They found substantial variation in B
across countries, with an average value of 0.88. This result suggests that population
is more evenly distributed between cities than would be predicted by therank-size
rule. Rosen and Resnick conelude that the rank-size rule is only a first approximation
to the relationship between rank and size.
National Population
City
Tokyo, Japan
Mexico City, Mexico
New York, USA
Sao Paulo, Brazil
Buenos Aires, Argentina
Santiago, Chile
Montevideo, Uruguay
Population
Share of Country's
Population
19,037,361
16,465,487
15,627,553
15,538,682
10,759,291
4,227,049
1,157,450
15.76%
20.97
6.53.
11.46
35.47
34.87
39.36
e,
'
[J
J
:i
l
I
J
1
I
1
I
1
1
l
l
I
l
J
113
1670 (about 10 percent ofEngland's population). During the late sixteenth century,
intemational trade involving England grew rapidly, a result of military victories
against the Spanish, improvements in shipping technology, the discovery of huge
markets in Asia and the Americas, govemment policies promoting trade, and massive
emigration to the New World. This rapid growth in trade was a big factor in the rapid
growth ofLondon. In the case ofBuenos Aires, between 1887 and 1914 exports from
Argentina rose by 400 percent, and the population of Buenos Aires-the principal
trading city-increased by 1.1 million (265 percent).
The Effect of Trade Restrictions
The more complex theories of large prirnary cities require a more detailed analytical
framework (Krugman, 1996). Considera nation that has two cities-P and S-and
a fixed workforce to divide between the two cities. The firms hire workers and sell
their output to workers and consurners. Let's take the perspective of a worker who
must pick a city to live in, either Por S. What are the trade-offs for the worker, given
that P is a large primary city, that_ is, most of the workers are in P?
l. Bad news:higher commuting costs. A larger city has longer distances between
workers' bornes and the production site, and this is obviously bad news for the
worker. The long commuting distance tends to pull the worker toward city S.
2. Good news: higher nominal wages. If most workers are in city P, so are rnost
consumers, so firms in city P have lower transportation costs (for shipping their
output to their customers) than their counterparts in city S. These lower costs
could lead to higher pro:fits in city P, but competition for workers between firms
will generate higher wages instead. The higher wage will pull the worker toward
city P. \
3. Good news: lower net price of goods.- Most of the goods consumed in city
P are produced in that city, while most of the goods consumed in city S are
produced in P and shipped toS. As a result, the net price of goods (price plus
transport costs) will be lower in city P, and this tends to pull workers toward
city P.
'
The worker will choose city P-the primary city-if the good news dominates the
bad news, that is, if the benefits of higher wages and lower net prices domnate the
highei commuting costs. This is certainly possible, suggesting that the large prirnary
city is a feasible and sustainable outcorne.
What happens to the sustainability of this primary-city outcorne if we introduce
intemational trade? From the worker's perspective, the presence of exports and
imports will make both bits of good news less powerful. If :firms export sorne of
their output, the transportation costs associated with serving dornestic consumers
will diminish in importance. Although :firms in the larger city (P) will still have
lower transportation costs than firms in the small city and thus pay higher wages,
the wage premium will be srnaller. In the extreme case, a firm that exports all its
output will_ pay the same wages in the two cities. On the import side, if consumers
buy import goods instead of sorne domestic goods, the gap between the net prices
114
in the two cities will become less important. In the extreme case, a consumer who
buys all imports won't care if net prices are lower in the primary city.
In this model, the introduction of trade makes the primary-city outcome less
likely. Trade weakens the forces pulling the worker toward the primary city but
doesn't affect the force pulling the worker toward the smaller city (lower commuting
costs), so the balance may be tipped in _favor of the smaller city. Krugman (1996)
uses this logic to suggest that nations with liberal trade policies are less likely to
have large primary cities. Therefore, one possible explanation for a giant city in a
developing nation is that the nation has restrictive trade policies.
As is often the case in econornics, there are confiicting theories concerning the
effects of trade on urban concentration. The simple theory of scale econornies in
trade activities suggests that international trade contributes to the development of
urban giants, while the Krugman theory predicts that trade restrictions encourage
the development of such cities. Andes and Glaeser (1995) conclude that although
there is a negative relationship between trade and urban concentration, it may be that
urban concentration is causing low levels of trade, rather than low levels of trade
causing urban concentration. The ultima,te resolution of this debate will come from
additional empirical work.
'
115
to extract resources. At the same time, there was poltica! unrest in Rome and the
surrounding areas. The traditional aristocracy responded by distributing free grain
to the residents of Rome, and perhaps just as important, staging the infamous-and
very expensive-Roman circuses (50 per year at their height). These redistribution
policies encouraged people to move to Rome. When Julius Caesar restored stability
and reduced the grain distribution around 45 B.C., the growth of Rome slowed.
SUMMARY
l. Market-oriented firms are involved in monopolistic competition or spatial com-
2.
3.
4.
5.
6.
petition. Each has a local monopoly, but its monopoly power is limited by entry
and competition from other firms. In equilibrium, every firm produces the quantity such that marginal revenue equals marginal cost and each firm makes zero
economic profit.
The market area of an industry depends on the quantity produced per firm, per
capita demand, and population density.
Central place theory, which is based on market-area analysis, predicts that the
regional system of cities will be hierarchical, with a small number of large cities
and a large number of small cities. The set of goods sold in smaller cities will
be a subset of the goods sold in larger cities.
Ifthe assumptions ofthe simple central place theory are relaxed to allow imperfect substitutes (comparison shopping) and complements (one-stop shopping),
the equilibrium number of cities may decrease, but the urban hierarchy will not
be disrupted.
Central place theory is not applicable to resource-oriented firms and firms oriented to\.vard local inputs. The introduction of such firms into the central place
model may disrupt the urban hierarchy.
The rank-size rule suggests that the product of city rank and city size (population)
is constant. A study of the rank-size relationship for 44 countries suggests that
population is more evenly distributed between cities than would be predicted
by the rank-size rule.
1
_f
l. The discussion of the market area of music stores assumed that store owners
Bob, Tammy, and Dick had the same production costs. As a result, households
patronized the store nearest their homes. Suppose that Bob discovers a new way
of marketing CDs that cuts his production cost (and store price) in half, from
$6 to $3. Tammy and Dick continue to sell CDs for $6.
a. How does the decrease in production cost a:ffect Bob's market area?
b. What is the net price at the border between Bob's and Tammy's market area?
c. Will each household still patronize the firm closest to its residence?
2. Consider the market area of food stores in a region described by the following
assumptions: (i) The per capita demand for food is 30 units; (ii) Population
f
1
1
J
1
{
116
Part 1
3.
4.
5.
6.
7.
8.
density is 40 people per square mile; (iii) The land area of the region is 100
square miles; (iv) The output of the typical food store is 6,000 units.
a. How many food stores will there be in the region?
b. How large is the market area of the typical food store?
Consider the market areas of hardware stores in two independent regions, Low
and High. The average production-cost (APC) curves of the two regions reach
their mnimum points at the same level of output (1,000 units), but because
Low has lower input costs, the minirnum point of its APC curve has a lower
average cost (its average cost of 1,000 units is $1 instead of $2). Will the market
area in Low be larger, smaller, or the same as the market area in High? lf you
don't have enough information to answer the question, indicate what additional
information you need and how you would use it.
Consider a regional economy with two cities. The two cities have the same
population, but the average household income in city H exceeds that in city L.
The residents of H and L consume Y. The characteristics of Y are as follows:
The average production-cost (APC) curve is U-shaped; The APC curve is the
same at alllocations; Y is a standardized product, with no clase substitutes or
complements; all households have the same tastes for Y; Y is inferior (negative
income elasticity).
a. How will the size of the average market area of Y in city H compare to the
size of the average market area in city L?
b. How would your response to (a) di:ffer if the APC curve were a horizontal
line?
Suppose that you intend to purchase the franchise rights for a pizza parlar. The
franchiser has divided your region into two areas of equal size: H is a highincome area, and Lis a low-income area Suppose that the income elasticity of
demand for pizza is zero: the consumption of pizza is independent of income.
The income elasticity of demand for land is 1.0. Your objective is to maximize
the quantity of pizzas sold.
a. Which of the two franchises will you choose?
b. How would your response to (a) change if the income elasticity of demand
for pizza is 1.5?
In the example based on the data in Table. 5-3, the implicit assumption is that
.travel costs were the sarn.e in the two cities. Suppose that the opportunity cost
of travel is higher in the wealthy city. How might this change the market areas
in the wealthy city?
Consider the city of Vaudeville, where initially all video consumers travel to
video outlets to rent videos. Suppose that when the video outlets offer home
delivery, all video consumers switch to home delivery.
a. Under what circumstances is the switch to home delivery rational?
b. In what type of cities would you expect home delivery to be ef:ficient?
In 1930, most rural children walked to school. By 1980, most of them rode
school buses. Would you expect the average "market area" of rural schools to
increase or dcrease as rsult of the school bus?
117
9. The introduction of parcel post decreased transportation costs and increased the
market area ofthe typical firm. Can you think of any good for which a decrease
in transportation costs is likely to decrease the market area?
10. In the city ofMetro, retail firms locate eitherin the central area ofthe city orin one
ofthe suburban subcenters. There are two retail goods sold in the city, A and B.
The per capita demands for the two goods are equal. The scale economies associated with the production and sale of A are large relative to the scale economies
associated with B. Where in the city will the firms selling A and B locate?
11. Considera city with a uniform distribution of population where every household
consumes the same number of video rentals. The city is two miles long and two
miles wide. The mayor recentiy stated her policy for the location of video r ntal
outlets: "A video outlet reaches the minimum point of its average productioncost curve atan output of 1,000 units. Since the total demand for video rentals is
4,000 units, we should have four video outlets. Since the distribution of population is uniform, the video outlets should be distributed uniformly throughout the
city, with every outlet at the center of a one-square-milemarket area." Comment
on the mayor's policy. Will_it lead to an ef:ficient distribution of video outlets?
12. In the city of Zone, the actual number of grocery stores is less than the number
that would occur in the absence of zoning. A recent survey of grocery prices
suggests that Zone consumers pay less for groceries than people in a similar
city without zoning. For example, the prices of Spam and Velveeta are S percent
lower in Zone.
a. Is there any reason to doubt the validity of the survey?
b. Suppose that the survey is valid. Does the zoning policy make the residents
of Zone better or worse off?
13. A deve oper has requested a permit to build a drugstore in a rapidly growing
part of your city. Your mission is to figure out whether the new drugstore is
appropriate. What information do you need, and how would you use it?
14. Explain why poor areas of cities typically are served by small grocery stores,
not by large grocery.:.$tore chains.
1. Considera regi6n in which all firms are market-oriented (all employment is in
retail outlets). There are no shopping extemalities, and all retail activities are
subject to the same degree ofintemal scale economies. In the last 10 years, the
region's rank-size curve (with city rank on the horizontal axis and city size on
fue vertical axis) has become fl.atter.
a. What changes in market areas are lurking behind the fl.attening of the ranksize curve?
b. Provide an explanation for the changes in the market areas lurking behind
the fl.attening of the r_ank-size curve.
16. Mr. Wizard, a regional planner, recentiy made the following statement: "If my
assumptions are correct, all cities in this region will eventually be identical.
They will be the same size and will sell the same set of goods."
a. Assuming that Mr. Wizard's reasoning is correct, what are his assumptions?
b. Are Mr. Wizard's assumptions realistic?
118
Part 1
17. Sorne people claim that state capitals (e.g., Sacramento, California; Salem,
Oregon; Olympia, Washington) are boring cities. Specifically, it is claimed that
these cities have fewer goods than one finds in other cities of equal size. After
checking a map to see where each of these capital cities is located, use central
place theory to explain why they might be considered boring.
REFERENCES
l. Andes, Alberto F., and Edward L. Glaeser. "Trade and Circuses: Explaining
Urban Giants." Quarterly Joumal of Economics (1995), pp. 195-227.
2. Berry, Brian J.L.,and William L. Garrison. "The Functional Bases of the Central
Place Hierarchy." Economic Geography 34 (1958), pp. 145-54. An empirical
analysis of central place theory.
3. Christaller, Walter. Central Places in Southem Germany, trans. C.W. Baskin.
Englewood, NJ: Prentice Hall, 1966.
4. Krugman, Paul. "Urban Concentration: The Role of Increasing Returns and
Transport Costs." Intemational Regional Science Review 19 (1996), pp. 5-30.
5. Losch, August. The Economics of Location. New Haven, CT: Yale University
Press, 1954.
6. Rosen, Kenneth T., and Mitchell Resnick. "The Size Distribution of Cities: An
Examination of the Pareto Law and Primacy." Joumal of Urban Economics
8 (1980), pp. 165-86.
CHAPTER 6
-ANONYMOUS
In
tbis chapter, we'lllook at the causes of urban economic growth as well as the
consequences of growth. Our discussion will explain why sorne cities grow at a
healthy rate, while others stagnate. As we explore the causes of urban economic
growth, we'11 draw on sorne of the insights concerning the location decisions of
firms from Chapter 3 and also look at sorne of the techniques economists use to
predict future urban economic growth.
What is urban economic growth? We can measure the size of an urban economy
in several ways, the most natural of which is the total value of output (goods and
services) produced in the urban area. The total value of output is computed by
multiplying the quantity of each product by its price, and then sumrning up the
monetary values for each product. For example, if a city produces lO ton of steel
at a price of $300 peton, 20 haircuts at $10 each, and 10 pizzas at $8 each, the
total value of output would be $3,280..Altematively, we can measure the size ofthe
economy as the total income eamed iri the metropolitan area. Because the revenue
from selling products is paid out to the people who supply the inputs to production
(workers, landowners, entrepreneurs, and the owners of capital) the total value of
production equals total income. fu our example, the total amount paid to the people
who produce steel, haircuts, and pizza would be $3,280. Economic growth can be
defined as either an increase in the total value of output oran increase in total income.
We can use a production function to show the connection between inputs to
the production process and total output. The production function for steel could be
written as
,..
4
f
:e
.;(
, ..
!(
q = f(k, l, r, t),
where k is the amount of capital used in steel production (for exmple, machines,
119
i _
120
Part 1
= F(K, L, R, T),
where the capitalletters indicate that we are considering the entire urban economy.
Urban economic growth results from technological progress or increases in the
quantities of inputs used by firrus in the metropolitan area.
The production-function approach is a useful starting point for a discussion
of urban economic growth because it tells us that a city's economy can't grow
unless firms in the city use more inputs or adopt better technology. However, this
approach doesn't explain why urban growth occurs. The question is, What causes
firrus in the metropolitan area to ernploy more inputs or adopt new technology? In
a market-based econorny, profit-seeking firms make the input and output decisions
that determine the size of the urban econorny. Entrepreneurs hire workers, buy land
and raw materials and borrow rnoney to build production facilities-al! with the idea
of producing products to sell. To explore the reasons for urban economic growth,
we must examine the decisions of profit-seeking firrus.
The most natural way to explore firms' input and output decisions is with a rnodel
of the urban labor market. The demand for labor comes from firms in the city; the
supply oflabor comes from households living in the city.The model of the urban labor
market allows us to explore both the causes of urban growth and its consequences.
This approach focuses our attention on total employment as one measure of the size
of the urban economy. We can use the rnodel to show the effects of various changes
in the economic environment on total employment and wages in the city.
121
Round 1
Round2
Round3
Total
$1,000
600
400
$600
360
240
$360
216
144
$2,500
1,500
1,000
$1,000 increase in steel sales. In round 1, the increase in steel sales is paid out to
steelworkers, the owners of land under the steel mili, and the suppliers of capital
to the steel industry (investors) in the forrn of wages, land rent, and interest. If all
the input suppliers live in the city, local income increases by $1,000. Suppose that
60 percent of any income earned by input suppliers is spent on local goods such as
food, housing, and entertainment. In this case, local consumption increases by $600,
while the other $400 is spent on goods imported to the city.
The first-round increase in local consumption is the second-round increase in
local income. The firms producing housing and entertainment sell $600 worth of
goods and services, and pay this amount to their input suppliers-workers, investors,
and landowners. If these income earners spend 60 percent of the extra income on
local goods, the second-round increase in local consumption is $360. In round 3,
the $360 in local spending becomes income for other local workers, and they spend
$216 (equal to 60" percent of $360) on local goods and services. This spending and
respending of income continues forever, but every spending round is smaller than
the previous one because 40 percent of income is spent on imports.
The i.:J.crease in the city's total income exceeds the original increase in export
income. The marginal propensity to consume locally (m) is the fraction of income
spent locally (0.60 in Table 6--1). If the original change in exp01t sales is !::.X, the
increase in total income (!::.Y) is the sum of the income changes from a series of
spending rounds:
!::.Y = !::.X +m!::. X + m2 !::.X
+ m3 !::.X + .
The change in total income is the sum of an infinite series, and can be written as
1
_ _
= $2,500
1 0 60
The income multiplier is defined as the change in total income per unit change in
export sales:
!::.Y
1
-=-!::.X
1-m
122
For example if a city has a total of 5,000 jobs, including 2,000 exportjobs and 3,000
localjobs, the employment multiplieris 2.50 (equal to 5,000 divided by 2,000). The
idea behindthis calculationis that 2,000 exportjobs support a total employment level
of 5,000 jobs, ora total of 2.5 total jobs per export job. This method of computing
the employment multiplier is imprecise because the ratio oftotaljobs to exportjobs
is likely to change over time as a city grows or shrinks.
123
Labor supply
1,000
1
1
"'
1
1
1
1
1
1
1
1
1
1
1
1
Labor demand
25
. 75
50
Number of laborers in city (thousands)
The labor market is in equilibrium if supply equals demand. The equilibrium wage is $1,000 per
month, and the equilibrium number of laborers is 50,000.
The secnd reason for a negatively sloped demand curve is the output effect. As
the city's wage increases, production costs increase, increasing the prices charged by
the city's firm.s. As the prices of exports increase, the quantity of exports demanded
decreases, so exporters need fewer workers. In the local sector, as the prices of local
goods increase, city residents substitute imports for the relatively expensive local
goods, so local firms need fewer laborers. This is the output effect: an increase in the
city's wages increases prices and decreases output, decreasing the quantity of labor
demanded. The demand curve is negatively sloped because an increase in wages
generates both a substitution effect andan output effect.
What determines the slope of the demand curve? The deniand curve will be
relatively fiat-with a relatively large change in quantity demanded from a change in
the wage-ifeitherthe substitution effect orthe output effectis large.The substitution
effect will be large if firms can easily substitute nonlabor inputs for labor. The
output effect will be large under the following circumstances. First, if labor costs
are responsible for a relatively large share of total costs, a small increase in the wage
will cause a relatively large increase in the output price and thus a relatively large
decrease in total production. Second, if the price elasticity of demand for the output
is relatively large, a given increase in the output price will cause a relatively large
decrease in total production.
.
.
124
What causes the demand curve to shift to the right or the left? The following
factors determine the position of the curve:
l. Demand for exports. An increase in the demand forthe city's exports increases
export production and shifts the demand curve to the right: at every wage, more
workers will be demanded.
2. Labor productivity. An increase in labor productivity decreases production
costs, allowing export firms to cut prices, increase output, and hire more workers.
Similarly, when local workers become more productive, local producers can
underprice imports, sell more output, and hire more workers.
3. Business taxes. An increase in business taxes (without a corresponding change
in public services) increases production costs, increasing prices and decreasing
output, so the demand curve shifts to the left. The demand for labor decreases
as export firms lose customers to exporters in other cities and local firms lose
customers to importers.
4. Industrial public services. An increase in the quality of industrial public services (without a corresponding increase in taxes) decreases production costs and
increases output, so the demand curve shifts to the right. The demand for labor
increases as export :firms gain customers from exporters in other cities and local
firms gain customers from importers.
5. Land-use policies. Industrial firms require production sites that (a) are accessible to the intracity and intercity transportation networks and (b) have a full set
of public services (water, sewerage, electricity). By coordinating its land-use
and infrastructure policies to ensure an adequate supply of industrialland, a city
can accommodate existing :firms that want to expand their operations and new
firms that want to locate in the city.
Figure 6-2 shows the multiplier effect of an increase in export sales. Suppose
that an increase in the demand for exports increases the demand for export workers
by 10,000. The city's demand curve wil! shift to the right from D 1 to D2 : ata wage
of $1,000 per month, an additionallO,OOO export workers will be demanded. This
is the direct effect of an increase in export demand. If the employment multiplier is
2.5, every exportjob supports 1.5localjobs, sothe demand curve shifts to the right
by an additional15,000 workers (from D 2 to D3 ). This is the multiplier effect of an
increase in export demand. Total labor demand increases by 25,000 (2.5 times the
increase in the demand for export laborers).
The Labor-Supply Curve
Consider next the supply side of the urban labor market. The supply curve is positively sloped because of the migration effect: An increase in the wage increases
the relative attractiveness of the city, causing the migration of workers from other
cities in the region. One of the assumptions underlying the supply curve is that the
number of work hours per Jaborer is :fixed; changes in the wage do not affect the
number of work hours per laborer. This assumption is consistent with empirical evidence concerning the elasticity ofwork hours withrespect to the wage; in aggregate,
125
.....-.
l
. ,
'.ifi..'.
Original demand
--
on
1,000
10
-------------- --
----
"
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
50
60
75
Number oflaborers in city (thousands)
If export employment increases by 10,000 jobs, the dernand curve shifts to the right in two steps. The
shift from Dto D2 is the direct effect of the increase in export employment. Ata tixed wage, total
employment rises from 50,000 to 60,000. The shift from D2 to Dref!ects the multiplier effect of the
increase in export employment. If the employment rnultiplier is 2.5, local employment increases by
1.5 times the increase in export employment, increasing labor demand by an additional!5,000 jobs.
: -
er
Se.
t,
j
te
,,
tls elasticity is close to zero. Another assumption underlying the supply curve is
that the labor-force participation rate is unaffected by changes in the wage. In other
words, an increase in the wage does not increase the fraction ofthe city's population
in the workforce. Given these two assumptions, an increase in the wage increases
the supply of labor because more workers move to the city, not because existing
workers work more hours or because more of the city's current residents join the
workforce.
Why is the supply curve positively sloped, and what determines the slope of
the curve? An increase in total employment (and population) increases the total
demand for most goods, causing increases in the prices of land, housing, and other
goods. A growing city must offer higher wage to compensare its workers for its
higher cost ofliving. The results ofHanillton and Schwab (1985) and Bartik (1991)
suggest that the elasticity ofhousing prices with respect to total employment is about
0.35; a 10 percent increase in total employment increases housing prices by about
3.5 percent. According to Bartik (1991), the elasticity of the cost of living with
. respect to city size (including al! price changes) is about 0.20; a 1O percent increase
in total employment increases the cost of living by about 2 percent. This means that
to keep real wages constant, the city's nominal wage must rise by about 2 percent
for every 1O percent increase in employment.
A number of empirical studies have estimated the responsiveness of wages to
totill emp1oynient. The results of:Studies by Treyz and Stevens (1985), Robak (1982),
126
and Bartik (1991) suggest that thewage for a given occupation increases at about the
sarne rateas the cost of living.In other words, the elasticity of the nominal wage with
respect to total employment is about 0.20; a 1O percent increase in total employment
increases the wage by 2 percent. Alternatively, the elasticity of labor supply with
respect to the wage is about 5.0: a 2 percent increase in the wage causes a 10 peri::ent
increase in labor supply. It's important to note that this is the labor-supply elasticity
for an individual city. The elasticity is relatively large because workers migrate to
cities that offer relatively high wages. The national supply elasticity is much lower
than the local supply elasticity because there is less migration between nations than
between cities.
What causes the supply curve to shift to the right or left? The position of the
supply curve is deterrnined by the following factors:
1. Environmental quality. An increase in the quality of the environment (better
air or water quality) increases the relative attractiveness of the city, causing
migration to the city. The supply curve shifts to the right; at every wage, more
workers are willing to work in the city.
2. Residential taxes. An increase in residential taxes (without a corresponding
change in public services) decreases the relative attractiveness of the city, causing
out-migration thatshifts the supply curve to the left.
3. Residential public services. An increase in the quality of residential public
services (without a corresponding increase in taxes) increases the relative attractiveness of the city, causing in-migration that shifts the supply curve to the
right. Eberts and Stone (1992) suggest that labor supply is responsive to changes
in local infrastructure.
\
Figure 6-3 shows the effects of an increase in export sales on the urban labor market.
The labor-demand curve shifts to the right by 25,000 workers, refiecting both the
direct effect and the multiplier effect of an increase in 10,000 export jobs. As the
population of the city increases, the prices of housing and land increase, requiring
an increase in the wage to compensate workers for the higher cost of living. In other
words, the city moves up its supply curve. The equilibrium wage rises from $1,000
per month tb $1,300, and the equilibrium number oflaborers increases from 50,000
to 66,000.
Figure 6-3 suggests that predicting the effects of an increase in export employment is tricky. The simple approach is to use the employment multiplier to predict
the change in total employment from a projected change in export employment. In
the numerical example, the predicted change in total employment from this method
would be 25,000 (2.5 times 10,000). This approach provides an estirnate ofthe horizontal shift of the demand curve, not the change in equilibrium employment. To
accurately predict the change in total employment, one must also know the slopes of
the supply and demand curves. The slope ofthe supply curve indicates how rapidly
the wage increases with city size, and the slope of the demand curve indicates how
rapidly the quantity of labor demanded decreases as the wage increases.
127
Labor supp1y
1,300
g, 1,000
---------
---------j
";',
1
1
1
1
1
1
1
1
1
D3
1
1
1
1
25
50
66 75
Nutober oflaborers in city (thousands)
An iocrease in exports shifts the demand curve to the right by 25,000. As the city grows,
the prices of housing and land increase, requiring higher wages to compensate workers
for the higher cost of living. The equilibrium wage increases from $1,000 to $1,300,
and the equilibrium oumber oflaborers increases by only 16,000.
Figure 6-4 shows the effects of arightward shift ofthe supply curve.Suppose that
the city i.ffiproves its residential public services. For example, the city could irnprove
its public-safety programs or alter its transportation system to decrease cornrnuting
costs. In Figiu:e 6-4, the labor-supply curve shifts to the right: At each wage, more
people are willing to work in the city. The shift of the supply curve increases equilibrium employrnent and decreases the equilibrium wage. Figure 6-4 is consistent with
the empirical evidence provided by Eberts and Stone (1992) concerning the effects
of irnprovements in iocal infrastructure on wages and total employment. Workers
accept lower wages because the city provides a superior rnix of local public goods.
1
J
128
FIGURE 6-4 Improvement in Public Services Shifts.the Supply Curve and Increases
Total Employment
$
Original supply
Supply with
improved
public services
1,000
&,
,.,"
=
875
Demaod
50 55
Nwnber of laborers in city (thousaods)
An improvement in local public goods iocreases the relativo attractiveness of the cicy, shifting the
supply curve to the right The equilibrium number of laborers increases to 55,000, aod the
equilibrium wage falls to $875 per month. Workers accept a!ower wage because the city provides a
more efficient set of local public services.
129
The lesson is that if there are localization econornies, an increase in labor demand
by one firm may cause other firms in the industry to hire more workers too. The
same logic applies to an increase in the quantity of labor demanded resulting from
a decrease in wages. Suppose a city's wages drop, allowing a firm in a particular
industry to underprice its competitors in other cities. As a result, the firm will hire
more workers and produce more output. If the industry is subject to localization
econornies, other firms in the industry will produce more output and hire more
workers too. Localization economies make the hiring decisions of firms in an industry
interdependent.
The samelogic applies to urbanization economies. While localization econornies
are efficiencies resulting from the . growth of a particular industry, urbanization
economies result from the growth of the entire metropolitan economy. If there are
urbanization economies, a software start-up firm generates benefits for firms outside
the software industry (lower costs of intermediate inputs, benefits from labor pooling, knowledge spillovers) and ihese benefits willtend to jncreaseprduction and
labor demand in other industries.
What about learning in cities? As we saw in Chapter 1, a large city provides
an environment that is conducive to learning by workers and entrepreneurs. As a
city grows, the learning opportunities expand. Learning increases laborproductivity,
resulting in lower production costs, and this allow firms in the city to underprice
their competitors in other cities. When firms expand their operations to produce more
output, they hire more workers.In other words, learning shifts the labor demand curve
to the right.
Subsidy Programs
Many cities try to attract export firms by offering special subsidies for new firms.
Levy (1985) describes severa! types of subsidy programs, including the following:
l. Tax abatement. In sorne cities, new firms are exempt from local property taxes
for sorne period of time, often up to 1O years. Sorne cities offer tax abatements to
all new dev!)Oplllents, while others offer abatements only to firms that are assumed
to be particularly sensitive to tax differentials.
130
2. Industrial bonds. Sorne cities issue tax-free industrial bonds to finance property
developrnent. Tbe local govemment uses the revenue froni the bonds to purchase
the land, and leases the property to a private firm. Because the interest incorne frorn
industrial bonds is not subject to federal taxes, the bond buyer accepts a relatively
low interest rate (for example, 8 percent instead of 12 percent). Therefore, the lessee
pays less than the rnarket interest rate on the rnoney borrowed to finance the project.
The use of industrial bonds was sharply curtailed by the Tax Reform Act of 1986.
3. Government loans and loan guarantees. Sorne cities loan rnoney to developers, and others guarantee loans frorn private lenders to developers. In both cases,
developers borrow rnoney at a relatively low interest rate; either the city charges
an interest rate below the rnarket rate, or the city decreases the risk associated with
a private loan, allowing the developer to borrow private rnoney at a relatively low
interest rate.
4. Site development. Sorne cities subsidize the provision of land and public services for a new developrnent. Tbe city purchases a site, c!ears the land, builds roads
and sewers, and then sells the site toa developer ata fraction ofthe cost of acquiring
and developing the site.
How do these subsidy programs affect the urban labor rnarket? Any policy that
decreases production costs will increase labor dernand and increase equilibriurn employment, everything else being equal. In the case of tax subsidies, not everything
else is equal. Tax revenue supports local public services, so a community with low
taxes is likely to have inferior public services. Tbe empirical evidence cited in Chapter 4 suggests that if a city cuts taxes and decreases its spending on public services
(e.g., highways, education, public safety), it is unlikely to grow, and rnay in fact
shrink. In contrast, if the city cuts taxes and decreases its spending on redistributional programs to the poor, the city is likely to grow. In other words, the effect of
a tax cut depends on what type of services are cut along with taxes. If taxes are
used to finance public services used by businesses, a tax cut is unlikely to stirnulate
economic growth.
What are the fiscal implications of these subsidy programs? Another objective
of a subsidy program is to decrease the local tax burden. If a tax cut or a subsidy
makes the city more attractive to firms, the city's tax base will increase, increasing
total tax revenue. As the city grows, it will also spend more on local public services
(roads, schools, police, fue protection). Tbe subsidy program will be beneficia!from
the fiscal perspective if the increase in tax revenue exceeds the increase in the cost
of public services.
a pollution tax of $ 00, steel producers pay $100 for every ton of pollution
generate. The pollutJ.on tax affects both sides of the urban labor market.
131
they
l. Shift of demand curve. The tax increases the production costs of steel producers. In addition to paying for labor, capital, and land, a firm pays $100 for every
ton of pollution. The increase in production costs increases the price of steel,
decreasing steel production and decreasing the demand for labor. In Figure 6-5,
the demand curve shifts to the left: at every wage, less labor is demanded.
2. Decrease in pollution. The tax decreases air pollution for two reasons. First,
steel producers will install pollution-control equipment as a means of decreasing their pollution taxes, so the amount of pollution generated per ton of steel
will decrease. Second, the increase in the price of steel decreases total steel
production.
3. Shift of supply curve. The improvement of the city's air quality increases the
relative attractiveness.ofthe city. People sensitive to air quality will move to the
city, shifting the supply curve to the right.
!y
:]-
:s,
-.
a1
1
1L'
v
)-
!S
"
)
"('
:e
l
,.
-
C1
<
S
[t
,,
Figure 6-5 shows one possible outcome of the abatement program. Since supply increases and demand decreases, the program decreases the equilibrium wage.
Since the rightward shift of the supply curve is large relative to the leftward shift
of the demand curve, equilibrium employment increases. The supply shift will be
relatively large if households are responsive to changes in environmental quality,
meaning that a large number of households will migrate to the city as air quality
improves.
How does the abatement program affect the distribution of employment between
the polluting industry and the clean industry? As the wage falls, the production costs
of both industries decrease. For the steel industry, the decrease in the wage partly
offsets the increase in pollution taxes. The abatement program is likely to generate
a net increase in production costs, so the polluting industry is likely to decrease its
total workforce. In contrast, the clean industry will simply pay lower wages, so its
production costs '(lill decrease and its total employment will increase. In Figure 6-5,
the increase in employment in the clean industry more than offsets the
decrease in
employment in the steel industry, so total employment increases.This occurs because
households are relatively sensitive to pollution, so that rnigration to the cleaner city
causes a large decrease in the wage.
Of course, the abatement program could actually decrease total employment. If
households are not very responsive to improvements in environmental quality, the
supply curve will shift by a relatively small amount, generating a relatively small
wage reduction. In this case, the increase in employment in the clean industry will
not be large enough to offset the decrease in employment in the steel industry, so
total employment will decrease. In general, the abatement program will decrease
total employment if the supply response (migration that decreases wages) is small
relative to the demand response (the decrease in labor demand from the polluting
industry).
.,....
';
132
Initiallabor supply
Labor supply in
cleaner city
1,000 ----------------
1
1
687-------
------------ 1---
50 55
Number of laborers in city (thousands)
A pollution tax increases production costs, decreasing the demand for labor. It also improves
environmental quality, increasing the supply of labor and decreasing the wage. The outward shift of
the supply curve is Jarge relative to the inward shift of the demand curve (people are relatively
sensitivo to air pollution), so total employment increases. The increase in employment in the clean
industry more than offsets the decrease in employment in the polluting industry.
133
Lz =
134
depends on weather, fashion, and the tastes for outdoor winter sports (e.g., skng,
ice fishing, snow shoveling). If a city's residents consume a relatively large amount
of rnittens, the location quotient L2 will overestimate the number of export workers.
Conversely, if city residents consume a relatively small amount of rnittens, L2 will
underestimate the number of export workers.
2. Nationalself-sufficiency. Another assumption of the location-quotient approach
is that the nation neither imports nor exports rnittens, so the share of national employment in rnitten production is the share required for local self-sufficiency. If the
nation exports rnittens, the share of national employment in rnitten production will
overstate the share required for self-sufficiency, so the location quotient will underestimate the number of export workers. Conversely, if rnittens are imported, the
location quotient will overestimate the number of export workers.
3. lndustry grouping. The location quotient is typically computed for a set of
consumer goods, not for a particular product. For example, a city rnight compute
a location quotient for the apparel industry, which includes rnittens, pants, dresses,
and children's clothes. Suppose that al! the city's apparel workers produce rnittens
for export; no other apparel products are produced in the city. Suppose that the rnitten industry is large enough that the cicy's location quotient for apparel is 1.0: The
city uses the same share of its workforce in rnitten production as the nation uses
in apparel production. The city's location quotient suggests that the city does not
export apparel products, when in fact it exports rnittens and imports other apparel.
Grouping goods into broad categories causes the location quotient to underestimate
export employment. The results of Tiebout (1962) suggest that location quotients
systematically underestimate export employment.
B =
L -1
. T;
L
For example,if L is 4.0 and T is 800, B is 600; three-fourths ofthe industry's workers produce for export. The city's total export employment is estimated by sumrning
the export employment across individual industries. The employment multiplier is
total employment divided by export employment.
135
city's re1ative attractiveness to the firrns producing that good. The city's econornic
forecaster could estimate the national demand forrnittens for the next decade,1ooking
at the trends in weather, winter sports, and fashion. If the city's relative attractiveness is not expected to change, the city's rnitten production should increase at the
same rateas national rnitten consumption: a doubling of rnitten consumption should
double the city's rnitten production and emp1oyment. Ifthe city's relative attractiveness increases over time, its rnitten production and emp1oyment should increase at
a re1ative1y fast rate.
INPUT-OUTPUT ANALYSIS
An altemative to an econornic base study is an input-output study. Unlike the econornic base study, the input-output study generales a complete accounting of the
transactions between firms and households in the economy. This approach has two
advantages over the econornic base study. First, instead of assurning that every industry has the same multiplier, it derives a multiplier for each export industry Second,
instead of using the 1ocation-quotient approach to guess the volume of exports, it
measures exports directly.
To explain the workings of an input-output study, considera city that produces
and cbnsurn:es three goods: computers, electrical wire, and local consumer gods.
136
Computers and wire products are prciduced for both export and local consumption.
By definition,local merchants (e.g., restaurants, dry cleaners, grocery stores) produce
goods for local consumption.
Computer
Firms
Wrre
Producers
Local
Merchants
Households
Exports
Total
$300
$ !50
$ 180
o
o
$1.370
600
2,500
2,000
350
$2,500
920
$3,600
$2,000
1,000
2,500
3.600
1,970
$1.970
o
o
$ 400
o
1,000
600
$2,000
600
100
$1,000
Computer Firms
Wire Producers
Local Merchants
Households
0.00
0.20
0.00
0.50
0.30
0.30
0.00
0.00
0.60
0.10
0.06
0.00
0.00
0.80
0.!4
0.05
0.00
0.69
0.00
0.26
1.00
1.00
1.00
1.00
137
input coefficients for the computer industry: Por every dollar worth of computer
production, the computer industry uses 20 cents worth of input from the wire industry, 50 cents worth of labor, and 30 cents worth of irnported materials (e.g.,
silicon chips and plastic). The second and third columns show the input coefficients
for wire producers and local merchants. The fourth column shows that households
spend 5 percent of their income on home computers, 69 percent on local goods, and
26 percent on irnports.
l"
'
'{
l nf
put
Ll
.1 1
rj
Jllt.
vu)
)'- t
l :1
tr-l.
m-
. .t. r;
. ./l.(.
.J
:<-.
l,vJO
lnitial effect
, < <
<
Round 1
Round2
Round 3
$1.20 in wire
1 lO
2500
3,vv0
1 '0
$6 in compurers
$3.00 in labor
$20in wire
$0.60 in computers
'
$12 in wages
$8.28 in local
$100 in computer
exports
<
<
$0.50 in wire
.S
$2.50 in computers
$50 in wages
<
$34.50 in local
$1.25 in labor
$1.73 in computers
$27.60 in wages
138
--
1.23
0.25
0.52
1.10
2.65
3.04
The spending and respending of income continues forever, but every spending
round is smaller than the previous one. The multiplier process eventually peters out
because both producers and consumers spend sorne of their budgets on irnports.
As shown in the fifth row in Table 6-3 (Irnports), there is a !eakage of 30 percent
in computer production, 10 percent in wire production, 14 percent in local goods
production, and 26 percent from consumers. These leakages weaken the multiplier
process, causing each spending round to be smaller than the previous round.
The input coefficients from Table 6-3 can be used to compute spending multi.pliers for the two export industries. Por a description of the techniques used to derive
the spending multipliers, see the references at the end of the chapter. In Table 6-4,
the first colurnn shows the multiplier effects of a $1 increase in computer exports.
Total computer sales increase by $1.23; wire production increases by $0.25; and
local sales increase by $1.17. The total multiplier effect is the sum of the effects on
the three industries: a $1 increase in computer sales increases total sales by $2.65.
Because the wire industry has a smaller irnport leakage (1O percent compared to
30 percent for the computer industry), it has a larger multiplier (3.04 versus 2.65).
Constant Multipliers
The multipliers derived from both the input-output and econornic base approaches
are assumed to be constant. In fact, a city's multipliers can change for three reasons.
139
l. City size and consumer products. Multipliers increase with city size, refl.ecting
the ability of a larger city to support a wider variety of firms. As explained in
Chapter 5, a larger city has enough demand to support activities with scale economies
that are large relative to per capita demand. For exarnple, if a city grows to the point
at which it can support its own Mexican restaurant, city residents will patronize the
local restaurant instead of traveling to another city. A larger fraction of increases
in income will be spent locally, so the income and employment multipliers will
increase.
2. City size and intermediate inputs. As the city grows, the demand for intermediate inputs grows, and more of these inputs will be provided locally instead of
being imported. For exarnple, if the computer industry grows to the point at which
the city can support its own silicon chip maker, computer firms will spend less on
imported inputs and more on locally produced inputs. Therefore, a given increase in
computer exports will generate a larger increase in total income.
3. Changes in input prices. Changes in input prices cause factor substitution
that changes the interactions arnong the economic sectors. Suppose that the city's
wage increases, causing computer firms to substitute imported capital equipment
for laborers. If the input coefficient for labor decreases from 0.50 to 0.40, only $40
of a $100 increase in computer exports will be paid to laborers. Because a smaller
fraction of export income will be spent locally, the multiplier effect of an increase
in exports will decrease.
Constant Wage
The economic base and the input-output approaches assume that the city's wage
is fixed, regardless of city size. In other words, they estimate the horizontal shift
of the city's.labor-demand curve, not the equilibrium change in employment. Suppose that an conomic base study computes a multiplier of 2.5 and projects an
increase in export employment of 10,000 jobs. Based on these numbers, the economic base approach would predict that total employment in the city would increase
by 25,000 jobs. This is incorrect because the city's labor-supply curve is positively
sloped, so an increase in labor demand increases the city's wage.
In Figure 6-3, total employment increases by only 16,000 jobs, not 25,000 jobs.
Because the economic base and the input-output approaches assume a fixed wage,
they overestimate the stirnulative effect of increases in exports.
140
141
It's importan!to note that public policy need not focus exclusively on increasing
the city's exports. As explained earlier, among the altemative strategies are import
substitution and increased trade within the metropolitan area. To the extent that public
policy encourages import substitution or intracity trade, it can increase employment
and income without increasing the city's exports.
Equilibrium
Equilibrium
Change
5.40%
87.50%
82.78%
120,809
105,708
5,708
100,000
20,809
5.33%
87.64%
82.97%
121,737
106,687
5,687
101,000
20,737
770
230
70
!60
-0.07%
0.14%
0.!9%
928
'1
j
l
1,000
142
.percentage of working-age people who were actively involved in the labor market,
either working or looking for work) was 87.50 percent. Suppose that the employment rate is defined as the percentage of working-age people who are employed.
The employment rate is computed by multiplying the participation rate by one minus
the unemployment rate. In Bartik's study, the average employment rate was 82.78
percent. The hypothetical city has a potential workforce of 120,809 people; given
the participation rate (87.50 percent), 105,708 participate in the labor market; given
the unemployment rate of 5.40 percent, 5,708 of the participants are unemployed,
leaving 100,000 people who are employed.
The seond column ofnumbers in Table 6-5 shows the predicted long-run effects
of a 1 percent increase in total employment (1,000 jobs). According to Bartik, the
long run is about six years: it takes about six years for all the marketsto fully adjust
to the change in employment. Six years after the increase in total employment, the
unemployment rate is 5.33 percent, the participation rate is 87.64 percent, and the
employment rate is 82.97 percent. Migration to the city increases the city's potential
workforce by 928 people.
How many of the new jobs are filled by newcomers, and how many are filled
by original residents who would otherwi:se not be employed? Since there are 1,000
new jobs and only 928 new potential workers, there are at least 72 new jobs for the
original residents.If all the newcomers are employed (newcomers have a 100 percent
participation rate anda zero unemployment rate), there will be only 72 new jobs for
the original residents. If the newcomers have the same employment rate as the original residents, however, only 770 of the newcomers will be employed (82.97 percent
. of 928), so there wil1 be 230 jobs left for the original residents. About a third of
the 230 jobs will be filled by original residents who were previously unemployed,
and the remaining two-thirds will be filled by original residents who previously
did not participate in the labor market. Figure 6-7 summarizes the results from
Table 6-5; 77 percent ofthe new jobs are filled by newcomers, leaving 7 percent for
FIGURE 6-7 Distribution of.J,OOO New Jobs between Original Residents
and Newcomers
Newcomers: 770 jobs
unemp1oyed: 70 jobs
143
original residents who were unemployed and 16 percent for original residents who
did not participate in the labor market.
The simple lesson from Table 6-5 and Figure 6-7 is that employment growth
causes in-rnigration and population growth. There are three implications from this
lesson. First, ifthe city's workforce is justas qualified forthe new jobs as workers in
other cities who have the option of rnigrating to the city, the original residents will
get about a quarter ofthe new jobs. Second, if the city's workers are, on average, less
qualified for the new jobs than workers in other cities, more of the jobs will be filled
by workers rnigrating to the city. Third, employment growth increases population
and thus increases the demands for housing, land, and p"ublic services. This suggests
that local govemments should coordinate their econornic development po!icies with
their policies concerning land use, transportation, and infrastructure investrnent.
... o;
:; e o.zoo
-----------------------------
" >-.
'Ca.Q
e "-
.!2
:;
<.S 0.150
"
"
o
-- -" 0.100
f3
g
=- ""
e
e oo.o5o
"-.J::
0.000 ..J.
Average
Less
educated
Younger
Black
Source: 1imothy J. Bartik, Who Benefitsjrom- State andLi;cal Economic DeVeloprrlerit.Policies? (Kalamazoo, Na:
Upjohn lnstitute, 1991).
144
the job hierarchy (and up the pay scale) from a 1 percentincrease in total emp1oyment.
The average worker is promoted to a job with a wage that is 0.16 percent higher
than the wage for the old job. A less-educated worker (3 years 1ess education than
average) is promoted to ajob thatpays 0.196 percentmore; ayoung worker (12 years
younger than average) is promoted to ajob that pays 0.188 more; a black worker is
promoted to ajob thatpays 0.215 percent more.
3. Increase in employment rate. As explained earlier, employment growth decreases the unemp1oyment rate and increases the participation rate, so it increases
the fraction of the working-age population that is employed. InTable 6-5, the 1 percent increase in employmentincreases the employment rate from 82.78 percent to
82.97 percent.
Figure 6-9 shows the combined effects of changes in real wages, occupational
rank, unemployment rates, and participation rates on real income per capita. For the
. average household, a 1 percent increase in employment increases real income per
capita by 0.40 percent. In other words, the elasticity of real earnings with respect to
total employment is 0.40. The most important factors in the increase in income are
the increases in wages from being promoted to higher-paying jobs and the increases
in labor force participation. The elasticities are Iarger for households that are less
educated, young, or black, because workers in these groups experience relatively
large benefits from promotion to higher-paying jobs.
----------------------------------------
;:
"a"
2 .Q 0.400
c.
S
"
""'"
S"
8
.::; .5
0.300
---
- "
e
e".S
"E
."s "
0.200
- c.
0.100
-
"
Less
educated
Younger
B!ack
Source: Timothy J. Bartik, Who Benii.&s from StaJe and Local Economic Development Policies? (Kalarnazoo, MI:
145
SUMMARY
l. We can use the urban labor rnarket to show the causes and consequences of
r:
1
146
No.
a. Suppose that the demand forthe city's exports increases. Use the conventional
supply-demand graph to show the effects of the increase in export demand
on the city's labor market. Label everything: the axes, the curves, the new
equilibrium wage (JY*), and the new equi!ibrium total employment (N*).
b. Suppose that the city institutes a growth-control program that holds the total
housing stock (total square footage) at its initiallevel (before the increase in
export demand). Use the same s pply-demand graph to show the effects of
the increase in export demand under the growth-control program. Label the
equilibrium wage under the growth-control program W* and the equilibrium
total employment N*.
c. Explain the differences between the market equilibrium and the growthcontrol equilibrium.
3. Consider a city that uses a business tax to finance the provision of industrial
services (e.g., roads, water, sewers). If the city does not provide these services,
the individual fi.rms must supply their own roads, water, and sewers. Suppose
that the city decides to cut business taxes and expenditures on industrial services
by the same amunt. Evaluate the effects of the new tax-expenditure policy on
the city's labor market. Will total employment in the city increase or decrease?
4. Consider tWo cities, Flexville and Rigid City, that have the same equilibrium
wage and the same equilibrium total employment. The export industries in
the two cities produce the same good. In F!exville, export fi.rms produce with
variable factor proportions; the amount of labor per unit of output depends
on the relative price of labor. In Rigid City, export firms produce with fixed
factor proportions; the amount of labor per unit of output is fi.xed. Suppose that
both cities find a way to increase the quality of residential public services (e.g.,
public schools, parks, !ibraries) without increasing taxes. Use two graphs, one
for each city, to show the effects of the improvements in public services on
(a) equilibrium employment, and (b) the equilibrium wage.
5. One of the assumptions lurking behind the city's labor-supply curve is that
an increase in the wage does not increase the fraction of the city's population in the workforce. Draw two supply curves, one under the assumption of a
fixed participation rate and another under the assumption that the participation
rate increases with the wage. Explain any differences between the two supply
curves.
147
6. The section on envirorunental quality suggests that a pollution tax may increase total employment. Suppose that instead of using a pollution tax to reduce
citywide industrial pollution by a total of 25 percent, the city requires that each
polluting firm decrease its pollution by 25 percent.
a. Is the uniform-reduction policy more or less ef:ficient than the pollution-tax
program? In,other words, will the total cost of abatement be larger or smaller
under the uniform-reduction policy?
b. Is the city more or less likely to grow under the uniform-reduction policy?
Demonstrate, using a graph.
7. Consider a city with the following characteristics: (i) The price elasticity of
demand for the city's output is -3.0; (ii) The elasticity of labor supply with
respect to envirorunental quality is 0.1O; (iii) Polluting :firms produce with :fixed
factor proportions; labor per unit output is :fixed; (iv) Polluting :firms initially
employ 25 percent of the city's export workers. Suppose that the city adopts a
new envirorunental policy that increases its environmental quality by 10 percent
and increases the price of the polluting industry by 2 percent.
a. Assume that the city's wage is fixed. Predict the percentage decreases in the
demand for workers in tlie polluting industry and the export industry.
b. By what percentage will the supply of labor increase?
c. Under what conditions (what values ofthe relevant elasticities) will the equilibrium number of workers increase? illustrate your answer with a graph.
8. Consider a city that has decided to impose a pollution tax on its polluting :firms.
Fill in the blanks to make the following statements correct, and then briefiy
explain your responses. The city is more likely to grow as a result ofthe pollution
tax if:The elasticity of the supply of labor with respect to environmental quality
is relatively
; The labor-supply curve shifts by a relatively
_
amoun( when envirorunental quality improves; The elasticity of demand for
labor (with respect to the wage) by the city's nonpolluting industries is relatively
; The elasticity of the supply of labor with respect to the wage is
relatively
(the labor-supply curve is relatively
).
9. In a report issuedjn September 1989, a consulting :firm estimated the economic
impacts of moving the Los Angeles Raiders (a professional football team) to
Sacramento. The economic base study was based on the following assumptions: (z) Total attendance at the Raider games will be 700,000 people per year;
(ii) The average ticket price will be $30; (iii) The average fan will spend $10
on food, merchandise, and parking; (iv) Based on (ii) and (iiL), the average fan
wi11 spend a total of $40; (v) Total "direct" spending will be $28 million per
year ($40 x 700,000). Using a spending multiplier of 2.2, the consulting :firm
estimated the total economic impact of the Raiders to be $61.6 million per year.
Critically appraise the methods used to compute the total economic impact of the
Raiders.
10. There are two ways to increase total employment: increased exports and import
substitution. In what way are increased exports and import substitution similar?
In what way are they dissirnilar?
148
11. Considerthe trade-offs associated with choosing a reference region for defining a
city's location quotient. The traditional approach is to compare the city economy
to the national economy. What if the denorninator of the location quotient was
defined as the percentage of state employment in rnitten production? Would the
measured location quotient be closer to or farther from the city's true location
quotient?
12. In Figure 6-2, the shift of the demand curve from Dz to D3, which refl.ects
the multiplier process, is parallel. The implicit assumption is that the employment multiplier is constant. Draw the demand curves under the assumption that
the multiplier increases with city size.
13. Suppose that you have been given the following data on the transactions within
a city's economy.
a. Fill in the transactions table.
b. Will this city have a larger or a smaller export multiplier than the city with
the transactions listed in Table 6-2? Explain.
Inputs
Computers
Wire
Local
Labor
Imports
Total
Computer
Firms
Wire
Producers
$350
500
o
o
1,000
600
Local
Merchants
$
o
o
2,000
300
2,000
Households
Exports
$ 300
$_
600
2,500
o
o
Total
$2,000
1,000
3,600
1,650
3,600
14. Suppose that wire production in the city with the input coefficients shown in
Table 6-3 increases by $100.
a. Compute the first, second, and third rounds of spending resulting from the
increase in wire sales, using the format of Figure 6-6 to report your results.
b. By how much does spending on wire, computers, and local goods eventualy!
increase?
c. What is the ultimate increase in total sales?
15. Consider a city that estimates its export multipliers from the input coefficients
from a 1965 input-output study.
a. Will the city overestimate orunderestimate the stimulative effects of exports?
b. If your answer to (a) is "It depends on ... , " make a list of questions that
you would ask Ms. Information, the most knowledgeable person in the city.
Note: She does not know the current input coefficients.
REFERENCES
l. Bartik, Timothy J. Who Benefits from State and Local Economic Development
Policies? Kalamazoo, MI: Upjohn Institute, 1991.
2. Eberts, Randall W., and Joe A. Stone. Wage and Adjustment in Local Labor
Markets. Kalamazoo, MI: Upjohn Institute, 1992.
149
150
PART TWO
.,.
'
_.-,,-_-
part of the book examines the spatial structure of cities, exploring the market
forces and government policies that determine land-use patterns witbin metropolitan
areas. Chapter 7 develops the basic concepts of land rent and land use, explaining
how the price of land is determined and how land is allocated among alternative uses.
Chapter 8 explains why, until the rniddle of the Twentieth Century, most cities were
monocentric, with most employment concentrated near the center of the metropolitan
area. Chapter 8 also explores the interactions between the urban land market and the
urban labor market. Chapter 9 explains the market forces that caused the traditional
monocentric city to be replaced by the modern city, with employment and population
dispersed thrqughout the metropolitan area. Chapter 1O discusses the role of local
governments in the urban land market, exploring the market effects of zoning and
other land-use controls.
...- ft'
,-:: :.._.;
151
CHAPTER 7
bis chapter introduces sorne b sic concepts of land rent and land use, setting the
stage for the discussion of urban land use in Chapters 8 through 11. We'!l address
three questions about the land market. Pirst, what determines the price of land?
Second, who benefits from pub!ic policies that increase the fertility or accessibility
of land? Third, does the land market allocate land efficient!y? It wi!l be useful to
define two terms, land rent and market value. Like other assets, land yields a strearn
of marketable services and thus a strearn of income. Por exarnp!e, agriculturalland
yields a stream of agricultura!output (bushels of corn), generating a stream of income
for the farme. Siin.ilar!y, a parking lot in the city yields a strearn of parking services,
generating a stream of income for the parking firm. When a landowner grants the
rights to use bis land to another individual ora firm, he charges land rent If a farmer
is granted the right to grow corn on a plot ofland, the rent rnight be $1,000 per acre
per year. If a firm is granted the right to operate a parking lot on a plot of land, the
rent rnight be $5,000 per acre per year.
What determines the market value of land? The market value of land equals
t.he present value of the stream o.f renta! income generated by the land. To explain
the concept of present value, consider an asset that generates R of income each year
and is expected to generate tbis income for n years. If the market interest rate is i,
the present value of the stream of earnings from the asset is
PV=
7:o (1 + i)'
Por example, if an asset is expected to generate $20 of net income per year, starting
today and lasting for a total of five years, and the interest rate is 10 percent, the
153
154
20
20
1.21 + 1.33
20
1.46
+
+
PV = 20 + 18.18 + 16.53 + 15.04 + 13.70 = $83.45
If the strearn of earnings 1asts forever, the equation for present value simplifies to
PV = R
i
==
0.10
$200
For exarnple, if the $20 annual income !asts forever, the present va!ue of the asset is
$200.
The present va!ue is the maximum arnount that an investor is willing to pay
for an asset, given an altemative investment that yie1ds i percent per year. Suppose
that the altemative is a savings account that yie!ds 1O percent per year. The investor
can either invest in an asset that yie!ds $20 per year forever or invest in a savings
account that yields 1O percent per year.At a purchase price of $200, the investor
is indifferent between spending $200 on the asset and investing the sarne arnount
in a savings account; in both cases, the annual income is $20. At a purchase price
less than $200, the investor prefers the asset to the savings account. For exarnple,
if the price is $100, the investor can make $20 per year by investing $100 in the
asset, compared to $10 per year by investing the sarne arnount in a savings account.
Similar!y, for a purchase price exceeding $200, the savings account is more1ucrative
than the a.sset.
The market value of land is the present va!ue of the annua! renta! payments
from the land. Land used for residentia!, commercial, and industrial activities can,
in principie, yield a constan!strearn of renta! income. In contrast with agricultura!
land, which can deteriorate with use, developed land does not deteriorate. Therefore,
the market va!ue equa!s the annua! rent divided by the interest rate. For exarnple,
if the annua! rent on a p1ot of 1and is $5,000 per acre and the market interest rate
is 1O percent, the market va!ue of land is $50,000 per acre. The market value of
land equa!s the present va!ue because the present va!ue makes an investor indifferent
between buying the land (spending $50,000 to eam $5,000 per year in land rent
forever) and putting the $50,000 in a bank account with a 10 percent interest rate
(earnings of $5,000 per year).
This book uses land rent-not market value-as the price of land. Most of the
other relevant econornic variables are defined as strearns of revenue or costs. For
exarnple, a household eams an annua! income, anda firm computes its annual profits
as its annual revenue less its annua! cost To be consisten!,the price ofland is defined
as the annua! payment in exchange for the right to use the land: the price of land is
synonymous with land rent. Given the simple relationship betWeen rent and va!ue,
it's easy to make the translatiort from land rent to market value: just divide the annual
rent by the market interest rate.
155
Figure 7-1 shows the conventional cost curves for one-acre plots of the three
types ofland. The marginal-cost curves (MC) are positively sloped, and pass through
the U-shaped average total-cost curves (ATC) at the minimum points of average-cost
curves. The cost curves include all the nonland costs of production, including the
costs of raw materials (seeds and fertilizer), capital (tractors), and labor. They also
include the.opportunity cost ofbeing a farmer, for example, the money the farmer
gives up by being a farmer instead of a steelworker.
The positions of the cost curves depend on the fertility of the land. A farmer on
relatively fertile land can produce the same amount of coro with smaller quantities
of the nonland inputs. Because the farmer spends less money on seeds, fertilizer,
tractors, and labor, his average-cost curves are lower. In general, the higher the
fertility, the lower the cost curves.
How much are farmers willing to pay fr the three types ofland? In Figure 7-1,
the national coro market generates an equilibrium price of $10; supply intersects
demand ata price of $10. Farmers are price takers and maximize profit where price
equals marginal cost. The profit-maximizing output on the high-fertility land is
220 bushels per acre, generating profit equal to the shaded area. In this example,
profit equals $1,320 per acre per year (total revenue of $2,200 less a total cost of
$880).
A farmer would be willing to pay up to $1,320 per year to use one acre of the
high-fertility land. Similarly, a farmer would be willing to pay up to $320 per year
for the medium-fertility land. For the low-fertility !and, production costs are so high
that coro production is not profitable ata price of $10, so a coro farmer would not
be willing to pay anything for the low-fertility !and.
......
"'
0\
$1
MC111
Supply
'
MCh
1
10
. <"
\; bnt
..
. -.,:
''
'
i --z--r-----;
ATC
4
Demand.
220
Q*
Com market
160
Comperyear
Comperyear
High fertility
Medium fertility
Low fertility
The equilibrium price or corn is $10 per bushel. Competition for Iand forces fanners to pay their surplus (total revenue less total nonland cost) to
lfmdowners. The high-fertility land has lower production cosls and higher pre-rent profit, so farmers are willing to pay more rent. The medium-fertility
land earns less rent because it has higher production costs. The low-fertility land genera tes no profits, so its rent is zero.
) . )
157
.-.:.
,
_
<
...
,
:.e!'-;
,.
ll<
l<..''-..',.,.,
-
-
l.
.
-
...
)
-
t
l
J
l
1
J
Competition among prospective farrners bids up the price of land to the point at
wbich economic profit is zero (accounting profit is normal). Farrners are willing to
pay up to $1,320 forthebigh-fertility land, and are forced by competition todo so. At
any rent less than $1,320 per acre, the landowner will be able to find another farrner
willing to pay slightly more to use the land. Similarly, the equilibrium rent on the
medium-fertility land is $320. Because the equilibrium land rents make econoinic
profits equal to zero, farrners are indifferent between different plots ofland. Although
the high-fertility land.has lower production costs, the savings in production costs are
offset by higher land costs.
In equilibrium, land rent equals the excess of total revenue over nonland costs.
This is the leftover principie: Because of competition among farrners for land, the
landowner gets the leftovers. This principie assumes that individual plots of land
have unique characteristics, but farrners are all the same. Competition amortg a
large number of farrners, each of whom has the same cost curves, bids up the price
of bigh-fertility land to the point at which economic profit is zero. If the farrner
on the high-fertility land pays less than the excess of total revenue over nonland
cost, the farrner would be evicted and replaced with another farrner willing to pay
the leftovers (total revenue less nonland cost) for the opportunity to earn normal
accounting profits.
The leftover principie does not hold if there are restrictions on entry and competition. Onerestriction on entry comes from patents.If farrner Tom holds the patent for
a particular farming technique, he has lower production costs than all other farrners.
For example, suppose that Tom can produce an acre's worth of com for a pre-rent
profit of $2,000, and .other farrners, using inferior techniques, generate a pre-rent
profit of only $500. The landlord is unable to charge Tom rent of$2,000 because the
threat of eviction is a hollow one: There are no other farrners with the same production costs, so there are no farrners willing to pay $2,000 per acre. Instead, Tom pays
only $500, allowing him to make an economic profit of$1,500. The landowner does
not get the leftovers because the patent restricts competition. Once the patent expires
and all fa.."'Ilers have access to the same technology, the landowner can increase land
rent and convert the economic profit into increased land rent.
l
1
J:
1
1
l
l
l
1
1
l
l
J
......
FIGURE 7-2 Effects of Irrigation Project on Land Rent and Com Price
$
$
Original
supply
JO
9
MC 1
New
r+--supply
----------
M Cm
..... .
......
-'
_-
_.O_;. '..jo.,'
-.
"Are,
; -:_: ; :, '
-.:S.i :- Re..
:: : - - : ;.-,) ' :
1
1
1
1
1
1
1
Demand
1
1
180
240
Cornmarket
140
Comperyear
Comperyear
High fertilily
Medium fertility
Low fertility
.<\n irrigation project decreases production costs for aH three types of Iand. H the price of corn stays at $10, the shaded areas show the land rent on the three
types of land. The supply of corn will increase because land is used more intensively (output increases on the high-fertility and medium-fertility land) and
marginal (low-fertility) land is brought in lo production. The increase in supply decreases the equiJibrium price of corn, so the rent on each plot of land is
159
il( ,
..
J(
....
'1.
(
!.(
. --
=-
:t
1
]
production costs are paid to landowners in the form of hi"her rent, so the benefits of
the irrigation project go to landowners.
"'
Will the price of corn be affected by the irrigation project? The project increases the supply of coro for two reasons. First, the project shifts the roarginal-cost
curves downward, increasing the profit-maxirnizing outputs ofthe high-fertility and
medium-fertility farms. Second, marginalland (low-fertility land) is brought into
production. For these two reasons, the supply curve shifts to the right, decreasing
the equilibrium price of corn. Therefore, corn consumers benefit from the irrigation project. As the price of corn decreases, the pre-rent profits of farmers decrease,
decreasing land rent.. In other words, consumers gain at the expense of landowners.
What determines the distribution ofbenefits between landowners and coro consumers? The general rule is that the smaller the geographical area covered by the
irrigation program, the larger the share of the benefits that go to landowners. Consider first an irrigation project that decreases the production costs of a single 50-acre
plot ofland. The project causes a trivial increase in supply and virtually no change
in the price of coro. Therefore, all the benefits go to the landowner. Consider next a
national irrigation project that decreases the production costs of al! corn farmers. The
project causes a large increase in supply (existing land is cultivated more intensively
and more land is brought under production), so it decreases corn prices significantly.
In this case, a large share of the benefits goes to consumers.
The benefits of the irrigation project are capitalized into the market value of
land. Since the project increases the annual rent, it increases the present value of the
stream of earnings from the land, increasing its market value. For example, suppose
that the annual rent on high-fertility land increases from $1,320 per acre per year to
$1,500.If the market interestrate is 1O percent, the equation showing the relationship
between rerit and value suggests that the market value of the land will increase from
$13,200 per a'tre to $15,000 per acre.
:1
l.
1
.t
1
l
l
1
MARKET INTERACTIONS
The demand for land is derived from the demand for outputs (e.g., coro, carrots,
housing, retail goods, manufactured goods). This section examines the interactions
between the land market and the output market. The discussion addresses a sort of
chicken-and-egg question about the land market: Is the price of land high because
the price of output is high, or is the price of output high because the price of land is
high?
The British Corn Laws of the 1800s restricted grain imports to Britain. The decrease
in the supply of imported grain increased the demand for domestically produced
coro. Figure 7-3 shows the effects of the Coro Laws on the coro market and the land
market.
-
(
t
t
J.
160
FIGURE 7-3 The Com Laws Debate
(A)
Corn market
$
Corn supply
Pz --- - ------
o
"o
i"f
1
1
1
1
1
e,
Cz
Quantity of corn
(B)
Landmarket
$
-"'
.S.
:
:
"O
Rz
.."-1'
D
Quantity ofland
The Corn Laws restricted grain imports, increasing the demand for
domestic corn and its price. Corn production increased from C 1 to
Cz, increasing the demand for land from D to Dz. The price of
land increased from Rto Rz. The price of land is bigh because
the price of corn (and the production of corn) is bigh.
r:
'
161
Coro market. The Coro Laws sbifted the deroand curve froro d1 to d 2 . The price
of domestic coro increased from P1 to P2 , and the quantity of coro produced
increased from C1 to C2 .
Land market. As domestic coro production increased, the demand for land
increased. In Figure 7-3, the increase in coro production sbifted the demand
curve for land from D to D2 Because the supply curve is perfectly inelastic
(they aren't making land any more), the increase in demand increased the price
of land from R 1 to R2 .
To summarize, the price of land is high because the price of coro is high. The
Coro Laws increased the price of coro, which stimulated the production of coro and
the demand for coro-growing land. Landowners responded by increasing the price
of land to a!locate the fixed resource among competing land uses. The lesson is
that high land prices are the result of high coro prices, not the reason for high coro
prices.
LAND. TAXATION
In 1880, Henry George proposed a 100 percent tax on renta! income. The proposed
tax was dubbed the "single tax" because it would have generated enough revenue
to support a!llevels of govemment at the time. The spirit of the single tax is best
expressed by George himse!f. The following is from an interview with David Dudley
Field (in the North American Review 1885):
in
Field: Then suppose Ato. be the proprietor of a thousand acres on the Hudson, chiefiy
farrning land, but at the same time having on it houses, barns, cattle, horses, carriages,
fumiture; how is he to be dealt with under your theory?
George: He would be taxed on the value of his land, and not on the value of his improvements and stock... The effect of our present system, which taxes aman for values
created by his labor and capital, is to puta fine upon industry, and repress improvement
162
The more houses, the more crops, the more buildings in tbe country, tbe better for us
all, and we are doing ourselves an injury by imposing taxes upon the production of such
things.
Field: Then you would tax the fanner whose fann is wortb $1,000 as heavily as you
would tax tbe adjoining proprietor, who, with the same quantity of land, has added
improvements wortb $100,000; is that your idea?
George: It is. The improvements made by the capitalist would do no hann to tbe
farmer, and would benefit the whole community, and I would do nothing to discourage
tbem.
Field: A large landlord in New York owns a hundred houses, each wortb, say, $25,000
(scattered in different parts of tbe city); at what rate ofvaluation would you tax him?
George: On bis houses, nothing. I would tax him on the value of the lots.
Field: As vacant lots?
George: As if each particular lot were vacant, surrounding improvements remaining
the same.
George proposed the single tax for both equity and efficiency reasons. On the
equity issue, George argued that land rent is determined by nature and society, not by
the efforts of landowners. As discussed earlier in the chapter, agricultura!land rent is
determined by the fertility of the soil and its accessibility to markets. Similarly, the
urban land rent is determined by its accessibility to other activities. In George's time,
cities were growing rapidly, causing rapid increases in land rent and value. George
argued that landowriers did nothing to deserve the increases in property value, so
that any windfall gains from urban growth should be taxed away.
On the efficiency issue, George argued that the land tax would eliminate the need
for taxes on improvements. The elimination of improvement taxes would stimulate
investment in houses, crops, and buildings. The land tax would not affect the supply
of land because the supply of land is fixed. The replacement of the improvement tax
with the land tax would increase the total wealth of society.
The single tax has been criticized for three reasons. First, the single tax would
decrease the net return to the landowner (net md rent) to zero, making the market
value of land zero. In other words, the government would essentially con:fiscate
the land. This strikes many people as inequitable. Second, if the net return on land
were zero, landowners would abandon their land, leaving government bureaucrats
to decide who uses the land. Unlike the prvate owner, who receives more income
if the land is used efficiently, the bureaucrat has nothing to gain from the efficient
use of land. Therefore, the government land market is less likely to allocate land to
its highest and best use. The third criticism is that it is difficult to measure land rent
(and the appropriate tax). Most land has structures or other improvements, and it is
difficult to separate the value generated by the raw land from the value generated by
the improvements.
163
'"
at less than 100 percent of its value. A partialland tax would be less confiscatory
than the single tax; like conventional taxes on labor and capital, the partial tax would
confiscate only a portion of the taxpayer's resources. In addition, because a partial
tax leaves landowners with a positive net retum, the land market will continue to be
run by those who have a private interest in allocating land to its highest bidder.
Another altemative toa pure land taxis the two-rate tax, or the split tax. Under
the conventional property tax, land and improvements are taxed at the same rate.
A 3 percent property tax is actually a 3 percent tax on land and a 3 percent tax on
improvements. Under a split tax, the tax rate on land may be 9 percent, while the tax
rate on improvements may be 1 percent. The split taxis widely used in Australia and
New Zealand. It is also used in sorne cities in Pennsy1vania. Pittsburgh implemented
a split tax in 1913, and six other cities, including Scranton and Harrisburg, have
adopted the split tax in recent years.
The replacement of the conventional property tax with the split tax would stimulate capital investment. The switch to the split tax would decrease the tax rate
on capital, encouraging capital improvements. Suppose, for examp1e, that Rhonda
would like to add a recreation room to her house. The new room would increase the
assessed value of her house by $20,000. Under a conventionaJ 3 percent property
tax, her tax liabilities would increase by $600 per year. Under a split tax with a
1 percent rate on improvements, her tax liabilities would increase by only $200 per
year. The tax penalty from the home improvement would be 1ower under the split
tax, so Rhonda would be more likely to improve her house.
164
.
reduce the tax rate on improvements. The decrease in the improvement tax decreased
the penalty on new construction and thus encouraged development. Although other
factors played a role in the city's commercial building boom of the 1980s, the tilting
of the graded-tax system toward lower taxes on improvements was an important
factor.
.'
SUMMARY
l. According to the leftover principie, the bid rent for land equals the difference
between total revenue and total cost. Competition for land ensures that the
landowner gets the excess of total revenue over total cost.
2. Land that is relatively fertile has relatively low production costs, s it commands
a higher rent.
3. A policy that increases fertility (e.g., an irrigation project) generates benefits for
both landowners and consumers.
4. The benefits of an irrigation project are capitalized into the market value ofland:
the increa.Se in rent increases the pre ent value of renta! income, increasing the
market value of land.
5. The demand for land is derived from the demand for output (e.g., coro, housing).
The price of land is high because the demand for output is high.
6. Henry George proposed the single tax, a 100 percent tax on renta! income.
method costs $150 per acre. Considera county where rice farmers are currently
willing to pay $500 per acre for land, and coro farmers (who do not sanitize their
fields) are willing to pay $300 per acre. The total output of the county is small
enough that the prices of rice and coro are unaffected by events in the county.
Suppose that field burning is outlawed in the county, forcing rice farmers to
switch to the alternative sanitizing method.
a. How does the field-burning law affect rice consumers, coro consumers, farmers, and landowners? In other words, who bears the cost of the pollutioncontrol program?
b. How would your answer to (a) change if the cost of the alteroative method
were $250 per acre?
c. How would your answer to (b) change iffield burning were outlawed in the
entire state of California?
2. Critically appraise the following statement: I would like to clear the air with
sorne facts about rice straw burning. Burning is the only economical way to
prevent stem rot in rice. This disease would drastically reduce the yield of rice
grown on the same land the next year. The California Departrnent of Agriculture
,_
165
'';
-,.'
r .:>''"c".!- i ,_,- 1 - e .6, .r' estimates the cheapest altemative to rice straw burning, which involves baling
and hauling it elsewhere, would cost about $150 per acre. The opponents of
straw buming suggest the savings ($150 per acre) go straight into the pockets
of growers. Actually, straw buming decreases the prices of Rice Krispies and
other rice products, so the savings go to consurners.
3. Consideran agricultura! economy with the following characteristics: (i) Al1 the
land in the region is initially used by tenant farmers to grow indigo; (ii) The
price of indigo is deterrnined in intemational markets; (iii) The tenant initially
pays the landowner 30 percent ofthe indigo harvest as rent; (iv) Output per acre
is 1,000 units per year and the price of indigo is $2 perunit; (v) The interest rate
is 10 percent per year.
a. Compute nonland cost per acre per year and the market value of land.
b. Suppose that the price of indigo drops to $1.90. Assurning that the tenant
continues to grow indigo on the land, compute the equilibrium rent in dollars,
in units of indigo, and as a percent of the indigo harvest
c. By how much does the market value of land drop as a result of the decrease
in the price of indigo (assurning the tenant grows indigo)?
d. How would your answer to (e) change if there is an alternative crop with the
same nonland costs and output per acre?
4. Suppose that Mr. Greengenes, a farmer and genetic engineer, develops a new
method for growing com that decreases the cost of growing com by $300 per
acre. Greengenes's landlord rejoices, saying, "According to the leftover principie, you will pay me $300 more in rent." Is the landlord correct? If not, is he
applying the leftover principie incorrectly, or is the principie wrong?
S. Consider Euphoric County, where a large share of the arable land is used to
grow M. Tlie production of M is illegal: there are severe penalties imposed
on M growers, but no penalties imposed on M consumers. Suppose that Mis a
competitive industry, with equilibrium profits equal to zero; total revenue equals
total costs. Included in the costs are the costs associated with engaging in illegal
activities (the opportunity cost of time spent in jail, legal costs, concealment
costs). Suppose that Euphoric County legalizes the production of M.
a. Depict graphically the effects of legalization on the equilibrium price and
quantity ofM. Explain your graph.
b. Depict graphically the effects of legalization on the price of land in Euphoric
County. Explain your graph.
6. The residents of mobile home parks own their dwellings and rent land from
absentee landowners. Considera city in which allland is cui:rently occupied by
mobile home parks. Suppose the city imposes a 50 percent tax on land, to be
paid (in legal terrns) by the person who occupies the land (the tenant, either a
mobile home owner or sorne other user). Who actually pays the tax?
7. What would be the effect of a partialland tax ($100 per acre) on land rent, land
values, and com prices?
8. As the flexible farmer approaches the marketplace, the farmer substitutes
nonland inputs for land. As a result, the land-rent.function of the flexible farmer
166
is steeper than the land-rent function of the inflexible fanner. What happens as
the flexible fanner moves away from the marketplace? Is the flexible land-rent
function steeper or flatter than the inflexible function? Explain.
REFERENCES
l. Alonso, William. Location and Land Use. Cambridge, MA: Harvard University
Press, 1964.
2. Alonso., William. "A Theory of the Urban Land Market." In Readings in Urban
Economics, ed. Matthew Edel and Jerome Rothenberg. New York: Macmillan,
1972.
3. George, Henry. Progress and Poverty. New York: Schalkenbach Foundation,
1954.
4. Milis, Edwin S. "The Value of Urban Land." In The Quality of the Urban
Environment, ed. H. Perloff. Washington, DC: Resourcesfor the Future, 1969.
5. Oates, Wallace, and Robert Schwab. "The Impact of Urban Land Taxation: The
Pittsburgh Experience."Cambridge,MA: LineoIn Institute ofLand Policy, 1992.
6. Ricardo, David. Principies of Political Economy and Taxation. 1821. Reprint.
London: John Murray, 1886.
CHAPTER 8
'
'..J
_,
'-;
:,
..
\.
bis chapter discusses land rent and land use in the monocentric or ;;_'d minated
city. The monocentric city was e dominant urban form until the early part of the
twentieth century. The key feature ofthe monocentric city was the heavy.concentration of employmet in the central core area. Mqst_people commuted inward to jobs
. in the city center. As we'll see later in the book, in the typical modero city employment is not concentrated in the central area, but is instead distributed throughout
the me)ropolitan area, with a large fraction of jobs in suburban areas. This chapter
explains the market forces behind the development of the monocentric city of the
nineteenth century.
Why study the monocentric city? Although few oftoday's large cities are monocentric,_the analysis of this kind of city is important for four reasons. _First, the
rp.onocentric cily was the dominan!urban form until the early part of the twentieth
ceniurj; so urban history is largely a history of the monocentric city. Second, many
-of today.ssmall and medium-size cities are still monocentric. Third, to understand
the transition from the monocentric city to the modero city, one must understand
..the forces behind the development of the monocentric city in the first place. Fourth,
many of the lessons from the monocentric model can be extended to the modero
cities.
As we'll see, the technology of transportation is an importan!factor in determining urban fomi:. The urban transportation technology of the nineteenth century
is incorporated in four assumptions underlying the monocentric mbdel:
l. Central export node. Manufacturing firms export their output from the city
through a railroad terminal at the city center, the city's central export node.
-2.. Horse-drawn wagons. anufacturing firms use horse-drawn wagons to transport their freight from their factories to the railro.d terminal.
167
,,
-
.. ''
168
3. Hub-and-spke stretcar system. Workers travel by streetcar from the residential areas to their jobs in the central business district (CBD). The streetcar
lines are laid in a radial pattern; the lines form the spokes that lead into the hub
(the CBD).
4. Agglomeration economies. The office industry relies on face-to-face contact;
employees from different office firms meet in the city center to transact business.
D11 er these assurnptions, the city center is the focal point of the entire metropolitan
area:"Manufacturers are oriented toward the railroad terminal; office firms are ori-- ented-toward the cetihl market area; h.ouseholds are orienteltoward employment
i!l the central core area; and retailers are oriented to thehub of the streetcar system.
J-
How much is a door firm willing to pay for an acre ofland?It depends on how far
the land is from the railroad terminal. The closer the production site to the terminal,
the greater the potential profit from the production site, so the more the firm will be
willing to pay for land. If u is the distance (in miles) from the terminal, the firm's
profit is total revenue (Pd D) rninus nonland costs (Cd), freight cost (td. D. u) and
rent (R):
rrd(u) = Pd D- Cd- td D u- Rd(u)
The door market is perfectly competitive with free entry. Competition for land
among potential producers will drive up the price of land to the point at which
econornic profit is zero. In other
words, the price of land equals- the difference
--------- --------- ........ -
169
Freight
Cost
o
.,,
1
2
3
4
5
6
7
8
$ 200
400
600
800
1,000
1,200
1,400
1,600
Total
Revenue
Production
Cost
Production
Site (acres)
Pre-Rent
Profit
$3,000
3,000
3,000
3,000
3,000
3,000
3,000
3,000
3,000
$1,000
1,000
1,000
1,000
1,000
1,000
1,000
1,000
1,000
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
$2,000
1,800
1,600
1,400
1,200
1,000
800
600
400
-1
;'
.'1 -:
.., ''
Bid Rent
forLand
$2,000.
1,800
1,600
1,400
1,200
1,000
800
600
400
J ., ; . :
'
-
e J-
' .J
' '
;)" ,
-1
_ between total revenue and total nonland costs. This is the.leftover principie: After
a firrn uses its revenue to pay its other input suppliers, the landowner gets whatever
is left over. If a particular firrn offered to pay a landowner less than the entire gap
between total revenue and total nonland cost, the landowner could find another firm
to outbid the first firrn. In other words, the leftover principie results from competition
between potentialland occupants. To derive the expression for land rent, we simply
set the profit expression equal to zero and solve for land rent R:
We can use a simple numerical example to show how to compute the bid rent for land
by the door industry. Suppose each firrn produces 50 tons of doors, using one acre
ofland and $1,000 worth of nonland inputs. The price of doors is $60 per ton and
the unit freight cost is $4 per ton. InTable 8-1, for a production site adjacent to the
terminal (u = 0), the firrn has no freight cost, and is willing to pay up to $2,000 for
a one-acre site; total revenue ($3,000 =50 tons times $60 per ton) minus nonland
cost ($1,000) is $2,000. The firrn will be forced by competition from other firms to
pay this amount for land.
As we move away from the central export node, freight costs increase, decreasing
the amount a firrn is willing to pay for land. One block from the node, the freight
cost is $200 (50 tons times $4 per ton per block transported), so the firm is willing
to pay up to $1,800 for an acre of land. As the distance to the node increases, the bid
rent for land decreases by $200 per block, the freight cost per block. The leftover
principie tells us that this is nota fluke: the decrease in bid rent per block equals the
extra freight cost incurred as we move one block farther from the export node. The
slope of the bid-rent function is the freight cost per block, td D = $4 50 = $200.
In Figure 8-1, the linear curve is the bid-rent curve under the assumption of
fixed factor proportions. The slope is equal to the. freight cost per block, which in
.. our example is $200 per plock. The negative slopre ects tJ1e fact that freight costs
increase as a firrn movesaway from the export node.
170
...
.' . \ . 5,000 p
- :) .
,.'
'
._,'
i.0.'"'
r ..1
, '. '--
4,000
'
@
-o
......
'
- '
.,
c,l
171
o
1.400
1,200
g_1,000
.s
.
'
'C
. 800
:z:
',,
--.-
t--... '
-----
600
1_.:_,_,,
400
200
0.30
0.40
0.50
0.60 0.70 0.80 0.90
Land (size of production site)
1.00
1.10
The isoquant shows tbe different combinations of land and nonland inputs needed te
produce a given quantity of output As tbe firm moves toward tbe export nade and, tbe
price of land increases, so tbe firm moves up tbe isoquant, substituting nonland inputs
forland.
the isoquant to point j. the firm can produce the same quantity of output with less
land (0.90 acres) but more nonland inputs ($1,050). At the upper end ofthe isoquant,
the firm can produce the same quantity of output with 0.30 acres of land and $1,560
worth of nonland inputs.
.
For a firm that engages in factor substitution, both the quantity of land and
nonland costs can vary with distance to the expbrtnode (u), so we denote the amount
of land used as Td (u) and nonland costs as Cd (u). The new expression for profit is
rrd(u) = Pd D- Cd(u)- td D u- Rd(u) Td(u)
Competition for land among potential fums will bid up the price of land to generate
zero economic profit, so the expression for the bid rent for land is
Rd(u) =
Pd D- Cd(u) - td D u
Td(u)
Table 8-2 uses our numerical example to compute the bid rent for a firm that
engages infactor.substitution (a flexible firm). Ata distance of seven blocks, the
r:
f
172
Part 2
Bid Rent with Flexible Production
Distance to
Nonland
Terminal
Freight
Total
Production
(blocks)
Revenue
Cost
Cost
Patterns
TABLES-2
o
2
3
4
5
6
7
8
$ 200
400
600
800
1,000
1,200
1,400
1,600
$3,000
3,000
3,000
3,000
3,000
3,000
3,000
3,000
3,000
$1,560
1,450
1,350
1,260
1,180
1,ll0
1,050
1,000
950
Size of
Production
Site (acres)
Pre-Rent
Profit
BidRent
for Land
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1.10
$1,440
1,350
1,250
1,140
1,020
890
750
600
450
$4,800
3,375
2,500
1,900
1,457
1,113
833
600
409
:flexible firm uses the same input combination as the infiexible firm (one acre and
$1,000 of nonland inputs), so the bid rentis the same ($600 per acre). Consider what
happens when the :flexible firm moves .one block closer to the node. Freight costs
will be lower at the more proximate location, so competition among firms will bid
up the price of land. Recognizing this, the firm will substitute away from the more
expensive input (land), occupying a smaller site and substituting other inputs for
land. In Table 8-2, at a distance of six blocks, the firm will occupy only 0.90 acres
and use $1,050 of nonland inputs. This is a sensible substitution because the firm
saves about $80 in land cost (reducing land consumption by 0.10 acres when land
would cost $800 per acre) and increases its nonland cost by only $50.
In the last column ofTable 8-2, a move from seven blocks to six blocks increases
land rent from $600 to $833. In contrast, the same move by an in:fiexible firm would
increase the bid rent by only $200, the savings in freight cost from moving closer
to the export node. As we move toward the export node, the bid rent increases more
rapidly for the :flexible firm because such a firm economizes on land, substituting
other inputs as the price of land increases. Factor substitution increases profits, and
according to the leftover principie, higher profits translate into higher bid rents for
land.
Figure 8-1 shows the bid-rent functions of the :flexible and infiexible fums. The
:flexible bid-rent function lies above the infiexible one for alllocations except u = 7,
the location where the infiexible finn just happens to pick the input proportions
that a :flexible firm would pick. As the :flexible firm moves toward the export node,
it substitutes nonland inputs for land, moving up its isoquant as the price of land
increases. For each point on the bid-rent functiOn, the!abe!is matched with a similar
label on the isoquant: Point I on the bid-rent function is associated with point i on
the isoquant; point J is associated with point j; and so on. The movement upward
along the bid-rent function (higher price of land) is associated with a movement
upward along the isoquant (less !and and more nonland inputs).
The :flexible bid-rent function is convex because of factor substitution. The infiexiblefirm uses the same input combination at alllocations; so its bid-rent function
173
simply reflects differences in freight costs and is linear, with a slope equal to transport
cost per mile. In contrast, the flexible furo engages in factor substitution, generating
savings in both transportation costs and production costs as the firrn moves toward the
export node. As a result, the flexible bid-rent function is convex; the curve becomes
steeper as we approach the export node.
Who will occupy land, flexible firrns or inflexible ones? Landowners will rent
land to the highest bidder, and flexibility translates into lower production costs,
higher profits, and a higher bid rent for land. An inflexible firrn will get land
only if the firm is lucky enough to choose the efficient input combination. In general, inflexibility means inefficiency, so competition for land eliminates inefficient
firms.
The Bid-Rent Function of Office Firms
Although firms in the office sector provide a wide variety of goods and services,
office firms share two importan!characteristics. First they gather, process, and distribute inforrnation. Because inforrnation becomes obsolete quickly, office firrns
must be able to collect and distribute it rapidly. Second, office firrns rely on faceto-face contact in the collection, processing, and distribution of inforrnation. For
example, accountants explain and interpret the information in accounting reports.
The loan officers of banks meet with prospective borrowers to appraise their creditworthiness. The investrnent advisors of finance firms meet with clients to assess their altitudes toward risk and their investrnent inclinations. In general, office
firms rely on speedy face-to-face contacts in collecting and distributing inforrnation. In contras!to door makers, who throw their doors into the back of a horsedrawn wagon, office firms transrnit their output in the minds and briefcases of their
employees. 'Suppose that office firrns in the city provide financia! services. The industry has
the following characteristics.
l. The office. Every finance firm is based in an office. The "output" of the firm is
financia! consultations, and each firrn produces F consultations per month.
2. Travel to city center. The employees of finance firrns travel from the office to
the city center (the hub of the streetcar system) to consult with clients. Each
consultation requires one trip to the city center.
3. Fixed price. The price of consultations (P 1) is fixed.
4. Competitive markets. There is free entry into the finance industry, so each
firrn makes zero econornic profit.
5. Factor substitution. Finance firrns engage in factor substitution; as the price of
land increases, they substitute nonland inputs for land. The nonland production
cost ofthe office (capital and labor cost) is C (u) and varies with the price of
land and distance to the city center.
The travel cost of a finance firm is determined by the opportunity cost of workers'
travel between
the office and
the clients in--the city center. Suppose a worker requires
..--- '.
.
174
Part 2
t minutes to walk one round-trip block, and the wage is W per minute. If the finn
provides F financial consults per month, the travel cost for a location u blocks from
the city center is
tWFu
For exarnple, if t = 3 minutes per block, W = $4 per minute, and F = 200 consultations per month, the monthly travel cost is $2,400 for a location one block
from the city center. The travel cost increases by $2,400 per block, to $4,800
for a location two blocks from the city center, $7,200 for three blocks away, and
so on.
The firm's total profit is total revenue less nonland production cost, travel cost,
and land rent. Ifthe finance industry is perfectly competitive, economic profit is zero,
and land rent equals the gap between total revenue and nonland costs (the leftover
principie). We can write the finance furo's bid rent for landas
P F- C(u)- t W F u
R(u) =
T(u)
.
The only real difference between this expression and the expression for the door
firm is in the transportation technology; the transport cost per mile of the finance
furo depends on the opportunity cost of the firm's workers (the wage).
Figure 8-3 shows the bid-rent function of finance furos. The bid-rent function
is negatively sloped because travel cost increases as the furo moves away from the
city center. The bid-rent function is convex because office furos engage in factor
substitution. As the firm approaches the city center, it substitutes nonland inputs
(capital and labor) for tlie relatively expensive land, producing the sarne output with
. less land and more of its nonland inputs. In other words, office furos near the city
center occupy taller buildings.
175
$6,000
"
&=
4,000
::;
t\
2,000
3
4
5
6
7
Distance from export nade (blocks)
. 8
The office land-rent function is negatively sloped because travel costs increase as
distance te the central marketplace increases. It is c01ivex because firms substirute
nonland inputs for landas the price of land increases.
..:..;._ ..
. .(
more centralland. The slope of the bid-rent function is deterrnined by transportation costs, and the finance industry has higher transportation costs because it uses
people-with high opportunity costs-to transrnit output (at $2,400 per block in
our exarnple). In cqntrast, the door industry transports its output with horsedrawn wagon, so its transport costs are relatively low ($200 per block in our
exarnple).
.
Does the land market allocate land ef:ficiently? In the terrns used by land developers, is land allocated to its "higheSt and best use"? The ffice industry, with its
higher transportation costs, occupies the land closest to the city center. This allocation is ef:ficient because the offi.ce industry has the most to gain from proxirnity to
the city center. To explain, suppose that a finance firm one block from the city center
swaps locations with a manufacturer five blocks from the city center. The land swap
increases the finance firm's travel costs by $9,600 (four blocks times $2,400 per
block), but decreases the door firrn's freight costs by only $800 (four blocks times
$200 per block). Office travel costs increase by more than freight costs decrease, so
total transportation .costs increase. The market allocation, which gives centralland
to the office industry, conornizes .on transportation costs..
176
FIGURE 8-4 Bid-Rent Functions and Land Use in the Central Business District
$
1 "
1
1
1
1
Office
Q Manufacturing
Tbe office industry has a relatively steep bid-rent function because !he travel cost of people exceeds
!he travel cost of freight, The office industry outbids manufacturers for laod near !he city center.
Centralland is occupied by !he activity with !h.e most to gain from proximity (decreased
i:ransportat{on costs).
r
1
177
178
0.30
Housing-price function
1
1
1
1
1
1
1
1
0.06
----------t---------1
1
1
1
1
1
1
12
15
per square foot. At a distance of six miles from the center, co=uting costs are
$120, so the household has $180 Jeft to spend on housing (18 cents per square foot).
In Figure 8-5, the slope of the function is two cents per mile.
The negatively sloped housing-price function is necessary for locational equilibrium, a situation in which no household has. an incentive to change its residentia!
location. To see this, suppose that the fiinction starts out as a horizontalline. If the
price of housing is 15 cents per square foot throughout the city, a household can get
a 1,000-square-foot dwelling anywhere in the city for $150 per month. A household
that starts out in a dwelling 1O miles from the city center has an incentive to move toward the city center because co=uting costs would be lower, but rent would be the
same. Other households have the same incentive to move closer to the center. As the
demand for housing near the city center increases, the price ofhousing near the center
increases; as the demand for suburban housing decreases, the price of suburban housing decreases. In other words, the movement of households toward the city center
transfotms a horizontal housing-price function into a negatively sloped function.
179
-.f
Good news: Commuting costs decrease by the change in distance times the
commuting cost, or -th t>u.
Bad news: Housing costs increase by the change in the price housing consumption, or f>Ph H.
.t{
''
' 1 ' .
'
'
The household will be indi:fferent between the two locations if the good news equals
the bad news, that is if
-th t>u = f>Ph H
Por example, suppose the move is just one mile: t,u = l.If th = $20 and H = 1, 000
square feet, the household will be indifferent between the two locations if the price
ofhousing increases by two cents per square foot. In Figure 8-5, the price of housing
increases by two cents per mile as the household moves toward the city center.
1(
-.
.l.
..'(
;{
.1
]
l
J
.1.
.{
1
1
:(
t
f
.(
'.
-.
'
1
1
J
180
..'
,'""
r
,-
.o
S 0.30
":;r
..
&
o;
j"j
..o..
"'
0.12
1
1
1
1
1
1
1
12
household substitutos other goods for housing, "the housing-price function is convex,
not linear.
Assumed Corisumption Pattern:
Distance to city c nter (miles)
Housing consumption (square feet)
3
400
600
750
12
1,000
can be rewritten as
-th /::,.u = t::,.ph H (u)
Ata given location (u), the change in cornrnuting cost (the change in u times the
transport cost per rnile) equals the change in the housing price times housing consumption. The equation can be rearranged to show the slope of tbe housing-price
function:
t:,.ph
t::,.u
th
H(u)
--=--In the numerical example, t is $20 and H(9) is 750, so the slope of the housingprice function at u= 9 is $0.0267 (20/750), compared toa slope of $0.02 under the
assumption of :fixed housing consumption. Since H (6) is 60, the slope at u = 6 is
$0.033 (20/600). As thehousehold moves toward the city center, housing consumption decreases, increasing the slope of the housing-price function.
181
. How rapidly does the price of housing decrease as distance to the city center
mcreases? The housing-price gradient is defined as the percentage chano-e in the
price of housing per mil e. Dividing both sides of the slope equation by Ph
!:::.Ph 1
th
!:::.u Ph
Ph H(u)
---=
In words, the housing-price gradient equals transport cost per mile divided by housing
expenditures. If the full cost of commuting (including monetary and time costs) is
$1 per round-trip rnile, the monthly commuting cost (for 20 workdays per month)
is $20 per round-trip mile. If the household spends $500 per month on housing, the
rent gradientis 4 percent per mile (20/500).
Consider first the possibility that housing is produced with fixed factor proportions. Each housing firm produces Q square feet of housing, using one acre of !and
and $K worth of capital. Once the firm erects a building, it can be used as a single
dwelling (with Q square feet of space), or divided into x units, each of which has
(Q/x) square feet ofliving space. For example, a building with 10,000 square feet
could be divided into 10 units, each with 1,000 square feet ofliving space.
According to the leftover principie, the bid rent for land equals the excess of total
revenue ov.er total nonland cost. In Figure 8-7, the total-revenue curve is negative!y
sloped because the price of housing decreases as distance increases. It is convex
because the housing-price function is convex, refiecting consumer substitution as
the price of housing changes. The residential bid rent equals the gap between total
revenue and nonland costs (K):
Rh(u)
= Ph(u) Q-
At u*, total revenue equalsnonland cost, so the bid rent for land is zero. The bid-rent
function is convex because the housing-price function is convex.
The bid-rent functionshown in Figure 8-7 is based on the assumption that
housing is produced with fixed factor proportions. Housing firms use the same
input combination at al1 locations, regardless of the price of land. In fact, as the
price of residentialland increases, housing producers substitute capital for land,
building houses closertogether or building taller apartrnent complexes. This response
decreases production costs and allows housing firms to pay more for land.
Factor substitution in the production of housing increases the convexity of the
bid-rent function. Recall that in the case of rr..anufacturing firms, factor substitution
transformed a linear bid-rent function into a convex one. The residential bid-rent
function is copvex because of consumer substitution, and the introduction of factor substitution makes it eve more convex. To summarize, the residential bid-rent
182
u
Miles to city center
The bid rent of the housing :firms equals total revenue per acre less the cost of Mnland inputs.
Total revenue (!he price of housing times square footage produced) decreases as the distance
to the city center increases because the housing-price function is negatively sloped. The cost
of nonland inputs is the sameat alllocations. The bid-rent function is convex because the
housing-price function (and the revenue function) is convex. At u, the cost of nonland inputs
equals total revenue, so the bid rent equals zero.
function is convex because ofboth consumer substitution (which malees the housingprice function convex) and factor substitution (which increases the convexity of the
rent function).
Residential Density
How does population density vary within the monocentric city? Density decreases
as distance to the city center increases for two reasons.
183
10-story apartment buildings, meaning that there is about 0.1O square feet of
land for each square foot of hous.g. Iri contrast, suburban housing could be on
lots with twice the square footage of the house.
Putting these two factors together, the city-center resident uses about 50 square
feet of land, while the suburbanite uses about 4,000 square feet. Therefore, in this
exarnple, population density is 80 times higher in the central city.
Activities are arranged according to their transportation costs: the higher the
transportation cost, the closer to the city center. As explained earlier, the activity
with relatively high transport costs has a relatively steep bid-rent function, and thus
locates closer to the marketplace. In the monocentric city, the market is the city
center, where office workers meet with clients and manufacturers load their output
onto ships or trains. The office sector, with the highest transport costs and thus the
steepest bid-rent function, occupies land closest to the center. Manufacturing, with
the next highest transport costs and thus the next steepest bid-rent function, occupies
the next ring of land. The residential sector, with relatively low transport costs and
thus a relativi:ly fiat bid-rent function, occupies the land farthest from the city center.
This spatial arrangement has two interesting features. First, offi.ce firms occupy
the central area of the CBD. As explained earlier in the chapter, office firms have relatively high transport costs and thus a relatively steep bid-rent function because office
output is transmitted by office workers, while manufacturing output is transported
by horse-drawn wagons.
The second feature of the monocentric city is that employment is concentrated in
the CBD, not distributed throughout the city. Why do all the manufacturers and office
firms 1ocate in the CBD? To explain this monocentric location pattern, consider a
door firm that is considering a move from the CBD to a suburban location. What are
the trade-offs associated with a move to the suburbs?
Higher freight costs. The firm will be farther from the central export nade, so
it will pay higher freight costs.
Lower wages. The firm will be closer to its workforce, so workers will commute
shorter distances. The wage compensates workers for commuting costs: the
longer the commuting distance, the higher the wage. When the firm moves
closer to its workforce, it decreases its workers' commuting costs, so the firm
can pay a lower wage.
184
FIGURE 8-"8 Bid-Rent Functions and Land Use in the Monocentric City
$
.----1
Residential bid-rent function
Agricultura! bid-rent function
u,
1
1
Office district
Manufacturing district
Residential district
185
'
\
Commuting Assumptions
Our simple model ofurban land use is based on a number of simplifying assumptions
about comrnuting. What happens ifthese assumptions are dropped?
l. Time cost of commuting. The model assumes there are no time costs of commuting, only monetary costs. Because comrnuting time comes at the expense of
work or leisure, there is an opportunity cost associated with comrnuting. Studies of
comrnuting behavior suggest that most people value comrnuting time at between
one-third and one-halfthe wage rate. The higher the opportunity cost of comrnuting,
the steeper the housing-price function and the residential bid-rent function.
2. Noncommuting travel. The simple model assumes that noncomrnuting travelfor shopping, entertainment, and other activities-is insignificant. If shopping and
entertainment destinations are distributed uniformly throughout the urban area, this
assumption migll.t be harmless; a move in one direction would el.ecrease travel costs
to sorne destinations but decrease travel costs to others, with a small net change
in total travel costs. In contras!,if noncomrnuting destinations are concentrated in
one particular location, for examp!e, the city center, a move toward the center will
18.6
decrease the cost of co=uting and other travel. This would make both the housingprice function and residential bid-rent function steeper.
3. Two-earner households. The simple model assumes that a single person from
each household commutes tothe city center. A switch to two-earner households
would double the savings in co=uting costs from moving toward the city center,
making the housing-price function and residential bid-rent function steeper.
187
1
2
3
4
5
6
7
0.12
0.10
0.08
0.06
0.04
0.02
0.01
240 .
200
160
120
80
40
20
40
40
40
40
40
40
40
24
20
16
12
8
4
2
20
20
20
20
20
20
20
centrallocations provide the best trade-off for the poor, while suburban locations
provide the best trade-off for the wealthy.
Consider a household that is considering a move from a centrallocation to a
location further from the city center. An outward move of one mile generates good
news and bad news.
Bad news: Comrnuting costs increase by t, the comrnuting cost per mile.
Good news: Housing costs decrease because housingis less expensive farther
from the city center. The savings in housing cost is equal to the change in the
price of housing per mile times housing consumption.
Table 8-3 shows a numerical example that supports the traditional theory of
income segregation. The second column shows the change in the price of housi..ng
per mile. Going from the city center, the price drops by $0.12 for the first mile,
$0.10 for the second mile, and so on. If a high-income household consumes 2,000
square feet of housing, a move from the city center to one mile out saves $240 in
h_9using costs ($0.12 times 2,000 square feet ofhousing). As shown in colurnn C, the
marginal benefit ofmoving out decreases, to $200 for the second mile, $160 for the
third mile, and so on, .If a one-mile move outward increases comrnuting cost by $40
(shown in colurnn D), the optimum location (where Marginal benefit = Marginal
cost) is 6 miles.
Suppose there is a low-income household with one-fourth the income of the
high-income household. Suppose the low-income household consumes 200 square
feet ofhousing. As shown in column E, the marginal benefit ofmoving out decreases,
from $24 for the first mile ($0.12 times 200 square feet), to $20 for the second mile,
$16 forthe third mile, and so on.If a one-mile move outward increases the comrnuting
cost of a low-income household by $20 (shown in column F), the household will
live 2 miles from the city center, where Marginal benefit = Marginal cost. Figure 8-9 shows the marginar benefit and marginal cost curves for both households
and the chosen Jocations.
The benefit and cost curves in Figure 8-9 are drawn under a key assumption
about .the responsiveness of housing consumption and comrnuting cost with respect
---,.?
i
188
200
!50
100
50
Marginal cost .
for low income
households
Marginal
high-income
households
If tbe income elasticity ofhousing is !argertban tbe income elasticity of commuting costs,!ow-income
households willlive close to tbe city center, while high-income households willlive fartber away.
189
200
Marginal benefit for
high-income bousebolds
150
100
Marginal benefit for
low-income housebolds
50
7
Ji the income elasticity of bousing is the same as the income elasticity of commuting costs,
low-income bouseholds and high-income households will cboose the same locations. In this example,
eacb cbooses a location six miles from the city center.
Other Explanations
Sorne altemative explanations of income segregation are based on factors excluded
from the sin;J.ple'monocentric model.
l. New suliurban housing. Suppose that the utility generated from a particular
dwelling decreases over time as the dwelling deteriorates and becomes obsolete. In
other words, people get less utility out of ar1 older house because it is less fashionable,
has higher maintenance costs, and is equipped with fewer modero gadgets. The
wealthy, who demand high-quality housing, occupy new housing instead of used
housing, As an urban area grows, it expands outward; and developers build new
housi.p.g for high-income J:iouseholds in the peripheral areas. The poor are left with
old houses in the central city.
2. Fleeing central-city problems. As explained later in the book, poverty contributes to three urban problems. First, crime rates are higher among the poor, in
part because the poor face a relatively low opportunity cost of comrnitting crime.
Second, fiscal problems are more likely in ajurisdiction with a large fraction of lowincome citizens. Third, students from poor fanrilies have relatively low achievement
levels and pull down the achievement levels of other students. To escape these problems, wealthy households fiee to the suburbs, leaving large concentrations of poor
households behind.
3. Suburban zoning. As explained in Chapter 10 (Land-Use Controls and
Zoriing), suburban governmeilts use zoriing to exciude low-income households.
190
Therefore, only the wealthy have the opportunity to escape the problems of the
central city.
!:::.u
H(u)
191
YiC\llffi S-11
u
Miles from city center
The wealthy bave a flatter bid-rent function, so the poor occupy central !and (land
less than u miles from the city center). The wealthy have a flatter bid-rent
function because they are sensitivo to crime, pollution, and the quality of schools,
. and the central cities have more crime and pollution, and inferior schools.
\
the wealthy have a flatter housing-price function and a flatter residential bid-rent
function. This is the traditional theory of income segregation.
Wheaton's results suggest that the slope ofthe housing-price function is roughly
inv.ariant with respect to income. If H and t increase at the same rate as income
increases, the two housing-price functions have the same slope, so the two residential
bid-rent functions have the same slope, Therefore, ifhigh-income households have a
relatively flat bid-rent function, it is not because of the trade-off between commuting
costs and housing costs.
The alternative explanations of income segregation suggest that the slope of
the residential bid-rent function is affected by other factors. Specifically, if central
cities have higher taxes, inferior schools, and more pollution and crime, households
are willing to pay more for housing and land in the suburbs. In other words, the
problems of the central city decrease the slope ofthe bid-rent function.If the income
elasticities of demand for safety, clean air, and education are relatively large, the
bid-rent function of wealthy households will be flatter !han the bid-rent function of
poor households. In other words, if the wealthy are willing to pay much more than
the poor for safety, clean air, and superior education, wealthy households will outbid
192
poor households for land in areas that are relatively safe and clean and provide highquality education. Wasylenko (1984) summarizes the empirical evidence supporting
these altemative explanations of income segregation.
Policy Implications
These altemative theories of income segregation suggest that public policy can affect the location choices of wealthy and poor households. A housing policy that encourages the renovation of central-city housing stock may cause sorne high-income
households to retum to the central city. Policies that decrease poverty decrease crime
rates, reduce fiscal problems, and improve central-city schools, encouraging highincome households to live in the central city. Similarly, policies that address the
crime and education problems directly increase the relative attractiveness of centralcity locations. Finally, policies that control exclusionary zoning allow the poor to
move to the suburbs.
193
Business bid-rent
function
Labor supply
1lo
Residential bid-rent
function
9
Distance from city center
:o":
10 --------Labor demand
100
Number of workers (1,000)
1
1
' '
j)
A'
The left panel ofFigure 8-12 shows the urban land market. The business bid-rent
function intersects the residential function at a distance of two miles, generating a
rectangulai..CBD with a width of two miles. This is the territory of the demand side
of the labor market; to compute the total quantity of labor demanded, we compute the
land area of the CBD and then multiply the territory by the density of employment
(workers per square mile). The residential bid-rent function intersects the agricultura!
bid-rent function at nine miles, so the residential area is a rectangle with a width of
seven miles. This is the territory of the supply side of the labor market: To compute
the total quantity of labor supplied, we compute the land area of the CBD and then
multiply the territory by the density of employment (workers per square mile).
The rightpanel ofFigure 8-12 depicts the urban labormarket, with a negatively
sloped demand curve anda positively sloped supply curve.
Why is the demand curve negative!y sloped? An increase in the wage increases
production costs and, according to the leftover principie, decreases the amount
businesses are wi!ling to pay for land. As a result, firms will be outbid for land
by other land users, decreasing the territory of the business sector and thus
decreasing the quantity of labor demanded.
Why is the supply curve positively sloped? An increase in the wage increases
the amount households are willing to pay for land. Households will outbid other
Jandusers,and as the territory ofthe residential area increases, the quantity of
labor supplied will increase.
194
In equilibrium, the wage is at a leve!($10) such that the quantity of labor demanded
(from the business district) equals the quantity supplied (from the residentia!district),
generating an equilibrium with 100,000 workers.
The Effects of a Streetcar System
Consider the effects of introducing a streetcar system into the monocentric city. In
the left panel of Figure 8-13, the streetcar decreases the unit commuting cost, so it
tilts the residential bid-rent function outward. The residential area expands into the
previously agricultura! area, increasing labor supply. In the right panel, this is shown
as a rightward shift of the labor supply curve. From the right panel, we know that
there will be an excess supply of labor at the original wage, so the wage will fall. As
the wage falls, there are two changes in the urban land market.
The residential bid-rent function shifts downward in a parallel fashion.
The business bid-rent function shifts upward because the decrease in wages
decreases production costs, increasing the amount firms are willing to pay for
land (the leftover principie).
These changes in the land market help close the gap between labor supply and labor
demand. The downward shift of the residential function decreases the territory of
FIGURE 8-13 Initial Effects ofthe Streetcar
Newlabor
supply
function
"..10
.
--------.
:
1
1
1
------
Labor demand
2
. 1
1
17
1
1
100
Number ofworkers (1,000)
Tbestreetcar decreases the unit cost of connuting, tilting the residential bid-rent function outward
and increasing the residential tertitory. The labor-supply curve Shifts to the right, causing excess
supply of labor.
195
Labor demand
function
3
13.5
: Distance from city cente: 1
1
150
Number ofworkers (1,000)
Equilibrium is restored with a larger business district (3 miles, up from 2 miles) and a larger
residential district (10.5 miles, up from 7 miles). The equilibrium wage is lower ($8, down from $10)
and the equilibrium number of workers is larger.
the residertiaf sector, decreasing the quantity of labor supplied. The upward shift
of the busi.J.ess function increases the territory of the business sector, increasing the
quantity of labor demanded.
Figure 8-14 shows the new equilibrium. In the left panel, there is a larger
business district (3 miles wide instead of2) anda larger residential area (10.5 miles
wide instead of 7). In addition, land rent is higher at alllocations in the city. The
streetcar increases accessibility in the residential district, and increases residen tia!
land rent as a result. There is an indirect effect on the business district; land rent is
higher there because wages are lower.
Does the streetcar make residents of the city better off? Although commuting is
more efficient in the streetcar city, residents are no better offbecause wages are lower
and land rent is higher. In a small open city, any initial advantage associated with
living in the streetcar city is quickly eliminated by in-migration, which continues
until the bene:fits of the streetcar are exactly offset by lower wages and a higher cost
ofliving.
Changes in Density
So far, we have assumed that both employment density and residential density are
:fixed. The iritroduction of the streetcar affected the territories of both sides of the
196
labor market, but did not affect the number of workers per square mile or the number
of residents per square mile. How would the analysis of the streetcar change if
densities were variable instead of fixed?
The introduction of the streetcar will cause changes in density that reinforce
the changes in the territories of the two sides of the urban labor market. Whenever
land rent increases, density will increase too; households will econornize on land,
living on smaller Jots; firms will econornize on land by producing on smaller production sites. Recall that the initial effect of the streetcar was an outward tilt of the
residential bid-rent function: Land rent is higher at alllocations, and in addition to
increasing the residential territory, density will increase too. Therefore, labor supply
increases as both the size and the density of the residential district increases. The
consideration of density changes means that the labor supply curve will shift by a
larger amount.
What about the wage effects of the streetcar? Recall that the excess supply of
labor decreases the wage, shifting the residential bid-rent function downward and the
business bid-rent function upward. The downward shift of the residential function
will decrease both the territory and density of the residential district, decreasing
the quantity of labor supplied and helping to close the gap between supply and
demand. Sirnilarly, the upward shift of the business bid-rent function will increase
both the territory and density of the business district, increasing the quantity of labor
demanded. In summary, the changes in density reinforce the changes in territories,
helping to clase the gap between labor supply and labor demand.
SUMMARY
l. In the monocentric city, manufacturers export their output in horse-drawn wagons to a central export node; office workers travel by foot from offices to a
central market area to exchange information; commuters and shoppers travel on
a hub-and-spoke streetcar system.
2. The bid-rent functions of manufacturers and office firms are negatively sloped
because of transportation cost, and convex because of factor substitution.
3. The office sector has higher transportation cost per mile, so its bid-rent function
is relatively steep and office firms occupy the central area of the city.
4. The housing-price function is negatively sloped because commuting costs increase with the distance to the city center and is convex because of consumer
substitution.
5. The residential bid-rent function is negatively sloped because the housing-price
function is negatively sloped, and is convex because of consumer substitution
(which makes the housing-price function convex) and factor substitution.
6. Employment is concentrated in the CBD because the cost of commuting from
the suburbs to the CBD (via streetcars) is low relative to the cost of moving
output from the suburbs to the city center.
7. In U.S. cities, the wealthy tend to locate in the suburbs, and the poor tend to
locate near the city center. This pattern is explained by the demand for new
r
'
197
'
i
1
l. In the city of Trekburg, manufacturers have two options Jor intracity freight.
They can use a conventional transportation system (the truck) ora matter transrnitter, which instantly transports the outputfrom the factory to the central export
node ("Beam it over, Scotty"). A transrnitter can be rented for $C per year, and
running the machine is costless. The transrnitter can transport output up to a
distance of two miles. All manufacturing output goes through the export node.
a. Draw the bid-rent function for a firm that uses the matter transrnitter, and
label itM.
b. On the same graph, draw the bid-rent function for a firm that uses the truck,
and label it T.
c. Will every manufacturer use the matter transrnitter? If not, where will the
firms using the truck be located?
2. Consider a traditional nineteenth-century monocentric city with a CBD radias of one mile. In 1869, buildings at the edge of the CBD are four stories
tall. In 1870, all the buildings in the CBD are destroyed by an earthquake,
and the mayor of the city announces that the maximum building height in
the rebuilt city will be four stories. In addition, business development will
be confim!d to a circle with a one-rnile radias; that is, the size of the CBD
is fixed. Suppose that the city is small enough so that events in the city do
not affect the equilibrium prices of its export goods (office or manufacturing
goods).
a. Draw the business bid-rent function befare the earthquake and label it R.
Draw the business bid-rent function after the earthquake (with the height
restrictions) and label it R.
b. Explain any differences between the two bid-rent functions.
3. Consideran office firm with the folfowing characteristics: The wage of executives is $120 per hour, and theexecutive takes four minutes to walk one block
(eight minutes to make a round trip); the price of output is $150, and the firm
produces 50 consultations (requiring 50 trips to the city center); at a location
four blocks from the city center, the firm occupies a one-acre site and spends
$1,000 on nonland inputs.
a. What is the travel cost per block?
b. How much is the firm willing to pay for land four blocks from the city center?
c. Given the available information, is it possible to compute how much the
firm. is willing to pay for)and one :block from the city center? If not, what
additional information do you need?
198
4. Consider.two monocentric cities: Rigid City, where of:fice firms produce with
fixed factor proportions, and Flexville, where office firms produce with variable
factor proportions. In each city, the CBD is a circular area with a radius of one
mile, and allland in. the CBD is used for of:fice space. At the edge of the CBD
there is 5,000 square feet of office space per acre, and the bid rent for office
land is $20,000 per year. Suppose that each city irnposes an annual tax of $1
per square foot of of:fice space, paid in legal terms by office firms. Assume that
the wage and the price of of:fice output are fixed. For each city, draw the office
bid-rent function before and after the new office tax. Provide numbers for the
bid rents (pretax and posttax) at the CBD edges. Explain any differences in the
effects of the tax in the two cities.
S. Depict graphically the effects of the following changes on the division of CBD
1and between office firms and manufacturers:
a. The unit freight cost decreases.
b. The price of of:fice output increases.
c. The opportunity cost of executive travel decreases.
6. Consider an industry that makes table tennis balls and competes with door
makers for land near the central expo:it node. Each table tennis ball firm produces
the same quantity ofoutput as a door.firm (five tons), sells for the same price
($160 per ton), and has the same production isoquants. Which activity willlocate
closer to the export node?
7. Consider a monocentric city in which the unit cost of commuting is $10 per
mile per month. A household located eight miles from the city center occupies
a dwelling with 1,200 square feet ata monthly rent of $600. Nonland cost per
dwelling is $200, and there are four houses per acre.
a. What is the price(per square foot) ofhousing at u= 8? What is the bid rent
at u= 8?
b. Assume that the demand for housing is perfectly inelastic. What is the price
of housing at u = 5?
c. Assume that housing firms do not engage in factor substitution. What is the
bid rent at u = 5?
d. How would your answers to (b) and (e) change if the demand for housing
is price-elastic and firms engage in factor substitution? Would the prices of
housing and land be larger or smaller?
8. Choose one word in each set of parentheses to make the following statements
correct, and then explain your choice of words.
a. The fiatter the demand curve for housing, the (more, less) curvature in the
housing-price function. In other words, the fiatter the housing demand curve,
the (more, less) convex the housing-price function.
b. The fiatter the demand curve for housing, the (more, less) curvature in the
residential bid-rent function.
9. Considera region with two cities: Lawland (L) and Violateville (V). The two
cities differ in their demand curves for housing; consumers in Lawlandhave negatively sloped demand curves; consumers in Violatevjlle have positively sloped
199
200
15. Suppose that the intracity truck is introduced into the traditional monocentric city, decreasing the cost of intracity freight. Assume that the city remains
monocentric, with all freight being exported through the central export node.
Depict graphically the effects of the intracity truck on the urban land market
(the business bid-rent function} and the urban labor market (labor supply and
demand).
16. Consider the city of Swampville, a city that recently drained a swamp near its
city center and thus increased the supply of developable CBD land by 1O percent.
Predict the effects of the draining of the swamp on
a. Land rent in the CBD.
b. Land rent in the residential district.
c. Employment density.
d. Residential density.
e. The equilibrium wage.
f. Total employment.
17. Relax the assumption of fixed output prices and depict graphically the effects
the height restrictions on the market. for the city's export goods. What are the
implications of the change in output prices on the business bid-rent function? a.
Land rent in the CBD and the residential district.
b. The demand for labor.
c. The supply of labor.
d. The equilibrium wage.
e; Residential density.
f. The city radius.
18. Suppose a city restricts the height of its residential structures. The maximum
building height ifor stories, the height that would normally occur ata distance
of five miles froiJ.. the city center.
a. Draw two residential bid-rent functions, one for the city in the absence of
height restrictions and one with height restrictions.
b. Discuss the effects of the height restrictions on the urban labor market.
c. How do the height restrictions affect land rent in the CBD?
19. Suppose a city imposes a gas tax of $1 per gallon and uses the increase in tax
revenue to decrease its income tax. The tax liability of the typical residen!is
unchanged by the change in tax policy.
a. Draw !Wo residential bid-rent functions, one before the gas tax and one after
the tax is imposed.
b. Discuss the effects of the gas tax on wages, total employment, and land rent
in the CBD.
20. Suppose a city builds a streetcar system and passes a law that prevents any outsiders from moving to the city. In other words, the city keeps its population at its
pre-streetcar leve!, so it is a "closed" city instead of an open one. Depict graphically the partial-equilibrium and general-equilibrium effects of the streetcar in
the closed city.
201
CHAPTER 9
Today's cities are very different from the monocentric cities ofthe 1800s and early
1900s. While jobs in the monocentric city were concentrated in the central core area,
jobs in fue typical modero city are spread throughout the metropolitan area. In the
largest 100 metropolitan areas in the United States, the median workplace is about
seven miles from the city center, and only 22 percent offobs are within three miles of
the center (Glaeser, Kahn, and Chu, 2001). This chapter explores the economic forces
behind the transition fr,om monocentric cities to modero cities, where employment
is split..between the central business district, suburban subcenters, and other sites
dispersed throughout the suburbs.
We start with the facts on the spatial distribution of employment and population
in modero cities. Then we look at the economic forces responsible for the decentralization of population, manufacturers, and offices. Finaly!, we take a elose look at
the development of suburban subcenters in Chicago, Los Angeles, and Houston.
45
45 --
45
,..,
r-
,.
/
',
... 40 t-
Tl
'
35 t.
'S
"1l
"'
29
30 --
,.2.6..,
'!< 25 -
ao
22
21
,----,
20
19
--
20
19
-
,-
e---
.
'S 15 -
""
11
10 1
,..9..,
1. .. '
..
1.
_l
_l
+" o
<:,'?'
#
o;'
iS'
cp<>
_1_
&'-
.$'1>
- .r
__l
' "'1>
_l
""
.._o'>
o
r-
..
1
1 .
L
_1_
,..
-...'l.J'?
$-"'"
_l
.,
&
"{o?;>"'
....'.b-
r,V
(1'>
e,">
Yo"''('
""
"1,
.,
Source: EdwfU"d, Glaeser, Mauhew Kalm, and Chenghuan Chu, "Job Sprawl: Employmenl Locmion in U.S. Metropolitan Areas," Brookings lnstitution Survey Series,
--,
<_">" .
"{
s"'"'
N
_]_
4,cfr
<!
_l
..
,----,
9"-s
VJ
r::
204
1948
!IJ1990
92
85
_)
,
.
75
70
.... 60
!::
8
;;=
"
50
,,
40
30
20
JO
oL
L__
Population
Source: 1948 data from John F. Kain, "The Distribution ad Movement of Jobs and Industry;' in The Merropoliran
Enigma, ed. James Q. Wilson (Cambridge, MA: Harvard Unive"ity Press, !968); 1990 data from U.S. Bureau of
the-Census, Joumey ro Work in the UnitedSra:es (Washington, DC: U.S. Govemment Printing Office), 1994.
seven miles). While 22 percent of jobs are within three miles of the center, about
20 percent of metropolitan residents live within three miles of the center.
Figure 9-2 shows the distribution of population and employment between central
cities and suburbs 1n 1948 and 1990. The percentage of the metropo!itan population
in central cities droppedfrom 64 percent in 1948 to 39 percent in 1990, and the
percentage of manuf.cturing employment dropped from 67 percent tci 45 percent
The losses in trade and service employment were even larger.
Table 9-1 provides a closer look at the distribution of employment between
central cities and suburbs. The employment sectors are listed in declining order
of importance. Three sectors had more than half of metropolitan employment in
central cities: public administration (62 percent), transportation, comrnunications,
and utilities (56 percent), and finance, insurance, and real estate (58 percent). The
three service sectors (professional and related services, business and repair services,
and personal services) had between 52 percent and 57 percent of their employment
in central cities.
Ta:ble 9-2 shows comrnuting patterns in U.S. metropo!itan areas. As shown in the
first row, about 40 percent ofmetropolitan workers comrnuted within suburban areas,
more than twice the percentage that commute from suburbs to central cities. The other
rows show commuting pattems within conso!idated metropolitan statistical areas
(CMSAs). In these large metropo!itan areas, the percentage of workers comrnuting
within suburbs is even higher. In CMSAs in the Northea3t, the percentage commuting
within the suburbs is more than four times the percentage commuting from the
suburbs to. the central city.
205
TABLE 9-1 Distribution of Emp1oyment between Central Cities and Suburbs by Type
ofEmp1oyment, 1990
Ernployment in Employment in Percent in Percent in
Central Cities Central Cities Suburbs
MSAs
Total
Professional and re1ated services
Manufacturing
Retai1 trade
--Finance, insurance, real estate
Other industries
PublicadrrWllstration
Who1esale trade
Personal services
93,117,895
21,648,685
15,586,909
15,351,455
6,993,936
6,748,723
5,649,399
4,782,462
4,672,853
4,489,663
4,308,269
2,885,541
51%
57
45
48
58
56
45
52
41
62
49
54
47,861,224
12,362,153
6,938,010
7,352,682
4,066,519
3,767,546
2,514,248
2,508,616
1,919,649
2,780,672
2,105,851
1,545,278
49%
43
55.
52 .
42
44
55 --'
48
59
38
51
46
Source: 1990 Census ofPopulation and Housing, Joumey ro Work in rhe Unired Srares rashington, DC: U.S. Government
Printing Office, 1994).
. .. :)
. '' .._:-'.:...
-
Central City to
Central City
Central City
to Suburb
Suburb to
Central City
32.5
30.7
39.4
23.8
20.6
31.0
7.6
8.0
4.5
8.2
9.6
10.0
19.8
16.3
9.9
16.6
28.6
19.2
Suburb to
Suburb
40.1
45.1
46.3
51.4
.41.2
39.8
Source: 1990 Census ofPopulati.on and Housing, Jov.mey ro Work in the United Srares (Washington, DC: U.S. Govcmment
Printing Office, 1994).
1
J
l
1
l
l
l
1
.l
:!
206
Average
Gradient
Percent of Population
within 3 Miles of City Center
1880
1890
1900
1910
1920
1930
1940
1948
1954
1958
1963
1.22
1.06
0.96
0.80
0.69
0.63
0.59
0.50
0.40
0.35
0.31
88%
83
78
. 69
61
56
53
44
34
28
24
over time, from 1.02 in 1880 to 0.31 in 1963. The third colurnn of the tab1e uses
the estimated density gradients to compute the percentage of the metropolitan population living within three miles of the city center. Between 1880 and 1963, the
percentage dropped from 88 percent to 24 percent.
The decentralization of metropolitan population is a worldwide phenomenon
(Anas, Arnott, and Small, 1998). Between 1801 and 1961, London's density gradient
decreased from 1.26 to 0.34, meaning that the percentage of its population living
within three miles of e city center dropped from 88 percent to 28 percent. In Paris,
the gradient decreased from 2.35 in 1817 to 0.34 in 1946. In cities throughout the
world, population has been shifting outward away from the city center.
Figure 9-3 shows recent data on density gradients for selected U.S. cities. Between 1970 and 1990, the density gradient of most cities decreased, with an average
reduction of 18 percent (from 0.142 in 1970 to 0.121 in 1990). Among the 79 metropolitan areas exarnined by Jordan, Ross, and Usowski (1998), the only area that experienced an increase in its density gradient was San Francisco, where the gradient
increased from 0.124 to 0.132.
1
l
)
'
207
FIGURE 9-3 Population Density Gradients in Se!ected Cities, 1970 and 1990
0.25 r
O Density gradient in 1970
llii!l Density gradient in 1990
0.2
r-
0.15 r-
0.1-
0.05-
Sourre: Stacy, Jo. John Ross, and Kurt Usowski, "U.S. Subut:banization in the 1980s," Regional Science
and Urban Economics 28, \1998), pp. 611-27.
...
support the idea that income growth encourages suburbanization (Anas, Arnott, and
Small, 1998).
Another possible factor in suburbanization is lower comrnuting costs. Over the
last 140 years, technological innovations have decreased the monetary and time
costs of comrnuting.In the 1800s, the development ofnew transit systems decreased
the cost of comrnuting from the suburbs to the central core. The most important
innovations were the horse-drawn streetcar (developed in the 1850s) and the conventional streetcar (developed in the 1890s). In the last 60 years, improvements
in the automobile and the intracity highway network have decreased the cost of
personal comrnuting. The automobile increased travel speeds, decreasing the time
cost of comrnuting. As we saw in Chapter 8, a decrease in unit comrnuting costs
increases the relative attractiveness of suburban locations, shifting the distribution
of population outward.
208
SUBURBANIZATION OF MANUFACTURING
The share of metropolitan manufacturing employment in central cities decreased
from about two-thirds in 1948 to less than half in 1990. Milis (1972) provides
evidence that the decentralization of manufacturing started long before 1948. What
caused the suburbanization of manufacturing employment?
,--- ,
'
--,
209
210
50
40
Panel A: If, as a firm moves away from the city center, freight costs increase
more rapidly than labor costs decrease, total cost will be mi.nimized at the city
center ($30, as shown by point i). Panel B: The intracity truck decreased the unit
freight cost, causing freight cost to increase less rapidly with distance to the city
center. As a result, total cost is now mioimizd. al a suburban location (5 miles
from the city center, with a cost of $20, as shown by point j). The truck changes
the outcome of the rug-of-war between the city center and suburban locations,
wbich have lower wages because they are closer to workers.
211
Rent
.'
. '-:'-:-'.(-- L
',,,;- , 1'
\)...
,..
'
employs 100 people and the firm's workers are distributed uniformly throughout a
streetcar city of 100,000 people. \Vhere in the city will the firm Jocate? Because
the firm draws laborers from the entire metropolitan area, it locates at a point that
is accessible to the entire urban area. In the hub-and-Spoke streetcar city, the city
center (the hub) is the most accessible location.
\Vhere.will the firm locate in a modero automobile city? In the auto city, employees can dpve'their cars from their homes to any location in the urban area. While
the firm might stilllocate in the city center, it could also locate along an urban highway ora suburban beltway because such sites are accessible to residents throughout
the metropolitan area. In general, the automobile loosened ,the ties to the streetcar
nodes in the core area, allowing the suburbanization of firms with S:(Jeciali ed labor
forCeS.
'
. . r( Se:' )
u.\ L -
212
1 , ' '/: ,
\J/
Atlanra
Basten
Chicago
Dalias
Denver
Detroit
Houston
Los Angeles
Mi ami
New York
Phi!adelphia
San Francisco
Washington
Average
Central Business
District
Secondary
Downtowns
Suburban Subcenters
(edge cities)
23.6%
37.4
53.9
20.5
30.4
21.3
23
29.8
!3.!
56.7
34.2
33.9
28.6
37.7%
9.9%
4.6
25.3%
18.8
19.5
40.3
29.4
39.5
37.9
25.4
16.6
6.2
8.9
13.9
27.!
o
4.5
4.2
o
o
7.8
4.5
7.2
3.2
8.8
12.5
6%
19.8%
Dispersed
41.2%
39.2
26.6
34.6
35.9
39.2
39.!
37
65.8
29.9
53.6
43.4
31.8
36.5%
Sou e: Roben E. Lang, ffice Sp-awl: The Evolving Geography of Business," The Brookings Insritution SurvSeries,
October 2000.
"
'
,r,
;..._ - - ........
_.. - - - _..;,...
,.;..
_....
- - - .:..
- - ...
... --
.4ol
"-
;..
-Jl-. , ,,-:"'
.J.&,
:; :
'
FIGURE 9-6 Percentage of Office S pace in Central Business Districts of Select U.S. Metropolitan Areas, 1999
60
'
50 -
..
u 40 -
'
p:.
37.7
.9.
.."'.
1l
":E
o
30 f.-
,3.10
28.
'8
1;',
l'l'
ll.20
::!:
e-
""'
p:!
10 -
,.."'
'?"
'
,_o'>
>o"'
.d. .
e,">'
:;:.
<:;"'
.,.,
<:;"'""
'
o'
<:;"
"'-
,_o'>
o<>"
"'"
' """
V'"
'
'
i;:
-,<;-
4_0/!c
;!-
::-."
<>.""
Source: Roberl E. Lang, "Office Sprnwl: The Evolving Geogrnphy ofBusiness," The Brookings lnstit,ltion Survey Series, Oclober 2000.
N
. ....
w
- r:..P
S ;
"' 4
'
.._o'>
.;;."C
.g
.,..,<>'
'?"
214
rP?i/
Percent of Suburban
Cluster Space Close to
Metropolitan
Area
Percent of Space
. in CBD
Numberof
C!usters
Freeway
lnterchange
Denver
Houston
Los Angeles
San Francisco
Seattle
Toronto
27
20
12
31
45
63
160
270
102
30
52
73
70
66
93
93
79
36
29
30
52
69
27
54
Ral Transit
10
25
Source: Gary Pivo, ''The Net of Mixed Beads: Suburban Office Development in SU Metropolitan Regions," Joumal ofthe
American PPuling Associatio"n, Autumn 1990, pp. 457-69.
215
to move to the suburbs. Her inputs (the client data) and her output (the income statement) can be sent over telephone lines, so she can do her accounting in the suburbs,
far from her CBD clients.
Electronic mail has caused the decoupling of sorne CBD finns. Many finns split
their operations into CBD activities and suburban activities. Considera finn whose
accountants condense information into reports for the firm's executives, who then
use the information in their interactions with other firms. In the absence of electronic
mail, the finn's accountants and executives must be in the same location to facilitate
the rapid turnaround of information. Given the need for face-to-face interactions
among the executives of different finns, the finn locates in the CBD. If the accounting information can be transmitted electronically, however, the accounting division
can move to the suburbs, and the firm can send information between the suburban
account ts and the CBD executives over telephone lines. Because electronic mail
facilitates intrafinn communication, it allows the finn to decouple its operations,
moving its support staff to the suburbs while keeping its executives in the CBD.
A second new communications method, the teleconference, is essentially a video
meeting. The teleconference is superior to the telephone conference because it allows
both visual and verbal communications. The teleconference decreases the need for
face-to-face contact within and between finns: instead of sendingemployees across
town to another of:fice, the finn can send them down the hall to the video room.
Because the video camera does not transmit as much information as a face-to-face
encounter, the teleconference is not a perfect substitute for face-to-face contact.
While the video camera focuses on one or two people in a meeting, a person who is
physically present can watch everyone in the room, picking up verbal and nonverbal
cues from a large number of people. In addition, the video camera cannot transrnit
tactile information. The tactile information provided by face-toface encounters is
sometimes useful; a firm, enthusiastic handshake sends a message that cannot be
easily conveyed through a video camera.
Because the teleconference is nota perfect substitute for face-to-face contact, it
is unlikely to cause substantial suburbanization of of:fice firms. Most of the face-toface contact that occurs in the CBD is between employees of different firms. The
subtle interactions between firms often require face-to-face contact. When people
from different firms meet, one of the purposes of the meeting is to establish trust
Because it is easier to establish trust in a face-to-face encounter, a video meeting is
inappropriate for sorne types of interactions.
216
_One option for a firm that locates in the suburbs is to locate in a cluster of other firms,
resulting in a suburban subcenter. The clustering of firms in suburban subcenters
_ results from agglomeration ecoi10nles in production. In a cluster fni.anufacturers
_ or of:fice fi.rms, the total. demand for sorne iptermediate inputs is large enough to
support externa! suppliers, who can exploit scale..ecoomies and thusproduce the
mtermediate inputs at alower cost. For example, manufacturers in a cluster may save
money by purchasing business services (such as repair, maintenance, accounting)
from a common supplier. Alternative!y, if there are scale economies in printing
glossy brochures and reports, a large cluster of of:fice firms can support a printing
firm, allowing each firm to save on printing costs. The same principie applies to
restaurants and hotels; one rule f thumb is that a cluster of offi.ce firms with total
of:fice space of 2.5 rnillion square feet can support a 250-room hotel.
Figure 9-7 shows land rent at different locations in a hypothetical city with a beltway (two arid one-half ffii.Ies from the city center) and two employment subcenters.
FIGURE 9-7 Land Bid Rent in Beltway City
JO
Rent
Distance
Distance
217
The land-rent surface has three peaks-the highest one at the city center and two
lower ones at the subcenter locations-and forms a circular ridge centered on the
beltway.
.,.
1
.
--
c:f
:f.
j
t
i
j
t
{
(
f
(
f
(
f
.L.
Giuliano and Small (1991) use data on the spatial distribution of activity in the
Los Angeles metropolitan area to show that there were 28 subcenters in Los
Angeles County and Orange County in 1980. They define a subcenter as a zone
where the employment density is at least 1O workers per acre and total employment
is at least 10,000 workers. In contrast, employment density in the CBD (downtown
Los Angeles) is 36 workers per acre and total employment is 469,000. In 1990, total
employment in the subcenters was about twice the employment in the CBD (23 percent versus 11 percent), and two-thirds ofthe metropolitan area's employment was
outside the center and the subcenters.
Table 9-6 shows the characteristics of the CBD and the 28 subcenters. The
subcenters are listed after the CBD in descending order with respect to total employment. The subcenters vary in employment density (workers per acre). The average
employment density ofthe subcenters is 17.7 workers per acre, compared to 36.0
in the CBD. Giuliano and Small exarnined the relationship between employment
density and distance to the CBD and found a negative relationship between density
and distance. This result is consisten!with the predictions of the model of the monocentric city. In the case of Los Angeles, there is a negative relationship despite the
fact that a relatively small fraction of total employment is in the CBD.
The subcenters vary with respect to employment ratios (jobs per residen!). The
average employment ratio of the subcenters is 1.58 jobs per residen!, compared to
1.47 in the CBD and 0.43 for the entire metropolitan area. Except for the airport
subcenters, which have few residents and thus large employment ratios, the employment ratios of the subcenters are relatively small. This suggests that the subcenters
are not isolated employment centers, but instead provide a mixture of jobs and
dwellings.
Giuliano and Small suggest that the subcenters can be divided into five types:
rnixed industrial, mixed service, specialized entertainment, specialized manufacturing, and specialized service. Most of the mixed-industrial subcenters started out
as low-density manufacturing areas riear a tra:nsport node (airport, port, or marina) and grew as they attracted other activities. Most of the rnixed-service subcenters are like traditional downtowns and provide a wide range of services. Many of
these subcenters functioned as independent centers before they were absorbed into
the metropolitan economy. The specialized-manufacturing subcenters include areas
near airports that produce aerospace equipment and older manufacturing areas. In
the service-oriented subcenters, 90 percent of employment is in service activities
such as medical care, entertainment, and education.
There are four basic conclusions from this study of the Los Angeles area. First,
the subcenters -differ in the mixes of goods and services they provide, suggesting that
the subcenters play diverse roles within the metropolitan economy. Second, many
218
..'..- '
r
,-
,-
TABLE 9-6 Characteristics of the CBD and Subcenters in the Los Angeles
,-
Metropolitan Area
Rank
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
. 24
25
26
27
28
29
Location
Downtown Los
Angeles (CBD)
Los Angeles West
Santa Monica
Hollywood
Los Angeles Airport
Orange County Airport
Glendale
Com.merce
Vernon/Huntington Park
San Pedro
Santa Ana
Ing1ewood
Pasadena
Long Beach Airport
Marina Del Ray
LongBeach
Van Nuys Airport
Burbank Airport
Hawthorne
Canoga Park/
Warner Center
Lawndale
L.A.East
Fu!lerton
Downey
Santa Ana South .
Shertnan Oaks
Burbank S.W.
Anaheim!Orange/
Garden Grove
Garden Grove/Stanton
Al! subcenters
Jobs
per Residen!
Distance to
CBD (miles)
Ernployment
(1,000)
Workers
per Acre
469
36.0
1.47
0.1
176
65
64
59
48
43
42
39
38
38
37
36
33
32
30
28
26
18
17
25.5
16.9
21.4
16.7
16.1
15.5
17.0
33.2
15.7
17.3
14.6
25.3
15.5
11.4
18.0
12.6
28.4
12.4
11.2
1.37
1.11
0.73
4.32
1,589.87
1.07
4.05
2.42
2.74
1.51
1.24
1.73
3,684.78
1.28
0.84
2.04
10.86
0.74
1.21
15.8
16.7
7.3
18.8
40.'7
12.3
9.8
4.9
23.3
32.9
14.7
12.1
23.3
14.0
25.3
22.1
16.5
13.5
27.4
17
16
16
15
14
13
13
11
17.1
37.3
11.4
17.3
12.2
11.9
18.0
11.3
1.36
2.30
4.97
2.38
!.76
1.04
1.92
1.06
20.5
6.8
27.3
14.8
37.4
18.6
14.1
30.2
10
994 (total)
5.60
26.6
17.7 (average)
12.9
1.58 (average)
Source: Reprinted from Regional Science and Urban Econ.omics 21 by Genevieve Giuliano and KeMeth Small, "Subcenters
in the Los Angeles Region," @ 1991 with perrnission from Elsevier Sctence.
of the subcenters are highly specialized, suggesting that there are large localization
economies (clustering of firms in the same industry to share input suppliers, save
on labor costs, and share information). Third, employment in the metropolitan area
is relatively dispersed; two-thirds of total employment is outside the CBD and the
subcenters. Fourth, employment density decreases as distance from the center increases, despite the fact that the center contains a relatively small fraction of total
employment.
r
r
,- ,
r
r
219
Subcenters in Houston
Mieszkowski and Smith (1991) explore the land-use patterns in the Houston
metropolitan area. and identify 10 subcenters. In 1985, total employment in the
individual subcenters was between 16,000 and 86,000, comparedto total employment in the CBD of 156,000. In 1985, the CBD contained about 10 percent ofthe
metropolitan area's total employment, leaving 23 percent for the subcenters and
67 percent for areas outside the CBD and subcenters. The corresponding figures for
1970 were 14 percent for the CBD, 11 percent for the subcenters, and 75 percent for
areas outside the CBD and subcenters.
Subcenters in Chicago
'
.
l
;(
:.{
..
:(
t
t
..
1.
'{
'
{
le
Edge Cities
The most recent trend in urban location patterns is the development of edge cities,
defined as new concentrations of office and retail activities outside the core areas of
metropolitan areas. Edge cities are typically centered on huge tracts of mixed-use
220
r-=
"
"
')
Subcenter
Municipality
OId satellite cities
Aurora
Elgin
Waukegan
OId industrial suburbs
Chicago Heights
Harvey
McCook
Post-World War 1I industrial
suburbs
Addison
Des Plaines
Franklin Park
Ni!es
Norridge
O'Hare
New industrial/retail suburbs
Northbrook
Palatine
Service and retail centers
Burbank
Evanston
Maywood
Edge cities
Naperville
OakBrook
Schaurnburg
Percent
1990
Employment Manufacturing
10,689
13,095
11,506
59
36
64
8
11
<1
15
38
1
8
2
10,862
9,880
31,109
62
43
64
2
4
4
22
5
.14
20
25
')
')
')
1
r)
-
'.
,-,
1
23,409
27,653
27,462
40,040
10,457
32,681
62
31
56
65
43
21
3
18
4
17
5
5
4
1
6
21
27
15
51
8
16
6
31
'1
"
43,030
19,385
31
19
21
45
23
9
11
8
14,028
19,551
26,749
1
4
7
23
13
ll
65
69
39
2
3
12
35,168
111,550
40,295
20
13
13
12
22
36
52
37
26
2
20
12
office space, and provide jobs, residences, shopping, and services for their inhabitants.GarreaU (1991) identifies 123 existing edge cities and 77 ernerging edge cities
in the 35 1argest rnetropolitan areas in the United States. In rnany cases, a single
developrnent cornpany or individual controlled the initial developrnent and building
of the edge city.
Table 9-8 shows sorne of the characteristics of severa! edge cities (Henderson
and Mitra, 1996). The amount of developed office space ranges frorn 4 rnillion
to 33 million square feet, and most of the office space is of the highest quality
(class A). By cornparison, the CBDs of Atlanta, Baltimore, Cleveland, Buffalo,
Detroit, Hartford, Indianapolis, and Milwaukee have between 2 million and 9 million
square feet of class A office space, and the CBDs of sorne of the largest rnetropolitan
areas (e.g., Washington, Los Angeles, B6ston, and Chicago) have between 17 million
and 40 million square feet of class A office space.
221
Employment in 1993
(thousands)
45
50
15
20
30
22
33
165
80
75
71
45
40
8
30
14
28
5
11
18
14
4
11
20
20
15
Source: Reprinted from Regional Science and Urbczn Economics 26 by Vernon Henderson, and Arindam Mitra, ''The New
Urban Landscape: Developers and Edge Cities," pp. 613-43,@ 1996 with permission from Elsevier Science.
What is the econornic relationsh.ip between a centrar business district and the
surrounding subcenters? The central business district provides better opportunities
for the face-to-face contact required in the production of many goods and services,
and suburban office firms depend on the central city for many financial and professional services. Swartz (1992) explores the corporate service linkages in New
York, Cbicago, and Los Angeles. As shown in Table 9-9, for all five corporate services studied (actuarial consulting, auditing, banking, investment banking, and legal
services), more thari half the suburban firms patronize central-City firrns. Tne percentage of suburban firrns patronizing central-city firms was the highest for banking,
r
.
222
TABLE 9-9 Percentage of Finns Using Corporate Services in the Central City
andSuburbs
Investment Legal
Actuaria!
Consulting Auditing Banking Banking Counsel
Percent of suburban firms patronizing
central-city firms
Percent of suburban firrns patronizing
suburban firms
Percent of suburban firms patronizing
firms outside metropolitan area
Percent of central-city firms
patronizing suburban firms
53
56
67
67
71
29
35
21
16
18
12
29
12
14
Source: Swiutz, Alex.. ''Corporate Service Linkages in Large Metropolitan Areas: A Srudy ofNew York, Los Angeles, and
Chicago." Urban Affai" Quanerly 28, pp. 276-96, 1992. Reprinted by permission of Sage Publications, Inc.
investrnent banking, and legal services. In contrast, few central-city finns purchase
corporate services frorn suburban finns.
The Swartz study suggests that submban office activity depends on the central
city for sorne business services that require face-to-face contact. Swartz reaches the
following conclusion:
Despite their ample supplies of office space, the suburbs of the nation's three largest
CMSAs do not constitute a self-sufficient outer city econornically autonomous from the
central city. Suburban office centers do not house the full range of business activities
found within the central city... A majority ofthese suburban companies remain dependen!on their metropolitan area's central cities for financial and professional services...
These finclings from a study of intrametropolitan linkages do not, however, suggest
that the suburbs are sirnply low-cost areas for basic ta-processing and other backoffice functions. Except for the Chicago area, where suburban companies rely almost
exclusively on the city of Chicago for corporate services, a significan!portion-usually
15 percent to 35 percent-of suburban-based firms do depend on suburban-based firms
for their service needs.
One interpretation of the Swartz study is that although suburban firrns depend on
central-city firrns for financial and professional services, the expansion and rnaturation of the suburban office sector malees thern less dependent.
Stanbach (1991) explores changes in central-city and suburban ernployrnent in
14 large rnetropolitan areas. He docurnents the growth in corporate services and
inforrnation processing in suburban areas, and reaches the following conclusion:
But the growth and maturation ofthe suburbs-especial!y insofar as the suburban development process has been strengthened by the development of rnagnet centers-must, of
necessity, alter the relationship between the central city and suburbs. In the new relationship, it would appear that central cities may become more vulnerable to competition, at
least in those activities for which their comparative advantage is marginal.
In the new relationship between central cities and suburbs, the central city will
continue to have an advantage in providing goods and services for which face-toface contact is required.
"'
223
SUMMARY
l. The truck contributed to the suburbanization of manufacturing employment
2.
3.
4.
S.
6.
because (a) the intracity truck replaced the horse-drawn wagon, decreasing
freight costs relative to commuting costs, pulling firms toward the low-wage
suburbs, and (b) the intercity truck freed manufacturers from the central export
node.
Severa! factors contributed to the suburbanization of population: higher income,
lower commuting costs, central-city problems such as crime and weak schools,
and high taxes.
In the last 30 years, there has been a dramatic decentralization of of:fice employment, assisted in part by advances in communication that have reduced the need
for face-to-face contact.
In the modero city, workers commute to jobs in both the CBD and suburban
subcenters. Workers are willing to pay more for housing and land near employment centers (everything else being equal), so the land-rent function has severa!
peaks, one near the CBD and one near each subcenter.
The clustering of firms in stiburban subcenters results from agglomerative economies in production. Firms in the subcenters share input suppliers, a labor
force, and information.
The monocentric, core-dominated city has been gradually replaced with the
multicentric, suburbanized city. The land-rent surface has local peaks at the city
center and the subcenters, and forms a ridge centered on the beltway.
l. Considera manufacturing firm that exports sorne of its output by train (through
a central export node) and the rest by truck (the truck exits the metropolitan area
using a suburban beltway). Suppose that production costs (including labor costs)
are the same at alllocations. The unit cost oftransporting output inward (toward
the central export node) is t;, and the unit cost of transporting output outward
(toward the beltway) is t0 The volume of output going to the central export
node is w;, and the. vcilume going to the beltwayis W 0 Under what conditions
will the firm's bid-rent function be positively sloped? How would your answer
change if wages were lower in the suburbs?
2. Consider a metropolitan area with a 1O-mile radius. Two members of the Dink
household work: Mr. Dink commutes to the city center (u= 0), and Ms. Dink
commutes to a suburban subcenter four miles due east of the city center.
a. Depict the household's housing-price functions if travel time per rnile is the
same in both directions (toward the city center and away from the city center),
and Mr. and Ms. Dink have the same opportunity cost of travel time.
b. Depict the household's housing-price functions if the travel time per mile
. of inward COilJ.Ill.uting (toward the city center in the morning rush hour and
away from tlie center in the evening iush hourY exceeds the travel time of
outward commuting (away from the city center during the morning rush hour
224
3.
and toward the center during the evening rush hour), and Mr. and Ms. Dink
have the sarne opportunity cost oftravel time.
c. Depict the household's housing-price functions if travel time per rnile is the
sarne in both directions, but Ms. Dink has a higher opportunity cost of travel
time.
Consider a city where a single finn sells pet rocks, a good with a large income
elasticity of demand. The firm islocated in the central shopping area ofthe city.
Suppose that households the city experience an increase in income.
a. Assume that the number of pet rock finns is fixed at one. What is the effect
of the increase in income on the finn's bid rent for land at the city center?
b. Is it realistic toassume that the number of petrock finns will remain constant?
c. If the number of finns is allowed to vary, what is the effect of the increase in
income on the retail bid-rent function?
Comment on the following statement: "Instead of investing money in traditional
radial mass-transit systems (hub-and-spoke systems), we should be investing in
circurnferential transit systems. We should build a circular transit system along
the urban beltways."
The traditional core-dominated city was a segregated city in the sense that all
ernployment was in one area (the core) and all residents lived in another area
(the satellite). The modero city is less segregated in the sense that ernployrnent
is dispersed throughout the city, in subcenters and along suburban beltways.
Mr. Wizard recently made the following statement, "If my assumptions are
correct, land use in the typical American city will soon be cornpletely integrated.
I predict that each rnanufacturer and each office finn will be surrounded by its
workforce: the metropolitan area will have dozens of 'company villages.' A
firm will be at the center of each village, and the firm's workers will occupy the
land surrounding the factory or office. Every worker will be employed in the
firm at the center of the village. That's integration." Assume that Mr. Wizard's
reasoning is correct. What are his assumptions? In other words, under what set
of assumptions will a system of company villages develop? Are Mr. Wizard's
assumptions realistic?
The developrnent of the internal-cornbustion engine caused fundamental changes
in u_rban land-use patterns. The transformation from the core-dominated city to
the modero suburbanized city took only about 50 years. Given the rapid pace
of technological change, it seems likely that sorne future innovation will cause
another transformation of cities. Given your knowledge of science fiction and
fact, describe an innovation that would cause fundamental changes in the spatial
structure of cities.
Table 9-6 shows the rank and total employment for the CBD and the 28 subcenters in the Los Angeles metropolitan area. How does the size distribution of
the employment centers compare to the size distribution of cities discussed in
Chapter 3 (Big and Small Cities)?
Telecommuting refers to a system under which employees work at home for
one or more. days per week and use personal computers and telephone lines
to interact with people in the of:fice. Consider an of:fice firm that must decide
4.
5.
6.
7.
8.
'
.1
JI.
ji
l.
1
J
l
(
1
J
1
1
1
1
1
'
225
226
227
15. McMillen, Daniel P., and John F. McDonald. "Suburban Subcenters and
Employment Density in Metropolitan Chicago." Joumal of Urban Economics
43 (1998), pp. 157-80.
16. Mieszkowski, Peter, and Edwin S. Mills. "The Causes of Metropolitan Suburbanization," Joumal of Economic Perspectives 7 (1993), pp. 135-47.
17. Mieszkowski, Peter, and Barton Sinith. "Urban Decentralization in the Sunbelt:
The Case ofHouston." Regional Science and Urban Economics 21 (1991).
18. Milis, Edwin S. Studies in the Structure of the Urban Economy. Baltimore:
Johns Hopkins, 1972.
19. Meses, Leon, and Harold Wliliamson. "The Location of Econoinic Activity
in Cities." In Readings in Urban Economics, ed. Matthew Edel and Jerome
Rothenberg. New York: Macmillan, 1972.
20. Pivo, Gary. "The Net of Mixed Bead: Suburban Office Deve1opment in Six
Metropolitan Regions." Joumal ofthe American Planning Association, Autumn
1990, pp. 457-69.
21. Stanbach, Thomas M. The New Suburbanization: Challenge to the Central City.
Boulder, CO: Westview Press, 1991.
22. Swartz, Alex. "Corporate Service Linkages in Large Metropolitan Areas: A
Study of New York, Los Angeles, and Chicago." Urban Affairs Quarterly 28
(1992), pp. 276-96.
1
CHAPTER 10
Tm.s
chapter explores the governrnent's role in the urban land rnarket. So far in the
book, we have assurned that land is allocated to the highest bidder, rneaning there is
no governrnent intervention in the urban land rnarket. In fact, local governrnents use a
number of policies to controlland use. Most cities have zoning plans that limit the
location choices of residents and firms. Sorne cities use zoning and other land-use
controls to limit population growth. This chapter addresses three basic questions
about land-use controls. First, why do cities controlland use? Second, what are the
market effects of land-use controls? Third, what are the legal foundations for zoning
and other land-use controls?
'
,;
1
229
--. ,:..:'
_;-<:
,
Business
bid rent
. - 1'
L)
')
' ....',
r.,
New labor
supply
Initiallabor
supply
'
Labor
demand
Residential
13
10
10 12
'
'
1 Distance from city center
1
1
90 120
Number of workers
1
1
In the initial equilibriwn the business district is 3 miles wide (area B) and the residential district is
9 miles wide (areas R and U), An urban service boundary at u= 10 miles reduces the territory ofthe
residential territory te area R. Tbe decrease in residentia] territory shifts the labor-supply curve te the
left, causing excess demand for labor and ultimately a higher wage ($13, up from $10).
employmeiitdensity (workers per unit ofland) and population density (residents per-unit of land) are coristant. Our final assumption is that the city is a rectangle, not a
circle.
Panel A of Figure 10-1 shows the initial land-use pattern in a monocentric
city. The residential ,bid-rent function intersects the agricultura! bid-rent function
at a distance of 12 miles frorn the city center, so the city is 12 miles long. An\
implicit assurnption of this model is that the city governrnent provides urban services )
":herever residents _outbid_ farmers, in _thicase to a distru:ce o12 _miles from e
c1ty center. The busmess b1d-rent functJ.on mtersects the res1dential b1d-rent functlonat a distance of 3 miles from the city center, so the CBD is 3 miles long and the
residential district is 9 miles long.
_,
Suppose that the city refuses to extend urban services beyond 1O miles from '
the city center. Unless residents can provide their own roads, sewage systems, and
schools, residential development outside the boundary is irnpossible. Land beyond
_u = 10 is rendered uninhabitable, so the market rent on land outside the boundary
drops to the agricultura! bid rent. The residential area is now only 7 miles long.
Panel B of Figure 10-1 shows the effects of the urban service boundary on the
urban labormarket. The service boundary decreases the size of theresidential district
from 9 miles to 7 niles; fiecreasmg the city's total labor supply. The leftWard shift
230
Business
bid rent
Residential
10
2.5
1
-- -----+---,,----,
R
of the city's labor supply causes excess demand for labor, causing the city's wage to
rise from $10 (point i) to $13 (point j).
The increase in the wage affects the urban land market. InFigure 10-2, the
increase in the wage shifts the residential bid-rent function upward (from the thin
line to the thick line). At the same time, the higher wage increases production costs
of firrns, shifting the bid-rent for business land downward (from thin to thick). As
a result, the two bid-rent curves intersect at 2.5 miles (down from 3 miles), so the
CBD shrinks from 3 miles to 2.5 miles. The loss of the business district is the gain
of the residentia! district, so the residentia! district is now 7.5 miles long.
r
.
:.
231
Who gains and who loses from the service boundary? People who own land outside the service boundary lose: for land between u = 10 and u = 12, the market rent
drops to the agricultura!bid rent, decreasing the market value of land. People who
own residentialland within the service boundary win; the increased demand forresidentialland (caused by the increase in the wage) bids up the price of land within the
boundary. People who own land within the CBD lose; the decrease in the demand for
CBD land (caused by the increase in the wage) decreases the price of business land.
Our model ofthe urban land and labor markets assumes that population density
and employment density are constant. In fact, one ofthe purposes of an urban service
boundary is to increase population density, generating a more compact metropolitan
area. If we modified the model to allow changes in density, population density
would increase because the increase in the wage would increase the price of housing
and land, causing residents to consume smaller quantities of housing, and housing
producers would produce housing with less land input per unit of housing.
Building Permits
Sorne cities control residential growth by lirniting the number of building permits
issued. By lirniting the number of new dwellings built per year, a city can control its
growth rate.
Consider a city that initially has no limit on the number of building permits
issued. Figure 10-3 shows the city's market for new housing: The supply curve
AG intersects the demand curve at point F, generating an equilibrium quantity of
100 houses per year andan equilibrium price of $50,000 per house. The city issues
100 free permits pr year, which are used to build houses with a market value of
$50,000. Deyelopers make zero economic profits (normal accounting profits), meaning that the cost of producing each house (the sum of land, labor, and capital cost)
is $50,000.
Suppose thatthe city lirnits the number ofbuilding perrnits to 60 per year. The
new supply curve forhousing isACE; the maximum number ofnew houses is 60, so
the supply curve becomes vertical at 60 dwellings. The new supply curve intersects
the demand curve at
equilibrium price of $70,000. In other words, the permit
policy increases the equilibrium price of housing by $20,000.
The perrnit policy also decieases the cost of producing housing. The permit
policy decreases the number of houses built, so it decreases the demand for land.
For example, if houses are built on quarter-acre lots, the permit policy decreases
the demand for lari.d from 25 acres per year (100 times 0.25) to 15 acres per year.
The decrease in the demand for land decreases the market price of land, decreasing
the cost of producing housing. This is shown by the housing supply curve: If 100
houses are built, the production cost is $50,000 per house; if only 60 houses are
built, the cost drops to $40,000 per house. Because the perrnit policy decreases the
number of houses built and decreases the price of land, it allows developers to build
houses at a lower cost.
The city must decide.how .to allocate the 60 building permits among its de. velopers. One option is to auction the perrnits to the highest bidders. What is the
monetary value of a building perrnit? A person with a building perrnit can make a
an
232
70 -----------
curve
Demand
curve
100
60
New houses per year
1
If the city decreases the supply ofbuilding permits from 100 to 60, the supply curve for
housing shifts from AG to ACE. The new equilibrium is point H (price of housing =
$70,000). Point C shows the cost ofproducing 60 houses ($40,000). The difference
between the price _:md the cost is the equilibrium price of the building permit ($30,000).
profit equal to the difference between the market price of a house ($70,000) and the
cost of producing the house ($40,000), so the monetary value of a permit is $30,000.
If the city auctions the permits, the market price would be $30,000.
Sorne cities allocate their building permits to development projects that promote
their development objectives. Ifthe city is interested in high-density housing, it could
allocate the permits to a high-density housing project. Alternatively, the city could
allocate the permits to a project in an area targeted for development. Many cities
use sophisticated point systems to rate and rank alternative development proposals.
Among the criteria: for which points are assigned are various design features (the
exterior appearance of the dwellings, the amount of open space within the project),
density, and the proxirnity of the project to existing public infrastructure (roads,
schools, sewer lines).
r
'
'
233
The first zoning ordinances were used by the wealthy to exclude the poor. In the
1880s, the city of San Francisco passed Iaws to segregate its Chinese population.
When explicit segregation was declared unconstitutional, the city passed a zoning
law that bannedlaundries from certain neighborhoods.The zoning law did not violate
the Constitution because it promoted public welfare by keeping an undesirable land
use (laundries) out of sorne residential areas. Because the Chinese operated most of
the city's laundries, the zoning law provided a legal means of segregation. In New
York City in 1916, the rapid growth of employment in offices and garment faetones
increased the number of low-wage women on the streets of rnidtown Manhattan.
The owners of exclusive Fifth Avenue stores feared that the increased traffic
of poor women would be bad for business, and proposed zoning laws that limited
building sizes and building heights. The law limited the growth of office buildings and garment factories, decreasing the number of clerical and garment workers
walking along Fifth Avenue. The zoning law was declared constitutional because
it promoted public welfare by controlling the nuisances created by tal!buildings
(shadows, blocked light, disruption of views).
There are many types of zoning, each of which has at least one purpose. This
section divides zoning into three types.The purpose of nuisance zoning is to separate
incompatible land uses. Cities use fiscal zoning to exclude households that would
not pay their share of the costs oflocal government. Design zoning is a sort of macroarchitecture: planners arrange activities within the city topromote the efficient use
of the city's infrastructure.
234
be set equhl to the marginhl extemhl cost of pollution, that is, the cost to society of
an additionhl unit of pollution. For example, if one ton of sulfur dioxide generales a
socihl cost of $2, the effiuent fee should be $2 per ton. The effiuent fee "intemhlizes"
the pollution extemhlity, forcing polluters to pay for pollution in the same way they
pay for labor, capitl, and raw materihls. A firm has an incentive to decrease pollution
because abatement decreases the firm's eff!uent bill. If the effluent fee equals the
marginhl extemhl cost of pollution, firms generate the optimum leve!of pollution.
In theory, a system of eff!uent fees would generate both the optimum leve!of
pollution and the optimum spatihl distribution of pollution.If the eff!uent fee varies
across space, a firm would pay a higher effluent fee for pollution generated near
sensitive areas (e.g., residentihl areas). For example, the effiuent fee might be $2
per ton of pollution if the firm is 10 miles from the residentihl area and $5 per ton
if the firm is 4 miles from the residentihl area. Under a system of spatihl effluent
fees, the firm would base both its production decisions and its location decision
on eff!uent fees. If the effluent fees are set equhl to the marginhl extemhl cost of
pollution, the firm would choose both the optimum location and the optimum leve!
of pollution.
To explain the differences between ioning and eff!uent fees, consider a rectangular city with the following characteristics:
l. Residents live in the westem partofthe city and cornmute eastward toa polluting
steel mill.
2. The longer the distance between the residentihl area and the mill, the higher
the wage: workers are compensated for cornmuting cost in the form of higher
wages. The firm's tothllabor cost equhls the wage (dollars per worker) times
the tothl quantity of labor.
3. The longer the distace between the mill and the residentihl area, the lower the
eff!uent fee. The firm's tothl pollution cost equhls the effiuent fee (dollars per
ton) times the amount of pollution.
4. The city starts with an industrihl zoning policy under which the steel mili is
located 1O miles from the residentihl area.
Figure 1shows thetothl cost of producing steel at different locations. The
positively sloped curve showslabor cost, which increases from $20 for a location
adjacent to the residentihl district to $63 in the industrihl zone (lO miles from the
residentihl area). The negatively sloped curve shows pollution cost, which decreases
from $60 at a location adjacent to the residentihl district to $5 in the industrial
zone. The variation in the pollution cost refiects the variation in the eff!uent fee. The
U-shaped curve shows tothl cost, the sum oflabor cost and pollution cost, which decreases from $80 for a location adjacent to the residentihl district to $53 for a location
4,2 miles from the residentihl area, then increases to $70 in the industrihl zone.
Suppose the firm starts in the industrihl zone. As the firm moves toward the residentihl district, there are benefits and costs. On the benefit side, labor cost decreases
as the firm moves closer to its workforce. On the cost side, the effiuent fee increases
because the firm's pollution affects more people, so pollution cost increases. Given
the shapes of the labor-eos!and pollution-cost curves, tothl cost is minimized at a
!.
..
235
60
20
1
1
1
1
4.2
6
Distance from residential area (miles)
10
Total cost is the sum of labor cost and pollution cost. As the pol!uting firm moves toward the
residencial district, pollution cost increases (more residents areatrected by pol!ution, so the
eff!uent fee increases), but labor cost decreases (commuting distances decrease, so the wage
decreases). Total cost is minimized ata distance of 4.2 miles from the residential area.
...
location 4.2 miles from the residential district. lf the city switches from a zoning
policy to an effiuent-fee policy, the steel producer would move from the industrial
zone to a location 4.2 miles from the residential district, decreasing its total cost
from $70 to $53.
The effiuent-fee policy is more efficient than the zoning policy for two reasons.
First, the effi.uent policy allows the firm to choose the location that rninirnizes production cost. From the soial perspective, the new location is more efficient because
the savings in commuting cost (as indicated by the labor-cost curve) exceed the
increase in pollution cost (as indicated by the pollution-cost curve). Second, the
effiuent-fee policy forces the firm to pay for pollution, so pollution decreases to its
optirnumlevel.
Will all the polluting firms leave the industrial zone when the city switches
from industrial zoning to effi.uent fees? A firm will stay in the industrial zone if the
pollution-cost curve is steeper than the labor-cost curve. In this case, a move toward
the residential district increases pollution cost by more than it decreases labor cost,
so total production cost increases. Nonetheless, the effiuent-fee policy is still more
efficient than industrial zoning because the effiuent-fee policy decreases pollution
to its optimuin leveL"
236
Why do cities use zoning instead of effiuent fees to control industrial pollution?
There are two principal reasons. First, industrial zoning is simple compared to a
system of spatial effiuent fees. To set the effiuent fees, the city would have to estimate
the marginal external cost of pollution for different locations in the city, and to collect
the fees, the city would have to monitor the polluting firms. It would certainly be
easier to put all the polluters in an industrial zone. Second, a switch from zoning to
effiuent fees may increase pollution in sorne neighborhoods. Although the factory
produces less pollution under the effiuent-fee policy, it may locate closer to the
residential district.Therefore, sorne neighborhoods may become more polluted. The
fact that pollution is at its optimum level is small consolation to !hose who breathe
the dirty air. The switch to effiuent fees would irnprove efficiency and generate a
net gain for society, so it would be possible, in principie, to compensate !hose who
breathe dirtier air. In fact, compensation is-rarely attempted, so local opposition to
effiuent fees remains.
Performance Zoning
Traditional zoning is based on a zoning map and a land-use list. The zoning map
shows the zoning classification for each plot of land in the city, and the Iist shows
the admissible land uses for each zoning classification. If an activity is designated
heavy industry, every firm involved in this activity must locate in the heavy-industry
zone, regardless of how much noise, odor, or smoke the individual firms actualy!
generate. The alternative to traditional zoning is performance zoning, under which
the citysets performance standards for each zone. For industrial uses, performance
. standards typically set upper lirnits for the arnount of noise, glare, odor, vibration,
gas, and smoke erriitted by the firms. Performance zoning is a sort of compromise
between traditional zoning and effiuent fees. Although performance zoning does not
force firms to pay for pollution, it does encourage abatement because it gives cleaner
firms more location options.
Retail Nuisances
. Retailers generate a number of externalities that affect nearby residents. The traffic
generated byretailers causes congestion, noise, and parking problems. A traditional
zoning plan establishes retail zones to decrease the exposure to these externalities.
For exarnple, zoning prevents the invasion of quiet residential neighborhoods by
shoppers and delivery trucks.
Under performance zoning, the city allows ;etailers in a particular zone as long
as they satisfy performance standards for parking, traffic, and noise. For each area,
the city may specify a minimum number of parking places, a rninimum speed for
traffic, and a maxirnum noise leve!.These performance standards force commercial
developments to provide off-street parking to control parking problems, signalization
and street irnprovements to control traffic problems, and berms and landscaping
to control noise problems. Performance zoning allows the mixing of commercial
and residential land uses because retailers take steps that shield residents from the
undesirable effects of commercial development.
237
Residential Nuisances
Most residential extemalities are generated by higb-density bousing. Suppose that
a d veloper builds a four-story apartment complex in a neigbborbood with singlefamily bornes. The apartment complex increases traffic volume, increasing congestion and noise, and increases the demand for on-street parking, causing a sbortage
of parking spaces. In addition, the tall apartment building is likely to deprive neigbbors of scenic views and sunlight. Under conventional zoning, the apartment building
would be excluded from the low-density neigbborbood. Conventional zoning shields
the residents of single-farnily bornes from extemalities by excluding bigb-density
bousing.
The altemative to conventional zoning is performance zoning. Under a
performance-based zoning policy, tbe apartment complex would be allowed if the
developer took the following steps:
l. Off-street parking. The developer provides enougb off-street parking to prevent parking problems.
2. Street improvements. T):e developer pays for street improvements that prevent
congestion problems.
3. Building design. The building is designed to prevent loss ofviews and sunligbt.
The developer uses landscaping to establisb a buffer between tbe apartment
building and the single-family bornes.
Tbe idea behind performance zoning is that the apartment building sbould be judged
on the basis of its actual effects on the neigbborbood, not on the simple fact that it
is high-density bousing. If a higb-density project does not generate extemalities, it
would be. allowed in the neigbborbood.
.
Fiscal ZoniJ.g
Tbe second type of zoning occurs because local governments finance public services
with property taxes. Under fiscal zoning, a city excludes bousebolds that would
impose fiscal burdens on local government. Suppose that a bousebold consumes
$3,000 worth of local services but pays only $2,000 in local property taxes. Sucb
a bousebold generates a fiscal burden because its tax contribution is less than the
cost of its local public services. Fiscal burdens may be generated by (1) bousebolds
living in higb-density bousing, (2) bousebolds living in the fringe areas of the city,
and (3) new co=ercial and industrial development.
High-Density Housing As will be explained later in the book, the most importan!
source of local revenue is the property tax. Because a bousebold's tax liability
increases with its bousing cbnsumption, a bousebold in a small bouse or apartment
pays arelatively small amount in taxes, and is therefore more likely to impose a fiscal
burden on local government: Local governments exclude bousebolds by zoning for .
low-density (higb-value) bousing.
.
.
Largeclot zoning establisbes a Iili.himum lot'sze for residential development.
Land and bousing are complementary goods, so the city can use restrictions on
238
lot size to indirectly establish a mnimum house value. Suppose that a city break:s
even on a house worth $125,000: the tax liability of such a house equals the cost
of supplying public services to the household living in the house. According to one
rule of thumb, the market value of land is about 20 percent of the total property
value (the value of the structure and the land). If the price of land is $50,000 per
acre, a half-acre lot ($25,000 worth of land) produces a house worth $125,000 (five
times $25,000). A mnimum lot size of one-half an acre ensures that the govemment
breaks even on al!new houses.
To explain the effects of large-lot zoning on the urban land market, consider
an unzoned city with 120 acres ofundeveloped land. As shown in Figure 10:-5, the
equilibrium price ofland is $20,000 per acre. At this price, the total demand for land
equals the fixed supply, and land is divided equally between single-family homes
(SFHs) and apartments (60 acres in each).
Suppose that the city passes a zoning ordinance that limts apartment buildings
to a total of 40 acres of land. What happens to the equilibrium price of land? In
Figure 10-5, the city moves up the demand curve for apartment land from point B
to point E: since only 40 acres are available for apartments, developers are willing
to pay $40,000 per acre for apartment land. At the same time, the city moves down
the other demand curve from point Ato point C; there are now 80 acres available
for single-family homes, and the surplus of land decreases the price per acre from
$20,000 to $15,000.
This example shows the effects of large-lot zoning on landowners and housing
consumers. Zoning makes land for apartments a relatively scarce comrnodity, so the
owners of apartment land gain from large-lot zoning. In contrast, zoning makes land
for single-family homes a more plentiful commodity, so the owners ofland destined
for single-family J:10mes lose. What about housing consumers? The increase in the
cost of apartment land increases the costs of apartments, so apartment dwellers lose.
In contrast, the land cost of single-family homes decreases, so home buyers gain.
Exclusionary and lnclusionary Zoning Large-lot zoning is often labeled exclusionary zoning because it excludes low-income households. The exclusion is both
direct and indirect. In the example above, large-lot zoning prevented the construction of 20 acres of apartments. Because low-income households are more likely to
occupy apartrnents, the decrease in the number of apartrnent buildings decreases the
number of low-income residents. In addition, the zoning policy increases the price
of apartments, indirectly excluding low-income households: the apartments that are
built are too expensive for many poor households.
In response to the exclusion problem, many cities have developed inclusionary
zoning policies. Under an inclusionary housing program, the local govemment forces
local developers to build dwellings for low-income households. The dwellings are
typically sold to low-income households for less than the cost of construction. By
providing subsidized dwellings for low-income residents, the local comrnunity partly
offsets the exclusionary effects offiscal zoning.Most dwellings built for inclusionary
purposes are not typicallow-income dwellings, but are medium-income dwellings
provided to the poor at the price of low-income dwellings.
40
1l"'
&
40
Demand for
SFH!nnd
30
30
-o, 20 .
:; .
;;::: 15
o
Demand for
1
1
partment land
1
1
1
1
1
1
-----------20
--------------,1 ---- e
.g
1
1
""
1
1
1
20
40
80
60
Land for single-family homes (acres)
10
1
----.----1---1
1
1
1
1
1
1
1
1
1
-B--
'1
1
1
1
20
40
60
80
Land for apartment buildings (acres)
A fixed amount of land (120 acres) is divided between single-family bornes ad apartments. In the absence of
Iarge-lot zoning, the price of land is $20,000 per acre, and land is divided equally between apartments and bornes
(60 acres for each type of housing; poinls A and B). If apartment acreage is limited to 40 acres (point E), 80 acres
are left for single-family homes (point C). The price of apartment land rises to $40,000, and lhe price of land for
single-family homes drops to $15,000.
'C
240
Fringe Land Use New housing is usually built on the fringe of the metropolitan
area. If the costs of supplying public services are higher in fringe areas, the tax
contribution of new housing may be less than the cost of public services at the
fringe. For example suppose the development of fringe areas would require the
expansion of a city's public services, generating a cost of $2,000 per new house,
compared to $1,500 for existing houses. If fringe houses are subject to the same
taxes as existing houses, there will be a fiscal burden of $500 per fringe house. There
are several options for dealing with the fiscal burden of fringe housing.
l. Zone the vacant land for agricultura! use. If the city sets a rninimum lot size
of 20 acres, no additional houses will be built on the land, so the expru:sion of city
services is unnecessary.
2. Impose a tax surcharge for new development. A tax surcharge (for example,
$500per house) would eliminate the fiscal burden of new housing, so the city would
be more likely to approve fringe development.
3. One-time impact fee or development tax. By paying an impact fee, the developer compensates the loc"al government for the fiscal burden of new housing. If the
city collects a $5,000 impact fee from the developer and invests the money in the
bank at a 1O percent interest rate, it can use the annual interest earnings ($500) to
cover the difference between the tax contribution and the public-service cost of new
housing. Who pays the impact fee? In a competitive market, the developer makes
zero econornic profits, so the fee is passed on to consumers (housing prices increase)
and landowners (land prices decrease).
241
Commercial and Industrial Development Insorne cities, connercial and industrial development generates fiscal burdens. If the city cannot pass on the cost of new
infrastructure to new firms and employees, it may restrict development by (1) limiting
the supply of connercial and industrialland and (2) restricting building heights.
The city of San Francisco recently adopted a zoning plan that restricts building
heights in its downtown area. The objective of the policy is to limit the number
of people working downtown. By limiting the number of downtown workers, city
officials hope to control the volume of traffic on the bridges into the city, all of which
are aiready congested. An altemative to the zoning policy would be to increase the
carrying capacity of the bridges, with the cost of the capacity expansion imposed
on new employers. Since this is impractical, the city has decided to use its zoning
policy to limit total employment.
Sorne cities impose impact fees on connercial and industrial developers, using .
the revenue from the fees to expand local transportation networks. Por example,
in the Westchester area of westem Los Angeles, developers pay a one-time fee of
$2,010 for each additional rush-hour trip generated by new office buildings. The
revenue from the impact fee is used to widen the roads used by the employees of the
new office buildings. Impact fees can reduce the fiscal burden of new development,
decreasing the opposition to development.
Design Zoning
The third type of zoning is a forro of macro-architecture. Justas an architect designs
an individual house, the planner designs a city, arranging activities to promote the
efficient use ofthe city's infrastructure (e.g., streets, sewage systems, water systems).
Residential an9. employment growth is directed to areas where infrastructure can be
efficientlyprovided. Design zoning is also used to preserve open space.
242
75 --------r-------,
i'
g 60
8.
50
"O
.S
'b
8 30 -------
Southern land
development zone
Northern land
preservation zone
In the initial equilibrium, the price of land is $50,000 per acre. If the north area is
rezoned for preservation, the demand for southern land will increase (increasing its price
to $60,000), and the demand for northern land wil! decrease (decreasing itS price to
$30,000). Under a TDR policy, northern !andowners receive development coupons that
can be used to override density limits in the development zone. The. re!axation of density
. limits increases the market value of southern land from $60,000 to $75,000, so southem
landowners are willing ro pay up to $15,000 fot an acre's worth of development coupons.
r
1
243
land decreases its price to $30,000 because the landowner has fewer development
options. When northem land is taken offthe residentia! market, the resulting shortage
of residentialland increases the price of southem land to $60,000. Under a TDR policy, northem landowners receive development coupons that can be used to override
the density linllts (10 dwellings per acre) in the development zone. The relaxation
of density lirnits gives southem landowners more development options, increasing
the market value of their land from $60,000 to $75,000. Southem landowners are
willing to pay up to $15,000 for an acre's worth of development coupons, so the
market value of an acre's worth of transferable development rights is $15,000.
The TDRs dinllnish the inequities caused by preservation zoning. In the absence
of the transferable rights, the owner of the preserved land suffers a loss of $20,000
per acre ($50, 000 - $30, 000). In contrast, if the allowable density is 20 dwellings
per acre, the southem landowner receives a gain of $25,000 per acre. Under the TDR
policy, the owner of the preserved land sells her development rights to the landowners
in the development zone for $15,000 per acre, so her net loss from the zoning policy
is only $5,000. In contrast, the landowner in the deve1opment zone pays $15,000 per
acre for the rights to build more than 10 dwellings per acre, so his net gain from the
zoning policy is only $10,000 p"er acre.
What determines the market value of the development coupons? The coupons
have a positive price because they allow landowners in the development zone to
exceed the normal density restrictions. The m:arket price depends on the power of
the coupons to override density restrictions: if the coupons allow a 100 percent
increase in density, they are obviously more valuable than they would be under a
15 percent density bonus. The market price also depends on the underlying demand
for high-density housing in the development zone. Ifhouseholds in the development
demand low-density housing, a developer won't need the development coupons, and
the market price of the coupons will be relatively low. In contrast, if there is a strong
demand for high-density housing in the development zone, the market price of the
coupons will be relatively high.
Open-Space Zoning
Sorne cities zone parcels ofland as open space, green belts, or agricultura! preserves.
This type of zoning provides city dwellers with open space by denying landowners
the full use of their land.
244
50
Marginal benefit of
open space
-'"
"
8_
"O
.S
..o...
"
of open space
1
1
1
1.
1
1
1.
60
Acres of open space preserved
100
The opportunity cost of open space is the value of land in an alternadve use (housing).
The optimum amount of open space is 60 acres (where the marginal-benefit curve
. intersects the marginal-cost curve). If the govemrnent does not compensate landowners
for open-space zoning, it will preserve 100 acres (the point at which the marginal
benefit of open space_js zero).
willing to pay $20,000 for undeveloped land, the marginal cost of open space is
$20,000 per acre. If the city purchases land for open space, it will outbid developers
for 60 acres of land. This is an efficient allocation of land; for the first 60 acres of
open space, the value of land as open space exceeds the value of land in housing.
For the 6lst through the 100th acres, the value ofland in housing exceeds the value
as open space, so 40 acres are developed.
Suppose that the city simply zones 1and for open space. From the perspective
of city officials, the marginal cost of open space is zero. The city preserves allland
for which the marginal benefit (from the-.demand curve) is positive, so it zones for
100 acres of open space instead of 60. This is inefficient because the 61st through
the 1OOth acres are more. valuable as housing l d.
245
'
'
JI:
)J
:1
:f
T
_(
t
f
.\
1
\
1
1
1
!
There are two lessons from Houston's experience without zoning. First, in the
absence of zoning, landowners have the incentive to negotiate restrictions on land
use. It seems that neighborhood extema!ities are large enough to justify the cost
of developing and enforcing restrictive covenants. This is the Coase solution to
extemalities (named after R. H. Coase): The parties affected by extemalities negotiate
a contract to salve !he extema!ity probiem. Second, in the absence of zoning, most
industrial firms cluster in locations accessible to the transportation network, and
most retailers cluster in shopping centers and retail strips. If a city were to drop its
zoning plan, it is unlikely that glue factories and pizza parlors would invade quiet
residential neighborhoods.
r
.
246
patterned after the Standard State Zoning Enab!ing Act, which was developed by
the U.S. Department of Commerce in 1926. Section 1 of the Enabling Act states:
Grant of Power. For the purpose of promoting health, safety, morals, or the general
welfare of the community, the legislativo body of cities and incorporated villages is
hereby empowered to regulate and restrict the height, number of stories, and size of
build.ings and other structures, the percentage of the lot that may be occupied, the size
of yards, courts, and other open spaces, the density of popu!ation, and the location and
use of buildings, structures, and land for trade, industry, residence, or other purposes.
247
Equal Protection
The second criterion forthe constitutionality ofzoning is equal protection. The equalprotection clause of the Fourteenth Amendrnent requires that alllaws be applied in
an impersonal (nondiscrirninatory) fashion.
Recent court cases have tested whether fiscal zoning violates the principie of
equal protection. Fiscal zoning excl des households who do not pay their share
of the cost of local public services. Local governrnents have been sued by people
outside the city, who clairn that fiscal zoning systematically excludes sorne types
of people from the city. The plaintiffs argue that zoning laws are not applied in an
impersonal manner, but instead treat sorne people differently than others; rich and
poor are treated differently, as are whites and blacks.
The federal courts have upheld the constitutionality of exclusionary zoning. In
the Euclid decision, the Supreme Court suggested that although a zoning ordinance
must generate sorne benefit for insiders (citizens of the community), the effects of
zoning on outsiders are unimportant. In Warth v. Se/den (1975), the court disrnissed
the clairns of outsiders because they did not prove that the zoning ordinance caused
specific personal damage. In Vzllage of Arlington Heights v. Metropolitan Housing
Corporation (1977), the court disrnissed the claims of outsiders because they did not
prove discriminatory intent on the part of zoning officials. In Ybarra v. Town of Los
Altos Hills, the court ruled that although zoning laws that discrirninate on the basis
of race are unconstitutional, zoning laws that discriminate on the basis of income
are legal. In general, the federal courts have adopted a noninterventionist approach
to exclusionary zoning.
Soli\e state courts have adopted a more activist role. In Southem Burlington
County NAACP v. Mount Laurel (1975), the New Jersey Supreme Court ruled that
Mount Lailiel's exclusionary zoning harmed low-income outsiders. The court directed the city to develop a new zoning plan under which the city would accommodate its "fair share" of low-income residents. The court established quotas for
communities to provide enough housing for low- and moderate-income workers to
live within reasonabe commuting distances of their jobs. The effects have been
minor, in part because the state legislature modified the quotas and even allowed
communities to buy and sell up to half their quotas (Milis and Lubuele, 1997). A
nearby high-income community could sell up to half its quota to Trenton ata negotiated price. The result was presumably to improve the housing, if not the schools
and security, of low-income residents, but it hardly contributed to the integration of
suburbs or to an increase in affordable housing close to suburban workplaces.
Other states have ruled on exclusionary zoning.The implication fromAssociated
Home Builders Inc. v. City of Livermore (California Supreme Court, 1976) is that the
courts will judge zoning on the basis of its effects on both insiders and outsiders. If a
zoning ordinance does not represen!a reasonable accommodation of the competing
interests of insiders and outsiders, it may be declared unconstitutional. In Oregon,
state lawrequires municipalities to plan and zone land for a diversity ofhousing types
and income levels. According to Fischel (1985), few state courts have followed
the lead oftheNewJersey court, so the Mount Laurel decision has not affected
248
exclusionary zoning practices in many states. Given the decisions from federal courts,
. exclusionary zoning is still considered legal,
Just Compensation
The third criterion for the constitutionality of zoning is just compensation. The Pifth
Amendment states "... nor shall private property be taken for public use, without
just compensation." This is the taking clause: if the government converts land from
private to public use, the landlord must be compensated.
How is the taking clause applied to zoning? Most zoning ordinances do not
actually convert land to public use, but merely restrict private use. Por example,
nuisance zoning prevents a landowner from building a factory in a residentia! area,
and large-lot zoning prevents a landowner from building high-density housing. By
restricting the use of private land, zoning decreases the market value of the property."
The policy issue is whether landowners should be compensated for the loss
of property value caused by zoning. Por example, if large-lot zoning decreases a
landowner's property value by $5,000, should the local government pay $5,000
in compensation? According to Pischel (1985), the courts have provided rnixed and
confusing signals to local zoning authorities. The courts routinely uphold zoning laws
that cause large losses in property values, suggesting that as long as the landowner
is left with sorne profitable use of his land, compensation is not required. The courts
have developed four rules to determine whether compensation is required.
l. Physical invasion. Compensation is required if the government physically occupies the land. The invasion rule is applicable only when the government actual!y
occupies the land. kdoes not apply to most zoning actions, in which the government
merely restricts private land use.
2. Diminution of value and reasonable beneficia! use. The origin of this rule
is Pennsylvania Coal v. Mahon (1922), in which Justice Holmes states, "... while
property may be regulated to a certain extent, if regulation goes too far it will be
recognized as a taking."In other words, compensation is required if zoning decreases
the property value by a sufficiently large amount. Unfortunately, the courts have not
indicated how far zoning must go before comp nsation is required. A related rule is
reasonable beneficia! use: If zoning leaves the;landowner with options that provide
a reasonable rate of return, no compensation is required.
3. Balancing means (benefit-cost). According to this rule, compensation is required if the cost of a zoning ordinance (the loss in property value) exceeds the
bene:fit to the local community. Because this iule requires compensation for inef:ficient zoning ordinances, it discourages inefficient zoning practices. Por example,
suppose that a zoning ordinance decreases property values by $50 and generates a
$30 bene:fit to the comrriunity. According to the balancing-means rule, compensation
must be paid. Taxpayers will oppose the ordinance because the cost of compensating
landowners exceeds the benefit generated by the ordinance. If the balancing-means
rule were used, voters would approve. only efficient zoning ordinances.
4. Harm prevention. According to this rule, compensation is not required if the
zoning ordinance prevents a harmful use of the land. In other words, zoning is not a
249
taking if it prevents the landowner from using land in ways that are detrimental to
the general public. The harm-prevention rule suggests that a landowner, like a car
owner, has limited property rights. The car owner has the right to drive her car, but
she must stop at red lights. Should the driver be compensated for the opportunity
cost of time spent waiting for the light to turn green? Since the traffic lights prevent
a harmful use of the car, compensation is not required. Sirnilarly, landowners have
limited property rights: If zoning prevents the landlord from building a polluting
factory inaresidential district, compensation is unnecessary because nuisance zoning
prevents a harmful use of the land. Most zoning ordinances are judged by a broad
interpretation of the harm-prevention rule: if an ordinance prometes public health,
safety, or welfare, compensation is usually not required.
SUMMARY
l. To control population growth, sorne cities use urban service boundaries to limit
2.
3.
4.
5.
6.
the land area available for deve!opment, and others limit the number of building
permits.
A zoning plan designates a set of admissible land uses for each plot of land in
the city. There are three types of zoning: nuisance zoning (reducing exposure
to pollution), fiscal zoning (excluding households that might impose a fiscal
burden), arid design zoning (arranging activities within the metropolitan area to
improve aesthetics or circulation).
Under an inclusionary zoning policy, developers provide subsidized new dwellings to low-income residents. The cost of inclusionary housing is borne by
housing consurners (higher housing prices) and landowners (lower land prices).
Fringe <;!evelopment often irnposes a fiscal burden on the city. The taxes from
new developrnent fa!!short of the cost of extending public services to the urban
fringe.
Under a system oftransferable development rights, the market value of a development coupon is determined by (a) the power ofthe coupon to override density
restrictions and (b) the underlying demand for high-density developrnent in the
development zone.
The zoning authority of!ocal governments comes from state governments.There
are three criteria for the constitutionality of zoning: due process, equal protection, and just compensation.
r
.
250
''According to the leftover principie, the landowner gets the leftover (the excess
of total revenue over total cost), so an increase in the price of housing should
increase-not decrease-the price ofland. Either the leftover principie is wrong
or the analysis of the permit policy is wrong."
3. Consider the building-permit policy depicted in Figure 10-3, with an equilibriurn price of permits equal to $30,000. Suppose that the city announces on
January 1 that 300 days later (October 2S) it will give the 60 permits to the first
60 licensed building contractors through the planning office door. Because the
police department expects a line to forrn outside the planning office, the police
chief announces that the following queuing rules wil! be enforced: (i) Nocuts:
When a person joins the queue, he or she goes to the end of the queue; (ii) No
substitutions: No one can reserve a place in !ine for anyone else.
To receive a permit, a !icensed contractor must be one of the first 60 people
in line and must remain in the line until October 28. Therefore, instead of an
equilibrium price for the perrnits, there is an equilibrium waiting time (time
spent in line). Suppose that 25 of the city's 100 licensed contractors have an
opportunity cost of $150 per day; 25 have an opportunity cost of $300 per day;
25 have an opportunity cost of $600 per day; and 25 have an opportunity cost
of $1,000 per day.
a. What is the equilibriurn waiting time?
b. Suppose that the city eliminates the no-substitution rule. Would you expect
the equilibriurn waiting time to increase, decrease, or not change?
4. Use Figure 10--4 to show the effects ofthe following changes on the optimum
location of the polluting factory. Do the changes cause the optimurn location to
move eloser to or farther from the residential area?
a. The unit commuting cost (cost per mile) increases.
b. Residents become more sensitive to pollution.
c. The polluting firm discovers a new method of pollution abatement that cuts
abatement cost in half.
5. Considera city where a polluting industry is separated from the residential area
by a 5-mile buffer. The following table !ists the wages and effiuent fees for
different locations within the buffer zone: the closer the firrn to the residential
district, the lower the wage and the higher the effiuent fee. Suppose that the firrn
uses 450 hours of labor per week and generales 20 units of pollution per week.
Assume that the quantities of output and ihput are constant. What is the social
cost of the segregation zoning po!icy? In other words, if the city prevents the
firrn from locating at its optimum location, what is the cost to society?
Commute Distance
(miles)
5
4
3
2
Effluent Fee
($ per ton)
!.O
2.5
4.5
7.0
10.0
Wage
($ per hour)
10.0
9.9
9.8
9.7
9.6
251
6. Suppose that a city eliminates all zoning. What types of retailers would you
expect to move from the retail zones to the residential zones?
7. Suppose that all local taxes are eliminated and all funding for local public
services comes from state governments. Would you expect any changes in local
zoning practices?
8. Suppose that you are a member of the zoning board in a growing city. A residential developer appears before the board with a proposal to develop a parcel of land at the city's edge. The proposed development would increase the
population of the city by 5 percent. The developer suggests that the development will increase tax revenue and the cost of public services by 5 percent,
so it won't generate a fiscal burden. Is the developer correct? What additional
information would you need to evaluate the fiscal effects of the proposed development?
9. In a certain city, all new housing is built on thecity's edge (lO miles from the
city center). The city institutes the followiirg policy: Every new house will be
subject toa $1,000 impact fee or development tax. The fee is paid by the firm
that builds the house.
a. Who will actually pay the development tax?
b. Will the development tax affect the radius of the city? Explain, using a
graph.
c. Would you expect the implementation of the development tax to change the
city's zoning practices?
10. In Figure 10-5, large-lot zoning increases the price of apartment land and decreases the price of land for single-farnily homes. Under what circumstances
(what values ofthe relevant elasticities) will the zoning policy increase the total
value of land in the city?
11. Consider a large metropolitan area in which there is a large number of suburban municipalities, each of which has sorne vacant land. In each municipality except municipality Z, high-density housing (apartments) is explicitly
outlawed. Suppose that Z starts out with no controls on residential1and use,
and its vacant land is expected to be divided equally between high-density
housing (apartments) and single-farnily homes (SFHs). The price of land for
apartments equals the price of land for SFHs. Suppose that Z institutes a largelot zoning policy that limits high-density housing to one-quarter of its available
land.
a. Depict graphically the effects ofZ's zoning policy on the prices of apartment
land and SFH land within the municipality.
b. Will the zoning policy increasor decrease the total value of land within
municipality Z?
12. Suppose that a city has recentiy proposed the rezoning of a 1O-acre paree!ofland,
. from low-density use (l dwelling per acre) to high-density use (20 dwellings
per acre).
a. Who will support the zoning change? Who will oppose the rezoning?
b. Describe a policy under which all parties will agree to the rezoning if in fact
the rezoning is efficient.
'
r
.
252
'
'
13. Ollie recently bought the Notel Hotel in downtown Portland and intends to de-
molish the hotel and replace it with an office building with 100,000 square feet of
office space. When the city planner hears about Ollie's plans, he quickly rezones
the hotel site as a historical preserve, preventing the demolition of the hoteL
a. You have been asked to estimate Ollie's loss from the rezoning of his land.
What infomiation would you collect, and how would you use it?
b. The city offers Ollie development coupons (TDRs) that permit the development of 100,000 square feet of office space somewhere else in the city. What
determines the market value of the coupons?
c. Under what circumstances will Ollie be worse off as a result of the zoning
and TDR policy?
14. Consider the North/South TDR example in the section on design zoning. Suppose that instead of issuing 10 development coupons per acre (250 in total), the
government issues 5 per acre (125 total).
a. Will the new policy increase or decrease the equilibrium price of TDRs?
b. Under what conditions;(what values ofthe relevant elasticities) will the typical TDR recipient be better off with 5 coupons instead of 1O?
15. Suppose that city U establishes an ui"ban service boundary five miles from the
city center. People who own land outside the boundary object to it and demand
compensation. Design a practica! scheme under which those who are harmed
by the service boundary will be compensated for their losses. Be specific. How
will the money for compensation be raised? Who will actually pay for the compensation package? How will you determine the amount of compensation per
landowner?
16. Using FigureJ0-7, compute the total welfare loss caused by a switch from the
optimum policy (purchase of land for open space) to open-space zoning.
' ity provides open space by purchasing land from landowners,
17. Suppose that a C
not by zoning land for open space. If someone suggests that the city buy another
plot of land for open space, the mayor appears on local television and solicits
contributions to the city's open-space fund. The city will purchase the land for
open space if the total contributions exceed the market price of the land. If the
contributions do not cover the cost of the land, the city will not buy the land and
will return the contributions to the individual donors. Will the city provide the
optimum amount of open space?
18. Consider the following quote:"Depending on the variable one controls, a zoning
policy may eitherincrease or decrease the price ofundeveloped land within and
around the zoned city." Explain this statement, using examples of the variables
that are controlled by the various zoning policies.
19. Consider two of the compensation rules developed by the courts: the dirninutionof-value rule and the balancing-means rule. Will these rules promote horizontal
equity (the equal treatment of equals)?
20. Comrnent on the following statement: "The courts should not be involved in
overseeing local zoning practice. Such matters are the responsibility of the legislative branch of government. Let democracy work."
'
253
Steven R. Schell. "The Need for a Standard S tate Impact Fee Enabling A et."
Joumal ofthe American Planning Association 54 (Winter 1988), pp. 7-17.
3. Siegan, Bernard. Land Use without Zoning. Lexington, MA: D. C. Heath, 1972.
4. Milis, Edwin S.; and Luan Sende Lubue!e, "Inner Cities." Journal O Economic
Literature 35 (1977), pp. 727-56.
'1. '.--_-'.
254
---== "
PART THREE
Urban Transportation
within a metropolitan area.This part ofthe book examines the two main components
of the urban transportation system. Chapter 11 describes the automobilelhighway
system, focusing on three problems caused by automobiles: congestion, pollution,
and highway accidents. Chapter 12 explores the economics of urban mass transit,
focusing on the commuter's choice of a mode oftravel (e.g., automobile versus mass
transit) anda city planner's choice of a mass-transit system (e.g., buses versus light
rail versus heavy rail). The chapter explains why so few commuters in the United
States use mass transit and why light-rail and heavy-rail transit systems are usually
less efficient than bus systems.
'
255
CHAPTER 11
The home is where pan of the fami/y waits unti/ the others
are through with the car.
-HERBERT PROCHNOW
Tbis
257
r
.
258
'2
--:--
50
40
"E 30
E
o
."
.o...
"E
z"
.e
--
20
10
O Walk
O Bus or rail
i!ll e ool
O Drive alone
Central-city residents
Suburban residents
1.96
3.93
1.45
4.72
23.!1
1.24
7.07
43.73
Source: U.S. Bureau of the Census, 1990 Census of Popularion. an.d Housirtg, SSTF20, Joumcy to Work in
the Uniced States, (Washington, D.C.: U.S. GovernmentPrinting Offi.ce,l994).
259
2. Monetary travel cost. The monetary cost of auto travel is 20 cents per mile.
3. Time cost. The opportunity cost of travel time is 10 cents per minute.
The total cost of a commuting trip is the surn of the monetary and time costs of
the 10-mile trip. The monetary cost is $2 (10 miles times 20 cents per mile). The
time cost depends on travel time: A trip that takes 30 minutes has a time cost of $:
(30 minutes times 10 cents per minute); a 20-minute trip has a time cost of only $2.
Therefore, the total cost of a 30-minute trip is $5 ($2 plus $3), and the total cost of
a 20-minute trip is $4.
Social trip
cost
H
12.80 ----------1
11.40
-----------, --
1
1
9.14
-- ------- -r----
1
1
1
:;;
o
"
"5"
7.13
Private trip
cost
1
1
5.48
4.80
1
1
1
1.82
1,000
1
(
(
., ..,.,._
{
---
.-..-.--
----------- 1 ------1---------1
1
1
'{_
1,200
1,400
1,600
Vebic1es per lane per hour
1,800
Drivers use the bighway as long as the marginal benefit (shown by the demand curve) exceeds the
private trip cost, so the equilibrium traffic volurne is 1,600 vebicles per lane per hour. At the
optimurn, the marginal benefit equals the marginal social cost (the social trip cost). The equilibrium
volume (1,600) exceeds the optimum volume (1,400) because drivers ignore the externa! costs of their
trips.A congestion tax of $4.34 intetrialiZes the co:ngestion extrnaliiy, generating the optimurn
traffic volume.
260
The demand curve shows, for each trip cost, how many travelers use the highway.
For example, if the trip cost is $12.80, there are 1,200 peoplefor whom the benefit
of the trip exceeds the cost, so the traffic volume is 1,200 vehicles per lane per
hour. As the cost ofthe trip decreases, there are more people for whom the benefi.t
exceeds the cost, so the city moves downward along the travel demand curve: there
are 1,400 vehicles ata cost of $9.14, and 1,600 vehicles ata cost of $5.48.
The demand curve is a marginal-benefit curve. For each traffic volume, it
shows how much the marginal traveler is willing to pay for the highway trip. Suppose
that the city starts with a trip cost of$9.15 anda traffic volume of 1,399. If the trip
cost decreases to $9.14, traffic volume increases to 1,400, meaning that the 1,400th
driver is wllling to pay $9.14 to make the trip; at any cost above $9.14, the trip
would not be worthwhile. The demand curve shows that the marginal benefi.t of the
10-mile trip decreases from $12.80 for the 1,200th driver to $1.82 for the 1,800th
driver,
When the 401st drive.r enters the highway, the congestion threshold is crossed. As the
highway becomes crowded, the space between vehicles decreases and drivers slow
down to maintain safe distances between cars. As more and more drivers enter the
highway, travel speeds decrease and travel times increase; the trip takes 12.8 minutes
TABLEll-1 Traffi.c Volume, Travel Time, Travel Cost, and Congestion Externalities
e
D
E
.F
G
H
Increase in Increase in
Traffic
Travel Time Total Travel Externa! Private
Social Marginal
Volume Trip Time perDriver
Time
Trip Cost Trip Cost Trip Cost Benefit
(vehicles) (minutes) (minutes)
(minutes)
($)
($)
($)
(demand)
A
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
12.0
12.0
12.8
14.8
18.0
22.4
28.0
34.8
.. - 42.8_
52.0
o
o
0.007
0.013.
0.019
0.025
0.031
0.037
0.043
0.049
o
o
4.2
10.4
19.0
30.0
43.4
59.2
77.4
98.0
o
'o
0.42
1.04
1.90
3.00
4.34
5.92
7.74
9.80
3.20
3.20
3.28
3.48
3.80
4.24
4.80
5.48
6.28
7.20
3.20
3.20
3.70
4.52
5.70
7.24
9.14
ll.40
14.02
17.00
31.10
27.44
23.78
20.12
16.46
12.80
9.14
5.48
1.82
- .... , ..
,_
,'
261
if there are 600 vehicles, 22Aminutes for 1,200 vehicles, and 52.0 minutes for 2,000
vehicles.
Columns C and D show the effects of the marginal driver on the travel times
of other drivers. For low traffic volumes (below 400 vehicles), an additional driver
do.es not affect speeds or travel times. For volumes above .400 vehicles, however,
an additional driver slows traffic and increases travel times. For example, the 600th
driver increases the travel time per driver by 0.007 minutes (colurnn C); when the
driver enters the highway, the trip time increases from 12.793 to 12.80. The increase
in total travel time is simply the extra time per driver (0.007) times the number of
other drivers (599), or 4.2 minutes (colurnn D). This is the congestion extemality:
The marginal driver slows traffic and increases travel time, forcing other drivers
to spend more time on the road. The congestion extemality increases with traffic
volume; the 1,200th driver increases travel time by 30 minutes (0.025 minutes times
1,199), and the 1,600th driver increases travel time by 59.2 minutes (0.037 times
1,599).
The externa! trip cost equals the monetary value of the congestion extemality.
The figures in column E are based on the assumption that the opportunity cost
of travel time is 10 cents per minute. For the 600th driver, the extemal trip cost
equals the increase in total travel time (4.2 minutes) times 10 cents per minute, or
42 cents. The extemal trip cost increases with traffic volume, rising to $3.00 for the
1,200th driver, $5.92 for the 1,600th driver, and $9.80 for the 2,000th driver.
Columns F and G show the private and social costs of travel. The private trip
cost is the travel cost incurred by the individual commuter, defined as the sum of
the monetary cost ($2.00) and the private time cost. To compute the private time
cost, multiply the trip time by the opportunity cost of travel time. The opportunity
cost is 10 c nts-per minute, so the private time cost is $1.20 for a volume of 200
vehicles, $1.28 for 600 vehicles, and so on. Therefore, the private trip cost is $3.20
for 200 vehicles, $3.28 for 600 vehicles, and so on. The social trip cost is the sum
ofthe private trip cost (column F) and the extemal trip cost (colurnn E). Figure 11-2
shows the cost curves for private trip cost and social trip cost. The social-cost curve
lies above the private"cost curve, with the gap between the two curves equal to the
extemal trip cost.
There are sorne altemative labels for private and social trip costs. An altemative
label for private trip cost is average travel cost, defined as total travel cost divided
by the number of drivers. Since each driver travels at the same speed and thus has the
same travel cost, the average travel cost equals the private trip cost. An altemative
label for social trip cost !s marginal travel cost, defined as the ic"lcrease in the total
cost of travel resulting from adding one more traveler. Since the social trip cost
equals the trip cost incurred by the marginal driver plus the extemal cost he imposes
on other drivers, the social trip cost is the same as the marginal travel cost.
Equilibrium versus Optimum Traf:fic Volume
....What is the equilibrium number.of drivers? A driver uses the higbway if.tlle.Jllarginal
benefit of a trip (from the demand curve) exceeds the private trip cost.InFigure 11-2,
262
the demand curve intersects the private-trip-cost curve at 1,600, so the equilibrium
number ofvehicles is 1,600 and the equilibrium private trip cost is $5.48. The 1,60Ist
driver does not use the highway because the marginal benefit of using the highway
is less than the trip cost.
What is the optimum number of drivers? The basic efficiency rule is that an activity should be increased as long as the marginal social benefit exceeds the marginal
social cost; at the optimum leve!,the marginal benefit equals the marginal cost. In
Figure 11-2, the marginal social benefit is shown by the demand curve, and the
marginal social cost is shown by the social-trip-cost curve. The demand curve intersects the social-trip-cost curve ata volume of 1,400, so the optimum traffic volume
is 1,400 vehicles. For the first 1,400 drivers, the social benefit of travel exceeds the
social cost, so theiruse ofthe highway is ef:ficient. For the 1,401st driver, the social
cost exceeds the benefit, so her use of the highway is inefficient.
The equilibrium volume exceeds the optimum volume because drivers ignore
the costs they impose on other drivers. An additinal driver slows traffic, forcing
other drivers to spend more time on the road. Suppose that Carla, the 1,599th driver,
has a private benefit of $5.49. From column Fin Table 11-1, the private trip cost for
1,599 vehicles is about $5.48, so Carla uses the highway. Her use of the highway is
inef:ficient because the benefit ofthe trip ($5.49) is less than the social cost ofthe trip
($11.40, equal to the sum of Carla's private cost of $5.48 and the externa! trip cost
of $5.92). Because Carla ignores the $5.92 externa! cost, she makes an inefficient
choice.
263
In this example, both cjtizens benefit from the imposition of congestion taxes.
. A more rigorous analysis of the benefits and costs would sliw that sorne people
264
in the city would be harmed by the congestion tax policy. Por sorne travelers, the
savings in travel costs and the transfer payment would not be large enough to offset
the congestion tax or the loss in consumer surplus.
A more rigorous analysis would also show that the winners' benefits exceed the
losers' costs.In other words, the move from the equilibrium to the optimum generates
a net gain for society. In Figure 11-2, the net gain is shown by the shaded area, the
area between the demand curve (the marginal-benefit curve) and the social-trip-cost
curve.
To explain the computation of the net gain from moving to the optimurn, consider
a small move toward the optimum. Specifically, suppose the city starts with the
equilibrium volurne (1,600 vehicles) and somehow persuades the 1,600th driver to
not use the road. What are the benefits and costs of diverting the driver?
Benefit: The total travel cost decreases by the social trip cost for the 1,600th
driver (about $11.40 at point J).
Cost: The driver loses the benefits of the highway trip; the willingness to pay
for the trip is shown by the demand curve (about $5.48 at point B).
The net benefit from diverting the 1,600ih driver is the difference between the social
trip cost (a benefit of $11.40) and the willingness to pay (a cost of $5.48), or $5.92.
To compute the net gain to society from moving to the optimum, we repeat
this thought experiment for the 1,401st through the 1,599th driver. The social gain
from diverting the 1,599th driver is slightly lower than the gain from diverting the
1,600th driver because the social trip cost is lower (the city starts lower on the cost
curve) and the willingness to pay is higher (the city starts further up the demand
curve). The net gain to society from moving from the equilibrium (point B) to the
optimum (point C) is lhe surn of the differences between the social trip cost and
the willingness to pay (from the demand curve) for the 1,600th through the 1,401st
drivers. In other words, it is the area between the two curves, shown by the area of
triangle CJB. Because there is a net gain from the move to the optimum point, the
govemment could, in principie, redistribute income from the winners to the losers
to ensure that everyone is made better off by the congestion tax.
265
Social trip
cost
Privare trip
cost
Peak-period tax
1
1
1
. Off-peak
demand
Peakdemand
1
1
1
1
v.
Traffic volume
During the off-peak period, the demand for travel is relatively low, generating a low traffic volume
(V., compared to Vp during the peak period), so the optimum congestion taxis relatively low.
266
r:
1
about 15 cents per mile, with higher taxes on the most congested roads. The average
peak-period trip is 10 miles long, so the typical peak-period cornrnuter would pay
about $1.50 per one way trip, or $3.00 per day. The congestion tax would decrease
traffic volume during the peak period by about 26 percent, increasing travel speeds
and reducing delays..
r,
Chapter JI
267
and 4 cents on the weekend. The system uses. a trip-toll method that, with the help
of a transponder in each car, determines when and where a vehicle enters and exits
the tollway; a central system matches entry and exit registrations and computes the
vehicle's toll for the trip. The toll road is open to occasional users without in-vehicle
transponders: the system photographs license plates and sends bilis to the registered
owners.
One approach that is gaining popularity involves changing the rules for highoccupancy vehicle (HOV) lanes. An HOV lane-sometimes known as a "diamond"
lane or an express lane-is a lane designated "for use by high-occupancy vehicles,
typically defined as a vehicle with at least three passengers. A HOT lane is a lane
that can be used both by high-occupancy vehicles and other vehicles that pay a toll
(HOT stands for "high occupancy and toll").
The first HOT project was along Riverside Freeway (California State Route
91), which connects emp1oyment centers in Los Angeles and Orange Counties with
rapidly growing areas to the east. Two HOV lanes that had been added in the median
strip of the freeway were switched to HOT lanes. The toll varies by time of day, with
the highest toll ($2.75 per trip) l;letween 5 A.M. and 9 A.M. on weekdays. Each car
has a transponder in its windshield for identification purposes and a corresponding
account that is billed each time the vehicle uses the HOT lane. Changing to a HOT
lane increased traffic volume on the old HOV lanes, with about 80 percent of users
paying the toll. The change also decreased traffic volume and increased speeds along
the regular lanes on Route 91, generating benefits for commuters who did not pay
the toll.
Another HOT project is along Interstate 15 in San Diego. The reversible facility
consists of an eight-mile stretch of two lanes in the freeway median, which are
accessible olly at the endpoints ofthe facility. Afee is charged for each trip, with the
toll varying "real time"from $0.50 to $4.00, depending on the leve!of congestion.
Each vehicle has a transponder and a prepaid account that is billed each time the
vehicle uses the HOT lane. The toll is highest between 7 A.M. and 8 A.M., and 4 P.M.
and5 P.M.
These recent exieriences with congestion pricing are prornising. Travelers respond to higher prices by changing their travel behavior in ways that decrease traffic volume and improve the efficiency of travel. The most frequent responses are:
(1) forming carpools, (2) switching to :mass transit, (3) switching to off-peak travel,
(4) picking altemative routes, and (5) combining two or more trips into a single trip
(Small and Gomez-Ibanez, 1998). In many other cities in the United States, Europe,
and Asia, policymakers are evaluating the merits of congestion pricing.
268
the efficacy of the alternative policies, it will be useful to list the four ways that the
congestion tax decreases traffic volume:
l. Modal substitution. The tax increases the cost of auto travel relative to carpooling and mass transit (buses, subways, light rail), causing sorne travelers to
switch to these other travel modes.
2. Time of travel. The tax is highest during the peak travel periods, causing some
travelers to travel at different times. Because work and study schedules are
relatively inflexible, commuters and students would be less likely to change
their travel times than other travelers (e.g., shoppers). Nonetheless, firrns would
have a greater incentive to change work schedules to allow their workers to
avoid costly travel during the peak period. The institution of flextime and the
rearrangement of shift times would cause sorne workers to change their travel
times.
3. Travel route. The congestion tax is highest on the most congested routes,
causing sorne travelers to switch to alternative routes.
4. Location choices. The congestion tax increases the unit cost of travel (travel
cost per mile), causing sorne commuters to decrease their commuting distances.
Sorne workers may move closer to their jobs, and others may switch to jobs
closer to their residences.
These four changes cause the city to move up the travel-demand curve as the cost
of travel increases.In Figure 11-2, the congestion tax decreases traffic volume from
1,600 to 1,400 because it changes travel modes, times, routes, and distances.
Gasoline Tax
,
One alternative to the congestion tax is a gasoline tax. The simple idea is that if
travel is more expensive, traffic volume decreases. The problem is that the gas tax
increases the costs of all automobile travel, not just travel along congested routes
during peak periods. In contrast with the congestion tax, which changes travel times
and routes, the gas tax does not encourage drivers to switch to other travel times or
routes. Suppose the governrnent wants to use the gasoline tax to decrease the peak
period traffic volurne to its optimum leve!. Wht is the required gasoline tax? To have
the same effect as the congestion tax in Figure 11-2, the gas tax must increase the
cost of travel by 43.4 cents per rnile ($4.34 per 10-mile trip). If the typical car gets
22 miles per gallon of gasoline (the average mileage in 1992), the required gas tax
is $9.55 per gallon (43.4 cents times 22); if gas mileage is only 10 miles per gallon,
the tax is $4.34 per gallon. Even if a gas tax at this leve!were feasible, it would be
inefficient because it would also increase the cost of off-peak travel by 43.4 cents
per rnile.
ParkingTax
A number of cities use parking taxes to discourage driving to central business district
jobs. In an experiment in Madison, Wisconsin, a tax surcharge of $1 was irnposed
)tK.:
,..'.
:
'.
'
)(
.
el
1(
1.
::(
:l
r
t
269
on drivers who arrived at parking garages during the peak travel period (7 A.M. to
9 A.M.) and left their cars for more than three hours. As explained by Parody (1984),
the surcharge decreased traffic volume because (1) sorne commuters switched tocarpools and mass transit and (2) sorne changed their travel times. In Washington, D.C.,
the govemment increased parking costs for its employees, causing sorne workers to
switch to carpools and mass transit (Miller arid Everett, 1982).
In Ottawa, Canada, the govemment increased parking rates for govemment
employees from zero to 70 percent of the commercial rate, causing (1) a 23 percent
decrease in the number ofworkers driving to work, (2) a 6 percent increase in the auto
occupancy rate (from 1.33 to 1.41), and (3) a 16 percent increase in bus ridership
(DiRenzo, Cima, and Barber, 1981). In a series of experiments in Los Angeles,
the elirnination of free parking decreased the number of solo drivers by between
18 percent and 83 percent. When one Los Angeles firm increased its parking fee
from zero to $28.75 per month; the number of solo drivers dropped by 44 percent;
when the firm increased the monthly fee to $57.50, solo driving decreased 81 percent
(Small, 1992).
There are three potential problems with using parking taxes to decrease congestion. The first is that the taxes must be imposed only on peak-period commuters;
drivers who travel during the off-peak periods should be exempt from the tax. As
shown by the Madison experiment, this problem can be solved by imposing a surcharge for drivers who arrive at parking garages during the peak travel periods.
Second, in contrast with the congestion tax, which increases the unit cost of travel
and decreases travel distances, the parking tax does not depend on the distance traveled. Therefore, there is less incentive for commuters to economize on travel cost
by living closer to their workplaces. Third, because much of the congestion problem
is caused by car that do not park in congested areas, the tax does not force all
peak-period travelers to pay for the congestion they cause.
Shoup (1993) suggests that the subsidization of parking by employers contributes to the congestion problem. In 1990, about 95 percent of American commuters who drove to work benefited from free parking at their place of work. In Los
Angeles, the average subsidy froin free parking provided by employers is $3.87 per
day. Shoup estimates that employer-paid parking shifts 25 percent of al! commuters
into solo driving and increases the number of cars driven to work by 19 percent. One
possible response to this problem is to "cash out" the parking subsidy. An employer
that provides free parking worth, say, $150 per month, would give its workers the
option of free parking or $150 in cash. When faced with the cash option, many
workers would take the cash and make other arrangements for getting to work, thus
decreasing traffic volume.
270
FIGURE 11-4 Effects ofWidening the Highway
Increase in consumer
surplus
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
vo
vw
Demand
(willingness to pay)
1
1
Traffic volume
The widening of the highway shifts the private trip cost to the right: the congestion threshold
increases, and trip cost is lower at every traffic volume above the original threshold volume. The
decrease in trip cost increases traffic volume from V0 to Vw. The benefit ofthe widening is the
increase in consumer surpus (the shaded area).
threshold. The decrease in trip costs increases tra:ffic volume as the city moves down
the demand curve from point D to E.
Underwhat conditions would the widening ofthe highway be efficient? One way
to measure the total benefit is with consumer surplus. As explained earlier, consumer
surplus equals the excess of the willingness to pay (from the demand curve) over the
actual cost of travel. Widening the highwaydecreases the trip costs of all drivers, so
it increases consumer surplus.
In Figure 11-4, the increase in consumer surplus is shown as the shaded area.
For the original drivers (Volume = V0 ), the consumer surplus per driver increases
by the savings in private trip cost (C0 - Cw). The increase in consumer surplus is the
rectangular area with a width of (C0 - Cw) anda length of V0 The decrease in trip
cost also increases tra:ffic volume from V0 to Vw. The consumer surplus of the new
drivers equals the area of the triangle DE F. The benefit of widening the highway
exceeds the cost if the increase in consumer surplus (the sum of the rectangle and the
triangle) exceeds the cost of widening the highway. This analysis ignores pollution
cost, which is discussed later in the chapter.
There are n:tanY anecdotes about increases in highway capacity that did not
result in less congestion and faster travel during the rush hour. The reason is that
271
the demand for rush-hour travel is highly elastic, with many commuters initially
deterred from using the congested highway because of slow travel speeds. This is
the phenomenon of "latent demand." In the language of Small (1993), there is a
"reserve arrny of the unfulfilled" that will switch toa previously congested highway
as soon as an increase in capacity increases travel speeds. This latent demand will
fill most or all of the new capacity during peak periods. Many of the new users have
switched from other routes or other times, so we must look beyond the rush hour
along the widened highway to see the effects of the increase in capacity.
The city could also improve the flow of traffic on the existing highway. Street
lights can be synchronized to keep traffic f!owing ata steady speed. Designating oneway streets and restricting on-street parking increase the carrying capacity of streets.
Sorne cities have placed stoplights on expressway on-ramps, thus smoothing the flow
of vehicles onto expressways. In the language of transportation engineers, the ramp
lights decrease the "turbulence" caused by entering vehicles, thus increasing travel
speeds. In Los Angeles, the installation of such a system on the Harbor Freeway
increased average travel speeds from 15 miles per hour to 40 miles per hour. A
similar system in Dalias increased travel speeds from 14 miles per hour to 30 miles
perhour.
j
FIGURE 11-5 Effects ofTransit Subsidies on Auto Volume and Transit Ridership
Auto and highway
Mass transit
$
Social trip _..
Original demand
curve
cost
:;,
:;,
.g-
Prvate trip
cost
f-<
.gf-<
J1
1
1
1
1
1
1
1
A*
A" A'
T'
T" T*
Transit ridership
In the optimum situation, auto drivers pay a congestion tax, so auto volume is A* and transit ridership is T*. If there is no congestion tax, the auto volume is A'
and the transit ridership is T' (the underpricing of auto travel shifts the demand curve for transit downward}. A transit subsidy increases transit ridership to T 11
and decreases auto volume lo A11 (the subsidy shifts U1e auto-demand curve downward}. The subsidy narrows the transit gap fas ter than it narrows the auto gap.
273
from A' to A", c!osing about orie-quarter of the gap between the equilibrium and the
optimum volume. The subsidy narrows-but does not eliminate-the gaps between
the optimum and equilibrium volumes: A" is stillless than A, and T" is stil!less
than T*.
Would it be possible to e!iminate both the auto gap and the transit gap? In
Figure ll-5, the transit gap is closing faster than the auto gap; the subsidy c!oses
two-thirds of the transit gap, but only one-quarter of the auto gap. In other words,
the subsidy increases transit ridership by more than it decreases auto volume. This
occurs because in addition to diverting sorne drivers from cars to transit, the subsidy
also attracts new travelers to transit. Por example, suppose that the subsidy causes
250 drivers to switch from autos to transit (A' - A"= 250) and also generates 280
new travelers. The total increase in transit ridership is 530 (T 1 '-T' = 530), the sum
of the diverted auto drivers and the new transit riders. Because the transit gap is
c!osing faster than .the auto gap, if the city decides to e!iminate the transit gap, the
city will be !eft with an auto gap. Altemative!y, if it decides to eliminate the auto
gap, the city will overshoot the transit gap: transit ridership will exceed the optimum
ridership. The city can close ont<-but not both-gaps.
The fundamental problem of the transit subsidy is that it underprices transit,
increasing ridership above its optimum leve!. In the absence of congestion taxes,
a transit subsidy may improve the efficiency of the transportation system, but the
subsidy wi11never be as efficient as a system of congestion taxes. The policy question
is whetherthe benefit ofthe transit subsidy (the diversion of drivers from underpriced
and congested roads) is larger or smaller than the cost (excessive transit ridership).
'-
This section shows how the govemment could pick both the optimum traffic volume and the optimum road width. The long-run analysis of highway pricing and
investment has four steps:
l. Derive the average total-cost curves for different highway widths.
2. Derive the long-run-cost curves (average and marginal).
3. Pick the optimum traffic volume and road width.
4. Pick the congestion tax that generates the optimum traffic volume on the optimumroad.
Average Total Cost for Two-Lane and Four-Lane Highways
The average total cost of travel is the sum of the trip cost and the average road
cost. Figure 11-6 shows the trip-cost curves for both the two-lane and the four-lane
highways; the cost curves are horizontal up to the point at which congestion sets in;
the congestion threshold is V2 vehicles per hour for the two-lane highway and V4
vehicles for the four-lane highway. The average road cost equals the total cost of
building the highway divided by the number of trips.
r:
.
274
. ".
Average total
cosr (2 lanes)
(4lanes)
(4lanes)
Vz
V'
v
Total traffic volume (vehicles per hour)
The average total cost of travel is the sum of private trip cost and average road cost. The average
total-cost curves are U-sbaped. If there are constan!retums to scale in road building, the two
average total-cost curves reach the sarne minimum cost e, but the four-lane higbway reaches !bis
cost with twice the tiaffic volume (V* instead of V'). The long-run average cost is composed of the
minimum points of tbe verage total-cost curves.
The average total-cost curves are U-shaped. An increase in traffic volume has
two effects on average total cost. First, an increase in volume spreads construction
cosi over more drivers, so the average road cost decreases (the road-cost effect).
Second, an increase in volume may increase congestion, increasing travel times and
the trip cost (the trip-cost effect). At traffic vohimes below the congestion threshold
(Vz for the two-lane highway and V4 forthe four-lane highway), there is no trip-cost
effect, so the average total-cost curve is negatively sloped. At high traffic volumes,
the average road cost is relatively low and congestion is severe, so the trip-cost effect
dominates the road-cost effect. Therefore, the average total-cost curve is positively
sloped. In Figure 11-6, the average total-cost curve for the two-lane highway reaches
its mnimum point at a volume of V' drivers, indicating that the road-cost effect
dominates for volumes up to V' drivers and the trip-cost effect dominates for higher
volumes. Forthe four-lane highway, the trip-cost effect dominates for volumes above
V* drivers.
The cost curves in Figure 11-6 are drawn under the assumption of constant
returns to scale in highway building. If there are constant returns to scale, a doubling
of consiruction cost (building four hinesinstead oftwo) doubles the carrying capacity
275
of the highway. The congestion threshold of the four-lane highway (V4 ) is double
that of the two-lane highway (V2 ). Similar!y, it takes twice as many drivers on the
four-lane road to reach a given private trip cost: To reach the private trip cost e',
the four-lane highway requires twice the traffic volume. The two total-cost curves
reach the same rninimum cost (e*), but the four-lane highway reaches this point
with twice the traffic volume (V* instead of V').
276
v v.
Total traffic volume (vebicles per hour)
The optimum traffic volume (V') is the volurne such that the marginal benefit oftravel (shown by
the demand curve) equals the long-run marginal cost oftravel (LRMC). The four-lane bighway
minimizes the average total cost of travel for V' drivers, so the optimum road width is four lanes. The
appropriate price of travel is C', so a congestion tax of T is added to the private trip cost of C'. Tbe
congestion tax equals \he average road tax, so congestion taxes cover the cost of the optimum road.
gap between the social trip cost and the private trip cost (C*-C'). Because the long
ron average cost is the sum of the private trip cost and the average road cost, the
gap between the LRAC and the private trip cost is the average road cost (C*-C').
Because the congestion tax equals the average road cost, a system of congestion tolls
pays for the optimum road.
Is this a fluke, or is there something systeinatic about congestion tolls and road
costs? The expressions for average road cost and the congestion tax are as follows:
Average road cost = Average total cst - Private trip cost
Congestion tax = Social trip cost- Private trip cost
Therefore, the average road cost equals the congestion tax if the average total cost
(ATC) equals the social trip cost (STC). STC is actually the shortrun marginal
cost of driving; it shows the increase in social cost from one additional driver. The
marginal-cost curve (STC) intersects the ATC curve at its minirnum point. If the
government builds the optimum highway, the market reaches the minimum point of
sorne ATC curve, so STC equals ATC. Therefore, the average road cost equals the
congestion tax, meaning that the revenue from congestion taxes covers the cost of
277
the road. The revenue from a congestion tax provides an important signa! to road
builders. If tax revenue exceeds the total cost of building the road, the road should
be widened. In other words, any surplus tax revenue should be spent on widening
the road. To explain, suppose that the city builds a two-lane highway and imposes a
congestion tax of $4.34 (see Figure 11-2). If the total revenue from the tax ($4.34
times 1,400 drivers) exceeds the total cost of building the road, the road should be
widened.
278
Under these two assumptions, the congestion tax tilts the bid-rent function downward; the slope increases and the bid rent decreases at alllocations. The gap between
Ro and R,increases with distance to the city center because the longer the cornmuting distance, the larger the tax liability and the larger the reduction in the bid rent
for land.
The third bid-rent function in Figure 11-8, R,, shows the actual effect of congestion taxes. It lies above R, for tw reasons:
If the govemment uses the congestion tax revenue to cut other taxes, residents
are willing to pay more for housing and land at alllocations.
The congestion tax decreases traffic volurne, increasing travel speeds and
decreasing travel times. The decrease in the time cost decreases the unit cost
of travel, partly offsetting the effects of the congestion tax on the slope of the
bid-rent function.
By comparing R,to the original bid-rent function R 0 , we can see that the congestion
tax increases b.id rents within u* miles of the city center and decreases bid rents
beyond u*. Locations near the city center become more attractive because the congestion tax bill is relatively low, so the bad news (increase in congestion taxes) is
dominated by the good news (decrease in other taxes and lower time cost). In contrast, locations far from the city center become less attractive because the congestion
279
Initial density
function
tax bill is relatively large. To summarize, the congestion tax increases the relative
attractiveness oflocations close to the city center, increasing the slope ofthe bid-rent
function.
Figure 11-9 shows the effect of congestion taxes on the residential density
function.The density function shows population density (people per acre) at different
locations. As explained in Chapter 8, density increases as one approaches the city
center: land is more expensive, and people economize on the relatively expensive
1and by consuming less of it (they live in smaller dwellings on Jess 1and), so there are
more residents per acre. The congestioil tax shifts the density function from D0 to De.
In Jocations where land rent riseS" (distance less than u*), households consume less
land, so density increases. Beyond u, where Jand rent falls, households consume
more Jand, so density decreases.
The change in the density function shows that the congestion tax generales a
more compact city. After the tax, the city has a larger fraction of its population living
within a given distance of the city center, so it has shorter commuting distances
and travel times. As explained earlier in the chapter, the congestion tax causes the
city to move up its travel-demand curve; the increase in travel cost decreases traffic
volume. One reason for the decrease in traffic volume is that workers have shorter
coin.muting distns.
280
'.
Another urban transportation problem is air pollution resulting from the emissions
of motor vehicles. Motor vehicles emit severa! pollutants that contribute to air pollution in a significan!way: volatile organic compounds (VOC), carbon monoxide
(CO), nitrogen oxides (NO,), and sulfur dioxide (S02 ). VOCs react with NO,in the
atrnosphere to form ozone (03) and also generate particulate matter. In the United
States, transport activities are responsible for about two-thirds of CO emissions,
about half of VOC emissions, and about two-fifths of NOx emissions (Small and
Kazimi, 1995). Motor vehicles powered by petroleum products emit carbon dioxide (C02), one of the "greenhouse" gases that are accumulating in the atrnosphere.
Most scientists believe the buildup of these gases will cause global temperatures to
increase, resulting in significan!changes in weather patterns.
In the United States, most of the regulations concerning vehicle emissions are
included in the Clean Air Act of 1963, amended in 1970, 1977, and 1990. The act
specifies emission standards for various pollutants, including CO, VOC, and NO,.
Over the last few decades, emissions have decreased dramatically, so a new car today
emits a small fraction of the amount emitted by cars made two decades ago (Small
and Kazimi, 1995).
281
The Clean Air Act also defines standards for ambient air quality, specifying
maximum concentrations of CO, 03 , NOx, S02 , and particulate matter. The individual states are responsible for meeting the standards. The states have used various
policies to decrease auto pollution, including vehicle inspection and repair programs, restrictions on parking, and "cash for clunkers" (paying people for old cars
and then crushing the cars). Arnendments to the Clean Air Act (in 1977 and 1990)
included one innovative feature for the control of stationary pollution sources. In
cities where the EPA standards were exceeded, new pollution sources were allowed
if two conditions were satisfi.ed. First, the new sources had to adopt the most effective abatement technology. Second, existing sources had to decrease their ernissions
to ensure that there was a net irnprovement in air quality. This policy set the stage
for the development of a market for pollution rights, in which new firrns purchase
pollution rights from existing firrns. Because this "offset" policy rnirnics a system
of effiuent fees, it generales effi.cient abatement activity.
282
r
1
total i:niles driven, thus decreasing air pollution. The problem with a gas tax is that
every driver would pay the same tax, regardless of how much pollution his or her
car generates. In other words, there would be no practica! way to impose a higher
tax on cars that generate more pollution. Such a tax would decrease pollution by
decreasing i:niles driven, not by persuading people to drive cleaner cars.
What is the marginal externa! cost of auto driving? Smal! (1997) estimates an air
pollution cost of between $0.007 and $0.037 per vehicle-rni!e, or roughly between
$900 and $4,400 over the life of the typical car. If the typical car gets 22 miles per
gallon, the associated gasoline tax would be between $0.15 and $0.81. Concerning
C02 ei:nissions and global wari:ning, one proposal is to irnpose a tax on the carbon
content of fuels such as gasoline. Under a carbon tax that would stabilize C02
ei:nissions at 80 percent of the 1990 leve!, the tax per gallon of gasoline would
be between $0.45 and $0.67. Adding the taxes for conventional air pollution and
greenhouse gasses, the associated hike in the gasoline tax would be between $0.60
and $1.48, which would generate gasoline taxes well below most European !evels.
Transit Subsidies
An alternative approach for controlling auto pollution is to subsidize mass transit
systems. Although both autos and mass transit generate air pollution, the pollution
per passenger is lower on mass transit. If there are no pollution taxes, auto travel
is .il.derpriced relative to transit. As explained earlier in the case of congestion
externalities, the underpricing of autos generates an excessive tra:ffic volume (the
equilibrium auto volume exceeds the optimum volume) and too little transitridership
(the equilibrium transit ridership is less than the optimum ridership). One response
to the nderpricing of autos is to subsidize transit, the substitute good.
As shown in FigUre 11-5, transit subsidies can be used to move the auto and
transit markets closer to their optimum points (A and T*). As explained for the
case of congestion externalities, the basic problem with transit subsidies is that
they decrease the cost of transit below the marginal social cost, causing too much
transit travel. The excessive amount of travel is costly because society's productive
resources (e.g., labor, capital, fue!) are used inef:fi,ciently: The transit system will use
an excessive number of drivers, mechanics, and buses, and consume an excessive
amoilnt of fue!. If the pollution cost per auto passenger exceeds the pollution cost
per transit rider by a wide margin, there are large air-quality benefits from transit
subsidies. The real question is whether the air-quality benefits dominate the resource
cost of the policy.
AUTOSAFETY
The third transportation problem is auto safety. Tra:ffic accidents generate two types
of costs. First, people involved in accidents are often injured or killed. Motor-vehicle
accidehts are the !eading cause of death forpeople ages 1 to 24; in 1993, 42,000
people died in motor vehic!e accidents in the United States, and another 2 i:ni!lion
283
were injured (Small, 1997). About a third of traffic deaths and about three-fifths of
the injuries occur in urban areas. Second, traffic accidents during rush hours disrupt
traffic flows and contributeto congestion.
This section explores the effects of various public policies on auto safety and
traffic accidents. Federal safety legislation requires automakers to install a number of
safety features (e.g., seatbelts, shatterproofwindshields). One policy issue concems
the effects of these regulations on highway death rates and accident rates. Another
issue concerns the effects of congestion and pollution policies on traffic safety. For
exarnple, policies that encourage the switch to smaller cars decrease the congestion
problem (smaller cars take up less space on the highway) and the pollution problem
(small cars use less fuel and generate less pollution). Unfortunately, smaller cars are
less safe (injuries are more severe in collisions involving small cars), so there are
trade-offs between safety, congestion, and pollution.
284.
FIGURE ll-10 Optimum Speed and Safety Regulations
$
20
Marginal benefit
16
12
45
46
47
48
49
51
50
Speed (miles per hour)
52
53
54
The optimum speed is the speed at wbich the marginal benefit of speed (value of decreased travel
time) equals its marginal cost (increased expected accident costs).ln the absence of safety regulations,
Duke chooses point A (50 mph). Safety regulations decrease tbe expected cost of collisions, sbifting
the marginal-cost curve downward, increasing tbe optimum speed to 52 mph (point B).
from 45 mph to 46 mph, he can spend more time dancing, and he is willing to pay
$16 for the additional dance time. The benefit curve is negatively sloped, refl.ecting
the dirninishing marginal utility of dancing.
Table 11-3 shows the expected cost of crashing at different speeds. As the speed
increases, the probability of crashing increases; the probability is only 0.001 for a
speed of 45 mph, but rises to 0.011 for a speed of 55 mph. Faster driving also causes
more damage to Duke and his car. If he crashes at a speed of 45 mph, his total
crash cost (medica! bilis plus auto-repair costs) will be $2,000. If he crashes ata
speed of 55 mph, his total crash cost will be $12,000. The expected crash cost
is the probability of crashing times total crash cost. The expected crash cost rises
from $2 for a speed of 45 mph (0.001 times $2,000) to $132 for a speed of 55 mph
(0.011 times $12,000). The marginal cost of peed is the change in the expected
crash cost for a one-unit increase in speed. An increase in speed from 45 mph to
46 mph increases the expected crash cost from $2 to $6, so the marginal cost of
speed is $4 for a speed of 46 mph. The marginal cost of speed increases with the
speed, reaching $22 for a speed of 55 mph. What is the optimurn travel speed? Duke
increases his speed as long as the marginal benefit of speed (the value ofthe increase
in dance time) exceeds the marginal cost (the increase in the expected crash cost). In
Figure 11-10, tbe optimurn speed (where the marginal-benefit curve intersects the
marginal-cost curve) is 50 mph. Table 11-3 showsthat a speed of 50 mph generates
a marginal cost of $12 andan expected crash cost of $42.
285
Probability
ofCrash
Crash
Cost ($)
Expected
Crash
Cost ($)
Marginal
Cost of
Speed ($)
Marginal
Benefit of
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0.009
0.01
0.011
2,000
3,000
4,000
5,000
6,000
7,000
8,000
9,000
10,000
11,000
12,000
2
6
12
20
30
42
56
72
90
110
132
o
4
6
8
10
12
14
16
18
20
22
17
16
15
14
13
12
45
46
47
48
49
50
51
52
53
54
55
Speed ($)
11
10
9
8
7
Source: Reprinted from Joumal oj Urban Economics by Arthur M. Sullivan, "The General Equilibrium Effects of
Congestion Externalities," pp. 80-104, 1983 with permission from Elsevier Science.
What are the effects of safety regulations on the expected crashcost? Suppose
that Duke is forced to buy a car with seatbelts and a shatterproof windshield. If he
crashes in his new car, he is less likely to suffer serious injuries, so the expected cost
of crashing at each speed decreases. In Figure 11-1O, the decrease in the expected
crash cost shifts the marginal-cost curve downward, increasing his optirnum speed
from 50 mph to 52 mph. Because the safety regulations protect Duke in an accident,
they cause him to take more risks, so the probability of crashing increases. The net
effects of the safety regulations are smaller than they would be if his speed did not
change; he has a lower injury cost if he crashes, but he is also more likely to crash.
...
286
.,
!
would increase the carrying capacity of the road system, reducing congestion. On the
other hand, travel in small cars is less safe. The policy issue is whether the decreases
in energy consumption, pollution, and congestionare worth the cost (the increase in
traffic injuries and deaths).
The occupants of small cars are more likely to suffer serious injuries in a collision. The state of New York has estimated the injury rates for differnt car types
(Meyer and Gomez-Ibanez, 1981). In a collision between two large cars (weight
greater than 3,500 pounds), the rate of serious injury is 5.2 per hundred rnillion
vehicle miles. In a collision between a large car and a small car (weight less than
2,000 pounds), the injury rate is 3.7 for the occupants of the large car and 13.8 for
the occupants of the small car. The combined injury rate for such a collision is 8.8
per hundred rnillion vehicle miles.
In a collision between two small cars, the injury rate is 9.7 for the occupants
of each car. In other words, a collision between two small cars is more lethal than
a collision involving a small and a large car. Are small cars more or less likely to
be involved in collisions? Studies of the accident rates of different car sizes have
generated rnixed results; sorne studies found higher collision rates for small cars,
while others found no significan!differences in accidentrates. Overall, it appears that
small-car occupants are just as likely to be involved in collisions and more likely to
suffer serious injuries. Therefore, travel in small cars is more dangerous than travel
in large cars.
A shift toward a small-car fleet is likely to increase injury and death rates. As
the fraction of small cars increases, the fraction of collisions involving small cars
increases, increasing the rate of serious injury. According to Meyer and GomezIbanez (1981), if the percentage of smal1 cars in the fleet increases from 20 percent
to 70 percent, the percentage of dangerous accidents (between a large anda small car,
or between two small cars) will increase by 50 percent to 95 percent The increase
in the fraction of dangerous accidents increases traffic death rates. These numbers
suggest that there is a considerable cost associated with the switch to smaller cars.
The policy question is whether the benefits (less congestion and pollution) are greater
than the cost
Meyer and Gomez-Ibanez point out two additional factors that must be taken
into account in predicting the safety effects of small cars. First, about 37 percent of
urban traffic deaths involves pedestria:ns and bicyclists. If small cars are less lethal
for pedestrians and bicyclists, there would be fewer pedestrian and bicycle deaths.
Second, about 11 percent of urban traffic deaths result from single-car crashes
(a car hitting a fixed object). Since small car,s are about as safe as large cars in this
type of accident, a shift to small cars is unlikely, to increase the number of single-car
fatalities. The switch to small cars is also likely to change driving habits. A driver
who switches to a smaller, more dangerous car is likely to drive more cautiously
to mitigate the risk of serious injury. If Duke switches to a small car, his expected
injury cost would increase, shifting his marginal-cost curve upward. As a result, he
would decrease his speed, partly offsetting the increase in the severity of accidents
by decreasing the probability of crashing. In addition, Duke would be more likely
to fasten his seatbelt; thus decreasing the severity of injuries.
287
SUMMARY
l. The three most popular commuting trips are between suburbs (40 percent of
commuters), within the central city (33 percent), and from a suburb to the
central city (20 percent).
2. About four of :five workers (and 9 of 10 suburbanites) commute by auto, truck,
orvan.
3. Automobile drivers base their travel decisions on private-not social-costs, so
the equilibrium volume exceeds the optimum volume.
4. A congestion tax intemalizes the congestion extemality, and could be implemented with a vehicle identification system under which every car is equipped
with an electronic device that allows sensors along the road to identify the car
as it passes.
5. In the long run, the optimum road width and traffic volume are determined simultaneous!y. The congestion tax will pay the fu!!cost of the optimum highway.
6. The econornic approach to the auto pollution problem is to impose pollution
taxes on new cars.
The marginal benefit (demand) is $22.12 for a volume of200 and drops by $2.48
for ev ry additional 200 drivers..
a. Usethis equation to derive atable like Table 11-l.
b. Wha{is the equilibrium traffic volume?
c. What is the optimum volume?
d. What is the appropriate congestion tax?
2. You have been assigned the task of estimating the extemal trip cost at a traffic
volume of 1,600.vehicles per lane per hour for a one-rnile section of an urban
freeway between two exits. You have traffic data for the relevant stretch of
highway for the last year. Outline your strategy for computing the extemal trip
cost. What inforrnation would you collect and how would you use it?
3. In estimating the congestion tax for peak-period travel, Keeler and Smal! (1977)
assumed that the imposition of a congestion tax would not affect traffic volume.
a. Depict graphically the model Qf auto travel underlying this assumption.
b. Did Keeler and Small overestimate or underestimate the congestion tax? In
other words, ifthe assumption of a fixed traffic volume were dropped, would
the estimated congestion tax increase or decrease?
4. Consider the following quote from Peabody and Associates, a transportation
consulting firrn: ' ccording to the Peabody principie, an increase in highway
capacity increases traffic volume by an arnount sufficient to leave the private
trip cost unchanged."
288
289
c. The legal speed limit decreases to 46 mph. The expected cost from speeding
tickets increases by $3 for every 1 mph over the speed limit: $3 for 47 mph,
$6 for 48 mph, $9 for 49 mph, and so on.
d. Highway improvements cut the probabilities of crashing in half.
10. Duchess, a resident ofHazard City, has the same marginal-benefit curve as Duke
(Table 11-3), but her marginal-cost curve is horizontal, with a marginal cost of
speed without air bags is $14.
a. What is her optimum speed?
b. Suppose that an airbag system would decrease the marginal cost of speed to
$8. Ifthe airbag system costs $300, should Duchess buy the system?
11. Chopperville is evaluating the merits of using helicopters to clear up highway
accidents. Suppose that an accident simply stops traffic from the time the accident occurs to the time the disabled vehicles are removed from the highway.
During rush hours, the t)'pical accident stops traffic for 7,000 cars. Under the
current tow-truck system, the typical rush-hour accident stops traffic for 20 minutes. Under a helicopter system, the typical rush-hour accident would stop traffic
for only eight minutes. Suppose that the cost of operating the tow-truck system
(including the costs of labor, fue!,and equipment) is $200 per accident and the
cost of operating the helicopter system (including the costs of labor, fue!, and
equipment) would be $3,000 per accident. Your task is to determine whether
the helicopter system is superior to the tow-trucksystem.
a. Is there enough inforrnation to determine which system would be superior? If not, what additional inforrnation do you need, and how would you
use it?
b. If possible, specify "reasonable" values for the missing bits of inforrnation
and compute the benefits and costs of switching to the helicopter system.
c. Suppose that it is possible, for each accident, to instantaneously determine
which driver is at fault and that the responsible driver must pay for either
a helicopter or a tow truck to clear up the accident (at a cost of $200 for
the truck or $3,000 for the helicopter). Describe a simple public policy that
would provide the incentive for drivers to freey! choose the socialy! efficient
method of clearing up the accident (either the truck or helicopter).
290
CHAPTER 12
Mass Transit
Tru.s chapter explores the problerns and prospects of urban rnass transit. In the
last forty years, the percentage of urban travelers using rnass transit decreased and
the mass-transit sector went frorn a rnarginally profitable private-sector enterprise to a
deficit-ridden public-sector operation. Tbis chapter explores sorne of the reasons
for tbis transformation and identifies sorne changes that could irnprove the urban
rnass-transitsystem. Here are sorne of the questions we'll consider:
l. How do commuters pick a travel rnode, and why do so few choose mass transit?
2. What are the relative costs of different transit systerns (heavy rl, light rail, and
buses) and how do they compare to the costs of an auto-based system?
3. Is there a dark sidto light rail?
4. Are bigh-occupancy vebicle lanes effective in increasing vehicle occupancy and
decreasing congestion?
5. How would the privatization of urban mass transit affect the features of the
transit system, ridersbip, and costS?
6. Do heavy-rail systerns like San Francisco's BART have large effects on land-use
pattems?
292
FIGURE 12-l Transit Modal Choice for Residents of Central Cities and Suburbs
4,500,0004,000,000 i3,500,000 f;;;
e
J:j
3,000,000 i-
:gc. 2,500,000 fe
"" 2,000,000 f;;;
-t
,,.
""'
500,000
O Heavy rail
li!J Light rail
O Bus
f- .
Centra!-city residents
Suburban residents
1,492,634
52,810
2,381,213
260,648
24,197
950,807
Source: U.S. Bureau othe Census, 1990 Cen.sus of Popularion and Housing, SSTF20, Joumey ro Work in
the Unired Srares. (Washington, DC: U.S. Govemment Printing Office, 1994).
three transit rnodes for residents of the central city and suburbs. About 61 percent of
central-city transit cornrnuters use the bus, and 38 percent use heavy rail (subways and
elevated trains). Suburban transit riders are less nurnerous; about 77 percent of them
use the bus and 21 percent use heavy rail. Although light rail systerns have received a
lot of attentionin the popularpress, ridership onihese newer systems is relatively low.
Transit ridership varies with the path of the cornrnuter trip. As shown earlier in
the book, about 40 percent of rnetropolitan wbrkers cornrnute from one suburb to
another. Less than 2 percent of these cornrnuters use mass transit. Transit usage is
highest among workers who cornrnute withi.n the central city; about one in six of
these cornrnuters use mass transit.
Table 12-1 shows trends in transit ridership. Total ridership in 1995 was about
45 percent of ridership in 1950. After steady increases in ridership between 1970 and
1990, the number of transit rides has decreased recently, with the largest reductions in
bus ridership. The trolley coach (a bus powered by overhead electric wires) reached
its peak in 1950 and has declined since then. The use oflight-rail systems (streetcars)
decreased steadily between 1940 and 1980, but has recently staged a recovery;
systems have recently been built or restred in Portland, San Jose, Sacramento,
;.,
293
Total Rides
Heavy Rail
LightRail
1940
1950
1960
1970
1980
1990
1995
13,098
17,246
9,395
7,332
8,567
8,873
7,763
2,382
2,264
1,850
1,881
2,388
2,346
2,033
5,943
3,904
463
235
133
176
251
Trolley Coach
Motor Bus
534
1,658
657
182
142
126
119
4,239
9,420
6,425
5,034
5,837
5,784
4,848
Source: American Public Transit Association, Transit Fact Book (Washington, DC, 1991).
Buffalo, San Diego, and Pittsburgh. In Canada, tbere are new light-rail systems in
Edmonton and Calgary.
There is substantial variation in transit ridership across metropolitan areas. Figure 12-2 shows the percent ofworkers who use public transportation formetropolitan
areas (CMSAs or MSAs) with at least 1 million workers. The cities are arranged,
left to right, in descending order of total employment. In the New York CMSA,
about 25 percent ofmetropolitaiJ. workers uses public transit. No other metropolitan
FIGURE12-2 Percentage ofWorkers Using Public Transportation, Selected Cities
30.0
25.0
24.8
"'
!::l
,s
:e
20.0
"" '
e
;;
"
15.0
: :
"o
..
:
o 10.0
"
"'
5.0
<
z""
<
<
<
<
u
::
1
,3
00
":E
u
00
:a=
:a'"
"
= %
J: "'"":a
=
""' "'
""
"'e
3 o. -'"
o
" """' "=" .!l>
:E u
,3
;;
"a
"
o
.e
"'
Source: U.S. Bureau,ofthe Census,J990 Censu:s -of Popularion.Housing, -Labor. Force CQaracrerisrics and
Joumey to Work. (Washington, OC: U.S. GovemmentPrinting Office, 1994).
-"
"
.e
eJ'
.5
Ji;
294
r1
area has a transit share greater than 14 percent. In fact, there are only three other
metropolitan areas where transit ridership is at !east 10 percent: Chicago, Washington D.C., and Philadelphia. Transit ridership is much higher arnong central-city
residents: 47 percent of workers who live in New York City use transit, as do 26 percent of workers in Chicago and 25 percent of workers in Philade1phia.
What are the elasticities of demand for mass transit?.By how much would transit
ridership increase when transit fares decrease or transit service improves? There are
four general conclusions from empirical studies of transit ridership:
l. Price e1asticity. The demand for transit is price-inelastic, with a price elasticity
between -0.20 and -0.50 (Beesley and Kemp, 1987). A common rule of thumb is
that a 10 percent increase in fares decreases ridership by about 3.3 percent, meaning
that the price e!asticity is -0.33. According to Small (1992), the price elasticity is
relatively large for off-peak trips and trips by high-income commuters.
2. Time elasticities. The demand for transit is more responsive to changes in travel
time. Por the line-haul portion of the trip (time spent in the vehicle), Domencich,
Kraft, and Va1ette (1972) estimate an elasticity of -0.39: a 10 percent increase in
1ine-haul time decreases ridership by abqut 3.9 percent. Por access time (time spent
getting to the bus stop or transit station), they estimate an elasticity of -0.71.
3. Value oftravel time. According to Small (1992), the average commutervalues
the time spent in transit vehicles at about half the wage: the typical commuter would
be willing to pay half of his hourly wage to avoid an hour on the bus or train. The
value of time spent walking and waiting is two to three times larger: the typical
commuter would be willing to pay between 1.0 and 1.5 times his hourly wage to
avoid an hour of walking or waiting time. The value of travel time increases less
than proportionately with income; a 50 percent difference in income generates less
than a 50 percent difference in the value of travel time.
4. Noncornrnutingtrips. The elasticities of demand for noncommuting travel are
higher than the e!asticities for commuting trips.
There are three principal implications from these empirical results. Pirst, an
increase in transit fares increases total fare revenue. A fare increase decreases ridership by a relative!y small arnount, so total revenue (fare times ridership) increases.
Kraft and Domencich (1972) suggest that dropping the price of mass transit to
zero-making transit free-would increase ridership by only about a third. Second,
a simultaneous improvement in service and fares may increase ridership. Suppose
that a transit authority increases the frequency and speed ofbuses and finances the improved service with increased fares. Because people are more sensitive to changes in
time cost than changes in fares, ridership may increase. Third, service improvements
that decrease walking and waiting time (more frequent service, shorter distances between stops) generate larger increases in ridership than improvements that decrease
line-haul time.
295
time and monetary costs). Suppose that a commuter has three travel options: the
automobile, the bus, and a fixed-rail transit system such as San Francisco's Bay
Area Rapid Transit (BART) or Washington, D.C.'s, Metro.
The commuting trip can be divided into three parts. The collection phase involves travel from the home to the main travel vehicle. The auto mode has no collection cost because the driver uses his own vehicle. The bus mode has a moderate
collection cost because the rider must walk from his home to the bus stop. The fixedrail system has the highest collection cost because there is a relatively long average
distance between transit stations, so riders must either wal.k a long distance or take
another mode (e.g., auto or bus) from the home to the transit station. The line-haul
phase is the part of the trip spent on the main travel vehicle. The heavy-rail mode
has the shortest line-haul time because it has an exclusive right-of-way, so it avoids
rush-hour congestion. Although the bus and the auto both travel on congested streets
and highways, the auto is faster because the bus must stop to pick up passengers
along the way. The distribution phase involves travel from the end of the vehicular
trip (at the downtown transit station, parking garage, bus stop) to the workplace. If
parking is available near the w rkplace, the auto has the shortest distribution time,
followed by the bus and the fixed-rail system.
Bus
BART
o
0.30
0.00
10
0.30
3.00
15
0.30
4.50
40
0.10
4.00
50
O.!O
5.00
30
0.10
3.00
0.30
0.00
5
0.30
1.50
9
0.30
2.70
1.00
0.00
1.00
.9.50 .
1.50
0.00
2.00
3.00
5.00
4.00
1.50
10.20
9.00
10.50
11.70
296
at one and One-half times her wage rate. If her wage is $12 per hour, she is willing
to pay $6 to avoid one hour ofin-vehicle time and $18 to avoid one hour ofwalking
and waiting time. Therefore, the cost of wal.king and waiting time is 30 cents per
minute, and the cost of in-vehicle time is 10 cents per minute.
l. Collection time cost. Carla walks to the bus stop or the BART station. Since
the bus stop is closer to her home, the bus has a lower collection cost.
2. Line-haul time cost. BART is the fastest mode (it operates on an exclusive
right-of-way), followed by autos (which travel on congested streets) and buses
(which travel on congested streets and stop to pickup passengers).
3. Distribution cost. Carla parks her auto in a company parking lot under her
office building, so the distribution cost of the auto trip is zero. The bus stop is
relatively close to the office, so the bus has a lower distribution cost than BART.
4. Monetary cost. The monetary cost of the auto trip is $0.20 per mile, or $2.00.
The bus fare is $1, and the BART fare is $1.50. Half of the $6.00 parking cost
is allocated to the morning co=ute.
The cost of driving is less than the cost of the bus and BART. Although the
monetary cost of driving exceeds the monetary cost of the bus by $4, the lower
time cost of the auto more than offsets its higher monetary cost. Similarly, the
auto is more expensive but faster than BART. The largest difference in time cost is
for collection and distribution costs, where the auto has a cost advantage of $4.50
over the bus ($3.00 for collection + $1.50 for distribution), and $7.20 over BART
($4.50 + $2.70). Since the auto has a lower total cost than both the bus and BART,
Carla drives to work.
What would it take to persuade Carla to switch from her auto to mass transit?
There are severa! possibilities:
l. Subsidized transit. If the bus and BART were free, Carla would still drive. To
get herto switch to the bus, she would have to be paid a bribe of 51 cents per bus
ride. A bribe of $1.21 would cause her to switch to BART. Transit fares must
decrease by relatively large amounts to offset the time-cost advantages of the
automobile.
2. Line-haul time. If the line-haul time of th'e bus decreased from 50 minutes to
less than 35 minutes, Carla would ride the bus. She would take BART if its
line-haul time decreased to less than three rin.utes.
3. Collection and distribution time. Carla would ride the bus if its collection
and distribution time decreased from 15minutes to Jess than 10 minutes and
would switch to BART if its collection and distribution time decreased from
24 minutes to Jess than 15 minutes.
4. Auto monetary cost. Carla would ride the bus if the car's unit cost increased
from 20 cents to 36 cents per mile. The imposition of a congestion tax of at
least 16 cents per mile would cause her to switch to the bus.Similarly, ifthe city
imposed a pollution tax of at least 16 cents per mile, Carla would stop driving.
5. Parking cost. Carla would ride the bus iftheparking cost rose to at least $4.50
($9 per day).
297
6. Wage. If Carla's wage dropped to $8, the bus would be less costly than the
auto, and if her wage dropped to $6, BART would be less costly than the auto.
As the wage decreases, the opportunity cost of travel time decreases, increasing
the relative attractiveness of the modes with relatively low monetary cost and
high time cost.
To summarize, to get Carla to switch to transit, the changes in eitherthe monetary
cost orthe line-haul time cost would haveto be relatively Iarge. On the other hand, she
would switch to transit with relatively small changes in collection and distribution
time costs. These conclusions are consisten!with the transit elasticities discussed
earlier in the chapter.
What type of commuters take mass transit instead of driving? There are five
possibilities:
l. Proximity to stops and stations. A person who lives near a bus stop or a rail
station has a relatively low collection cost for transit and is more likely to take
the bus or BART.
2. Low opportunity cost. For a person who has a low opportunity cost of travel
time (e.g., a worker with a low wage, a student, a retired person), the time cost
of travel is relatively low. As a result, the advantages of transit (lower monetary
cost) dominate the disadvantages (longer commuting time). Such a person is
more likely to choose transit.
3. Low walking cost. A person who enjoys walking has relatively low collection
and distribution costs and is more likely to choose transit.
4. Disutility of driving. A person who dislikes the hassle and anxiety of driving is
more likely Jo choose transit. In terrns of the numerical example, such a person
has a relatively high cost for in-vehicle auto time and is more likely to take the
bus or BART. In contras!,a person who considers driving a challenging forrn of
athletics will buy a pair of gloves and drive to work.
5. No automobile. Many of the poor do not have access toan auto, so their only
option is to use public transit.
298
also increases line-haul (in-vehicle) time: more time is spent picking up and dropping
off passengers, so the trip takes longer.
Consider next the design of a fixed-rail system. San Francisco's BART provides
a nice illustration of the trade-offs associated with the design of a rail transit system.
There are two basic design trade-offs:
l. Mainline versusintegrated system. BART is considered a mainline system
because it relies on other transit systems to collect its riders from residential neighborhoods. Riders must either walk, drive,or ride a busto the BART station. The
alternative to the mainline system is an integrated system, under which commuters
make the entire commute trip on a single transit mode.
2. Spacing between stations. BARThas widely spaced stations (about 2.5 miles
apart), so there are few stops on the way to the city center. As a result, travel time
from the suburban station to the downtown station (line-haul time) is relatively short,
and in-vehicle time cost is relatively low. On the other hand, the average commuter
must travel a relatively long distance from home to one of the widely spaced transit
stations, so collection cost (by foot, bus, or car) is relatively high.
BARTwas designed to compete with the line-haul portian ofthe automobile trip.
It achieves this objective, providing comfortable, speedy service from the suburban
stations to the city center. There is a trade-off, however: Collection cost is relatively
high because BART is a mainline system with widely spaced stations. Because
walking and waiting time is more costly than in-vehicle time, the negative attribute
(high collection cost) dominates the positive one (comfortable, speedy line-haul
travel), so the full cost of a BART trip is relatively high and BART has diverted a
relatively small number _of auto commuters.
\
299
c.
. s
" 6 --------------------
e:
c.
5 ------------------ --1
1
-------------c -------1----1
1
Initial demand
1
1
1
1
1
1
1
1
. 1
Demand with
diamond lanes
Vo
Traffic volume
The addition of diamond lanes (reserved for buses and carpools) affects travel in nondiamond lanes in
two ways: there are fewer vebicles in nondiamond lanes, so the demand curve sbifts to the left; there
are fewer lanes, so the trip-cost curve sbifts to the left. If a relatively small number of vebicles are
diverted to the diamond lanes, the net cost of travel in nondiamond lanes increases from $5 to $6.
300
lanes shift the demand curve to the left, decreasing trip cost for the original trip-cost
curve.
The bad news is that there are fewer lanes for the remaining autos; the diamond
lanes decrease the amount of road space, so the congestion threshold decreases and
the trip-cost curve shifts to the left. In this example, the shift of the demand curve is
small relative to the shift of the cost curve, so the trip cost increases from $5 to $6.
This was the result of the ill-fated diamond lane on the Santa Monica Freeway in
Los Angeles. The diamond lane shifted a relatively small number of commuters to
car pools, so 25 percent of the freeway's capacity was used by only 6 percent of the
vehicles (Dahlgren, 1995). After the resulting public outcry, the lane was returned
to general use.
Sorne HOV lanes are more favorable for commuters who continue to use generalpurpose lanes. If the shift ofthe demand curve in Figure 12-3 were large enough, the
demand shift would more than offset the cost-curve shift, leading to less congestion
anda lowerprivate trip cost. Houston's HOV systemis an example of a system thatis
favorable to nonusers..The system caused large increases in carpooling and bus ridership, so the HOV lanes diverted large volumes oftraffic from general-purpose lanes.
The average vehicle occupancy increased by 20 percent and congestion dropped by
4 percent (Richmond, 1998).
301
FIGURE 12-4 Costs of Alternative Transit Systems
$
BART nd feeder
bus
.56
Autos and
highways
10
22
Peak passengers per hour (thousands)
30
Tbe average cost of auto traffic is independent of auto volume, but the average costs of an integrated
bus system and BART decrease as volume increases. The bus system is more efficient than BART for
all volumes studied and more efficient rhan an auto-based system for volumes exceeding 1,100
passengers per hour
\
congestion tax equals the cost of building the optimum road, so congestion taxes
can be used to both intemalize congestion extemalities and pay for the roads. The
public cost also includes the cost of air and noise pollution.
The average cost cilrve is horizontal for two reasons. First, Keeler et al. assume
t.1.at the average operating cost and the average pollution cost (cost per mile) do
not depend on traffic volume. Second, the trip time (and prvate trip cost) does not
depend on traffic volume: as volume increases, the road is widened to accommodate
the increased traffic without any reduction of travel speeds. Given constant returns
to scale in highways, a doubling ofhighway capacity doubles traffic volume without
changing the trip time. Using data for 1972, Keeler et al. estimate that the full cost
of an auto trip with a six-rnile line haul is $4.15.
302
The average cost curve is negative1y s1oped for two reasons. First, as ridership increases, the fixed cost of administering the bus system is spread over more
riders. Second, as ridership increases, collection and distribution times decrease. An
increase in ridership decreases headways (the time between buses) and decreases
the space between bus stops, so riders spend 1ess time walking to and waiting for a
bus. In addition, an increase in ridership al1ows a busto fill up with a shorter collection route. Less time s spent picking up passengers, so in-vehicle time decreases.
According to Kee1er et al., if the tra:ffic vo1ume along the corridor exceeds 1,100
passengers, the integrated bus system is 1ess costly than the auto system.
'l
CostofBART
The BART option invo1ves the mainline heavy-rail system and feeder buses to bring
commuters to the BART stations. The time cost of the BART option is the sum of
walking and waiting time costs and in-vehicle time cost. The agency cost is the sum
of the operating and capital costs of the feeder buses and BART. Given the substantial
investment in tracks andrights-of-way, the capital cost ofthe BARToption is large.
The cost curve is negatively sloped for three reasons. First, the system has
a substantil fixed cost, which is spread over more riders as ridership increases.
Second, as ridership increases, BART headways decrease, decreasing waiting time.
Third, an increase in bus ridership decreases the time and monetary cost of feeder-bus
serv1ce.
BART is more costly than the bus system forthree reasons. First, given its design
as a mainline system with widely spaced stations, the collection and distribution
costs are relative1y high. Second, its capital cost is very large; the fixed cost of the
system was $1.6 billion.-According to Webber (1976), BART could buy enough
buses to carry its passengers for only $40 million (2.5 percent of BART's capital
cost). Third, BART has a surprisingly high operating cost: in 1975, the operating
cost was 15.7 cents per passenger mile, compared to 13.6 cents for a bus. Most
planners were surprised to discover that the large investment in capital equipment
did not generate large savings in operating cost.
-1
)
'
,_.1
System Choice
The Kee1er et al. study provides important information for transit planners. For
all corridor volumes studied (up to 30,000 passengers per hour), BART was more
costly than an integrated bus system. At a peak volume of 30,000 passengers per
hour, BART is 50 percent more costly than the bus system. BART's peak ridership
through the transbay tube is about half this amount. The lesson for planners is clear:
With the possible exceptions of New York City and Chicago, which have corridor
volumes exceeding 30,000 passengers per hour, an integrated bus system is likely
to be more effi.cient than a modero fixed-rail system like BART.
Experiences with new heavy-rail systems in other metropolitan areas are similar
to the BART experience. The Metro system in Washington, D.C., would have to
'--'
-'
'--)
'J '
303
charge one-way fare of $8 (Gordon and Ri_chardson, 1989) to cover its operating
and capital costs. The Red Line in Los Angeles would have to charge a fare of at
least $11 to cover all its costs. Ridership on the new systems in Atlanta, Miarni, and
Baltimore have fallen well short of levels required to malee heavy railless costly
an bus systems. e problems witb heavy-rail systems should not be surprising in
light of tbe concluswn by Keeler et al. tbat corridor volumes must exceed 30,000 to
malee a heavy-rail system competitive with a bus system.
Light Rail
In recent years, many medium-size cities have built light-rail transit systems. Light
.,
-..
"' '
- ;
,_,
:""",
.1(.
:{
1
-
_,
ral is the modem version of the trolley and streetcar systems tbat were built in tbe
late 1800s and early 1900s. The first modem light-rail system opened in Edmonton in
1978. Since then, new systems have opened in San Diego, Calgary, Buffalo, Portland,
Vancouver, Sacramento, and San Jose. Many other cities are planning to build new
systems or extend old streetcar and trolley systems.
How does light rail compare to buses and heavy rail? Gmez-Ibez (1985) has
studied the cost and performance of new light-rail systems in San Diego, Edmonton,
and Calgary, and has cometo the following conclusions:
l. Capital cost. The capital cost of light rail includes the costs of the vehicles, the
track, and the power distribution system. These costs malee light rail more costly than
a bus system, which uses less expensive vehicles and travels along regular streets and
highways. Lightrail is much less costly than heavy rail; the cost per mile is $7 million
to $20 million, compared to $50 million to $200 million per mile for heavy rail. Light
rail is less eicpensive because it typically uses surface streets (instead of requiring
subways or elevated guideways) and is designed for lower speeds.
2. Operating cost. Compared to a bus system, light rail has a lower labor cost
because it needs fewer operators; a three-car train can be operated by one person
and can carry about 10 times as many riders as a standard bus. Because light rail
has a higher maintenance cost on its vehicles and its right-of-way, however, its total
operating cost is equal to or greater than tbe operating cost of buses.
3. Performance. Most light-rail systems consist of one or two radial rail lines,
witb feeder buses for tbe collection and distribution phases of the transit trip: Like
tbe heavy-rail systerris, tbe light-rail systems have higher collection and distribution
costs tban a bus system. Light rai also has shorter line-haul times and provides a
more comfortable ride tban a bus.
Table 12-3 compares the cost of the San. Diego light-rail system to the bus
routes it replaced. The total cost per light-ral passenger was $3.11, compared to
$1.16 for the bus system. Light ral has a moderately higher operating cost and
a much higher capital cost. The light-rail system seems to have increased transit
ridership; ridership in the area served by tbe system increased from 3.4 million to
4.8 million per- year. Accqrding to Qmez-Ibez, much of the increased ridership
r
_ : :
r-
304
Bus Systern
Light-Rail Systern
$0.82
0.34
1.16
3,398
$1.02
2.09
3.11
4,845
$5.47
Source: Jos A. G6mezIbez. "A Dark Side to Light RailT Joumal ofthe American.
PlaMing Association 51 (Summer 1985), pp. 337-51.
was caused by population growth in the corridor served by light ral. If one assumes
that the entire increase in ridership was caused by the introduction of the light-ral
system, the cost per additional rider was $5.47. The San Diego experience suggests
that although light ral may increase ridership, it does so at a relatively high cost.
More recent studies of light-ral systems have provided additional evidence that
light ralis more expensive than a bus system. Richmoild (1998) examines light-ral
systems in 11 cities (Baltimore, Buffalo, Dalias, Denver, Los Angeles, Pittsburgh,
Portland, Sacramento, San Diego, San Jose, and St. Louis) and compares their performance to bus systems. Here are sorne of his conclusions.
l. Light rail has higher capital costs. Por example, the capital cost of the Long
Beach light-ral system was $881 million, compared to the $168 million in capital
costs that would have been required for an equivalent bus system(Richmond, 1998).
2. Light rail has J:p.gher operating costs. Most data reported on the operating
costs of light ral omits the costs of the feeder buses that bring riders to light-ral
stations. Ignoring these costs, the average operating cost for light ral is somewhat
higher or perhaps slightly lower than that for equivalent bus lines. Por example, the
cost per passenger mile for Portland's MAX is $0.38, compared to $0.32 for one
equivalent bus !ine, and $0.39 for another. Once the cost of feeder buses is inc1uded,
light ral is more expensive than equivalent bus lines.
3. Light rail takes passengers out of buses: Por the Blue Line in Los Angeles,
63 percent of riders were previously bus riders. In Portland, about 55 percent of
MAX's riders switched from buses to light ral.
The system requires feeder buses or sorne other system to collect passengers,
and this is expensive for transit authorities and bothersome to potential riders. In
the modem metropolitan area with dispersed employment and retail activities, it is
difficult to attract enough riders to make light ralless costly than a well-designed bus
system. In many cases, busways and other HOV systems would be less expensiveand more effective in increasing transit ridership (Richmond, 1998). In other cases,
simple and inexpensive changes in regular bus service (adding buses, changing routes
or schedules, decreasing fares) may be more efficient than big projects.
305
..
o
<.>
-"
-"
"..
60%
o.
..o..
""
"1::
50%
40%
"
34%
Q.
"
"
30%
20%
.;;j
"><
o
10%
..e
"
0%
Bus
Heavy rai1
Light rai1
. Source: American Pul;l]ic Transit Associati_o. T_ i! Fact Bo_ok, -(W hi':lgtoDC, 1997?.
306
Demand = Marginal
social benefit
R'
Ridership
R*
Mass tranSit is a natural monopoly: scale econonties generate a negatively sloped long-run average
cost curve. The optimum output is where the demand curve (the marginal-benefit curve) intersects the
marginal-cost curve. At the optimum ridership (R'), the price (P') is less than the average total cost,
so the system genetates a deficil The system would cover its cost ata price of P', but ridership (R')
would be below the optimufu leve!.
another rider is relatively small. The optimuin ridership (R*) is where the marginal
social benefit (from the demand curve) equals the marginal cost. To generate this
leve] of ridership, the price must be P*, which is less than the average total cost.
The deficit associated with marginal-cost pricing is shown by the shaded area. If the
transit system were forced to cover all its costs, it would charge a price of P' and
generate a ridership of R', which is less than the optimum leve].
The second justification of transit subsidies is explained in Chapter 11 (Autos
and Highways). Auto travel is underpriced because drivers do not pay the externa]
costs associated with highway congestion and air pollution. Therefore, transit is
overpriced relative to auto travel, and transit subsidies can be used to offset the
artificially low price of autos. As explained in Chapter 11, there are trade-offs with
transit subsidies: although the subsidies discourage auto use, decreasing congestion
and pollution, they also cause the underpricing of urban travel in general, leading
to an excessive amount of travel. The policy question is whether the benefit of
the subsidy (decreased congestioo and air pollution) exceeds the cost (too many
resources used in transportation).
'-
'
307
2. Higher wages. Wages for transit workers have increased relatively rapidly over
the last few decades.
3. Increase in transit mileage. Transit service was extended into low-density
suburban areas, where ridership is relatively low.
4. Decrease in labor productivity. Between 1970 and 1995, total transit ridership increased slightly (by 6 percent) while total transit employment more than
doubled. Between 1980 and 1995, ridership dropped by 9 percent but employment increased by 66 percent. The resu1t, as shown in Figure 12-7, is rapidly
declining-or more accurately, diving-productivity. Between 1970 and 1995,
the average productivity (the number of transit rides per employee) was cut
in half.
FlGURE 12-7 Average Productivity of Transit Workers, 1970-1995
60,000
::
50,000
-=
"
;;
"'
"o.
40,000
"O
;;
:;=
.o...
."
e
="
:u
"
.30,000
20,000
10,000
"O
-!9L70--.--:-l-,1:97:-::5----:!::!98::0:-----:1::!9::85:-.--!:-::9':::90::-- !9 9:;-5-
Source: American Public Transit Association. Transir Facr Book. (Washington OC, 1997.)
308
Why did the nurnber of passengers per employee decrease? According to Fielding (1987), an importan!factor was an increase in the fraction of transit trips taken
during peak periods. To explain the peaking problem, consider a city where transit
ridership is uniform throughout an eight-hour workday; the city can serve its transit
riders with a single bus running all day. Suppose that all the riders suddenly decide
to travel to work during two moming rush hours: traffic volume during the two rush
hours quadruples. The city will no longer be able to serve its transit riders with a
single bus and a single bus driver, but will be forced to buy three more buses and
hire more drivers.
What are the effects ofthe shift to peak ridership on the city's laborproductivity
and labor cost? Consider first the possibility that the city hires part-time drivers.
If the city hires drivers by the hour, productivity does not change: the city hires
four part-tirne drivers instead ofone full-time driver. On the other hand, if the city
must hire four full-time drivers (who will be idle for six hours per day), its labor
productivity decreases and its labor cost rises. In other words, if the shift to peakperiod travel is combined with restrictive work rules that prevent part-time workers
and split shifts, labor productivity decreases and labor cost increases.
r
1
309
310
311
e o cities with tbe most extensive legal jitney services are Atlantic City
and Mianu (Cervero, 1996). In Atlantic City, jitneys operate along Atlantic Avenue,
the city's main commercial strip, and serve 11 million passengers per year. Entry
is restricted, with the number of jitneys fixed at 190. Refiecting the substantial
profits that can be eamed, the market value of a jitney medallion (license) and the
accompanying vehicle was $160,000 in 1995. In Miami, jitrieys operate along 21
designated routes that are drawn to limit competition between jitneys and public
buses; no jitney route duplicates more than 30 percent of a bus route. The city of
Miami recently decreased the number of jitneys, a response to losses in revenue and
ridership experienced by the public bus system. The unsubsidized jitneys took riders
away from the subsidized buses, so the city reduced the number of jitneys.
make regular trips to common destinations (Cervero, 1996). Tbis regularity allows
the development of subscription bus services: riders pay in advance to ride buses that
travel between their neighborhocds and workplaces. Most subscription bus services
started out in the 1970s as private operations, but have gradually been replaced by
government prograrns. The appeal of government involvement is that government
grants cover a largefraction (up to 80 percent) ofvehicle costs. Many transit agencies
that operate subscription buses contrae!the services out to private operators, which
keeps costs down because the firms pay lower wages to their workers.
Among the metropolitan areas with extensive subscription bus services are
Cbicago, Houston, San Francisco, and New York. In San Francisco, Golden Gate
Transit established 22 commuter routes between suburban communities and the
downtown 'fina_;cial district, and contracts with firms to provide the services. In
New York City, private bus lines carry about 60,000 workers each day to jobs in
Manhattan.
The private operators have lower costs !han public buses, (e.g., about 30 percent
iower in New York)..
Summary of Paratransit
Paratransithas the potential to fill the gap in the urban transporta!ion system between
solo-ridertaxis andlarge public buses. In contrast with subsidized public buses, paratransit operations actual!y eam profits. In addition, paratransit does not simply siphon
riders from other mass-transit modes, but also causes sorne travelers to switch from
single-occupancy vehicles. In surnmaiizing the prospects for paratransit, Cervero
(1996) notes:
Given the fiscal cutbacks facing America's public transit industry today, the expansion of more entrepreneurial, commercial transportation services seems unavoidable.
While critics charge that the peor will suffer as a result, other remedies-like user-side
subsidies-are available for redressing inequities. Moreover the history of commercial
partransit 'ls certainly not ene ofigrioring pooi eighborhoods.
jneys and neighborhood car services, Iow-income areas have traditionally been their market base.
For
312
"""',
BRITISH EXPERIENCE WITH DEREGULATION
In Great Britain, the transit industry was deregulated under the British Transport Act
of 1985. As explained by Gmez-Ibez and Meyer (1990), the act relaxed controls
on entry into the transit industry, reorganized most public transit authorities as forprofit organizations, and introduced competitive bidding for certain transit services.
Local governments can continue to subsidize transit services as long as the subsidized
service is provided by the low bidder in a competitive auction. Although the London
area was not covered under the Transport Act, contracting out is used extensively in
metropolitan areas, and full deregulation is expected sometime in the 1990s.
Britain's new transit policy combined deregulation with substantial cuts in transit subsidies. For example, transit subsidies were cut by about 23 percent in the
metropolitan counties and about 6 percent in other counties. The decrease in transi t
subsidies generated pressure to increase fares and cut services. Gmez-Ibez and
Meyer (1990) and Small (1992) surnmarize sorne of the results of the new transit
policy:
l. Service. In the first full year following deregulation, ttal mileage was up by
3.3 percent in the metropolitan counties and 16.5 percent in other counties. The
use of minibuses (with.l2 to 25 seats) increased dramatically. Sorne unprofitable
services were curtailed; for example, sorne off-peak and low-density routes were
abandoned.
2. Fares. Transit fares increased by about 35 percent in real terms over a two-year
period.
1
1
313
ofthe intemal-combustion engine allowed the development of new travel modes (the
car and the truck) that were more flexible and less costly than streetcars and horsedrawn wagons. The new modes decreased the reliance of firms and households on the
central area of the city, allowing the decentralization of employment and residents.
This section discusses two types of changes in transportation: an increase in gasoline
cost and the construction of fixed-rail transit systems.
Gasoline Prices
Suppose that the real cost of gasoline increases. How will the increased cost of auto
travel affect housing and land prices? As explained in Part II, increase in transportation cost increases the slopes of the housing-price function and the residentia!bid-rent
function. Studies bySmall (1986) and Coulson andEngle (1987) confirm these theoretical predictions; an increase in the price of gaso!inincreases the relative attractiveness oflocations near employment centers, increasing the prices ofhousing and land.
How will an increase in the cost of gasoline affect the distributions of employment and population? The changes are likely to be relatively small for three reasons.
First, the cost of gasoline is a relatively small part of the cost of travel. Second,
other responses to an increase in gasoline prices (buying more fuel-efficient cars,
carpooling, switching to public transit) allow drivers to decrease travel cost without
changing residences or workplaces. Por these two reasons, the changes in the prices
of housing and land are likely to be relatively small. Even if the changes in price
were large, it would take a long time before there would be changes in population
density near employment centers because structures last a long time and are costly
to demolish. Small (1986) estimated that a $1 per gallon increase in the price of
gasoline would-cause a relatively small number of suburbari households to move to
the central city.
Fixed-Rail Systems and Land Use
One of the objectives of San Francisco's BART was to increase employment opportunities in the areas near transit stations. An early study of BART (Dyett and
Escudero, 1977) suggests that BART hada moderate positive effect on employment
in downtown San Francisco, but not much of an effect elsewhere. Similar!y, Webber
(1976) concludes that BART increased downtown employment but didn't affect employment near suburban stations. Webber suggests that office development in the
downtown area would have been about same if transit authorities had expanded bus
service instead of building BART. More recently, Cervero and Landis (1995) show
that BART increased employment in areas near transit stations, with most of the
growth in the downtown area.
A recent study of Atlanta's MARTA suggests that the heavy-rail system did not
increase employment in areas near transit stations (Bollinger and Thlanfeldt, 1997).
Although public employment in these areas grew, the gains in public employment
_ were offset by losses in private-sector empl()yment. This result refiects the policy of
shifting gove=ent employment to areas along raillines.
'
.. ..'f:
314
There are two main reasons why fixed-rail systems have such small effects
on employment. First, the typical new system does not significantly increase the
accessibility of sites near station areas because those sites were already accessible
by private automobile. In contrast, streetcars in the late 1800s made areas along the
spokes of -the hub-and-spoke system accessible to urban workers, causing dramatic
changes in land use. Similarly, the building of suburban highways in the 1950s and
1960s made suburban areas much more accessible, resulting in fundamental changes
in land-use patterns. The second reason for the smallland-use effects of fixed-rail
systems is that buildings in urban areas are durable, lasting at least 50 years. Even
if a new transit line increased the accessibility ofa particular area enough to make
highcdensity employment sensible, the transition from low density to high density
would probably occur slowly.
In sumnarizing the results from severa! studies of the effects of rai! systems on
land-use patterns, Altshuler (1979) makes two observations:
1. In a growing economy, rail transit contributes to the clustering of activities near
downtown stations. These clustering effects are usually negligible outside the
central business district.
2. Invesi:ment in rail transitis sensible only if it is used in concert with more powerfulland-use instruments such as zoning and property taxation. Ifthe government
uses its zoning and tax policies to generate high-density development, rail transit
provides an ef:ficient means of delivering a large number of workers to the dense
central area.
In general, there is evidence that fixed-rail systems cause only small changes in
the spatial distribution of employment. The primary effect of transit is an increase
in downtown employmet.
SUMMARY
l. Transit ridership is relatively unresponsive to changes in monetary cost (fares),
but responsive to changes in time cost, especially waiting and walking costs.
On average, conunuters value in-vehicle time at about half their wage rate and
collection and distribution time at between one and one and one-half times the
wage rate:
2. The time cost of a trip is the sum of collection cost, in-vehicle cost, and distribution cost. Conunuters choose the mode that minimizes the sum of monetary
and time costs.
3. Diarnond lanes cause a moderate number of conunuters to switch from autos to
buses. The decrease in the number of auto lanes increases congestion, increasing
the time cost of auto conunuters.
4. A study of alternative transport systems for the San Francisco Bay area concluded that (a) BART was more costly than an integrated bus system for all corridor volumes studied, and (b) Forcorridor volumes exceeding 1,100 vehicles
per hour, an integrated bus system was less costly than an auto-based system.
...'
'
-
'
'
'
315
5. In 1995, passen:ger fares covered only about 38 percent of the operating cost
of transit. Transit deficits have increased over the last few decades, a result of
decreases in fares and productivity and increases in wages, energy cost, and
transit mileage.
6. Deregulation of transit in the form of contracting arrangements has decreased
the cost of providing transit services in a number of cities.
7. Fixed-rail systems cause relatively small changes in location pattems. If the
govemment uses its zoning and tax policies to generate high-density development, rail transit can provide an efficient means of delivering workers to the
high-density areas.
BART
12
0.20
16
0.20
40
0.15
20
0.15
11
0.20
0.20
0.20
1.5
Auto
o
0.20
30
0.15
Monetary cost
316
3. Marta the transportation planner recently proposed that bus headways (the time
317
12. Considerthe effects of dropping a planner's bomb on the San Francisco Bay area.
A planner's bomb doesn't hurt any people, but destroys everything except the
BART infrastructure (tracks, vehicles, and station). Most important, it destroys
a11 buildings, so the metropolitan area must be completey! rebuilt. Design a set
of public policies that will ensure that in the rebuilt Bay area, BART ridership
will be high enough that BART is as efficient as an integrated bus system.
13. In the city of Phoenix, private companies provide bus services on low-volume
routes ata fraction of the cost of the public bus agency. In New York City and
Chicago, private companies provide bus service on high-volume (peak-period)
routes at afraction ofthe cost ofthe public bus agency. Ifprivate firms can underprice the public sector on high-volume and low-volume routes, what, if anything,
should be the role of the public sector in the provision of transit services?
14. Suppose that the purpose of cross-subsidization is to increase the we!fare of
people a!ong low-volume routes. Discuss sorne a!ternative means of improving
their we!fare.
REFERENCES
.l. Altshuler, Alan A. The Urban Transportation System. Cambridge, MA: Joint
318
11. Fielding, Gordon J. Managing Public Transit Strategically: A Comprehensive Approach to Strengthening Service and Monitoring Performance. San
Francisco: Jossey-Bass, 1987.
12. Giuliano, Genevieve, and Kenneth A. Small. "Alternative Strategies for Coping
with Traffic Congestion." University of California Transportation Center, Working Paper No. 188. Berkeley: University of California Transportation Center,
1994.
13. Giuliano, Genevieve, and Roger F. Teal. "Estirnating the Potential Cost Savings
ofTransit Service Contracting." Transponation Research Record 1108 (1987),
pp. 1-11.
14. Gmez-Ibez, Jos A. "A Dark Side to Light Ral?" Joumal ofthe American
Planning Association 51 (Summer 1985), pp. 337-51.
15. Gmez-Ibez, Jos A., and John R. Meyer. "Privatizing and Deregulating
Local Public Services: Lessons from Britain's Buses." Joumal of the American Planning Association 56 (Winter 1990), pp. 9-21.
16. Gordon, Peter, and Harry Richardson. "Notes from Underground: The Failure
of Urban Mass Transit." The Public lnterest 94 (1989), pp. 77-86.
17. Heseltine, P. M., and D. T. Silcock. ''The Effects ofBus Deregulation on Costs."
Joumal ofTransponation Economics and Policy 24 (1990), pp. 239-54
18. Keeler, Theodore E.; L. Merewitz; P. Fisher; and K. Small. The Full Costs
of Urban Transport. Monograph 21, part 3, Institute of Urban and Regional
Development, University of California, Berkeley, 1975.
19. Kraft, Gerald, and Thomas Domencich. "Free Transit." In Readings in Urban
Economics, ed. Matthew Edel and Jerome Rothenberg. New York: Macmillan,
1972, pp. 459-80.
20. Lave, Charles A. -"The Private Challenge to Public Transportation-An
Overview." In Urban Transit: The Private Challenge to Public Transponation,
ed. Charles Lave, Chapter l. Cambridge, MA: Ballinger, 1985.
21. McCullough, Williarn S.; BrianD.Taylor; andMartin Wachs. "Does Contracting
Transit Service Save Money?" Access 11 (Fall, 1997), pp. 22-26.
22. Mohring, Herbert. ''The Benefits ofReserved Bus Lanes, Mass Transit Subsidies, and Marginal Cost Pricing in Alleviating Traffic Congestion." In Current
Issues in Urban Economics, ed. P. Mieszkowski and M. Straszheim. Baltirnore:
Johns Hopkins University Press, 1979.
23. Morlok, Edward K., and Philip A. Vitori. "The Comparative Costs of Public
and Prvate Providers of Mass Transit." In Urban Transit: The Private Challenge to Public Transponation, ed. Charles Lave, Chapter 10. Cambridge, MA:
Ballinger, 1985; pp. 233-50.
24. Peskin, R.; S. Mundle; and P. K. Varma. "Transit Privatization in Denver: Experience in the Second Year." Transponation Research Record 1402 (1992),
pp. 17-24.
25. Pickerell, Don H. ''The Causes of Rising Transit Operating Deficits." Transponation Research Record 915 (1983), pp. 18-24.
26. Richmond, Jonathan E. "New Ral Transit Investments-A Review." Taubman
Center for State and Local Govemment, 1998.
,
!
319
27. Rosenbloom, Sanc!ra. "The Taxi in the Urban Transport System." In Urban
Transit: The Private Challenge to Public Transportation, ed. Charles Lave,
Chapter 8. Cambridge, MA: Ba!linger, 1985.
28. Smal!, Kenneth A. "Economics and Urban Transportation Po!icy in the United
States." Regional Science and Urban Economics 27 (1997), pp. 671-91.
29. Small, Kenneth A. Urban Transportation Economics. Philadelphia: Harwood,
1992.
30. Small, Kenneth. "The Effect of the 1979 Gasoline Shortages on Philadelphia
Housing Prices." Joumal of Urban Economics 19 (1986), pp. 371-81.
31. Walters, A. A. "Extemalities in Urban Buses." Journal ofUrban Economics 11
(1982).
32. Webber, Melvin W. "The BART Experience-What Have We Learned?" The
Public lnterest, Fa!!1976, pp. 79-108.
. '
1
PART FOUR
Urban Problems
Tru.s part of the book examines the econornics of three urban problems: po erty,
low educational achievement, and crime. In most metropolitan areas, these problems
are concentrated in a small nurnber of neighborhoods, most of which are in the
central city. As we'll see in Chapter 13, households tend to sort themselves with
respect to income, education leve!, and race, and this sorting contributes to the
spatial concentration of urban problems. Chapter 14 explores the problem of urban
povercy, examining the causes and consequences of concentrated poverty. Chapter 15
explores the reasons for low educational achievement in central-city neighborhoods
and evaluates the merits of policies that have been proposed to address the problem.
Chapter 16. develops an econornic model of crime, and uses the model to evaluate
anti-crime policies, explain why crime rates are higher in central cities, and explain
why crime raies dropped so rapidly during the 1990s.
321
CHAPTER 13
Household Sorting
This is the first of several chapters exploring the economics of urban problerns,
including poverty, crirne, poor educ;ation, and inadequate housing. Most of these
urban problerns are not spread evenly throughout rnetropolitan areas, but are instead
concentrated in sorne central-city neighborhoods. For exarnple, the poverty rate
within central cities is about twice the poverty rate for rnetropolitan areas (11.2 percent). In the Cleveland rnetropolitan area, researchers have identified 95 differentneighborhoods. In 1989, the poverty rates exceeded 40 percent in 8
neighborhoods and 30 percent in 19 neighborhoods. Sirnilarly, crirne rates are
very high in sorne central-city neighborhoods.
One reason for the spatial concentration of urban problerns is that households
tend to sort the:psefves into neighborhoods of persons with similar incorne, education, and race. The typical neighborhood does not have a perfect mix of households
of different socioeconornic backgrounds, but instead is relativel y hornogeneous. The
hornogeneity has two irnplications for urban problerns. First, to the extent that low
incorne causes other problerns such as crirne and inadequate housing, these problerns
will be concentrated as well. Second, there is evidence that a large concentration
. of low-incorne households rnakes poverty and other urban problerns even worse. In
other words, a low-incorne household will fare worse if it is surrounded by other lowincorne households. This chapter explores the social and rnarket forces that cause
households to be sorted in neighborhoods with respect to incorne, education, and race.
v; bro ,$e-le<.o.o'\.-
One reason for household sorting is that people sort thernselves with respect to their
dernands for local public goods such as education and public safety. The dernand for
local public goods varies across households, and c::itizens in the typical rnetropolitan
areas can choose frorn dozens of local municipalities, each with a different bundle of
local public goods. The variety of municipalities allows citizens to "vote with their
feet,"choosing the jurisdiction with the best combination of public services and taxes.
i
1
1
1
--
323
324
What about streets and public parks? One way to charge people for using city
streets would be to set up a systerrt of toll booths, one on each comer. Such a system
would be costly; the city would have to hire a large nurnber of toll collectors, and
road users would incur a substantial time cost in stopping at every comer to pay
the tolls. As explained in the chapter on autos and highways, there is a low-cost
system-the congestion tax-for charging drivers for using streets and highways.
There is little political support for sucb a system, so streets are likely to remain
nonexcludable. The same argument applies to parks. Although it might be possible
to install fences and turnstiies in city parks, such a system wuld be very costly.
.-,
-'-" .
325
What about public-safety services? For sorne safety activities, it would be easy
to identify the beneficiarles and send them a bill for the safety services rendered. For
example, the fire department could charge afee for putting out fires, and the police
could charge afee for responding to burglarles and robberies. Under such a system,
the fire and police departments would not provide services to people who could not
afford the service fees. This system of user charges is po!itically infeasible, because
it would violate most people's notions of equity and fairness. For other types of
safety activities, it would be difficult to identify the beneficiarles and charge for the
services. For example, when a city captures, prosecutes, and imprisons a criminal,
every potential crirne victim benefits. Other examples are crowd and traffic control.
For these services, it would be impractical to exclude people who do not pay.
The third characteristic of a local public good is that its benefits are confined
to a relatively small geographical area. Unlike national defense, which generates
benefits for the entire nation, most of the benefits of the local police force and
local fue department go to local citizens. Simi!ar!y, local citizens get most of the
benefits from local streets and highways. The appropriate size of the jurisdiction is
determined by the "loca!ness" of the public good (the geographical extent of the
benefits from the good): the more extensive the benefits, the larger the jurisdiction
required to contain all the beneficiarles.
Figure 13-1 shows three individual demand curves for parks, one for each
citizen. A demand curve shows how much a consumer is willing to pay for one
more unit of a product. Iri other words, the demand curve is also a willingness-topay curve anda marginal-benefit curve. In Figure 13-1, Sam has a low demand
for parks and thus has a low marginal-benefit for park acres (MBs)- For example,
point w indicates that he is willing to pay $20 for the 10th acre ofparks. In contrast,
Marian has a medium.demand and marginal-benefit curve (MBm), and is willing to
pay $30 for the 1Oth acre of parks (point x). Finaly!, Bertha has a high demand and
a high marginal-benefit curve (MBb), and is willing to pay $53 for the 10th acre
(point y).
We can use the marginal principie to determine the efficient level of park acreage.
This principle tells us that the efficient leve!of a particular activity is the leve!at
326
Park acreage
The marginal social benefit of park acreage is .the sum of the private benefits of
three citizens. The efficient park acreage is the arnount such that the marginal
social benefit equals the marginal social cost, whicb in this case is 70 acres.
which the marginal benefit equals the marginal cost. The social benefit of a given
park size is the sum of the marginal benefits of Sam, Marian, and Bertha: If there
is no congestion, the three citizens do not affect their fellow citizens' enjoyment
of the park; if there are no spillovers, there are no other potential beneficiaries
to worry about. For 10 acres of parks, the marginal social benefit is $103 (Sam's
marginal benefit is $20, Marian's is $30, and Bertha's is $53). This is shown by
point z. Similar!y, for other park sizes, we add the marginal prvate benefits to get
the marginal social benefit. In other words, the marginal-social-benefit curve is the
vertical sum of the individual marginal-benefit curves.
At the efficient leve!, the marginal social benefit of park acreage equals its
marginal social cost. In Figure 13-1, the marginal costis $60 per acre, so the optimum
acreage is 70 acres. For any amouD.t less than 70 acres, citizens in the city would
be willing to pay more than $60 per additional acre, so an increase in size would
increase social welfare. For examp!e, suppose the city starts at 10 acres. At this
point, the willingness to pay for parks (from the social-benefit curve) is $103 and
the cost is only $60, so an increase in park acreage would generate a large net gain.
In contrast, for any amount exceeding 70 acres, the willingness to pay would be !ess
than the social cost, so a smaller pak would be more efficient.
327
!O
55
ll5
Park acreage
lf each household pays $20 per acre to finan c. parks. che preferred amouncs are
O acres (point s forSain). 55 acres (poinc m for Marian), and 115 acres (point b
for Bertha). Under majoriey voting, voters will choose the preferred size of the
median voter (Marian).
328
so the 1arger size wins the e1ection. Marian's preferred size wou1d also win in an
e1ection against the 1arger size (115 acres): Sarn and Marian wou1d vote for 55 acres.
Marian's preferred size wou1d win both e1ections because she is the median voter,
de:6.ned as the voter that splits the rest of the voting public into two equal ha1ves.
The prefeired budgetofthe median voter wins an e1ection against a smaller size (as
Berthajoins the median voter to make the vote 2 to 1) and any 1arger size (as Sarn
joins the median voter to make thevote 2 to 1). This is the median-voter result: If
the spending 1eve1 for a public good is deterroined by a majority vote, the winning
budget is the preferred budget of the median voter.
In Figure 13-2, the size chosen by voters (55 acres) is 1ess than the optimum
size (70 acres). ls this an artifact of the e1ection sequence, or wou1d 55 acres win
a direct e1ection against the efficient 1evel? In an e1ection between 55 acres and
70 acres, Sarn wou1djoin Marian in voting for 55 acres because 55 acres is c1oser to
Sarn's preferred size. Therefore, the preferred size of the median voter will defeat
the efficient size in a direct e1ection.
The power-and inefficiency-of the median-voter resu1t can be shown by considering the ef[ects of changes in Bertha's preferences. Suppose that her marginal
benefit curve shifts upward by 100 percent: for every size, her wi1lingness to pay for
parks doub1es. Because the marginal-socia1-benefit curve is the vertical sum of tbe
individual marginal-private-benefit curves, tbe social-benefit curve would also shift
upward, and the optimum park size wou1d increase by a 1arge arnount. Nonethe1ess,
voters wou1d still choose 55 acres because Sarn will join with Marian to defeat any
size above Marian's preferred size. The sarne is true of changes in the other nonmedian benefit curve MBs. The voting outcome depends only on the position oftbe
marginal-benefit curve of the median voter.
'.
329
Household Sorting
3. No interjurisdictional spillovers. There are no spillovers (externalities) associated with local public goods: al!the benefits froln local public goods accrue
to citizens within the municipality.
4. No scale economies. The average cost of production is independent of output.
5. Head taxes. A municipality pays for its public goods with a head tax: if you
have a head, you pay the head tax.
To extend the ear!ier exarnple of park acreage, considera metropo!itan area that
has nine citizens divided into three municipalities. Initially, the municipalities are heterogeneous, with one citizen of each type. Each municipality has one Sarn-type citizen (preference for a small park), one Marian-type citizen (preference for mediumsize park), and one Bertha-type citizen (big park). In this case, each heterogeneous
municipality will choose 55 acres, the preferred size of its median voter (Marian).
Two of three citizens in each municipality would prefer a different park size.
Under the Tiebout process, households can sort themselves into municipalities
according to their demand for parks. Suppose the three Sarns get together to forrn one
municipality forpeople withre!atively low demand forparks (municipality S).In Figure 13-3, aggregating the marginal benefits ofthree Sarn-type citizens, the marginal
FIGURE 13-3 Citizens in a Homogeneous Municipality Choose the
Efficient Leve!of the Local Public Good
$
Marginal social
benefit = 3 times MB,
20 p.-f-.,=----1
1
1
1
10
Park size (acres)
Municip3.J..ity S has three Sarn-type households (low demand for parks), so tbe
marlo!inal social-benefit curve is three times the marginal private-benefit curve. 1f
costs are divided equally between the three citizens, they will al!agree to a park
size of 10 acres. The inefficiencies of majoricy voting are eliminared by sorting of
households into homogeneous municipalities.
.
. . .
Urban Problerns
social benefit of a given park size is three times the-marginal prvate benefit of the
typical citizen. The marginal-benefit curve intersects the marginal-cost curve at a
park acreage of 1O acres, and this is the efficient size for the municipality.If the peracre cost of the park is divided equally between the three households, the marginal
prvate benefit for each citizen equals the marginal cost at the efficient size (10 acres).
The three households have identical preferences for parks, so there is unanimous
support for the efficent park size.
The same analysis applies to decision making in the two other homogeneous
communities. Each municipality has three households with the same margina!benefit curve, so an equal sharing of park costs causes unanimous support for the
efficient park size. In municipality M, the three Marians vote for a medium-size park
(55 acres).In municipality B, the three Berthas votefor a big park (115 acres). Under
the Tiebout system, the inefficiencies of majorty voting are avoided by sorting of
households into homogeneous communities.
Citizens in the Tiebout world sort themselves into municipalities according to
their demands for local public goods. The same logic applies to school districts and
other jursdictions providing public goods. Before citizens in a Tiebout metropolitan
areas vote with ballots, they vote with their feet, forming homogeneous jurisdictions
where the decisive median voter has the same preferences as all other voters. As
a result, the citizens avoid the inefficiencies that can occur in jurisdictions with
differing demands for local publicgoods.
330
Part 4
331
FIGURE 13-4 Sorting with Respect to Demand for
Local Public Gooda
G.
Land will Q.e occupied by the highest bidder. InFigure 13-4, low-income households outbid high-income households for land in municipa!ities where the quantity
of local services is less than G*. In contrast, high-income households have a higher
bid for municipa!ities with G > a. High-income households occupy land with large
G because they have a steeper bid-rent function. The slope indicates how rapidly
the bid rent increases as the quantity of local pub!ic goods increases. The household
with the steeper bid-rent function has the most to gain from higher levels of G.
What determines the slope ofthe bid function? The slope is defined as the change
in the bid rent (per unit of land) for a one-unit increase in the local public good.
.
.B
Slope =
G
6
The change in the bid rent equals the household's marginal benefit of G (the willingness to pay for a one-unit increase in G) divided by the amount ofland consurned
by the household.
.B
Marginal benefit of G
-=
t:,. G
Land consumption
Table 13-1 shows a numerical example of the slope of the bid-rent function. In the
first column, a low-income household has a marginal benefit of$20 and consumes
332
TABLE 13-1 Slope ofthe Bid-Rent Function for Local Public Goods
Marginal beoefit of G
Land consumption
Slope = Marginal benefitfLand
Low-Income Household
High-Income Househo!d
$20
$120
3 units
$40 per unit of land
2 units
$10 per unit ofland
,
r'
two units of !and. Dividing the marginal benefit of G by land consumption, the slope
of the bid-rent function is $10: the household is wi!ling to pay $10 per unit of land
for a one-unit increase in G. Similary!, if a high-income household has a marginal
benefit of $120 and occupies three units ofland, the change in the bid rent is $40 per
unit of land.
How does income affect the slope of the bid-rent function? Local public goods
are typically "normal" goods in the sense that increases in income increase the
utility-maxirnizing consumption leve!of the local public good. Therefore, a highincome household will have a !arger marginal benefit than a low-income household:
the numerator in the expression for the slope of the bid-rent curve is larger for
the high-income househo!d. In addition, "land is a normal good, so a high-income
household will consume more land than a low-income household: the high-income
household also has a !arger denominator. Since the high-income household has both
a larger numerator Oarger marginal benefits) and a larger denominator Oarger land
consumption), an increase in income has an ambiguous effect on the slope of the
bid-rent curve. In theory, the high-income household could have a steeper bid-rent
curve or a flatter one.
The high-income household will have a steeper bid-rent function ifthe marginal
benefit is more income-e!astic than the demand for land. In Table 13-1, suppose
the high-income household has twice the income of the low-income household.
As shown by the first row of the table, the marginal benefit of the high-income
household is six times the marginal benefit of the low-income household ($120
compared to $20), indicating that the marginal benefit is highly income-e!astic.
In contrast, the land consumption of the high-income household is only 1.5 times
the land consumption of the low-income household (three units compared to two).
Because the marginal benefit is more incomecelastic than the demand for land, the
high-income household has a steeper bid-rent function. The slope is $40 (equal to
$120 divided by three units of!and), compared to $10 (equal to $20 divided by two
units of land).
Our analysis of the allocation ofland with differing levels of the local public good
is similar to our earlier analysis of the allocation of land with differing accessibility.
In Chapter 8, we saw that land c!osest to the city center was occupied by the land user
with the steepest bid-rent function, that is, the user with highest unit transportation
cost. The allocation of land near the city center to the user with high transport costs
is efficient because it gives proximate land to the user with the most to gain Oower
transport costs) from proximity. In the context of local public goods, land with the
largest quantity of the local public good is allocated to the household that has the
most to gai.l (the largest marginal benefit)per unit of land.
'
333
We've seen that households sort themselves to get different levels oflocal public
goods, resulting in sorting according to income. Income sorting happens because the
demand for local services depends on income, so when households sort themselves
to get different quantities of local public goods, they end up sorting themselves
with respect to income. In contrast, sorting does not happen with respect to other
household characteristics such as the color of hair because hair color doesn't affect
the demand for local public goods.
Part 4
334
Urban Problems
Big-House
Municipality
Small-House
$3,000
50
$3,000
$3,000
50
50
50
. $200,000
1.5%
$300,000
1%
$100,000
3%
Mixed
Municipality
Spending per household
Number of small
houses ($100,000)
Number ofbig
houses ($300,000)
Average house value
Property tax
Municipality
municipality with the characteristics shown in the first column of Table 13-2. The
municipality is homogeneous with respect to the demand for local public goods: for
each household, the desired spending is $3,000 per household. If the municipality
initially finances local public goods with a head tax of $3,000 per household, there
is no incentive for any household to move: each is getting the.level of local public
goods it wants, and paying for what it gets.
335
What about the households who live in small houses? After the big-house folks
leave, the original municipality will have only small houses, each with a property
value of $100,000 (shown in the third column of Table 13-2). To pay for $3,000
worth of local public goods per household, the property tax rate must rise to 3 percent. In the absence of large-lot zoning, the small-house households would try to
move into the big-house municipality, buying a $100,000 house and paying only
$1,000 in property taxes (1 percent of $100,000). But large-lot zoning makes such
a move impossible; the only way to get the lower property tax rate is to buy a big
house.
.
336
Parents are willing to pay more for housing in neighborhoods that have favorable
peer groups in schools:
What sort of neighborhoods have favorable peer groups? The best students come
from families that are involved in schools and encourage their children to learn. In a
general statistical sense, the quality of school peer groups increases with the educa. tion leve!and income of the parents. Of course, sorne middle-income neighborhoods
have better peer groups than sorne high-income neighborhoods because the quality
of the peer group depends on income, education, and also other household characteristics.
",
Cbapter !3
Household Sorting
337
Low-Income
Household
High-Income
Household
$40
2units
$20 per unit of land
$90
3 units
$30 per unit ofland
lf' ',
,-.-.
;{
'
' ( .
- t ,
l
1
s
Quality of peer group: average test score
Tbe. bid rent for land depends on the quality of peer groups (as measured by
test scores) in local schools. If the hid rent of the high-income household is
steeper, they will outbid low-income households for land with the highest
quality peer groups.
t
1
1
(
1
1
1
1
1.
1
1
1
{
low-income households in the neighborhood with lower quality peer groups. Inother
words, sorting with respect to income (and education) continues.
The new households live in neighborhoods with similar households because the
income elasticity of the marginal benefit of test scores (better peer groups) exceeds
the income elasticity of demand for land. This result is similar to the earlier result
concerning sorting motivated by different levels of local public goods. If higherincome households have a larger marginal benefit for local public goods or peer
groups, they willlive in the communities with the higher leve!of public goods or
peer groups as long as their demand for land isn't too much larger. The demand
for land matters in allocating land because the right to local public goods and peer
groups is tied to thconsumption of land.
338
'
'
'
120
.
.
o
.e
w
w
I,C)
17
25
33
41
49
57
65
73
Rank of neighborhood
Source: Center for Urban Poverty nnd Social Change, Case Weslem Reserve Universily. http://povertycenler.cwru.edu
81
89
.wo.,.
60
:l
'O
E" so
o
o
;
'O
40
""'
t;
;
-:
30
"
."
."
..
o'
20
"
""'
10
17
25
33
41
49
57
65
73
81
89
Rank of neighborhood
Source: Center for Urban Poverty nnd Social Chnnge. Case Weslern Reserve University. hllp://poverlycenler.cwru.edu
_._,..
r -
341
RACIAL SORTING
So far in this chapter, we've seen that households sort themselves into neighborhoods
according to the demand for local public goods, housing consumption, and school
peer groups.In this part of the chapter, we'11 explore sorting with respect to race, also
known as racial segregation.In the United States, more than two-thirds of the blacks
living in metropolitan areas reside in central cities; leaving one-third of metropolitan
blacks for suburban areas. For whites, the fractions are reversed: One-third live in
central cities, leaving two-thirds for the suburbs.
Segregation Facts
One way to quantify the degree of racial segregation is the index of dissimilarity.
This index shows the proportion of one racial group that must relocate to achieve a
completely integrated spatial configuration, one in which each census tract replicates
the racial composition of the metropolitan area as a whole. In 1990, the average
dissimilarity index for U.S. metropolitan areas was 69 percent, indicating that to
achieve complete integration, 69. percent ofblacks (or whites) would need to relocate
(U.S. Census Bureau, 1998). As shown in Figure 13-8, the degree of segregation
FIGURE 13-8 Dissimilarity Indexes for Different MSAs
100
90
80
..,
70
.5 60
><
"C
- 50
"
]
40
30
20
10
o
"'
=-a
=-.Si
"'
8
o
"
5
g"..,'
::;:
o;
::i
'
'
<
:";':
"'.
o
o
"g'.
e::.
...1
-=
"
..:E
o
"'
"C
"
.o..
u
"
"
"
"
..,
-o
"'
:2
...1
Ol
.Si
e::.
:;
"'
"'
':
o.
"
:a
"
..:
"C
"
"'
:.
"
:".a,
::;:
Sourc: U.S. Bureau of the Census, Residencial Segregarion, Summary Tables and Residenrial Segregarion. Detailed
Table.s (Washington, DC: U.S. Governmeot Printing Office, 1998).
T
.
342
..
80
:a
"o"
"
o
60
-"!
.o"."..''
o
':;
40
"'
20
00
nnnQnnnnnnnnRDfi[
19
13
25
31
37
43
49
55
61
67
73
Rank of neighborhood
Source: Center for Urban Poverty Rnd Social Change, Case Weslem Reserve University. http://po.,.ertycenter.cwnl.edu
79
85
91
344
black buyers are not given information about dwellings in predominantly white
neighborhoods. Second, brokers provide inferior service to black households by
(1) misrepresenting the housing market, (2) delaying transactions, and (3) giving
!ess help in arranging financing. According to Yinger (1998), there is conclusive
evidence that blacks are treated differently from otherwise identical whites: blacks
are shown fewer dwellings, steered into certain neighborhoods, and given less advice .
. and assistance on financing options. Por example, among black renters, one in ten
was excluded altogether from housing made available to white renters, and one in
fourlearned about fewer vacant dwellings.
Another factor is exclusionary zoning in suburban communities. As explained
earlier in the book, suburban governrnents use zoning to exclude low-income households from their jurisdictions. Among the exclusionary practices are minirnurnsquare-footage requirernents for lots and dwellings, prohibition of rnultifarnily units,
rnaxirnum densities, requirements for two-car garages, and development fees. Exclusionary zoning is rnotivated in part by fiscal concerns: If low-incorne households
do not pay their "fair share" of the costs of local public services, suburban governrnents have an incentive to exclude thern. Since black households have lower
incornes, on average, than white farnilies, exclusionary zoning has a larger effect on
black households.
Until recentiy, federal housing policies have indirectly encouraged segregation.
Historically, federal housing policies have encouraged the poor to live in poor centralcity areas. Most public housing was built in poor areas, and until recently, housing
vouchers (coupons used to help pay for private housing) could be used only in the
city where the recipient lived when the voucher was issued. Recently, the federal
governrnent made most housing vouchers portable, so more recipients can use thern
to rent suburban ho sing.
Neighborhood Transition
In the 1970s and 1980s, many neighborhoods in U.S. central cities were transformed
frorn segregated white neighborhoods into neighborhoods with black majorities.
What factors affect the speed of neighborhood transition?
An irnportant factor is the attitude toward'integration. A white household that
prefers a segregated neighborhood is likely to leave the neighborhood as its minority population grows. As shown in Table 13-4, the number of departing whites
increases with the minority's share oftheneighborhood population. These numbers
TABLE 13-4 Racial Mix and Transition
Percent of Neighborhood
That Is a Minority Group
7
24
41
64
20
33
60
1
1
l
345
explain why many neighborhoods change rapidly from highly segregated white
neighborhoods to highly segregated minority neighborhoods. An increase in the
share of rninority households decreases the attractiveness of the neighborhood to
many white households, causing white flight and further increases in the rninority
share of households.
Galster (1990) studied the racial transformation of neighborhoods in the
Cleveland metropolitan area during the 1970s. He explored the factors that determine the rate at which the black share of a neighborhood's population increases
over time. The most important factors are (1) the initial black share of population,
(2) white attitudes toward segregation, and (3) the proxirnity of the neighborhood to
other neighborhoods where blacks are in the majority.
Figure 13-10 shows Galster's results for three types of neighborhoods. The
curves show the gains in the black share of population (in percentage points) over
the 1970-1980 decade for different initial black shares. Curve A shows this relationship for a neighborhood that (1) is not adjacent to a black-majority neighborhood
and (2) has "average" white preferences for segregation. For example, a neighborhood that starts the decade as 10 percent black gains 5.5 percentage points over the
FIGURE 13-10 Changes in the Racial Composition of Cleveland Neighborhoods:
1970-1980
40
o
00
"; !';
....
s"'
Affmnative marketing
30
"=
".",
"
:0
-e
Average preferences
"'="
10
22
37
30
Percent b1ack in 1970
50
60
Source: Figure from "Neighborhood Racial Change, Segregationist Sentiments, and Affirmative Marketing
Policies:' by George C. Ga!ster in Joumal oj Urban Economics 27, pp. 344-61, 1990 Elsevier Science (USA),
reproduced by permi.ssion o_f tbe publisher.
T
.
346
decade, so the neighborhood is 15.5 percentblack at the end ofthe decade. Along the
positively sloped portion of the curve, the gain in the black share increases with the
size of the initia! b1ack share. Por examp!e, if the initia! share is 20 percent, the gain
is 9.5 percentage points, so the 1980 share is 29.5 percent. Por the type of neighboi:hood shown by curve A, there is a black majority by the end of the decade if the
initia! black share exceeds about 37 percent (point X).
Curve C in Figure 13-10 shows the changes in the black share of population for a neighborhood that (1) is adjacent to a black-majority neighborhood and
(2) has relatively strong white preferences for segregation. Por a given initia! share,
this neighborhood experiences more white flight than the average neighborhood, so
the gain in the black share is relatively large. For example, for an initia! share of
1O percent, the gain is 20 percentage points, so the black share at the end of the
decade is 30 percent. This type of neighborhood is more likely to have a black majority by the end of the decade than the average neighborhood: Por any initia! share
greater than 22 percent (point Z), the black gain is large enough to generate a black
majority.
During the 1970s, two municipa!ities in the Cleveland area (Shaker Heights
and Cleve!and Heights) used public policy to encourage racially ba!anced neighborhoods. The idea was to encourage blacks to move into white neighborhoods, but to
prevent the development ofblack-majority neighborhoods. There were a number of
features of this "affirmative marketing" po!iy, including (1) information and persuasion campaigns to attract blacks, (2) policies to maintain the quality of housing
and public services (especia!ly education) to discourage the departure ofhouseholds
.who place a high va!ue on the quality of housing and public services, and (3) special broker services for members of the "underrepresented" racial group, typically
whites. To prevent the white flight that rnight drop the number of white households
below 50 percent, the municipalities catered to white households.
The third curve in Figure 13-10 shows the gain in the black share for a neighborhood that (1) engaged in affirmative marketing and (2) is not adjacent toa black
majority neighborhood. Por a neighborhood with a sma!l initial black share, the
black gain exceeds the gain experienced by the average neighborhood (shown by
curve A). Por example, for an initial share of 10 percent, the black gain in the affirmative marketing neighborhood is 24.5 percentage points, compared to 5.5 points in
the average neighborhood. In contrast, for a large initia! black share, the affirmative
marketing neighborhood experiences a relative!y sma!l gain. Por example, for an
initia! share of 28 percent, the black gain is 3 percentage points, compared to 12
points in the average neighborhood. In the neighborhood with affirmative marketing,
the b!ack share of the population is stable if theinitia! share is about 30 percent.
The success of affirmative marketing depends on whether or not the neighborhood is adjacent to a black-majority neighborhood. If it is not, the black share of
population by the end ofthe decade never exceeds 30 percent, so the po!icy achieves
its objective of a racia!ly balanced neighborhood. In contrast, the policy is ineffective in neighborhoods that are adjacent to black-majority neighborhoods; if such a
neighborhood starts the decade with a black share of at least 12 percent, it will have
a black majority by the end of the decade.
r-
347
50
46
"a
.e
":""0"'
30
22
01
:.,
20
"
10
o
.e
Mild
preferences
Averag
preferences
Average
preferences
Not adjacent to
black majority
Strong
preferences
Adjacent to black
majority
Source: Figure from "Neighborhood Racial Change, Segregationist Sentiments, and Affirmative
Marketing Policies," by George C. Galster in Joumal of Urban Economics 27, pp. 344-61,@ 1990
Elsevier Science (USA), reproduced by permission of the publisher.
Figure 13-11 sum.marizes Galster's results for neighborhoods that did not adopt
affumative marketing policies. Por each type of neighborhood, the figure shows
the minimum initial black share that generales a black majority by the end of the
decade. The threshold share is larger for neighborhoods that are distant from blackmajority neighborhoods and have relatively weak white preferences for segregation.
The smallest threshold (22 percent) is for a neighborhood that is adjacent to a black
majority area and has strong white preferences for segregation.
SUl.YlMARY
l. Households sort themselves with respect to the demand for local public goods
and, because local public goods are supported by the property tax, with respect
to the demand for housing. This sorting contributes to sorting with respect to
income because the demands for local public goods and housing increase with
income.
2. Households sort themselves with respect to the willingness to pay for favorable
peer groups in schools. If the income elasticity of the marginal benefits of
peer groups exceeds the income elasticity of demand for land, high-income
households will have better school peer groups.
3. Among the factors contributing to racial segregation are household preferences,
income differences, racial discrimination in real-estate markets, exclusionary
zoning, and housing policies.-
348
Chapter !3
349
Household Sorting
a. Assume that Mr. Wizard's reasoning is correct. What are bis assumptipns?
b. Suppose that the equilibrium number of municipalities tums out to be nine,
not three. Where did Mr. Wizard go wrong? What set of assumptions would
have given him the correct prediction?
7. Metro is a metropolitan area in wbich al! households demand the same spending
per pupil on public education. Education is produced with constan!retums to
scale. Because schools are financed with the property tax, there are dozens
of school districts. According to Peabody, a fiscal analyst in Metro, "If my
assumptions are correct, if we replace the property tax (a tax on market value)
with a tax based on square footage (e.g., $2 per square foot of living space per
year), the equilibrium number of school districts will decrease to one." What
are Peabody's assumptions?
8. Under the traditional public school system, parents must send their children
to the neighborhood school. In recent years, many people have suggested that
parents be able to send their cbildren to the public school of their choice, not necessarily the neighborhood school. Suppose tbis change is implemented. Discuss
the effects on the Tiebout shopping and sorting process. Would there be more
or less income segregation?
9. In the example of the Tiebout model with a property tax, wealthy households
established an exclusive suburb with a property tax rate of 1 percent, spending
on public goods equal to $3000 per household, and a rninimum lot size of
0.50 acres. Discuss the effects of the following changes on the property tax rate
and the rninirnum lot size.
a. The market value of houses is $200,000 (and the market value of land is
$120,000)
b. The market value of land is $100,000 (and the market value of houses is
$300,000).
350
6. Hamilton, Broce W. "Zoning and Property Taxation in a System of Local Governments." Urban Studies 12 (1975), pp. 205-11.
7. Harnilton, Broce W. "Capitalization of Intejurisdictional Differences in Local
Tax Prices." American Economic Review 66 (1976), pp. 743-53.
8. Heikkila, Eric J. "Are Municipa!ities Tieboutian Clubs?" Regional Science and
Urban Economics 26 (1996), pp. 203-26.
9. Kain, John F. "Housing Segregation, Negro Employment, and Metropolitan
Decentralization." Quarterly Joumal of Economics 82 (1968), pp. 175-97.
10. Kain, John, "Black Suburbanization in the Eighties: A New Beginning ora
False Hope?" In American Domestic Priorities. Ed. John Quigley and Daniel
Rubinfeld. Berke!ey and Los Angeles: University of California, 1985.
11. Milis, Edwin S., and Luan Sende Lubue!e. "Inner Cities." Joumal of Economic
Literature 35 (1997), pp. 727-56.
12. Ross, Stephen, and John Ymger. "Sorting and Voting: A Review of the
Literature on Urban Public Finance." Handbook of Regional and Urban
Economics, Volume 3. Edited by Pau! Cheshire and Edwin Milis. Amsterdam:
North Holland, 1999.
13. Tiebout, Charles. "A Pure Theory of Local Expenditures." Joumal of Politica/
Economy64 (1956), pp. 416-24. The original statement ofthe Tiebout model.
14. U.S. Bureau of the Census, Residential Segregation, Summary Tables and Residential Segregation, Detailed Tables (Washington, DC: U.S. Government Printing Of:fice, 1998).
15. Yinger, John. "Evidence on Discrimination in Consumer Markets." Joumal of
Economic Perspectives 12, no. 2 (Spring 1998), pp. 23-40.
CHAPTER 14
urban Poverty
Thls
chapter considers the problem of poverty in urban areas. The poverty rate
inside U.S. metropolitan areas is 11.2 percent, compared to 14.3 percent outside
metropolitan areas. The main problem ofurban poverty is not the high metropolitan
poverty rate, but instead the extremely hlgh poverty rates in central cities. For the
United States as a whole, the poverty rate for central cities is about twice that for
areas outsice central cities. Also, withln central cities, poverty is concentrated in a
relatively small number of neighborhoods.
Sorne recnt research provides a clase look at poverty at the neighborhood
level. Jargowsky [1996] defines a hlgh-poverty neighborhood as a census tract with
a poverty rate of at least 40 percent. In 1990, there were 3,714 hlgh-poverty neighborhoods in the United States (6 percent of the nation's census tracts), and about
five-sixths ofthese were in metropolitan areas. Between 1970.and 1990, the number
of hlgh-poverty census tracts in metropolitan areas more than doubled, along with
the number ofpeople living in them. As shown in Figure 14-1, between 1970 and
1990 the number of hlgh-poverty census tracts in Milwaukee increased from 6 to
50, and the poverty area spread outward. In 1990, several dozen additional census
tracts were on the borderline, with poverty rates just below 40 percent.
Figure 14-2 shows the poverty rates for 95 neighborhoods that were studied in
the Cleveland metropolitan area. Each neighborhood is a small set of census tracts
that local policymakers consider a neighborhood unit. For the metropolitan area,
19 of the neighborhoods, housing about 15 percent of the area's population, have
poverty rates exceeding 30 percent. For the central city, 19 of the 36 neighborhoods have poverty rates exceeding 30 percent, and 42 percent of the central city's
population lives in these poor neighborhoods.
351
352
O Never O 1990
Source: Paul A. Jargowsky, PoYerty andPlace: Gherros, Barrios. and che American City@ 1997 RusseU
:'
70
60
/
50
"
E
40
5
o
p.
30
20
10
64
28
73
Rank of neighborhood
":!1'
Source: Center for Urban Poverty and Social Change, Case Western Reserve Universily. hUp://povertycenler.cwru.edu
82
91
T
.
354
One fact about poverty that is often neglected in policy debates is that over half
of poor households have someone who works at least part-time, and one in five has
a full-time, year-round worker. A four-person household with a full-time worker is
classified as poor if the worker's hourly wage is less than $8.51 per hour. In other
words, a full-time job may not be enough to escape poverty. Either the wage must be
..
:.:
-"'--"---
40 -
...
35 -
30.4
.--
288
;..:..:..
,,
23.6 22.8
r-r-
25 t-
20
r:-
30 L1
nA
..
'
..'
22A
i..
..
r---
1-
16.5
.--
15 -10
t-
r-
11.8
.--
.,<"
.Y<$,
"
o'>
16.4
-
143
.,
9.8
4>.""
11.2
-
'
1
1
.
1
1
-<$-'"''{-
1:?
,},
(,<.
o< :'
0"'
:4-
'b
fb-v
'l.J<.""\-
<_/
<;,<>o"'
.,<'
.fi'
t.
.fi'
i.'
v
""'
.fi'
, #>
vO
i.'
"
J>
;,
- 'lJ
.,:f>
:.'
lo
9.2
,-
5.8
r:-
;,
'6
eJ
'Qo
'b
<>"''t!!
. 1
4:>"<$, ,,:p'>
< ;
<'
..
.;
.:-
..
1' 1
'"
4:>"<1,
83
ru
'1>(}'
;::._ fb.v
15,
J'"
;;,"
:4-
4:>"<$,
!>
.p
roo/
"
<o
$;.ro'fJ
'f:>"r
<"'
i:,
rf>'"iJ
o;:y
15
<
.;, ..
<ovo
o<S'
0
,:,o<i.
"
,><, '1>"e;
-$-"
..
o" O"
o"' <$'
,:,o<i.
"
o"
:
.'l.Js
>:'
. :::.,V"'
"'..$"
. '1>"""
""
o"'
(H
Source: U.S. Bureau of the Census, Poverty in tlle United States, 1999 (Washinglon, D.C., 2000).
..._;'
'<''
"."
r,.,<S'
"
.," '
;:
.,.,'1>"
.," '
;:
tJ
iti'
'<$'
,<5"
. :<>
,<5"
"i)
a::o
()
2.8
n
1
'-"
?:::>:>tt;
'li'
356
quite a bit higher than the minimum wage or the household must have more than one
worker. In fact, many households escape poverty by having two low-wage eamers.
357
thus affordable), but slow. Because low-income households have low opportunity
costs for travel time, they tolerate slowness in exchange for low monetary cost of
transportation. In contrast, automobiles are expensive but fast, making them relatively attractive for high-income households. Given their greater reliance on public
transportation, low-income households are attracted to the city center, where high
population density ma.ke it possible to support efficient public transportation.
There is also evidence that low-income households are attracted to central cities
because they have more generous redistribution policies. Compared toa low-income
householdin the suburbs, alow-incomehouseholdin the centralcity is 9.7 percentage
points more likely to live in subsidized housing and 23 percent more likely to receive
substantial goverilment income transfers (G!aeser, Kahn, and Rappaport, 2000).
Across metropolitan areas, the greatest central-city concentrations of low-income
households occur where central-city redistribution programs are the most generous
relative to the programs of suburban governments.
We've seen that two sorts of public policies-public transportation and redistribution programs-attract Jow-income households to central cities. An alternative
theory is that Jow-income households-in particular black households-are forced
to live in the central city, far from suburban employment opportunities. This spatial
rnismatch between residences andjob sites contributes to central-city poverty.
358
21%
23%
20%
Black>20%
10%
E
-a
j
o
;;
4%
0% q.-- -----L-- -L--1%
-10%
Entire CMSA
Black < 20%
-20%
Black>40%
1%
1! -30%
5" -40%
-50%
-60%
-32%
O Total
Manufacturing
Manufacturing, rransportation,
cornmunication, utilities,
construction, public adrninistration
-61%
-70%
Source: Author's calculations based on Steveo Rafael. "The Spatial Mismatch Hypothesis and Black Youth
Joblessness: Evidence frorn the San Francisco Bay Area." Joumal of (Jrban Economics 43 (1998), pp. 79-111.
'
t
._
Cbapter 14
359
Urban Poverty
T
.
360
'
361
.g 0.90 r-
0.88
..
"S,
0.80 r:
'(; 0.70 r-
r-
0.60
"
"C
.9 0.50 r-
..
0.49
"0.40
0.29
.e
0.30 -
"'
&l 0.!0 :"@ 0.20 ;;;
Dropping out of
higb scbool
ldteness
Smgle
motherhood
Source: Author'salculati.ons based on reSulrs ih David M. Cutler ru:td Edward L. Glaeser: "Are Ghettos Good or
Bad?" QuanerlyJoumal of Economics (1997), pp. 827-72.
T
.
362
single motherhood is 0.29, meaning that the probability of being a single mother
increases by 2.9 percent for every 10 percent.increase in segregation. These elasticities suggest that segregation has powerful effects on the economic prospects of
black households, leading to lower employment rates and higher rates of single
parenthood.
Why does segregation generate these adverse outcomes? One reason is that
young people living in a highly segregated, low-income environment have fewer
contacts with positive role models-educated and successful people-and are less
likely to be successful themselves (Cutler and Glaeser, 1997). The lack of contact
with positive role models increases the frequency of adverse outcomes: the less
frequent the contact with educated adults, the higher the high school dropout rate
and the incidence of single motherhood, and the lower the share of black youths
attending and completing college.
Racial discrimination in the real estate market. This barrier could be lowered
by more rigorous enforcement of the nation's fair-housing laws.
Exclusionary zonin:g. This barrier could be lowered by encouraging local governments to eliffiinate practices that discriminate againsthousing for low-income
households.
Public housing policies oriented toward the central city. This barrier could
be eliminated by making all housing assistance portable.
The most recent wave of welfare reform, which will force most welfare recipients
into prvate employrnent, would increase the payoff from eliminating these barriers.
There is an old debate among urban polioymakers concerning the wisdom of
encouraging the suburbanization of central-city black households. One concern is
that only the most educated and skilled blacks' wil1 move to the suburbs, and their
departure will worsen conditions for those who stay. To address this concern, policies
should be designed to give all central-city residents the option of moving closer
to suburban employment centers. The Cutler and Glaeser study of the effects of
segregation (1997) suggests that the poorest and least educated blacks are harmed
the most by segregation, and thus have the most to gain from moving into a less
segregated envirorunent.
One alternative to a dispersa! policy is a development policy in which firms are
encouraged to locate in the central.city. Under the black-capitalism approach, the
government encourages the development ofblack businesses. The experiences with
black capitalism have not been encouraging (Bates and Bradford, 1976). Another
'
363
approach is enterprise 'zones, areas of the city where firrns (1) pay lower taxes,
(2) receive subsidies for worker training, and (3) are exempted from many local
regulations. The limited experience with enterprise zones suggests that they are not
very effective in luring firrns to central cities (Boarnet and Bogart, 1996; Papke,
1993 and 1994; Dowall, 1996).
There are two basic problems with the development approach. First, for many
types of activities the central city is not a profitable location. The suburbanization
of jobs was caused by fundamental changes in the economy, for example, the development of the automobile and the truck, changes in production technology, and
innovations in information technology. Given the advantages of a suburban location,
a firm in the central city may be unable to compete with its suburban counterparts.
Second, poor households in central cities may not have the skills required for jobs
available in the central city. As a result, policies that attempt to foster development
of central cities are unlikely to reduce poverty.
Theories of Discrimination
The traditional theory of discrirnination suggests that market forces limit the extent
of wage discrirnination (Becker, 1957). Considera city with two types of workers,
black and white, who have the same marginal revenue product (MRP), defined as the
marginal physical product of labor times the market price of the product produced.
, Suppose that white workers are paid their MRP ($6) but blacks are paid only $5.
If you were a profit-maxirnizing firm, what would you do? If you hire black workers,
you will have lower production costs than your all-white competitors so your profits
will increase. What's good for you is good for other capitalists, so other firms will
hire the undervalued black workers. As the demand for black workers increases and
the demand for white workers decreases, the gap between black and white wages
narrows.In equilibrium, whites and blacks have the same wage. Sirnilarly, we would
expect gender discrirnination (lower wages for women) to disappear.
Under what conditions would wage discrirnination persist? If a white firm can
keep other firms from entering its industry, it can protect itself from the competition
of color-blind firms. However, bigotry js not costless. The white monopolist pays
relativelylllgh wages to its white workers, and its higher production costs force its
T
.
364
'
consumersto pay relatively high prices.1n addition, profits are lower, so stockholders
pay for bigotry in the fotrn of lower dividends and stock prices. Job discrirnination
may persist if (1) bigoted firms are protected fromcompetition and (2) consumers
and stockholders are willing to pay for their bigotry. Similarly, gender discrimination
may persist.
The theory of statistical discrimination suggests that the black-white wage differential will persist if there is imperfect inforrnation about worker productivity
(Arrow, 1973, and Phelps, 1972). Suppose that an employer must hire one of two
workers, a black or a white. If the firm knew which worker had the higher MRP, it
would obviously hire the more productive worker. Such inforrnation is not general!y
available, however, so the employer must guess which worker has the higher MRP.
Suppose that because of differences in education and work experience, the average black is less productive than the average white, although sorne blacks are more
productive than sorne whites. If the employer has no other inforrnation about the
productivity of the applicants, the best choice, in a statistical sense, is the white
worker. White workers may be chosen over equally productive (or more productive)
black workers because black workers arless productive on average. In this case,
discrimination occurs because of imperfect inforrnation, not bigotry.
All
All workers
Al! males
All females
White workers
Whitemales
Wbi te females
B1ack workers .
Black males
B1ack females
Hispanic workers
Hispanic males
Hispanic females
B1ack maJe as percent
of white maJe
B1ack female as percent
of white female
Sorne College
Time, No
Bache!or's
Degree
Degree
Education
Levels
Nota
High School
Graduate
High School
Graduate Only
$26,792
33,251
19,414
27,556
34,276
19,647
20,537
23,876
17,485
18,262
20,312
15,310
$14,013
16,748
9,790
14,234
17,023
9,582
12,956
14,877
10,739
13,068
14,774
9,809
$21,431
. 26,333
15,970
22,154
27,467
16,696
17,072
19,514
14,473
. 18,333
20,882
14,989
$22,392
28,458
16,152
22,898
29,206
16,125
20,275
24,894
16,627
18,903
21,705
15,699
$36,980
46,1ll
26,841
37,711
47,016
26,916
29,666
36,026
25,577
30,602
35,109
25,338
70%
87%
71%
85%
77%
89%
112%
87%
103%
95%
Chapter 14
Urban Poverty
365
white males, and black females earn 89 percent as muchas white females. The racial
gaps are much smaller forcollege graduates: black male graduates earn 77 percent
as much as white graduates, and black female graduates earn 95 percent as muchas
white graduates.
These earnings gaps result from severa!factors, including differences in number
of hours worked, age, work experience, location, quality of education, and racial
discrimination. The questions are: How much of the earning gap is caused by racial
discrimination, and how much is caused by other factors?
There is sorne disagreement among economists about the relative importance of
discrimination. Based onrecent studies of discrimination, Schiller (1995) concludes
that only about a quarter of the earnings differences are caused by discriminatory
practices. Darity and Mason (1998) suggest that black men experience a loss in earnings of 12 percent to 15 percent as a result of labor-market discrimination. Heckman
(1998) notes that the black-white earnings difference has decreased substantially
over the last severa! decades, and is now small enough that
most of the disparity in eamings between blacks and whites in the labor market of the
!990s is due 10 the differences in skills they bring 10 the market, and notto discrimination
within the labor market.
Although there is consensus among economists that the discrimination penalty has
been decreasing over time, there is sorne disagreement about the actual size of the
penalty.
WELFARE REFORM
In 1996, theoverhaul offederal antipoverty programs ended decades ofpolicy based
on the notion tJ:tat poor farnilies are entitled to cash and in-kind assistance. The Personal Responsibility and Work Opportunity Reconciliation Act of 1996 eliminated
the entitlement of poor farnilies to receive cash assistance. The federal government
now provides block grants to states to provide Temporary Aid to Needy Families
(TANF), with restrict:lons on what recipients must do to qualify for assistance and
how long they can receive it.
After a maximum of 24 months of assistance, a recipient must participate in
work activities, defined as employment, on-the-job training, work experience,
community service, vocational training, or child care services for individuals
providing community service. Single parents must participate at least 30 hours
per week, and members oftwo-parent families must participate at least 35 hours
per week. There are exceptions to this rule for the parents ofvery young children.
After a total of 60 months of cash assistance (consecutive or nonconsecutive),
assistance is stopped. States can allow exceptions to this rule, with the restriction
thatno more than 20 percent ofthe state's recipients exceed the 60-month limit.
The federal government also provides incentives for states to reduce welfare
. dependence. The federal goveniment will cut the state's block grant ifthe state fails
366
to meet ambitious targets for (1) increasing the fraction ofwelfare recipients who
work or (2) reducing the number of people on welfare. Another incentive is $1 billion
available for performance bonuses to reward states for moving welfare recipients
into jobs. The general thrust of this welfare-reform package is to change the welfare
system into one that requires work in exchange for time-lirnited assistance. There
are also block grants for child care subsidies to help working parents.
Although welfare reform reduced welfare case!oads, so far it has had rnixed
effects on income and poverty rates. Por the average former recipient, the net effect
of higher earnings and lower welfare payments is a net increase in income. But
the number of farnilies in deep poverty (income less than half the poverty leve])
has increased, in part because about two-fifths bf former welfare recipients are not
working (Sawhill, 2001). Among former recipients who are working the average
wage is only $7.00 per hour, a wage that is not high enough to lift a single-earner
farnily above the poverty line. Although the poverty rate for children has decreased,
the incomes of the poorest fifth of the nation's population have continued to falL
In 1999, about 700,000 farnilies were significantly worse off in 1999 than their
counterparts in 1995.
Looking to the future, three questions remain. First, if the economy stops growing, how would a recession affect caseloads and poverty rates? Second, will former
welfare recipients acquire human" capital in their new jobs, allowing them to eam
higher wages? If so, the early gains in ei:nployment will translate into larger income
" 1
367
gains for the fonner recipients. Third, how will welfare refonn a.ffect the structure
of families? One policy objective was to decrease the number of children in singleparent households. So far, teen pregnancy rates have dropped, but the incidence of
single-parent families has not changed. Finaly!, how can welfare policy be adapted
to address the problem of concentrated urban poverty?
Market demand
Initial supply
Supply after
welfare reform
...
7.00
"
"'
6.50
230 270
Quantity oflabor (1,000 hi:mrs)
We!fare reform will increase the nuunber of low-skilled workers in the market, shifting the
supply <:urve te the right and decreasing the wages .of Jow,skilled workers.
368
For the least educated and uilskilled workers (high school dropouts and the least productive high school graduates), wages could drop by about 6 percent (Holzer, 1996).
Women from female-headed households who worked befare welfare reform could
see their earnings drop by about 8 percent (Bartik, 1998). The largest losses in earnings will be experienced by low-skilled women, because they will face additional
competition as many former AFDC recipients join the workforce.
SUMMARY
l. Poverty rates are highest among racial rninorities, children, female-headed
households, high school dropouts, and central-city residents.
2. Sorting with respect to income explains why poverty is concentrated in certain
neighborhoods, but not why the poor neighborhoods are in the central city.
3. One reason for central-city poverty is that low-income households are attracted
to public transportation systems and redistribution programs.
4. A second reason for central-city poverty is the spatial rnismatch: A disproportionate number of black workerslive in the central city, far from suburban
employment opportunities.
5. Sorting with respect to income makes poverty worse because low-income communities spend less on housing and get less education per dallar spent.
6. Segregation makes poverty worse: Black households in more segregated cities
earn less income and are more likely to be idle, have lower high school graduation
rates, and young women are more likely to become single mothers.
7. Discrimination in labor markets leads to lower wages for racial minorities.
8. The general thrust 0f the recently adopted welfare-reform package is to change
the welfare sys'tem into one that requires work in exchange for time-limited
assistance.
'
Chap<er 14
369
Urban Poverty
3. Consider a segregated city in which all blacks live in the central city. In the
suburb (eight miles from the central city), manufacturing firms in industry X
employ black workers. The unit cost of commuting (including monetary and
time costs) is 40 per rnile and blacks in the central city earn $80 per day. The
price elasticity of demand for X is -2.0. Firms produce X with fixed factor
proportions, and labor is responsible for 80 percent of production costs.
a. Compute the daily earnings of a black suburban worker.
b. Compute the effects of racial segregation on black suburban ernployment:
Predict the percentage increase in total black employment that would result
from a decrease in the black commute distance from eight miles to zero.
4. The section on job discrimination suggests that the market penalizes racist and
sexist firms because such firms are underpriced by color-blind and sex-blind
firms. As a result, there arelirnits on the extent of wage discrirnination.
a. Whatif all firms (capitalists and managers) are color-blind, but white laborers
refuse to work with black laborers? Will wage discrirnination (lower wages
for blacks) persist?
b. What if firms and employees are color-blind, but white consumers refuse to
purchase goods produced by blacks? Who bears the cost of bigotry? Will
wage discrirnination persist?
5. In the state of Florida the implementation of a welfare-reform package that
forced many welfare recipients to work occurred at the same time as an increase
in the rninimum wage from $4.25 to $4.75. Depict graphically the rnarket effects
of these two changes on the market for low-skilled labor. To sirnplify matters,
assume that the old rninimum wage was actual]y the rnarket-equilibrium wage.
.
370
Part 4
Urban Problems
371
24. Papke, Leslie. "What Do We Know about Enterprise Zones?" In Tax Policy and
the Economy, ed. James Poterba. Cambridge, MA: MIT Press, 1993.
25. Phelps, Eclmund. "The Statistical Theory of Racism and Sexism." American
Economic Review 62 (1972), pp. 659-61.
26. Rafael, Steven. "The Spatial Mismatch Hypothesis andBlack YouthJoblessness:
Evidence from the San Francisco Bay Area." Journa/ of Urban Economics 43
(1998), pp. 79-111.
27. Sawhill, Isabel. "From Welfare to Work: Making Welfare a Way Station, Nota
Way ofLife," Brookings Review, Summer 2001, pp. 4-7.
28. Schiller, Bradley, The Economics of Poverty and Discrimination. 6th ed.
Englewood Cliffs, NJ: Prentice-Hall, 1995.
29. U.S. Bureau ofthe Census, CurrentPopulationReports, Series P60-201. Poverty
in the United States, 1999. Washington, DC: U.S. Govemment Printing Office,
2000.
iili ;..
\ :i}
,,; ..
:'t.:
?J
1
).:, .
' .
'1' ::.
.1'
J:.
r_
:!)
'J :
., .
Tl;:
'
CHAPTER 15
The Urban
Education Problem
Human history becomes more and more a race between
education and catastrophe.
-H.G.WELLS
bis chapter explores the causes and onsequences of low educational achievement in central cities. We know from earlier chapters that education is a key factor in
deterrriining a person's employment prospects and earnings potential. Soit is not surprising that central-city neighborhoods with low levels of educational acbievement
also have bigh poverty rates and bigh crime rates. The urban education problem is
that students in many central-city neighborhoods are poorly educated, contributing
to the problems of poverty and crime.
We can use Figure 1-5-1 to compare educational achievement in central cities to
acbievement outside'central cities. The figure shows the percentage of eighth graders
who scored at or below the basic leve!on the NAEP math exam. The "urban" school
districts are ones for wbich at least 75 percent of students live in central cities. Por
al! but three states, a larger percentage of urban students score below the basic leve!,
with substantial differences in many states. Por examp!e, in California 59 percent
of urban students. scored below the basic leve!, compared to 44 percent of nonurban
students. The percentages are 61 percent and 25 percent for New York, 53 percent
and 24 percent for Massachusetts, and 91 percent and 37 percent for Maryland.
Figure 152 shows the differences in educaiional achievement witbin a particular
central city, New York. The graph shows, for each of the city's 312 schools, the
percentage of students in the lowest performance leve!. Starting with the bighest
ranked schools shown on the left side of the graph, for the top 128 schools fewer
than 20 percent of students perform at the lowest leve!. At the other end, for the
bottom 15 schools, more than 80 percent of students are at the lowest leve!. Por the
32 lowest ranked schools, over half of students perform at the lowest leve!.
One measure of the educational attainment of a community is the share of the
population that has graduated from high school. In Figure 15-3, the horizontal axis
measures the percentage of the adult population that did not complete bigh school for
372
)
'-
._;
'
- ...:,.,
.S
.1
.e:
.;:
o
00
373
374
""
100
,----
"e
"E
80
.g
""
;;;
" 60
"="
"O
40
.2
.o...
Ci
"
20
150
School raok
90
80
70
60
""2
::
.2:>
"
50
40
30
20
10
..,..
o
10
"+:
20
30
40
50
60
70
80
375
376
Consider first the inputs controlled directly by the school. Under the supervision of the local school district, the school designs a curriculum, purchases instructional equipment (e.g., building, books, science labs, computers) and hires teachers.
Teachers vary in educational background, experience, and quality. The school also
controls class size, and can increase the teacher input per student by reducing class
size.
Consider next the role of the student's home environment. Educational achievement is infl.uenced by the home environment ofthe child in three ways. First, parents
set the rules of the household, establishing an environment that is either favorable or
unfavorable to educationPor example, an unfavorable environment is one in which
children watch television instead of reading books or doing their homework. Second,
parents can motivate their children by encouraging reading, helping with homework,
and rewarding success. Third, parents can provide instructional materials such as
books and home computers, encouraging independent learning.The quality of the
home environment depends in part on the income and education leve!of the parents.
The children of wealthy and educated parents learn more because they receive more
encouragement and assistance at home, and also pick up verbal and quantitative
skills in everyday interactions with their parents. In contrast, children from poor
and less educated farnilies learn less because there is often less encouragement to
achieve, and less learning from parents. In addition, children living in poverty are
often inalnourished, inhibiting their ability to learn.
Considernext the role ofthe student's peer group. A childlearns more ifhe or she
is surrounded by smart and motivated children. Smart peers promote achievement
because of cooperation (children learn from one another) itnd competition (children compete with one another). Motivated peers promote achievement because the
teacher can spend .Iess -time disciplining and motivating students, and more time
teaching them. In addition, an unmotivated student provides an undesirable role
model for other students.
Empirical Results
Dozens of studies have explored the relationship between educational inputs and
output. The key question is, Which inputs are most productive in increasing educational achievement? There is widespread agreement that the most important factors
are the home environment and the peer group. In general, the higher the income and
education leve!of the parents, the betterthe home ei:J.vironrnent and the higher the
child's educationalacliievement.
What abm.it the effect of peer groups? The same parents who provide good home
environments provide good peers as their child's classmates. As a result, educational
achievement in a school is positively related to the average income and education
leve!of the farnilies patronizing the school. There is sorne evidence that the largest
peer group effects are experienced by low achievers: the students at the bottom ofthe
class have the most to gain from adding smart and motivated students to the class.
There is also evidence that these peer effects are most important in the middle and
upper grades (grades 5 through 12).
f -
377
There is also agreement that teachers differ in their ability to teach students. In
other words, teachers are not like b!ackboards, but vary in effectiveness. The most
effective teachers are the ones that are smart, motivated, innovative, and flexible. A
study ofinner-cityschools found that during a single acadernic year, a student with
a high-quality teacher outperforms a child with a low-quality teacher by up to one
full grade leve!(Hanushek, 1992). For exarnple, consider a child who starts third
grade at grade leve!2.0, right on schedule. With a high-quality third-grade teacher,
the child could move up to grade leve!3.2 by the end of.the year and be ahead of
schedule; with a low-quality teacher, the child could move up slightly, to grade leve!
2.2, and not be ready for fourth grade. More recentiy, Rivkin, Hanushek, and Kain
(1998) provide evidence of substantial variation in teacher productivity, with the
largest difference arnong math teachers.
Although teachers differ in effectiveness, the list of characteristics that makes a
productive teacher has eluded researchers. The process of teaching requires subtle
skills that cannot be easily measued, so it is difficult to predict which teachers will
be the most productive. In looking for teacher characteristics that explain productivity differences, researchers have focused on education leve!(years of postgraduate
study), comrnunication skills (verbal ability), and experience (years of teaching).
The results of the studies are as follows:
378
schools in South Africa, Case and Deato (1999) show that-smaller classes lead to
higher achievement. In a study ofhigh school students in lllinois, smaller classes were
associated with higher graduation rates anda larger fraction ofgraduates attendin b
college (Sander, 1993). Rouse (1998) shows that schools in low-income areas in
Milwaukee that have relatively small classes have higher achievement levels in math
and reading.
Do schools differ in their production of education? Rivkin, Hanushek, and Kain
(1998) show that there are large differences among schools in their effects on student
achievement. The difference are not related to leadership, organization, or financia!
conditions. Instead, the differences result from the irnpact of teachers. The most
effective schoo1s are the ones with the best set of teachers, not the best organization
or instructiona1 equipment.
Alternative Approach: School Resources and Market Earnings
The traditional studies of the education production function use test scores as a
measure ofthe output of education. An alternative approach 1ooks at the relationship
between schooling and future earnings, using the wages earned by adults as a measure
of the output of their earlier education. The production question is: Does the wage
earned by an adult depend on the quantity ofresources used in her K-12 education?
Por example, consider two adults who differ only in the class sizes they had during
grade school.If school resources matter, we would expect the adult who had smaller
classes to have a higher skillleve1 and thus earn a higher wage.
Card and Krueger (1996) summarize the results of severa! studies that examine the relationship between schoo1 resources and earnings. There is evidence that
school resources nia.tter: students who were educated in schools with relative1y high
spending levels earn higher wages. The estimated elasticity of wages with respect to
per pupil spending is about 0.15, meaning that a 10 percent difference in spending
generates a 1.5 percent difference in wages. The elasticity of wages with respect to
c1ass size is about 0.07, meaning that a 10 percent difference in class size generates a
0.7 percent difference in wages. As Card and Krueger note, sorne studies suggest that
the relationship between educational spending and wages is weak or nonexistent, so
the issue has not been resolved.
Private versus Public Schools
We can gain sorne insights into the education production function by comparing
public schools to private schoo1s. In general, educational achievement is higher in
private schools. Why?
According to Murnane (1986) differences in peer groups explain half the difference between achievement leve1s of public and private schools. To quantify the
peer effect, he poses two questions
How would the test score of the typical public-school student change
student transferredfrom the publicschool toa prvate school?
if the
379
The answer is that the student's score would increase from 24.4 to 26.1. By transferring to a private school, the typical student would experience a 7 percent increase
in achievement.
How would the test score of the typical public-school student change if the
student, along with his public-school classmates, transferred from the public
school toa private school?
The answer is that the student's score would increase from 24.4 to 25.1, an increase
of about 2.9 percent. If the student brings his public-school peers along with hirn to
the private school, he experiences less than halfthe achievement gain that occurs ifhe
leaves his public-school peers behind. The other half of the achievement difference
is presumably caused by differences in curriculum, teachers, educational rnethods,
and discipline. An importan!advantage of private schools is that they have more
fiexibility in expelling troublernakers.
More recent studies suggest that the differences in achievement between public
and private schools are srnall, at least as rneasured by test scores. Goldhaber (1996)
finds no difference in achievement between comparable students in public and private
schools, and Gamoran (1996)" finds a very srnall advantage for students in private
schools. Rouse (1996) found a small private-school advantage in mathernatics, but
no difference in reading.
Although the differences in achievernent between public and private schools
appear to be srnall, the differences in educational attainment rnay be large. Achievernent is usually measured by test scores in a particular grade, while attainrnent is
measured by the years of schooling. The key measures of attainrnent are high school
graduation rates, college attendance rates, and college graduation rates. Neal (1997)
found that urban minority students attending private (Catholic) high schools have
rntich higher graduation rates than comparable students in public high schools. In
addition, pnvate-school students are more likely to get a college degree. These higher
attainrnent rates translate into higher earnings after high school; on average, students
who attend private high schools earn 8 percent more than comparable students who
attend public schools (Neal, 1997).
380
. 7,000
Lowest-spending clistricts
O Highest-spending clistricts
6,000
S. 5,000
g_
..
:6
4,000
e:
3,000
>
<
Kentucky
New Jersey
Source: Jeffrey KaiZ, "The Search for Equity in School Funding,"' Goveming, August 1991, pp. 20-22.
as rnuch as the low-spending districts. In the 1980s, spending ratios above 1.5 were
comrnon in other states (Katz, 1991).
In other words, children from poor school districts must receive the same quality of
education as children from wealthy school districts. The courts also directed state
governrnents to equalize spending on primary and secondary education.
The rnost frequently cited court case is Serrano v. Priest, decided in 1974 by the
California Supreme Court. When John Serrano complained about the low quality
of the local high school's education program, the school's principal suggested that
Serrano move his farnily to a school district where per pupil spending was higher.
Instead of moving, Serrano sued the state of California, arguing that spending inequalities were unconstitutional:
381
Baldwin Park
Beverly Hills
$4,169
3.34%
$272
$49,501
2.55%
$1,535
Source: James Guthrie, "United S tates School Finance Policy 1955-1980." In Schobl Fin..ance
Policies aru! Practices, ed. James Guthrie (Cambridge, MA: Ballinger, 1980). p. 14.
To illustrate the spending disparities, Serrano cited the differences between two
schooldistricts in the Los Angeles area, Beverly Hills andBaldwin Park. As shown in
Table 15-1, Beverly Hills used its larger property tax base to spend more per student
while charging a lower tax rate.In ruling in favor of Serrano, the court ruled that the
property tax system violated the equal-protection clause of the state's constitution.
The court ordered the state legislature to develop a financing system under which
per pupil spendfug would vary by no more than $100.
Limits on Reform
Two court cases established the limits on education reforrn. In Mclnnis v. Ogilvie,
the plaintiffs argued that the Constitution guarantees the right to equal educational
outcomes rather than equal educational spending. They argued that government
should eliminate inequalities in educational achievement, spending more on lowachieving students to bring them up to the leve] of high achievers. The court ruled
against the plaintiffs, arguing that there was no practica! way to enforce an equalachievemet standard. Given this decision, the objective of reforrn efforts has been
to equalize spending, not achievement.
Another case, SanAntonio lndependent SchoolDistrictv. Rodriguez, established
the limits for federal involvement in school finance. The court ruled that education
is not a fundamental right guaranteed to U.S. citizens. This ruling was based on
the fact that education is not mentioned in the U.S. Constitution. Consequently,
the equal-protection clause of the Constitution does not apply to education, and
variation in per pupil spending is not proscribed by the Constitution. There are two
implications from the San Antonio decision. First, any reforrn of the school finance
system must come from state governments, not the federal government. Second,
although the court system may promote equalization of spending within individual
. states, the courts will do nothing to promote equalization of spending across states.
Foundation Grants
Most states use foundation grants to help narrow the gaps in education spending
across school districts.The foundation grant per pupil is deterrnined by the following
formula:
Grant = Foundation leve!- Foundation tax rate Local property value per pupil
382
T
'
Grant
Education Spending
$4,000
5,000
$2,000
2,000
$6,000
7,000
Suppose a state picks a foundation leve! of $5,000 and a foundation tax rate of
1.5 percent (0.015). For a school district with $200,000 of local property value per
pupil, the foundation grant is $2,000:
Grant = $5,000- 0.015 $200,000 = $2,000
In effect, the foundation grant equals the difference between the foundation leve!and
the amount of money the school district could raise locally if it were to impose the
foundation tax rate. A district with a lower tax base will receive a larger grant; for example, a district with only $100,000 of property value per pupil would receive a grant
of $3,500. Note that the foundation grant is independent ofthe school district's actual
tax rate; it depends only on the local tax base and the state's foundation variables.
We can use a simple exarnple to illustrate how a foundation plan affects the tax
and spending options of a local school district. Table 15-2 shows spending options
with different tax rates. The district has $200,000 of property per student, so perpupil
spending increases by $2,000 for every percentage point of the property tax. One
option is to pick a tax rate of 2.0 percent, which generales $4,000 of local revenue
per pupil. If the foundation grant is $2,000, the district has a total of $6,000 per pupil.
If the district picks a tax rate of2.5 percent, it would collect an additional $1,000 in
local revenue; because the grant doesn't depend on the district's tax rate, education
spending increases by onJy $1,000.
How does a foundation grant affect spending in the school district? As explained
earlier in the book, a loe;al school district is likely to pick the preferred budget of the
median voter. Figure 15-5 shows the budget lines and indifference curves ofMarian,
the median voter. The horizontal axis measures per pupil education spending and
the vertical axis measures spending on other goods. The budget line B C shows the
initial trade-offbetween education and other goods:Marian has a total of $10,000 per
pupil to spend on education and other goods and every dollar spent on education
decreases spending on other goods by one dollar. Before the foundation grant, Marian
maxirnizes utility at point i (Education spending = $4,000), and this is the point
chosen by the local school district.
The foundation grant shifts Marian's budget line outward from BC to BDF.
Point D is in the new budget set because Marian can use the $2,000 grant to get
$2,000 worth of education while spending all ofher own money ($10,000) on other
goods. The points along the line connecting D and F are in the new budget set;
starting from point D, Marian faces a dollar-for-dollar trade-offbetween education
and other goods. The shift of Marian's budget line moves the. utility-maxirnizing
point from i to f. In other words, the grant increases education spending by $400
(to $4,400) and spending on other goods by $1,600 (to $7,600).
'
383
7,600
u2
e
2,000
4,000 4,400
Spending on education ($ per pupil)
F
$
A $2,009 foundation grant shifts the budget line of the median vorer from BC to B D F and the
utility-matimizing point moves from i to f. One-fifth ($400) ofthe granr is spenr on education,
and the remaining four-fiftbs is spent on the other goods. The school district cuts the tax rate from
2.0 percenr (raising $4,000 in local revenue) to 1.2 percent (raising $2,400), allowing citizens to
spend part of the grant on other goods.
How can the foupdation grant increase spending on other goods? The district
can fund its $4,400 education prograrn with the $2,000 grant and only $2,400 oflocal
tax revenue. The grant decreases the local contribution to education from $4,000 to
$2,400, so the school district cuts local taxes by $1,600 and allow its citizens to
spend that much more on other goods. From the perspective of local citizens, the
foundation grant is equivalent to an increase in consumer income, which increases
the demand for al1 "normal" goods, including education and other goods. By cutting
taxes, the school district is simply responding to the demands of its citizens.
In principie, foundation grants can be negative for sorne wealthy school districts,
meaning that a school district would transfer money to the state government. In our
exarnple, if a wealthy district has a tax base per pupil of $400,000, the foundation
grant would be negative:
Grant
384
T
'l
'
Under apure foundation system, the district.would transfer $1,000 to the state. The
transfer would shift the budget line of the median voter to the left, decreasing the
desired spending on education and other goods.
How does a foundation plan affect spending inequalities? The largest grants are
given to the school districts with the lowest property values, so the greatest stimulus
for education will occur in low-wealth districts. If the foundation plan is pure in the
sense that high-wealth districts are taxed, education spending in these districts will
actually decrease, and the spending gaps will narrow even more.
Guaranteed Tax Base or Power Equalization
A more aggressive approach to equalization involves an attempt to equalize the
revenue-generating capacities of school districts. Under a guaranteed tax base
(GTB) plan (a.k.a. district power equalizing), a state specifies a "guaranteed" tax
base per pupil, meaning that each school district has access to the same effective tax
base. The grant per pupil is detennined by the following formula:
GTB Grant =Local tax rate (Guaruteed tax base per pupil
-Local tax base per pupil)
The. tax base per pupil is the same as the property value per pupil. Once the state
picks the guaranteed tax base, it's up to each school district to pick a tax rate, which
then determines how much tax is collected locally and how big the grant is.
We can use a simple example to illustrate how a GTB plan works. Suppose
the guaranteed tax base is $300,000 and the local tax base is $200,000, so the gap
between the two tax bases is $100,000. Table 15-3 shows the implications of two
different tax rates. With-the lower tax rate (2.0 percent), the district raises $4,000
locally and gets a giant of $2,000, for a total of $6,000. With the higher tax rate
(2.5 percent), local tax revenue is $5,000 and the grant is $2,500, for a total of
$7,500. Because a tax-rate hike increases the grant as well as local tax revenue, the
local cost of getting an additional $1,500 in school spending is only $1,000.
The GTB plan decreases the opportunity cost of spending on education. Before
the grant program, there was a dollar-for-dollar trade-offbetween spending on education and other goods; every dollar spent on education required one dollar of local
taxes, so every dollar of education decreased spending on other goods by one dollar.
Under the GTB plan, the trade-off is lower.ln Table 15-3, increasing the tax rate
from 2.0 percent to 2.5 percent ncreases local tax revenue by $1,000 (and decreases
local spending on other goods by the same amount), but increases spending on
education by $1,500. On average, the school district gets a dollar's worth of education
Tax Rate
2.0%
2.5
Grant
Total
$4,000
5,000
$2,000
2,500
$6,000
7,500
..-
385
:::::
8. 6,800
.g
g,
6,000 --------------------------
.,S
lnitial budget
Une
e
\
4,000 4,800
Spending en education ($ per pupil)
A GTB grant tilts the budget line of the median voter, refiecting the decrease in the opportunity cost
of education spending. The. utility-maximizing point moves from i te g. Ha!f ($800) of the grant is
spent on education, and the other ha!f is spent en other goods. The school district cuts the tax rate
from 2.0 percent (raising $4,000 in local revenue) te 1.6 percent (raising $3,200), allowing citizens
te spend part of the grant en other goods.
by sacrificing about 67 cents' worth of other goods. In this exarnple, the GTB plan
is equivalent toa matching grant with a match rate of one-half; an additional $1,000
of local taxes increases the grant by $500.
How does the GTB plan affect the choices ofthe median voter? In Figure 15-6,
the plan pivots Marian's budget line f;:om BC to BD. The slope decreases because
the grant decreases the opportunity cost of education spending from $1 to 67 cents.
Marian chooses point g instead of point i, so school spending increases from $4,000
to $4,800. As in the case of the foundation plan, part of the GTB grant is spent on
other goods, with spending on these goods increasing by $800. In this example, the
GTB grant is split equally between education and other goods.
A GTB grant has a larger stimulative effect than a foundation grant because the
GTB grant has both an income-effect.and a substitution effect. Like the foundation
386
'
grant, the GTB grant ncreases Marian's real income, increasing her desired spending
on al! 'normal" goods, including education (the income e:ffect). The GTB grant
also decreases the opportunity cost of education spending from $1 to 67 cents,
causing Marian to substitute education for other goods (the substitution e:ffect). In
Figure 15-5, one-fifth ofthe foundation grant is spent on education. In Figure 15-6,
one-half of the GTB grant is spent on education.
Under a pure GTB plan, wealthy districts would transfer money to the state
govemment. If a district's tax base exceeds the guaranteed tax base, the local revenue
will exceed guaranteed revenue, and the surplus revenue will go to the state. Since
the district must transfer tci the state a fraction of each dollar collected, the GTB plan
increases the opportunity cost orprice of education. In a wealthy district, the income
and sulistitution effects ofthe GTB plan combine to decrease speilding on education.
Would apure GTB plan equalize spending on education? The GTB plan provides
each school district with the same e:ffective tax base. If every community chooses
the same tax rate, education spending would be the same in al!districts. If, however,
communities chose different tax rates, spending inequalities would persist. Under a
pure GTB plan, any remaining di:fferences in education spending are the result of
differences in chosen tax rates, not di:fferences in tax bases.
'
/ -1
387
in income taxes and a Jarge increase in the sales tax (Courant and Loeb, 1997).
The state now determines educational spending in al!but the 28 wealthiest school
districts. Spending per pupil increased in small rural districts, and decreased in poor
urban areas and rich suburban areas. The other states with complete state control of
school finance are California, Hawaii, Florida, and Wisconsin.
A recent study shows that the greatest disparities in education occur at the low
end and high end of income distribution (Terman and Behrman, 1997). In 1990,
schoo1s in the wealthiest 10 percent of communities spent about 21 percent more
per pupil than the national average. In contrast, schools in the poorest 1O percent
spent 11 percent less than the national average. Spending per pupil in the remaining
80 percent ofhouseholds was within 6 percent of the national average.In other words,
most of the spending inequality ciccurs in the poorest and wealthiest communities.
','
How segregated is thU.S. school system? In 1995, about two-thirds ofblack students attended schools that hada majority of minority students, and about one-third
attended schools in which at least 90 percent of students were rninorities. The figures for Hispanics are a bit higher; 74 percent were in minority-majority schoo1s
and 35 percent in .schoo1s where minorities made up 90 percent of the students.
Segregation is most pronounced in urban schools. In the 10 1argest schoo1 districts,
white students comprised more than 22 percent of the student population in only
one district (Institute on Race and Poverty, 1998). In 1995, the student body of the
entire Chicago schoo1 district was only 11 percent white; the student body ofDetroit
school district was only 6 percent white.
There is also segregation inside schools. Under the ability-tracking system,
students are segregated with respect to ability and achievement: the high achievers
are put in one class and the low achievers are in another class. Because minority
students often are behind white students in the early grades, ability tracking in the
1ater grades contributes to racial segregation within the schoo1s.
388
Court-Ordered Desegregation
The courts have outlawed both methods of segregation. In Brown v. Board of Education ofTopeka (1954), the Supreme Court declared that explicit racial segregation
is unconstitutional. This Brown case reversed Plessy v. Ferguson (1896), which established the constitutionality of "separate but equal" schools. The Brown decision
requiredthe desegregation of schools only in areas where local govemments pursued
a policy of explicit segregation. Most ofthe cities forced to desegregate their schools
under the Brown decision were in the South.
The mandate for desegregation was extended in the 1970s to northem cities
that used attendance boundaries and other subtle techniques to promote segregation
(Clotfelter, 1979). The courts ordered severa! large northem and westem cities to
desegregate their schools. The objective was to integrate the schools to the extent
that would have occurred in the absence of earlier govemment policies that had
encouraged segregation. In other words, the courts did not require the cities to
develop truly integrated school systems, only ones in which govemment policy had
389
a neutral effect. The segregation that resu!ts from the individuallocation choices of
households (the Tiebout shopping process) is considered constitutional.
How did white households respond to forced desegregation? One ofthe lessons
from the history of desegregation policy is that it often increases ratherthan decreases
segregation. There are powerful forces behind segregation that are not neutralized
by court-ordered desegregation, and households cancounteract desegregation plans
by moving to another school district. In the South, many central cities had marbled
spatial distributions ofhouseholds: black areas were interspersed within white areas.
The cities used explicit segregation to send whites and blacks living in the same area
to different schools. When the courts outlawed explicit segregation, many centralcity white families moved to the predominant!y white suburbs to keep their chi!dren
in all-white schools.
C!otfelter (1979) exarnined the various responses to desegregation in Baltimore
and Atlanta and carne to the following conclusions:
The fieeing of whites to the suburbs decreased the demand for central-city
housing and decreased housing prices. In stable neighborhoods, the price of
central-city housing decreased by about 4.7 percent for a 1O percent increase in
the proportion of blacks in local public schools.
Families with school-age children were the most likely to move to the suburbs
when central-city schools were integrated.
The same phenomenon occurred in other cities that used busing to integrate their
schools; sorne white households moved from desegregating school districts to predominantly white school districts.
Another possible response to desegregation plans is to switch to private schools.
C!otfelter (l97) hypothesizes the following relationship:
\
E= f(S, 1, B)
390
-r
.
'
. '1
One policy option is to increase the funding for central-city schools. The simplest approach would be to adjust equalization formulas to account for the higher production
costs of central cities. A more aggressive approach would be to increase spending
in central cities beyond the leve!required to equalize real spending across schools.
Given the powerful infl.uence of home environments and peer groups, equalizing
educational. spending will not equalize educational achievement.
What would be the most efficient way to spend additional money for central-city
schools?
Hire better teachers. As shown by the studies of the education production
.)
function, teachers vary in their effectiveness, and a central city that used extra
money to hire better teachers would experience an increase in achievement.
Reduce class size. As we saw earlier in the chapter, there is evidence that
smaller classes have higher achievement levels, with the largest gains in achievement for low-income students.
. 1
In both cases, spending more money in central-city schools could partly offset the
Should the govemment subsidize students who attend prvate schools? Under apure
education-voucher system, each child would be issued a voucher or coupon that
could be used to pay for either public or prvate schooling. The face value of the
voucherwould be equalto the current costper pupil for public schools (for example,
. '
391
$5,000), iillowing a family to choose either the public school or a private school
charging up to the public cost per pupil. The school would collect vouchers from
its patrons and redeem them from the state govemment. To qualify for the voucher
prograrn, a school would teach basic cognitive skills and civics, and admit students
without regard for race, sex, or religion.
Would a voucher plan increase educational achievement? The advocates of
education vouchers view the public-school system as an inflexible and inefficient
monopoly. They argue that vouchers would :inake private schools competitive with
public schools and thus force public schools to compete with private schools for
students. One possible consequence is that public schools would be more concerned
about efficiency and more responsive to parental concerns. If so, achievement could
increase, even for the students who remain in public schools.
The opponents ofvouchers suggest that wider school choice is likely to increase
segregation with respect to income, race, and academic ability. Although a parent
who receives a voucher can pick any school-public or private-not all parents
will exercise this option. The experience with school-choice prograrns suggests that
parents with the most education and the highest incomes are more likely to switch
schools (Levin, 1997). The children in these farnilies are above-average in terms of
achievement and parental involvement, so the movement of these children to other
schools makes the peer groups at their original schools less favorable. As a result,
students of low socioeconomic status will have less favorable peer groups and thus
lower educational achievement and attainment.
A recent study ofvouchers generates sorne predictions about the effects of vouchers on different types of students (Epple and Romano, 1998). The study assumes
that private and public schools are equally effective, and that peer group effects are
important in educational achievement and parental choices. The sirnulated effects of
a $2,000 unv rsal voucher system (covering only part of the cost of private schools)
are as follows.
. l. The share of students in public schools drops frorn 90 percent to 70 percent.
2. A large number of high-ability students switch from public schools to private
schools, and the average ability leve!of public schools drops by about 16 percent.
3. There is a large per capita benefit for high-ability, low-income students, most of
whom switch to private schools and benefit from a more favorable peer group.
4. There is a small per capita loss for low-ability, low-income students, most of
whom stay in public schools .and suffer from a lessJavorable peer group.
5. The total gains experienced bY high-ability students exceed the totallosses of
low-ability students by a small margin, so there is a net gain in achievement.
392
segregation that would occur with a universal program. In fact, if the face value of
the voucher is high enough, it would encourage prvate schools to accept more
low-income students, promoting integration rather than segregation.
How do targeted vouchers affect achievement? Using data from Milwaukee,
Rouse (1998) shows that students in the voucher program experenced faster gains
in math test scores, but no differences in reading scores. One fascinating result of
the Rouse study is that low-income students attending special public schools with
reduced classes did just as well or better than the students in the voucher program,
who attended prvate schools.The students in small public classes had about the same
gains in math scores, but larger gains in reading scores. Since the prvate (voucher)
schools had smaller classes than the regular public schools, this suggests that the
achievement advantage ofvoucher (prvate) schools comes in part from their smaller
classes.
One ofthe arguments in favor ofvouchers is that they would increase competition
between schools and thus increase student achievement. Achievement will increase
if competition causes schools to invest in the most productive inputs to the education
production function. As shown by Rivkin, Hanushek, and Kain (2000), most ofthe
differences between schools is explained"by differences in the quality of teachers,
not by differences in school organization or other education resources. What real!y
matters is teachers, so vouchers couldincrease achievementifthe greatercompetition
for students causes schools to hire better teachers.
Another input that affects achievement is class size. As we saw earlier in the
chapter, students in smaller classes have higher achievement !evels, with the largest
gains for low-income students. Of course, it costs money to reduce class size. If the
benefits associated with smaller classes exceed the cost, then reducing class size
is an ef.ficient way to inorease achievement. The question is, Will the competition
fostered by voucheisresult in smaller classes and higher achievement?
. '
SUMMARY
l. The production function summarizes the rel tionship between education inputs
and achievement. The most important inputs are the home environment, the peer
group, and teachers.
2. Teacher productivity increases with verbal i;kil!s and with the first few years of
experence, but is unaffected by graduate education.
3. Achievement increases as class size decreases, with the largest achievement
gains for low-income students.
4. Education spending vares considerably across school districts, a result of differences in desired spending and property tax bases.
5. A foundation grant is negatively related to the district's tax base per pupil. The
grant would increase spending on education and decreases taxes, so households
can spend part of the grant on other goods.
6Under the guaranteed tax-base (GTB) plan, every school district has access to
the same effective tax base. The puie version of GTB would decrease spending
393
$200.000
$4.000
$20,000
b. If the school district has 1,000 students, how much does the foundation plan
cost the state?
c. What is the bang (increase in education spending) per buck of the foundation
grant?c
394
t .,
1
'
. 1
5. Compute. the price elasticity of demand -for education given the following
facts: (i) the local school district has a property value per student of $300,000;
(ii) the state's guaranteed tax-base (GTB) is $400,000; (iii) as a result of the
GTB plan, the school dis.trict increases spending per student from $4,000
to $4,200.
6. As the budget analyst for a state. government, your task is to estimate the fiscal
impact of a guaranteed tax-base plan. Assume the following: (i) price elasticity
of demand for education is -0.80; (ii) The guaranteed tax base is $400,000;
(iii) the typical school district has 100 students, a property tax base of $300,000
per student, and initially spends $3,000 per student. Compute the effects of the
GTB on the following:
a. Spending per pupiL
b. The local contribution to education (per pupil).
c. The state contribution to education (per pupil) and the total budgetary cost
of the GTB plan.
d. The bang (iocrease in education spending) per buck (budgetary cost).
7. Use the median-voter model (explained in Chapter 19) to predict the effects of
desegregation on spending (per pupil) in public schools. Will spending increase
or decrease?
8. One ofthe criticisms of a voucherprogram for education is that it would increase
spending inequalities and achievement inequalities. Designa new voucher plan
that would narrow-rather than widen-spending and achievement differences
across schools. How would you change the conventional voucher proposal to
decrease spending and achievement inequalities?
9. Consider Creditland, a state that recently implemented a system of income
tax credits fo:i: .private schools. The characteristics of the state and the program are as follbws: (i) Each farnily has one child in school; (ii) The govern
ment spends $4,000 on each child in the public schools; (iii) tuition in private
schools is $3,500; (iv) the tuition tax credit is 50 percent of private-school
tuition.
a. What additional information do you need to compute the fiscal effects of the
tax-credit program?
b. . Make up sorne numbers and compute the fiscal effect.
c. Choose the word in parentheses that makes the following statement correct
and then explain your choice: "The net budgetary cost of the tax-credit plan
increases as the price elasticity of demand for private education (increases,
decreases) in absolute value."
REFERENCES
l. Card, David, and Alan B. Krueger. "School Resources and Student Outcomes:
An Overview of the Literature and New Evidence from North and South
Carolina:" Joumal of EconomicPerspectives 10, no. 4 (Fall1996), pp. 31-50.
2. Case, Aruie, and Angus Deaton. "School Inputs and Educational Outcomes in
South Africa." Quarterly Joumal of Economics 114 (1999), pp. 1047-84.
1 .
-,
395
396
John Quigley and Daniel Rubinfeld. Berkeley and Los Angeles: University of
California, 1985.
21. Murnane, Richard J.The Impact of School Resources on the Leaming of Inner
City Children. Cambridge, MA: Ballinger, 1975.
22. Murnane, Richard J. "Understanding the Sources of Teaching Competence:
Choice, Skills, and the Lirnits ofTraining." Teachers College Record 84 (1983),
pp. 564-69.
23. Neal, Derek. "The Effect of Catholic Secondary Schooling on Educational Attainment." Joumal of Labor Economics 15 (1997), pp. 98-123.
24. Rivkin, Steven G., Eric A. Hanushek, and John F. Kain. "Teachers, Schools, and
Academic Achievement. NBER Working Paper Number 6691 (1998).
25. Rouse, Cecilia. "Private School Vouchers and Student Achievement: An Evaluation of the Milwaukee Parental Choice Program." Quarterly Joumal of Economics 113, no. 2 (1998), pp. 553-602.
26. Rouse, Cecilia. "Schools and Student Achievement: More Evidence from the
Milwaukee Parental Choice Program." Federal Reserve Bank of New York Economic Policy Review (March, 1998), pp. 61-76.
27. Sander, William. "Expenditure and Student Achievement in illinois: New Evi. dence." Joumal ofPublic Economics 52 (1993), pp. 403-16.
28. Summers, Anita, and Barbara Wolfe. "Do Schools Make a Difference?"
American Economic Review 67, no. 4 (1977), pp. 639-52.
29. Terman, Donna, and Richard Behrman. Financing Schools: Analysis and
Recommendations. The Future ofChildren 7 (1997), pp. 4-23.
CHAPTER 16
Urban Crime
Tru.s
chapter presents the economic approach to crime and crime prevention. The
economic approach assumes that both criminals and victirns are rational in the sense
that they base their choices on the expected benefits and costs of alternatives. A
rational person cornmits a property crime if the expected benefit of the crime exceeds
the expected cost. Sirnilarly, potential victims use their resources to prevent crime if
the expected benefit of prevention exceeds the expected cost. The economic approach
leads logically to a discussion of the optirnum amount of crime: because crime is
costly to pr vet, it is rational to allow sorne crime to occur.
"\Vhy study crime? The analysis of crime is an important part ofurban economics
for three reaso'ns. First, most crime occurs in metropolitan areas, and crime rates are
highest in central cities. Second, households are sensitive to crime rates, so their
location decisions are affected by local crime rates. In other words, crime affects
the spatial distribution of people within and between cities. Because households
are attracted to areas with low crime rates, an area with a relatively low crime rate
has relatively high housing prices. Third, the relatively high crime rates in central
cities have contributed to the suburbanization of population; roany households have
moved to the suburbs to escape central-city crime. A recent study suggests that each
additional central-city crime increases the number of people who relocate to the
suburbs by at least one (Levitt, 1996).
CRIMEFACTS
Most crime data come from the FBI's Uniform Crime Reports. The FBI collects data
from local police departrnents on seven index crirnes. The index crimes are divided
into personal crimes and property crimes.
397
398
1970
1980
1990
1995
1999
5.0
9.5
85.2
59.5
7.8
18.6
176.9
187.2
10.2
36.8
298.5
251.1
9.4
41.2
424.1
257.0
8.2
37.1
418.3
220.9
5.7
32.7
336.1
150.2
Property crime
Auto tbeft
Larceny
Burg1ary
182
1,024
504
457
2,124
1,152
502
3,167
1,684
658
3,184
1,236
561
3,045
988 .
420
2,551
770
1,870
3,949
5,950
5,820
5,278
4,266
Personal crime
Murder
Rape
Aggravated assault
Robbery
Sou.rce: U.S. Federal Bureau of Investigation, Crime in the United States, 1995 (Washington, DC: U.S.
Government Printing Office, 1995).
Personal crime. The victim of a personal crime is placed in physical danger. For
sorne personal crimes, the criminal's objective is to injure the victim (homicide,
rape, aggravated assault). For other personal crimes, the objective is to steal
property, but the criminal uses a show of force to coerce the victim (robbery).
. Property crime. Property crimes are crirnes of stealth, not force. Examples are
burglary (illegal entry of a building), larceny (purse snatching, pocket picking,
and bicycle theft), and auto theft.
The FBI data provide a partial picture of the crime scene. Among the crimes
omitted in the Unifo.rm -Crime Reports are disorderly conduct, shoplifting, arson,
employee theft, and drug-related offenses (possession andsale of narcotics, public
drunkenness, drunk driving). Table 16-1 lists the crirne rates for the seven index
crimes for the last few decades. The crirne rates are expressed as the number of
crimes per 100,000 population. Most of the reported index crimes are economically
motivated. The four economic crirnes (robbery, burglary, larceny, and auto theft)
make up about 90 percent of total reported crimes. The total crime rate rose from
1,870 in 1960 to 5,950 in 1980, and then fell to 4,266 in 1999.
Because the FBI data include only the crimes that are reported to the police,
they provide im incomplete picture of crirne. About two-thirds of al!crimes-and
over half of violent crirnes-are not reported to police and do not appear in the
FBI data. Since 1973, the Justice Department has collected crime data from its
semiannual victimization surveys. The surveys indicate that the overall leve! of
crirne has decreased since its peak in 1981.
399
TABLE16-2 Victimization Rates in 1993
Robbery
Wbite
Black
Central cities
Suburban areas
Rural areas
Assault
5
13
42
51
11
55
5
39
3
37
Theft and burglary: Rate per 1,000 households
Theft
241
250
292
241
190
Household
Burglary
57
86
84
47
54
Auto Theft
17
34
34
17
7
Robbery and assault: Rate per 1,000 persons age 12 and over
Sourct: U.S. Bureau of Justi.ce, Criminal Vtctimization in the United States, 1993 (Washington, DC, 1995).
Income. The victimization rates for crimes of violence (rape, robbery, and
f.
t:
.
f
t
1
1
1
1
1
assault) decrease as income increases: the poor are more likely to be victims
of violent crime. For example, a person with an income of $7,500 is twice as
likely to be a victim of violent crime as a person with an income of $50,000.
In contrast, victimization raies for property crime increase with income. For
example, compared to a person with an annual income less than $7,500, a
person with an income above $50,000 is 50 percent more likely to be a victim
of property crime.
Place ofresidence. Table16-2 shows victimization rates for different locations.
Crime rates are higher in metropolitan areas than in nonmetropolitan areas, and
higher in central cities than in suburbs.
Race. Table 16-2 shows the victimization rates for whites and blacks. For all
five crimes listed, blacks are victimized more frequently than whites, with the
biggest differences for robbery and auto theft.
There is substantial variation in crime rates across neighborhoods within metropo!itan areas and central cities. Figure 16-1 shows the number of property crimes
committed by teenagers in the 95 neighborhood areas ofthe Cleveland metropo!itan
area. (Data on total crime are not available for the entire metropolitan area.) While
24 of the neighborhoods had fewer than 10 crimes, 30 neighborhoods had at least
50 crimes, and 10 neighborhoods had at least 100 crimes. Figure 16-2 shows the
property crime rates (per 100,000 population) and poverty rates (in percent) for the
35neighborhoods in the central city of the Cleveland metropo!itan area. In general,
the higher the poverty rate, the higher the crime rate.
1
1
t
l
1
1
...L
""
"g
u
."
5
[
10
22
82
94
Neighborhood
Source: Center for Urbfm Poveny and Social hange, Case Western Reserve Universily. hup:/lpoverlycenler.cwru.edu
401
g 12,000
]. 10,000
gg""
8,000
&"
6,000
>.
4,000
2,000
10
20
30 . 40
50
Poverty rate in percent
60
70
80
90
Source: Center for Urban Poverty and Social Change, Case Western Reserve Universicy. http://povc:rtycenter.cwru.edu
in the highest. A one-unit increase in the crime rate (an additional percentage
point) decreased the average property value by $336. The vandalism rate varied
from less than 1 incident to 30 incidents per 1,000 population. A one-unit increase in the vaD.<;lalism rate (an additional incident per 1,000 population) decreased
the average property value by $117. In another study, Thaler (1977) estimated
the elasticity of property values with respect to crime rates as -0.067: a 10 percent increase in the crirne rate decreased the market value of housing by about
0.67 percent.
402
Part 4
Urban ProblenlS
prevention
$39 billion .
Criminal justice
systern
$74 billion
')
The econornic mcidel of crime assumes that the criminal commits a crime if the
expected benefit <if the <;rime exceeds its expected cost. The model of the rational
criminal is relevant for econornically motivated crimes (property crimes), but not
for crimes of passion imd violence.
EL= Ps Loot
If the probability of success is 80 percent and the resale value of the stolen property
is $600, the expected value of the loot is $480.
On the cost side, Boris faces the possibility ofbeing sent to prison for committing
the burglary. The expectedvalue ofthe penalty for burglary is the probabi!ity ofbeing
imprisoned (P) times .the opportunity cost of time spent in prison. The probability of
'
')
403
TABLE 16-3
-:
'
Burglary activity
Expected loot
Loot
Probability of success (P,)
Expected loot (EL= P, Loot)
Expected cost
Probability ofpunishment (P; = Pa Pp)
Opportunity cost of prison time
Daily wage (W)
Expected workdays per year (D)
Forgone income per year (I = W . D)
Cost of lost freedom per year (F)
Opportunity costperyear (C = I + F)
Expected cost (EC = P . C)
Expected benefit ofburglary (B =EL- EC)
Legal acriviry
Daily income if employed (W) .
Net retumfrom burglary (R = B - W)
$ 600
0.80
$ 480
0.02
$
40
!50
$6,000
$2,000
$8,000
$ 160
$ 320
$
40
$ 280
being imprisoned equals the probability ofbeing arrested (Pa) times the probability
ofbeing sentenced to prison once arrested (Pp).
P = Pa Pp
404
high enough to overcome their under!ying aversin to it. One way to incorporate
the aversion to crime into the benefit-cost analysis is to estmate a person's anguish
cost, defined as the cost associated with engaging in antisocial activity. If Gabriel
has a strong aversion to crime, his anguish cost may be $10,000, meaning that he
will commit a crime only if the net return exceeds $10,000. IfLucifer is less averse
to crime, his anguish cost is only $10, so he will commit a crime if the net retum
exceeds $10. Returning to the example in Table 16-3, Boris will commit the burglary
if his anguish cost is !ess than the net return of the burglary ($280).
The model of the rational criminal provides three reasons why sorne people commit crimes. First, sorne people arerelatively ski.IIful at committing crime and escaping
punishment. For such people, the expected loot is relatively large and the expected
cost of crime is relatively small. Second, sorne peple have a relatively low opportunity cost of time spent committing crirue and time spent in prison, so the expected
cost of crime is re!atively low. For example, the poor may have a lower opportunity
cost because they are unemployed or earn low wages. Third, sorne people have less
respect for society and are less averse to committing crimes, so they need a relatively small net return to make crime worth,wliile. For most people, the moral anguish
associated with crime is large enough to make crime unattractive at almost any price.
405
600
500
------------------------
es,
. 400
00
Supply with
moderate aversion
1
1
1
1
1
1
1
1
1
1
1
1
1
----------
--------1
1
1
."e
";; 300
1
1
1
1
1
1
1
!:!
" 200
l.
100
1
1
1
1
20
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
60
80
40
Number ofburglaries per month
100
The supply curve shows the number ofburg1aries for different net returns (R) on crime. It
has a positive intercept because of anguish cost and is positivo!y sloped because an increase
in the net return increases the number of burglars and increases the number of burglaries per
burgiar. An iJlcrease in the aversion to crine shifts the s pply curve upward.
...
population sbift the supply curve. Young males cornmit more than their share of
crimes. Therefore, a society with a relatively large number of young males will have
relatively high crime rates.
OPTIMUM AMOUNT OF CRIME
Because crime prevention is costly, it may be more efficient to experience a crime
than to prevent it. In other words, the optimum amount of crime is positive. How
much crime should be we tolerate?
..,.., ,
'--'
o ECA
8181_1
406
.
n
F
'
Total cost
Victim cost
Prevention cost
1
1
1
.1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
100
B* .
Number ofburglaries
The total social cost otburglaries equals the sum of the cost incurred by victims and the cost of
prevention. Victim cost increases ata constant rate ifthe cost per crime is constant. The prevention
cost increases rapidly as the number of burglaries decreases, a result of diminishing marginal retums.
Total cost is minimized with B* burglaries (point G).
Although society can decrease crime by decreasing the net return on burglary,
crime prevention is costly. There are a number of options for crime prevention:
l. Hardening the target. Victims can decrease the expected loot by decreasing
the probbility of successful burglary (e.g., installing locks and using guard
dogs).
2. Increased probability of arrest. The police can increase the expected cost of
crime by increasing the probability of ari:est (e.g., hiring more police officers).
3. Increased probability ofimprisonment. The criminaljustice system can increase the expected cost ofcrime by increasing the probabilities of conviction
(e.g., hiring more investigators to gather evidence on crime).
4. Increased severity of pmshment. The criminal justice system can increase
the expected cost of crime by lengthening prison sentences (e.g., building more
prison cells).
.,
1
_.
407
-,:.;
....- ':.
-ll?
(
':[
l
f..
'C
1
{.
!
1
l
l
1
1
1
408
FIGURE 16-6 . Marginal Victim and Prevention Costs and the Optimum Crime Leve!
$
Marginal
prevention cost
1
1
1
1
1-------,_
A*
B*
Number of ctimes
100
The optimum ctime leve!is where the marginal victim cost equals the marginal prevention cost
(B' for burglaries and A for armed robberies). The optimum nurnber of robberies is smaller
because robbery has a hlgher victim cosL
The optirnum amount of crime is shown by the intersection of the marginal victim cost curve and the marginal prevention cost curve. Starting with 100 burglaries,
the marginal victim cqst for burglary exceeds the marginal cost of prevention, so it
is sensible to deter the IOOth crime. Prevention is efficient as long as the marginal
victirn cost exceeds the marginal prevention cost, so the optirnum burglary rate
is B*.
The marginal-cost curves explain why optirnum crime rates vary across crimes.
Suppose burglary and armed robbery have the same prevention cost, but di:fferent
victim costs; the victirns of armed robbery are often injured during the crime, so
armed robbery has a higher marginal victim cost. In Figure 16-6, the optimum
crime rate forrobbery (A*) is less than the optlmum crime rate for burglary: more
resources are invested in robbery prevention because the savings in victirn cost
per robbery are higher. Optirnum crime rates also di:ffer because of differences in
prevention cost: crimes with higher marginal prevention costs have higher optimum
crime rates, everything else being equal.
The same .argurnents can be used to explain why crime rates are higher in
sorne central-city areas. These areas have large concentrations of poor and unemployed people, so the opportunity cost of crime is relatively low and the number of
potential criminals is relatively large. Given the plentiful supply of criminals, the
marginal prevention cost is high. If the rn ginal victirn cost is the same throughout
the metropolitan area, the optirnurn crime rate is higher in the central city.
409
6 Urban Crime
CRIME PREVENTION
This section discusses sorne of the details of crime prevention, exploring the effects
of various "carrot" and "stick" approaches to preventing crime. The first part of
the section discusses carrot policies: By increasing the value of legal opportunities,
the government increases the incentives to choose legal activities over crime. The
second pait shows how potential victims can protect themselves from crime. The
final four parts examine stick policies: By increasing the likelihood and severity of
punishment, the government deters crime by increasing the expected cost of crime.
-.,.
Part 4 Urban Problems
410
1.. ,-
Protection
Cost
11
Total
. Cost
Probability
of Theft
Expected
Theft Cost
030
0.10
$30
$30
10
2
21
26
24
0.02
80
0.00
.-.
80
,--,
Protectionand Insurance
How does insurance affect Alice's spending on protective measures? Suppose that
she insures her $100 bicycle against theft, and there are no costs associated with
filing a claim (no transactions cost and no deductible payment). If so, the total cost
of leaving her bike unprotected is zero: Every time her bike is stolen, she collects
$100 from her insurance company and then buys a new bike. Therefore, A!ice will
leave her bike unprotected.
. This is the moral-hazard problem. If A!ice buys insurance, she reduces her
own protective measures and is more likely to be victimized. The moral-hazard
problem explains why many people do not lock their cars. According to the FBI,
411
80 percent of stolen cars are leftunlocked, and 40 percent have the keys in the ignition. The moral-hazard problem also explains why most insurance policies have
deductibles; the deductible increases the property owner's expected loss from theft,
encouraging the owner to invest in protective measures. Recently, sorne insurance
companies have tried to control the moral-hazard problem by providing free homesecurity systems to their policyholders.
THEPOLICE
One measure of the effectiveness of police is the arrest ratio, defined as the number
of arrests divided by the number of crimes committed. The arrest ratios for property
crimes such as burglary, larceny, and auto theft are in the range of 0.13 to 0.18,
and the arrest ratio for robbery is about 0.27 (Levitt, 1998). Arrest ratios for violent
crimes are much higher, (e.g., 0.42 for aggravated assault and 0.82 for murder).
One way to deter crime is to have arelatively high arrest ratio. From the criminal's
perspective, the arrest ratio is the probability of being arrested for a crime, and the
higher the probability, the lower the expected payoff from crime. According to a
recen!study, the elasticity of crime rates with respect to the arrest ratio is around
-0.30 forrobbery, burglary, and larceny (Levitt, 1998a).This means that a 1O percent
increase in the arres!ratio. will decrease crime rates by about 3 percent. In earlier
studies, the estimated elasticities for property crimes were around -0.50 (Ehrlich,
1972; WJ.lson and Boland, 1976; Bartel, 1979; Witte, 1980).
Other studies explore the relationship between spending on police activities
and criminal activity. Tauchen, Witte, and Griesinger (1994) shows that additional
spending for police decreases the likelihood that youths will commit crime. A 1O percent increase. in police spending increases the average length of time a youth is a
law-abiding citizen by about 4.7 percent.
Pollee Patrol
The police spend a lot of time patrolling streets and sidewalks. The idea behind
patrol is that the police should be in a position to respond quickly when crime
occurs. Does patrol deter crime? A number of studies show that car patrol does not
have a significan!effect on crime rates. In a famous experiment in Kansas City, the
city increased patrolling activity in sorne neighborhoods and decreased patrolling
in others. There were no significan!changes in either arres!ratios or crime rates in
any of the neighborhoods, suggesting that patrolling does not decrease crime. Other
experiments have cometo the same conclusion (Sherrnan, 1983).
Why doesn't car patrol work? One reason is that crime victirns are slow to
report crimes to the police. According to Sherrnan, the typical victim takes about
50 minutes to report a crime to police. Suppose that an increase in patrolling could
decrease the response time of police (defined as the period of time between receiving
. areport ofthe crime and arriving on the scene) from 10 minutes to 2 minutes. Ifthe
typical victim takes 50 minutes to report the crime, the police could arrive on the
412
Part 4
Urban Problems
crime scene 52 minutes after the crime is committed instead of 60 minutes. In either
case, the criminal's trail is aiready cold, so the improved response time is unlike!y to
increase the arrest ratio. On the other hand, if the victim took only 1 minute to report
the crime, an increase in patrolling could make a di:fference: the police would arrive
3 minutes after the crime instead of 11 minutes and would have a better chance of
capturing the criminal.
What about foot patrol? A number of studies show that foot patrol can increase
arrest ratios and deter crime. Police on foot patrol are constantly gathering infomlation about the people and businesses in their territories. The foot-patrol officer knows
who is arguing with whom, who is sudden!y living beyond his means, and whose
children are running amok. As a result, the foot-patrol officer is better equipped to
prevent crirnes and to solve crimes once they occur.
Pollee Investigation
.-.
.,
413
27 Plead Guilty
7 Go to Tria!
6 Disrrssed
S Convicted .o
in Tria!
27 Plead Guilty
o.
20 Incarcerated
9 Imprisoned
(more than 1 year)
12 Probation
11 Jailed
(less than 1 year)
Source: Brian Forst, "Prosecution and Senlencing," Chapter 10 in Crime and Public Policy, ed. James Q. Wilson
(San Francisco: Institure for Contemporary Stuclies, 1983).
prison sentence longer than ayear. The process by which the 40 cases are whittled
down to nine imprisonments is as follows:
l. Dismissed. Of the 40 cases accepted by the district attorney, 6 are disrnissed
because the evidence is insufficient, the offense is trivial, or the defendant disappears afterbeing released on bond, personal recognizance, or third-party custody.
2. Plea bargaining. Of the 34 remaining cases, 27 of the defendants plead guilty,
usual!y to a lesser charge, and 7 cases go to tria!.
3. Tria! acquitta!. Of the seven cases that go to tria!, two result in acquitta! and
five result in a guilty verdict.
4. Probation. Of the 32 defendants found guilty (27 from plea bargains and 5 in
tria!s), 12 receive probation (supervised release).
5. Jail versus prison. Twenty convicts are incarcerated. Eleven serve time in jai!s
(receiving sentences of less than one year), and nine are imprisoned (receiving
sentences of more than one year).
414
415
The parole system is based on the principie of rehabilitation. Under the parole
system, a convict stays in prison until the parole board declares thathe is rehabilitated.
The rehabilitated convict is released and serves the remainder ofhis prison term under
the supervision of a parole officer. A parolee cannot get married or own a car, and
must receive the approval of his parole officer before borrowing money or changing
his residence.
Most parole boards have large caseloads and base most parole decisions on
simple rules ofthumb. One popular rule is that a convict is released after serving either
three years or one-third of the prison sentence, whichever is shorter. Studies of the
parole system have generated rnixed results. In sorne states, parolees have lower rates
of recidivism than convicts who serve out their entire sentence. In other states, there is
no measurable difference between the recidivism rates ofparolees and other convicts.
416
g
>.
: R* ------------------
.e
aversion
.9
"
il
R'
B'
B" B*
Number ofburglaries
The increase in the severity of punishment decreases the net retum from R* to
R', causing a move from C to D.If a looger prison term hardens the criminal,
her aversion to crime will decrease, shifting the supply curve to the right and
causing a move from point D toE; crime rises close to its original leve!(B" is
clase to B*). If the criminal also Jearns more in prison, she will have a higher
probability of success when she gets out, causing a move from E to F
(increasing the crime rate to its original leve!).
Source: Brian Forst.,.."Pro.secution acd Sentencing," Chapter 10 (Table 1) in Crime and
Public Policy,\ed. James Q. Wllson (San Francisco: Institute for Contemporary Srudies,
1983.)
417
'..
Which is more important, deterrence or incapacitation? The answer to this question tells us whether a recent innovation in criminal sentencing-"Three Stri.kes
and You're Out"-is likely to be efficient. Under the three-stri.kes law, a person
convicted for a third felony is given a wiry long prison sentence (e.g., 25 years or
perhaps life imprisonment). About half the states have three-stri.kes laws, but just a
handful have implemented 'the law in a significant way. California has used the law
tosentence 40,000 people-about a quarter of its prison population-to lengthy sentences (Carelli, 1999). Georgia has sentenced about 2,000 people to lengthy terrns
under its three-strikes law.
To think about the effects of three-strikes laws, consider two extremes. First, if
prisons have no deterrent effect but only an incapacitation effect, the three-strikes
law will be ineffective and inefficient. By the time a person is convicted of a third
felony, he will probably be late in his criminal career, and will be committing fewer
and fewer crimes each year. There will be few crimes avoided by having a relatively
old criminal in prison, so the cost ofthe three-strikes law ($23,000 peryear for many
years) is likely to exceed the benefit (the decrease in the social cost of crime).
At the other extreme, suppose the only effect of prisons is deterrence. In this
case, the threat of longer prison sentences under the three-strikes !aw will decrease
the expected payofffrom crime for allcriminals, decreasing crime rates. Ifthe threestrikes law deters enough crime, there could actual!y be fewer people in prison. Ifthe
law has a large deterrent effect, its benefit (the decrease in the social cost of crime)
is likely to exceed its cost (extra expenses for the prison system).
A recent study suggests that deterrence is more important than incapacitation
(Levitt, 1998). Eor property crimes, deterrence is responsible for about three-fourths
of the decreases in crime caused by prisons. For example, if the number of burglary
arrests increases by 10, the deterrent effect will be 23 fewer burglaries, but the incapacitation effect is just 1less burglary. If the number of auto-theft arrests increases
by 10, a total of 5 auto thefts are avoided through deterrence, but only 1 is avoided
through incapacitation.
Although it is too early to determine the actual effects of three-strikes laws, the
early experiences are consistent with the notion that deterrence is more important
than incapacitation (Levitt, 1998). In the state of Washington, which has had a
three-strikes !aw since 1993, violent crime has dropped, but the prison population
is expected to grow by only 10 percent over the next 10 years. During the first
year of California's three-strikes law, violent crime dropped by 7.2 percent and the
projections for the prison population V{ere revised downward by 25 percent.
418
A Numerical Example
The principie of margirral deterrence can be explained with a simple example. Suppose that a state initially has a one-year penalty for burglary anda two-year penalty
for armed robbery. Robbery carries a greater penalty because robbery has a higher
social cost; the victims of robbery are often injured or killed, while the victims of
burglary simply lose their property.
Ernil is a career criminal who must decide whether to comrnit a robbery or a
burglary. He will comrnit the crime that provides the larger net benefit. His options
are sl:iown in Table 16-5. Robbery generales a greater expected loot ($2,000 versus
$1,700), but also carries a higher expected cost ($1,600 versus $800). The expected
net benefit of burglary is $900 compared to $400 for armed robbery, so Ernil picks
burglary rather than robbery.
TABLE16-5 Marginal Deterrence
Expected 1oot
Probability of punishment
.Pr:ison senten.ce
Annualopporrunity cosr-of p;ison.
Expected piison cost
Net benefit
Burglary
Burglary
Armed Robbery
(1 year)
(2 years)
$2,000
0.10
2 years
$8,000
. $1,600
$400
$1,700
0.10
1year
$8,000
$800
$900
$1,700
0.10
2 years
$8,000
$1,600
$100
419
The state's penalties for robbery and burglary provide a marginal deterrent for
robbery, the more serious crime. In deciding on the leve!of crime, Ernil weighs
the marginal benefit and marginal cost associated with upgrading from burg!ary
to robbery. The marginal benefit of upgrading is the increase in the expected loot
($300). The marginal cost is the increase in the expected prison cost ($800). Since
the marginal benefit is less than the marginal cost, Ernil sticks with burglary. Society
deters Emil from the more serious crime by imposing a larger penalty on robbery.
Suppose that the state passes a law that equalizes the penalties for burglary and
robbery at two years. The increase in the burglary penalty decreases the expected
net benefit from burglary to $100 ($1,700- 1,600), but does not affect the net
benefit from robbery. Since the net benefit of robbery now exceeds the net benefit
of burglary, Emi!switches to robbery. The equalization of penalties eliminates the.
marginal deterrent. The marginal benefit of moving from burglary to robbery is
still $300, but the marginal cost of upgrading (the increase in the expected penalty)
is zero. If Emil receives the same penalties for the two crimes, he will commit the
more !ucrative crime. The equalization of penalties eliminates the marginal deterrent
e:Efect, so Emil switches to the crime with a higher social cost. .
o
FIGURE 16-9 Response toan Increase in the Burglary Penalty
$
$
Initial supply of armed
robberies
Supply of
burghrries
R'
R"
B"
B'
Number of burglaries
A' A"
N umber of robberies
In the initial equilibrium, both burglary and robbe.ry have a net retum of R', so there are B' burglaries and .A' robberies. A decrease
in the net return of burglaries to R11 decreases the number of burglaries to B" (movement down the supply curve) and increases
the number ofrobberies toA" (shift ofthe supply curve and unchanged nel retum to robbery). The increase in robberies is less than
the decrease in burglaries because sorne burglars "go straighl."
-;-
421
to estimate the effects of changes in penalties on the social cost of crime. To do so,
they would need infonnation on the social cost of different crimes and the crirninals'
responses to the change in penalties. The principal questions are:how many crirninals
would be deterred; how many would upgrade toa higher-level crime; and how many
would downgrade toa lower-level crime?
14....
12
12 f.-
,.:.:.
11
,...-
f.-
10
r-:
f.-
C'"""
.
..
f.-
7
,...-
.
.
4 f.-
1
1
2 f.-
25,000 to
49,999
50,000 to
99,999
1,000,000
and more
Source: Edward L. Glaeser and B. Sacerdote. "VIhy Is There More Crime in Cities?" NBER Wqrking
Paper#5430, 1996.
422
Part 4
Urban Problems
arrest rate (the percentage of crimes that result in an arrest) and city size for two
reasons. First, it is more difficult for police to identify a culprit when the pool of
potential suspects is large. Second, lawful citizens in impersonal big cities may be
less inclined to help their neighbors and the police in crime-control efforts.
3. A higher concentration of female-headed households is responsible for 50 percent of difference in crime rates. It's not clear why higher rates of single parenthood
havesuch a powerful influence on crime rates. Glaeser and Sacerdote speculate that
children raised in single-parent families may have fewer job skills and less powerful
ethical restraints on criminal behavior.
Since 1970, the correlation between city size and crime has weakened, and in recent
years, there were dramatic reductions in crime rates in big cities.
'f ---,
1 --,
CJ Violent cnme
Property crime
700
.ll
600
IJILII n
11
n
1 1
1 1
1 1
1 1
1 1
1 1
1 1
1 1
p,'>
...._o;
o,'>
1 1
l. 1
11
"o' 1 1 1 1 1 1 1 1 1 1 1 1 1 1
'B
&o 4001-e..
o
o
o
-D I I I-IEII I I II II I I
-"2001
u 100
o
.:..
N
.....
i i ii i i ii
!!><:>
!!.>...
o',bey
"'
Source: FBI Uniform Crime Reports.
!!>">
!b"
!bb
;;
.._ ,o ,
!b'b
p,<:>
A'
o',Y
p,">
p,b
p,'\
p,'b
o,">
424
Will the enforcement of drug Jaws increase total property crime? In Figure 16-12, the illegality of drugs increases production costs and shifts the supply .curve to the Jeft. The illega!ity also increases the cost of consurnption, shifting
the demand curve to the Jeft. Because the supply side faces higher penalties, the
supply shift is larger and the equilibriurn price increases. If the demand for drugs is
price-inelastic, the increase in price will increase total expenditures on drugs. If drug
mega!
supply
mega!
demand
P"
Legal
supply
P'
------------- ----------1
1
1
1
1
1
1
K"
K'
.Kilos of drugs
When a commodity is declared illegal, the cost of producing and selling it increases, shifting the
supply curve upward. Consumer penalties increase the net cost of consuming the good, shifting
the demand curve downward. The equilibrium moves from point .B to C. The supplier penalty is
large relativo to the consumer penalty, so the supply shift is larger than the demand shift, and the
price increases from P' to P". Tbe equilibrium quantity decreases from K' to K".
'
425
addicts support their habits with property crirne, the increase in total expenditures
increases property crirne.
Si!verman and Sprull (1977) estimate the effects of changes in drug prices on
property crime. They estimate that the price elasticity of demand for heroin is about
-0.27, so a 50 percent increase in the price of heroin would decrease the quantity of
heroin demanded by about 13 percent and increase total drug expenditures by about
30 percent. Because only a portion of property crirne is drug-related, the resulting
increase in property crirne (14 percent) is !ess than the increase in drug expenditures.
There are two irnplications from this analysis. First, there is often a trade-offbetween drug control and property crime. Because a supply-side policy increases total
expenditures on drugs, property crime is 1ikely to increase. Although drug enforcement policies decrease drug consumption and thus decrease the social cost associated
with drug abuse, they also increase property crime. The difficult policy question is
whether the benefit (the decrease in the social cost of drug abuse) exceeds the cost
(increased property crime). The second implication is that demand-side po!icies do
not have the sarne trade-offs. Since a demand-side policy simply shifts the demand
curve downward, decreasing both the price and quantity of drugs, it decreases total
expenditure on drugs and thus decreases drug-related property crime.
SUMMARY
l. Crirne victimization rates vary with income, place of residence, and race, with
generaly! higher rates for racial minorities and the residents of central cities.
2. The optimum arnount of crirne is the leve!at which the marginal victim cost
equals _the marginal protection cost.
3. A rational person cornrnits a property crirne if the net return on crime is large
enough to offset her underlying aversion to crime. The net return depends on
the monetary value of the loot, the probability of success, the probability of
imprisonment, the opportunity cost of prison time, the length of the prison
sentence, and the income forgone in a legal job.
4. The supply curve for crirne shows the relationship between the net return and
the number of crirnes. The net return to crime is determined by the wages of
legal opportunities, protection efforts by property owners, arrest rates, and the
certainty and severity of punishment.
S. The crirne supply curve shifts in response to changes in demographics, such as
a change in the share of the population in the crime-prone years, 18-24.
6. The elasticity of crime with respect to the probability of arrest is about -0.30.
7. About 9 in 100 felony arrests result in prison sentences. About one-third of
arrests lead to convictions, butmost convicts receive probation or shortjail terms.
S. The principie of marginal deterrence is that the penalty should be scaled to
the severity of the crirne, thus providing a marginal deterrent for more serious
crirnes.
_
9. The prison system has four purposes: rehabi!itation, deterrence, incapacitation,
and retribution.
426
$1,000
0.70
0.03
$ 20
100
$3,000
2 years
Compute the values ofthe following variables such that Boris is indifferent between burglary and a legal job. In computing the threshold value for each variable, assume that the values ofthe other variables are the same as in Table 16-3.
a; The probability of success (Ps)b. The value of the loot
c. The probability _of imprisonment (P;).
d. The opportunity cost per year (C).
e. The length of the prison sentence.
3. Considera city with a wealthy area anda poor one. The two areas have the same
horizontal MVC (marginal-victim-cost) curve and the same negatively sloped
MPC (marginal-prevention-cost) curve. The two curves intersect ata crime rate
B*.The crime rate in the wealthy area is Bw and the crime rate in the poor area is
Bp- Given the currentallocationofpolice between the two areas, Bw < B* < Epa. Use a graph to depict the current situation.
b. Is the. current allocation of resources efficient? If not, how should resources
be reallocated? Put sorne numbers on your graph and use the graph to show
that the reallocation of resources will generate net gains for society.
c. Suppose that the two areas have the same MVC curve but different MPC
curves. Depict a situation in which a higher crime rate in the poor area
(BP > Bw) is efficient.
4. The city ofEvilville suffers from two types of crimes, marijuana consumption and burglary. Now let us suppose that you must .decide how to allocate the city's crimecfighting resources between the two types of crime. The
marginal-prevention-cost (MPC) curves are linear, and for a given crime rate, the
427
Protection
Cost.
Probability
ofTheft
0:30
0.10
0.02
0.00
11
24
80
Expected
Theft Cost
Total Cost
. 7. Critically appraise this statement: ''The social costOfrobbery is about six times
the social cost ofburglary. Given the principie of marginal deterrence, the length
of the prison term for robbery should be six times the length of the prison term
for burglary."
8. Consider the analysis of marginal deterrence shown in Figure 16-6. Under what
circumstances (what values for the social cost of burglary and robbery) would
the increasd penalty for burglary decrease the total social cost of crime?
9. Considera city in which there are only two types of crimes, burglary and armed
robbery. Choose one ofthe words in each set ofparentheses to make the following statement correct and then explain your choices. An increase in the penalty
for burglary will decrease the total social cost of crime if
a. The supply curve for burglary is relatively (steep, fiat).
b. The cross-elasticity of the supply of armed robbery with respect to the net
return on burglary is relatively (small, large).
c. The social cost ofburglary is (small, large) relative to the social cost of armed
robbery.
10. In the city of Swiftsure, thereis only one crime, armed robbery. The city recently
decided to double the probability of imprisonment from 10 percent (10 percent
of armed robberies result in a prison sentence) to 20 percent. Because the city
has a fixed prison capacity, it also cut the average prison sentence in half, from
3 years to 1.5 years.
a. Would you expect these changes in the city's crime policy to increase, decrease, or not affect the city's crime rate?
b. Would you expect these changes to increase, decrease, or not ajfect the total
cost of crime (the sum of victim and prevention costs)?
428
11. Suppose that Gotharn City doubles the required prisonsentence for burglary (its
only crime). Consider the following quote frorn Robin: "If rny assurnptions are
correct, the longer prison sentence will increase the crime rate."
a. What are Robin's assurnptions?
REFERENCES
l. Bartel, Ann P. "Wornen and Crime: An Econornic Analysis." Economic Inquiry
.-.
'
429
430
22. Levitt, Steven D. "Why Do Increased Arrest Rates Appear to Reduce Crime:
Deterrence, Incapacitation, or Measurement Error?" Economic Inquiry 36
(1998), pp. 353-72.
23. Lipsey, Mark. "Juvenile Delinquency Treatment: A Meta-Analysis Inquiry into
the Variability of Effects." In Meta-Analysis for Explanation: A Casebook, ed.
Thomas D. Cook et al. Beverly Hills, CA: Sage Foundation, 1992.
24. Marvel, Thomas B., and Carlisle E. Moody. "Prison Population Growth and
Crime Reduction." Joumal of Quantitative Criminology 10, no. 4 (1994),
pp. 109-40.
25. Mendel, Richard A. "Prevention or Pork? A Hard-Headed Look at YouthOriented Anti-Crime Programs." Washington, DC: American Youth Policy
Forum, 1995.
26. Phillips, Llad; Harold Votey; and HaroldMaxwell. "Crime, Youth, and the Labor
Market." Joumal of Political Economy 80 (1972), pp. 491-504.
27. Phillips, Llad, and Harold Votey. Crime and Public Policy. Beverly Hills, Ca!if.:
Sage Publications, 1979.
28. Reynolds, Morgan. Crime by Choice: An Economic Analysis. Dallas: Fisher
Institute, 1986.
29. Sherman, Lawrence. "Patrol Strategies for Police." In Crime and Public Policy,
ed. James Q. Wilson. San Francisco: Institute for Contemporary Studies, 1983.
30. Silverman, L. P., and N. L. Sprull. "Urban Crime and the Price of Heroin."
Joumal ofUrban Economics 4 (1977), pp. 80-103.
31. Spelinan; William. "The Lirnited Importance of Prison Expansion," Chapter 4
in The Crime Drop in America, eds. Alfred Blumstein and Joe!Wallman. New
York: Cambridge University Press, 2000.
32. Tauchen, He!eil;_ Ann. Dryden Witte; and Harriet Griesinger. "Criminal Deterrence: Revisiting.the Issue with a Birth Cohort." Review of Economics and
Statistics 76 (1994), pp. 399-412.
33. Thaler, Richard. "Econometric Analysis of Property Crime-Interaction between Police and Criminals." Joumal of Public Economics 8 (1977), pp. 37-51.
34. Wilson, James Q. Thinking about Crime. New York: Basic Books, 1975.
35. Wilson, James Q., and Barbara Boland. "Crime." Chapter 4 in The Urban
Predicament, eds. William Gorham and Nathan G!azer. Washington, DC:
Urban Institute, 1976, pp. 179-230.
36. Witte, AnnD. "Estimating the Economic Model ofCrime with Individual Data."
Quanerly Joumal of Economics 94 (1980), pp. 57-84.
PART FIVE
Housing
Tru.s part of the book explores the economics of the urban housing market and
evaluates the merits of various housing policies. Chapter 17 explains why housing is di:fferent from other products, focusing on its heterogeneity, immobility, and
durability. The chapter also explains why low-income households typically occupy
used housing and why many poor households are homeless. Chapter 18 describes
the market effects of various housing policies, including public housing, subsidized
private housing, and housing vouchers that allow recipients to choose where to live.
431
'.
'
.,)
CHAPTER 17
Tru.s is the first of two chapters on urban housing. It explains why housing is a
unique co=odity and focuses on five features that rnake housing different frorn
other goods. First, the stock of housing is heterogeneous; dwellings differ in size,
location, age, fl.oor plan, interior features, and utilities. Second, housing is irnmobile:
it is irnpractical to rnove a dwelling frorn one location to another. Third, housing is .
durable: if properly rnaintained, a dwelling can be used for rnany decades. Fourth,
housing is expensive: to purchase a dwelling, the typical household rnust borrow
a large surn of rnoney. Fifth, rnoving costs are relatively high: in addition to the
substantial rnonetary cost of rnoving, there are also costs associated with leaving the
old neighborho_od-with its schools, stores, and friends-behind.
The rin,ique nature of housing has sorne irnportant irnplications for housing
rnarkets. Because housing is heterogeneous and irnrnobile, the housing rnarket is split
into a set of distinct but related housing subrnarkets. The durability of housing rneans
that property owners rnake decisions about rnaintenance and repair and sornetimes it
is rational to sirnply abandon older housing. The durability ofhousing also rneans that
dwellings will filter down over time to households with progressively lower incomes.
The high cost ofhousing rneans that rnany households rent, rather than own, housing.
The high cost of rnoving means that households tolerate sorne dissatisfaction with
their housing circumstances and make big changes when they rnove. The high cost
of housing also contributes to the problem of homelessness.
434
Part 5 Housing
in the quality and efficiency of kitchen equipment and uti!ity systems (heating, air
conditioning, plumbing, electrical). Other differences occur in the interior design
(type of flooring, windows, cabinets) and structural integrity (the durability of the
foundation and the roof). To summarize, each dwelling offers a different combination
of size, layout, utilities, interior design, and structural integrity.
Because housing is immobile, one component of the housing bundle is the
residentiallocation. A house buyer purchases both a dwelling and a set of site characteristics. One site characteristic is accessibility: sites differ in their access to jobs,
shopping, and entertainment. Another characteristic is the provision of local pubtic services: metropolitan areas have dozens of local governments, each of which
provides a different combination of taxes and public services (schools, fue protection, police services). A third characteristic is environmental quality: sites differ
in air quality and noise levels (from cars, trucks, airplanes, and factories). A final
site characteristic is the appearance of the neighborhood (the exterior features of
neighboring houses and lots). To surnmarize, housing is consumed along with a
residential site, so the housing bundle includes severa! site attributes, including access to different facilities, tax liabi!ities, public services, environmental quality, and
neighborhood characteristics.
435
elementary school. Por example, considera four-bedroom house located three miles
from the city center that has a two-year-old roof. The pollution level is three units
below the average and the average test score of students is two points above the
average. Such a house sells for $85,100, the base price of $70,000 plus $10,000 for
the extra bedroom, $2,000 for the shorter commute, $400 for the four extra years of
roof life, $1,500 for the relatively low pollution leve!, and $1,200 for the relatively
high average test score.
Kain and Quigley (1975) used data from the St. Louis housing market in the
1960s to estimate the dollar values of different housing attributes. Among the attributes they exarnined were the size and quality ofthe dwelling, the exterior quality of
nearby dwellings, and distance to the central business district. Figure 17-1 shows the
percentage changes in monthly rent and house value (for owner-occupied housing)
for one-unit changes in each of these attributes. Por example, a one-unit increase
in interior quality (a measure of the quality of fioors, windows, walls, ceilings,
stairways) increases monthly rent by 2.1 percent and increases market value by
5.6 percent. A one-yearincrease in dwelling age decreases rent and value byrelatively
small amounts.
Neighborhood Effects
The quantity of housing services produced by a particular dwelling depends not only
on the characteristics of the dwelling, but also on the characteristics of its neighborhood. When one homeowner improves the appearance of his house by painting it or
repairing its broken windows, the neighborhood becomes a more desirable place to
live, so the market values of surrounding houses increase. This is the neighborhood
effect: Positive changes in the exterior appearance of one house cause spillover
benefits (inl::reases in market value) for surrounding houses.
The hedonic study of Kain and Quigley also indicates the importance of neighborhood characteristics. The exterior quality of nearby dwellings was measured on
a scale of 1 (bad) to 5 (excellent). As shown in Figure 17-1, a one-unit increase in
the quality of adjacent dwellings increased rent by 3 percent and market value by
5.3 percent. A one-unit increase in the quality of dwellings on the block increased
rent by 6 percent and market value by 2.9 percent. In general, the evidence from
hedonic studies suggests that neighborhood effects are highly localized; the spillover
benefits from positive changes in the exterior appearance of one house are confined
to dwellings within a few hundred feet of the house.
t3
0\
8
:::
."
S' 20.0
15.0
S
"
10.0
5.0
O.Of---'---'
50
-
N
Interior quality
Age
umero
Number of baths
Exterior quality
.
of adacent
Exterior quality
fromCBD
,
'
_j '_) _) } ) _) :_)
437
Disamenities such as air pollution, noise pollution, and proximity to toxic wastes
decrease property value. Pa!mquist (1992) shows that a doubling of highway noise
decreased housing prices by 4.8 percent in upper-middle-class neighborhoods and
3 percent in lower-middle-class neighborhoods. In contrast, prices in lower-class
neighborhoods were unaffected by noise levels. Kohlhase (1991) explores the implications of identifying a site in an urban area as a toxic waste site. The designation of
a site as a Superfund cleanup site reduces property values, with the largest reductions
for properties closest to the site.
Choosing a Dwelling
..
. .
.:.i..c
How does a household choose among alternative dwellings, each of which provides
a different bundle of characteristics? The household must find the dwelling with
the best combination of features at the best price. Most consumers do not have
access to a hedonic study of their housing market. As they shop, they gather their
own information about the implicit prices of location, size, and design features.
Eventual!y, the household choo_ses a bundle that maximizes the household's utility
subject to its budget constraints.
The household chooses the dwelling that provides the best affordable combination of features. Consider a housing market in which dwellings differ in only two
respects: size (square feet of living space) and a local amenity such as quiet surroundings (low levels of highway noise). Figure 17-2 shows the housing choices
for three different households. The budget lines show the combinations of size
and quiet that exhaust a household's housing budget. AB is the budget line for
middle-income households S and L, and CD is the line for H, a high-income
household, The-position of the budget line is deterrnined by the household's income:
the higher the income, the larger the set of affordable combinations of size and air
quality. The slope of the budget line is determined by the irnplicit prices of size and
quiet: the larger the price of quiet relative to the price of size, the flatter the budget
line.
The indifference curves in Figure 17-2 show the different combinations of quiet
and size that generate the same utility leve!for a given household. Three indifference
curves are shown, one for each household. The indifference curves are negatively
sloped, reflecting the household's subjective trade-off between quality and size;
as the local amenity decreases (as noise increases), the household needs a larger
dwelling to achieve the same utility level.
The housing choices reflect the households' preferences and incomes. Given the
shape ofits indifference curves, household L (amiddle-income household) chooses
a large dwelling in a relatively noisy location. Although household S has the same
income and housing budget as L, S chooses a smaller, quieter dwelling. Its choice is
rational because S has relatively strong preferences for quiet, as shown by the shape
of its indifference curve. Household H has a higher income and a larger housing
budget, so it chooses a large dwelling in a quiet area. The high-income household
faces the same monetary trade-off between size and quiet (the same relative prices)
but it has more money to spend on both size and quiet.
438
Part 5 Housino
FIGURE 17-2 Choosing among Dwellings That Differ in Size and Noise Leve!
B
H
We've seen that different housing choices reflect differences in income and
preferences for different housing characteristics. Our example considers housing size
and noise, but the same analysis is applicable for other housing characteristics such
as the quality ofthe dwelling, the exterior quality of nearby dwellings, air pollution,
proximity to amenities, and proximity to jobs, shopping, and entertainrnent. In each
case, a household picks the dwelling that provides the best bundle of characteristics,
given its income and its preferences for housing characteristics.
439
and another market for large ones.Similarly, sorne households look for houses accessible to central-city jobs, while others look for houses in high-quality school districts.
Final!y, sorne households look for new houses, full of modem conveniences, while
others look for older houses with old-world charrn.
Although the housing market is segmented into submarkets, the submarkets
are related because consumers have sorne fiexibility in their housing choices. As
the relative price of three-bedroom houses increases, sorne households switch from
the three-bedroom market to the two-bedroom market. Similarly, as the price of
older houses increases relative to the price of new ones, sorne households switch
to the new-housing market. In other words, dwellings from different submarkets
are imperfect substitutes, so households move between the submarkets in response
to changes in the relative prices of different types of housing. The implications of
housing submarkets are discussed later in the chapter.
DURABILITY OFHOUSING
Housing is more durable than most goods. If a dwelling is maintained properly, it
can last 100 years or more. Although dwellings deteriorate over time, they do so at a
relatively slow rate. The durability of housing has three implications for the housing
market. First, the landlord can control the rate of physical deterioration by spending
money on repair and maintenance. Second, there is a large supply of used housing on
the market every year. The general rule of thumb is that new construction in a given
year is between 2 percent and 3 percent of the total housing stock. Over the course
of a decade, new dwellings provide about 20 percent to 30 percent of the housing
stock, so between 70 percent and 80 percent ofhouseholds live in dwellings that are
at least 1O :Years old. The final irnplication is that the supply of housing is re1atively
inelastic. The market is dorninated by the stock of used housing, so changes in price
cause relatively small changes in the quantity supplied.
r
.
440
Part 5 Housing
Total revenue
Total cost
in year 30
Total cost
in year 20
340 -------------------- -
190 ------------
Q
Quantity of housing services (Q)
Tbe optimum qu.ntity of housing services is the Q at wbich profit (total revenue less
total cost) is maxinllzed. In year 20, profit is maxinllzed at Qzo. Tbe aging of the
dwelling increases maintenance cost, sbifting the cost curve upward and decreasing the
optimum Q.
The variable costs inc!ude repair and maintenance costs. The cost curve is positively sloped because variable costs increase with the quantity of services; it costs
more to maintain a dwelling ata higher quantity of services. The cost curve is convex
from below, esult of climinishing returns to maintenance: as quantity increases, it
becomes progressively more costly to maintain the property at the given quantity.
The straight line in Figure 17-3 shows the reladonship between total revenue (rent
per month) and the quantity of services. The implicit assumption is that the price
per unit quantity is constant; a dwelling with twice the quantity of housing services
commands twice the rent.
Profit equals the difference between total revenue and total cost. In year 20
(20 years after the dwelling was built), the landlord has the lciwer of the two cost
curves. Profit is positive above Q 1, and raches its maximum level at Q20 . At this
quantity, the slope of the revenue curve equals the slope of the cost curve. The
marginal benefit (change in total revenueper unit change in quanticy) equals marginal
'
441
cost (the change in. total cost per unit change in quantity. Therefore, the profitmaxirnizing Jandlord chooses an output of Q 20 units of housing services.
The optimum quantity changes over time, a result of changes in variable costs
or changes in demand.
l. Changes in variable costs. As the dwelling ages, it becomes progressively
more costly to maintain ata given quality leve!. In Figure 17-3, between years 20
and 30, the fixed cost does not change, but maintenance cost increases, causing an
upward tilt ofthe total-cost curve. By year 30, the optimum quantiiy drops from Q2o
to Q3o, and the landlord's profit falls.
2. Changes in demand. Suppose that the demand for housing in aparticular area
drops, decreasing the equilibrium price of rental housing. In Figure 17-3, the totalrevenue function would tilt downward (pivoting from the origin), decreasing the
profit associated with every quantity. As the revenue function tilts downward, the
optimum quantity of housing services decreases.
The Retirement Decision
'.
'
"
'
What happens when the profit from rental housing is negative? Suppose that because
of changes in costs or demand, the total-cost curve lies above the total-revenue curve
for all quantity levels. Renta! housing is clearly unprofitable, Jeaving the landlord
with three options: conversion, boarding up, or abandonment.
l. Conversion. A dwelling can be converted to nonresidential use, such as an
office, a store, or a parking Jot. Conversion is profitable if the alternative activity
generates enough profit to offset the cost of conversion to the nonresidential use. If
conversioh. wold require the demolition of the dwelling and the construction of a
new buildirig, conversion cost is relatively high, and conversion is unlikely.
2. Boarding up. A dwelling can be boarded up and taken offthe market temporarily. The board-up option is profitable if two conditions are met. First, if the landlord
expects the market rent to increase sometime in the future, the expected future profit
is positive. Second, if the landlord's carrying cost (the opportunity cost of keeping
his money in housing instead of a bank account) is relatively Jow, the cost associated
with waiting for the market rent to increase is relatively low. This type of temporary
retirement was common during the Great Depression.
3. Abandonment. The owner can disown the dwelling, walking away from the
property. Abandonment is profitable if the alternative uses (retail, commercial, industrial) do not generate enough profit to cover the cost of converting the property
from residential to nonresidential use. If conversion cost is high, the market value
of the property is zero, so there is no reason to keep title to the land.
Retirement results from three types of changes that decrease the profitability of
rental housing. On the demand side of the market, a decrease in average income ora
decrease in population lowers the demand for housing in certain areas. As the market
rent falls ' the -total-r.e.'venue .c..u.rv. e-"t.i.lts do. wnward, de.. cre..asing .profit. On the supply
side, an incre:se in the supply frental housing decreases market rents and profits.
-
'
442
Part 5 Housing
Finally; as tbe dwelling ages, increases in maintenance cost decrease profits for a
given total-revenue curve, increasing tbe likelihood of conversion or abandonment.
Total cost
--------------------------
280
---------------------
Total revenue
after decrease
indemand
210
190
162
100
1
1
1
1
1
1
1 .
1
1
443
Suppose that the demand for housing decreases, tilting the total-revenue curve
downward. The profit-maximizing quantity drops to Q 2 , and pretax. profit drops to
$48 ($21O - $162).The decrease in profit decreases the market value ofthe property;
an investor is willing to pay less for the property because it yields less profit. In a
perfect tax world, the governm nt would decrease the assessed value of the property
to retlect its lower market valu, thus lowering the landlord's tax liability. If the tax.
stays at $60, however, the posttax profit is negative ($48 - $60 = -$12), so the
landlord is likely to abandon the property. To prevent abandonment, the government
would have to cut the tax liability to less than$48.
White (1986) explores the effects of various factors on abandonment in New
York City and concludes that the property tax is by far the most importan!factor.
She estimates that the elasticity of abandonment with respect to the property tax is
1.65; a 10 percent increase in the property tax increases the frequency of abandonment by 16.5 percent. For example, if the average assessed va!ue of properties in
the Brownsville section ofBrooklyn were cut by $1,000 (a 6 percent reduction) the
resulting decreases in property taxes would lower the abandonment rate from 17 percent per year to 14.8 percent. Given this large elasticity, a tax. cut would generate a
fiscal surplus for the city. Althoi.Jgh the tax.liability per property would decrease, the
direct revenue loss would be offset by (1) an increase in the number of properties on
the tax rolls and (2) a decrease in the number of properties that the city must either
take over or demolish.
Because abandonment is the ultimate neighborhood externa!ity, it contributes
to the problems of central-city neighborhoods. Abandoned buildings provide targets
for vandals and graf:fiti artists, and quickly become eyesores. Even worse, they often
become the temporary homes and retail outlets for transients and drug dealers, so
they contribute-to crime. For these reasons, abandonment decreases the relative attractiveness of a neighborhood, decreasing the rent that other landlords can charge for
their properties. Abandonment feeds on itself, transforming livable neighborhoods
into unlivable ones.
444
Pan 5 Housing
rnaintenance and repair, slowing the deterioration rate and decreasing the nurnber of dwellings retired frorn the housing stock.
3. Remodeling used dwellings. Sorne landlords upgrade their dwellings, increasing the quantity of housing services generated.
Because the bulk of housing is used housing, the supply response will be relatively
large only if the second and third responses (!he used-housing responses) are relatively large.
How elastic is the supply ofusedhousing? Because dwellings deteriorate slowly
over time, a decrease in the deterioration rate has only a small effect on the housing
market. Although an increase in price slows down the deterioration process, the
process is slow in the first place. In addition, rernodeling is extremely expensive,
so it takes a very large price hike to make modifications worthwhile. For these two
reasons, housing supply is inelastic for relatively long periods of time. In other
words, the supply side of the housing market is sluggish; it takes suppliers a long
time to respond to an increase in the dernand for housing. In the rneantime, the price
of housing rernains relatively high.
The same argument applies to a decrease in housing dernand and the resulting
decrease in rnarket price. A decrease in price decreases the incentives for maintenance, so dwellings deteriorate at a faster rate and more dwellings are retired.
Even the fastest deterioration rate is relatively slow, so the decrease in the quantity
supplied is relatively small for relatively long periods of time. Although dwellings
can be converted to other uses, the high cost of conversion inhibits this response.
Because a decrease in price causes relatively small changes in a large part of the
rnarket, supply is relatively inelastic for a long period of time. Therefore, the price
of housing will be relativ ly low for a long period of time.
Whatis the priceelasticity ofthe supply ofhousing? It is difficult to estmate the
supply elasticity because it is difficult to rneasure the quantity of housing services.
The existing studies of housing supply suffer from a number of statistical problems
(see Olsen, 1987 and Quigley, 1979), so theirresu!ts must be interpreted with caution.
Ozanne and Struyk (1978) estimate that the supply elasticity of used housing is
between 0.20 and 0.30. In other words, a 10 percent increase in the market price
increases the quantity of used housing on the market by between 2 percent and
3 percent. Over a 10-year period, miw construction provides only about 30 percent
of the housing stock, so their estimate applies to 70 percent of the housing stock for
a 10-year period. De Leeuw and Ekanern (1971) estimate that the long-run supply
elasticity for renta! housing is between 0.30 to 0.70. In other words, the available
evidence suggests that the supply. of housing is relatively inelastic over relatively
long periods of time.
445
process through which a dwellingpasses from one use to another. The fi.ltering
process has two basic features:
l. Decrease in housing services. The quantity of housing services produced by a
particular dwelling decreases over time. The decrease in quantity results
from physical deterioratien, technological obsolescence, and changes in housing
fashion.
2. Decrease in occupant income. The dwelling is occupied by households with
progressively lower incomes. As the quantity of service decreases, the dwelling
is occupied by households that demand progressively lower quantities ofhousing
services, typically households with lower incomes.
The fi.ltering model can be used to address two questions. First, why do the peor
occupy used housing instead of new housing? Second, do the peor benefit from
subsidy policies that encourage the building of new housing for the wealthy?
Why Do the Poor Occupy Used Housing?
Most poor households occupy used housing. To explain why the peor occupy leftover
housing, considera city with two types ofhouseholds, high income and low income.
Over time, income is increasing for both types of households. Each dwelling in the
city can be described by its size (square feet of space) and its quality, and the quality
of each dwelling decreases over time.
We can use Figure 17-5 to explain the mechanics offi.ltering in the housing market. Consider first the high-income household, whose choices are shown in Panel A.
Given the household's 1992 budget line and its preferences for housing size and
quality, the. hosehold maximizes utility at point H. Between 1992 and 2002, the
household's income increases, shifting its budget line to the northeast. Given its
higher income, the household maxirnizes utility at point l, with a larger and higher
quality dwelling. In the meantime, the household's original dwelling has deteriorated, and now is situated at point U, with the same size but a lower quality.
The high-inconie household has two options. First, it could buy a new house,
built from the ground up with the intention of meeting the household's demand for
a larger, higher-quality dwelling. The new-house option is shown by highest budget
line in Panel A; it allows the household to reach point l. The second option is to
upgrade its original house and i;1crease its size and quality. It is actually very costly to
modify an existing dwelling, so ihe household will get a smaller quantity of housing
service for a given budget. This is shown by the budget line for upgrading, which lies
below the budget line associated with rtew housing. Although the household has the
same income in both cases, the greater expense associated with remodeling means
that the household will get less house from a given budget. Given the two budget
lines, the high-income household will reach a higher utility leve!in a new dwelling.
Consider next the low-income household, which also experiences an increase in
income over the decade. In Panel B ofFigure 17-5, this household starts at point L,
.and its dwelling deteriorates over the decade, moving from p_o_int L to point M.
Like the high-incom househld, the lowcinconie household could either upgrade
Q\
&
&
Budgel: 2002 new
Budgel:1992
Budget: 1992
Size
Budgel: 2002new
Size
The circles indicate dwellings in 1992 nnd the squares indicate dwellings in 2002. Dwellings deteriorate in quality over the decade, as indicated by the arrows.
Panel A: The high-income household is at point H in 1992. Income increases, and the honsehold can eilher upgrade the old house and reach point J or build a
ilew house and reach point /. Given the higher utility level from new housing, the household vaca tes the old dweJiing.
Panel B: The low-income household starts at point L; an increase in income shifts the budget line associated wirh new housing to_the northeast, making point N
possible. The u sed housing vacated by the high-income househoJd is shown by point U. If the price is low enough, rhe used dwelling will be in lhe budget set,
and will generate a higher utility level. The household moves to lhe used house and irs old house (point M) Jeaves the markel.
447
its old dwelling or build a new one. Assuming that upgrading is more expensive
than building a new dwelling, we can ignore the budget line for upgrading, Jeaving
us with the budget line for a new house. The Jow-income household has one more
option; there is used housing (vacated by the high-income household) on the market,
as shown by point U.
Under what conditions will the household be able to afford the used housing
at point U? It depends on the household's income and the price of the dwelling
vacated by the high-income household. Suppose that, given the household's income,
the price of the used dwelling is Jow enough that point U is actually affordable.
In other words, point U is in the household's budget set, along with aJl the points
below the budget line associated with new housing for the low-income household.
The question for the household is,
Which option generales higher utility, new housing that perfectly matches the household's preferences (a point along the new,housing budget Jine) or the used dwelling
(point U)?
Given the household's indifference curves, it will be better off in used housing. Given
its income and the price of new and used housing, utility is higher at point U than
at point N.
We've seen that it is rational to build new housing for high-income households
and use leftover dwellings to house low-income households. This conclusion is based
on two realistic assumptions. First, it is more costly to upgrade housing than to build
new housing; the high cost ofupgrading causes the high-income household to vacate
its old house. Second, the supply of Jeftover housing is Jarge enough that the price is
Jow relative to the cost of new housing. Given the relatively low price, used housing
generates a )l.iglier utility leve!for Jow-income households. By picking used housing
instead of nw housing, a low-income household compromises on its ideal dwelling,
but the Jow price makes the compromise a sensible one..
uy
448
Part 5
Housing
Initial equilibrium
13
20
40
60 80 100
Quantity of medium-quality dwellings
Construction subsidies shift the supply curve for
medium-quality (new) dwellings outward. In
equilibrium, the price drops from $30 to $24 arid the
quantity increases from 40 to 60.
---
20
40
60
80 100
Quantity of 1ow-quality dwellings
The decrease in the price of medium-quality
dwellings increases the supply of low-quality
dwellings: dwellings fi!ter downward and
decrease the demand for low-quality dwellings
as consumers switch from the low-quality
market to the medium-quality market. The price
decreases from $20 to $13 and the quantity
decreases from 80 to 60.
Figure 17-6 shows the initi.al equilibrium in the housing market. In the mediumquality submarket, there-are 40 dwellings ata price of $30 per square foot. In the
1ow-quality submarket, there are 80 dwellings at a price of $20 per square foot.
Suppose that the city subsidizes the constructi.on of new medium-quality housing. In Figure 17-6, the subsidy shifts the supply curve to the right from S1 to
S2 , increasing the quanti.ty supplied at every price: at the original price of $30,
there would be 80 dwellings instead of 40. Since the quanti.ty supplied exceeds the
quanti.ty demanded, the price of medium-quality housing decreases. As the price of
medium-quality housing decreases, two things happen.
Demand effect. The decrease in the relative price of medium-quality housing increases the quanti.ty demanded. Sorne households move from low-quality
dwellings into medium-quality dwellings, causing movement downward along
the demand curve D 1
Supply effect. As the price ofmedium-quality dwellings decreases, the quanti.ty
supplied decreases as sorne landlords downgrade their dwellings to the more
lucrati.ve low-quality subrn:arket. The market moves downward along the supply
curve Sz, decreasing the quanti.ty of medium-quality housing supplied.
In equilibrium, the subsidy increases the quanti.ty of medium-quality housing and decreasesits price. In Figure17-6; the number of medium-quality dwellings increases
.from 40 to 60, and the price decreases from $30 to $24.
449
Substitution effect. As consumers oflow-quality dwellings move into mediumquality dwellings, the demand curve for low-quality dwellings shifts to the left.
This is shown in Figure 17-6 as a shift from D3 to D4 . Because the two types
of housing are substitutes, a decrease in the price of medium-quality housing
decreases the demand for low-qua!ity housing.
Filtering effect. The downgrading of dwellings from the medium-quality submarket to the low-quality submarket shifts the supply curve oflow-quality housing to the right. This is shown in Figure 17-6 as a shift from S3 to S4.
The new equilibrium occurs where the new demand curve (D4 ) intersects the new
supply curve (S4). The subsidy program for medium-quality housing decreases the
price of low-quality housing from $20 to $13 and decreases the quantity of lowquality dwellings from 80 to 60.
The construction subsidies also cause the retirement of 40 dwellings. Since the
population is fixed at 120 hou eholds, every new house causes one retirement. The
number of households in the medium-quality submarket increases by only 20 (from
40 to 60), leaving 20 of the 40 new houses to filter down to the low-quality market.
The number of households in the low-quality submarket decreases by 20 (from 80
to 60), so there are 40 extra houses in the low-quality submarket. Therefore, the
40 lowest-quality houses are retired from the housing stock.
450
Part 5 Housing
Cr = V Ur + dr + mr- gr)
where ir is the interest rate paid by the property owner, dr is the depreciation rate for
renta! housing, m, is the maintenance cost, and gr is the rate of capital gain (increase
in market value). Suppose that V= $50,000, i= 10 percent, dr = 1 percent, and
mr =4percent, and g =O percent. Ifso, the annual costis $7,500, the sum of$5,000
in capital cost, $500 in depreciation cost, and $2,000 in maintenance cost. In a world
without laxes, the landlord charges an annual rent of $7,500, thus covering her costs
and making zero econornic profit.
451
What happens if the market value ofthe dwelling increases over time? If the
dwelling appreciates over time instead, g, will be positive and d, will be zero. For
example, if the market value increases by 8 percent per year, g, is 0.08, and the
landlord earns $4,000 per year in appreciation. In this case, an annual rent of $3,000
is enough for the property owner to break even. The sum of the capital gain of $4,000
and rent is equal to the total cost of $7,000 ($5,000 in capital cost plus $2,000 in
maintenance cost). The equilibrium rent is lower because the landlord earns money
by owning the property, and therefore needs less rental income to generate zero
econornic profi.t.
The Cost of Homeownership Consider next the cost of awning a home. In the
absence of laxes, the homeowner incurs the same types of costs as the landlord
(capital, depreciation, and maintenance). If the homeowner borrows money to buy
his home, the capital cost is the annual mortgage-interest payment. If he pays cash
for the house, the capital cost is the opportunity cost of investing his money in a
house rather than a bank account. In either case, the capital cost is the market interest
rate (i) times the value of the house (V). The homeowner also incurs depreciation
cost (d timesV) and maintenance cost (m timesV). The maintenance cost includes
the opportunity cost of time spent fixing faucets, painting walls, and keeping the
books. The annual cost of ownership is
Co = V Uo +do
+ mo - go)
where i 0 is the interest rate paid by the homeowner, do is the depreciation rate for
owner-occupied housing, m 0 is the maintenance cost for owner-occupied housing,
and g0 is the ca_pital gains rate.
\
-.
--.,'
452
Part 5
Housing
rate. In other words, the renter externality increases the market rent on residential
property.
The cost ofrenting is relatively high for two other reasons. First, renta! property is a riskier investrnent (it has a higher default rate), so lenders charge higher
interest rates on loans for renta! property. Second, sorne horneowners use work on
the house (rnaintenance, repair, and yard work) as a hobby, so they have a lower
rnaintenance cost than a landlord. Horneowners often use work on the house as a
rneans of accurnulating wealth in their spare time. Por these reasons, the annual cost
of horneownership is often less than the annual rent on an equivalent renta! dwelling,
even without considering the tax advantages of horneownership.
What type of household rents instead of owning? Three factors tend to push
sorne households toward renting. The first is low incorne; to qualify for a home
loan, the household rnust have enough incorne to pay the rnortgage. The second is
rnobility; given the substantial transactions cost of executing ahorne loan, frequent
rnovers tend to rent rather than own. The third is a distaste for work on the horne and
yard; if a person is unwilling to do everyday rnaintenance and small repairs, renting
is relativelyattractive.
' Jr; ,
'-''1\f[c-c
/3 1B Li1]---'-lE:,-:.
-A
Income
$20,000
30;000
50,000
_ . 453
I\JQR_ -F
Marginal
Tax Rate
Mortgage
Interest
Tax
Benefit
Nonitemizer
. $ 5,000
7,000
11,000
$1,050
3,080
3.50
6.16
15%
28
Benefit as Percent
oflncome
454
Part 5 Housing
0.12
H*
Housing consumption
H'
The optimum housi"ng consumption occurs at point e, where the marginal social benefit (shown
by the demand curve) equals the marginal social cost. If the marginal tax rate is 28 percent, the
mortgage subsidy decreases the marginal private cost of housing from $1 to 72 cents, increasing
housing consumption from H to H'. Consumers pay less !han !he marginal social cost of
housing, SO !bey co sume!OC much housing and divert investment from other uses.
'.
cost drops from $1 to 72 cents, the household increases its housing consumption from
H* to H'. This is inefficient because the household's benefit from increased housing
consumption (from the demand curve) is less than the opportunity cost of spending
on housing (from the marginal social-cost curve). By spending more on housing,
society has less to spend on factories, machines, and schools. The household makes
an inefficient choice because the govemment pays 28 percent of its housing bill.
Struyk, Mayer, and Tuccillo (1983) suggest that the mortgage deduction has
contributed to recent declines in U.S. productivity. The idea is that the deduction
encourages the investment in housing at the expense of industrial investment, so
there is less capital equipment (factories and machines) per worker. Rosen (1979)
estimated the effectofthe mortgage deduction on housing consumption in the 1970s,
before the 1981 and 1986 tax changes. He estimated that the deduction increased
housing consumption by about 14 percent, suggesting that a substantial amount of
capital is diverted to residential uses.
455
deduction for homeowners is neutral with respect to the consumer's choice between
owning and renting?
To explain the bias toward ownership created by the mortgage deduction, consider Fred the homeowner and Bamey the landlord. All dwellings in city B are
identical rock hovels that sell for $2,000. Fred has two housing options. First, he can
borrow $2,000 from the bank to purchase a home. If the interest rate is 1O percent,
the mortgage-interest payment for a perpetualloan is $200 per year, so the cost of
ownership in the absence of tax benefits is $200. Second, he can rent an identical
dwelling from Bamey. Since the houses are made of stone, there are no maintenance or repair costs, so Bamey's only expense is the capital cost of $200 per year.
The market rent ($200) is the same as the cost of ownership, so Fred is indifferent
between owning and renting.
The mortgage deduction creates a bias toward homeownership because it allows
Fred the homeowner to deduct expenses that do not generate taxable income. Contrast the mortgage deduction for Fred (homeowner) with the mortgage deduction for
Bamey (landlord). Bamey's deduction is sensible because Bamey declares his rental
income as taxable income. He declares $200 of rental income per house, and deducts
$200 in mortgage cost. His taxable income from his property is zero because his deduction equals his rentalincome.In contrast, Fred does not declare any rental income
from his property, but still deducts his mortgage cost. Fred's mortgage deduction
does not offset rental income, but simply subsidizes the cost of homeownership.
The government could e!irninate the tax bias toward homeownership in one of
two ways. The simple and obvious one is to eliminate the interest deduction for
homeowners. An alternative is to force Fred to declare his imputed renta! income
as taxable income. The irnputed rental income is the income eamed from owning a
dwelling and renting it to yourself. Alternatively, it is the money you could earn if
you rented your dwe!ling to someone else. Fred's imputed rental income is $200 per
year. If he were to declare $200 of irnputed rental income and then deduct his $200
mortgage cost, the two items would cancel one another, and the ownership bias
would disappear. His taxable income would be the same whether he rents or owns,
so he would be indifferent between renting and owning.
456
Part 5 Housing
Second, what is the appropriate fonn for the ownership subsidy? Does the present
system, under which the wealthiest households receive the largest subsidies, internalize the externality in the most efficient manner? There have been a number of
proposals to modify the mortgage deduction. One option is to replace the deduction
with a tax credit equal to a fixed percentage of the household's mortgage cost. For
example, if the tax credit is 15 percent, every household would receive a tax cut equal
to 15 percent of its mortgage cost. Under a tax credit, all taxpayers would receive the
same percentage subsidy for homeownership. A second option is to place a ceiling
on the amount of mortgage interest that can. be deducted. This would decrease the
subsidy to the wealthiest households.
457
.
"O
o
o
"'
.e
"o
';;
..o..
..=o,
c.
=o
S1 S2 S
Heusing censumptien (square feet)
An increase in incerne shiftS the budget !ine upward. In fue absence ef meving cost, the heuseheld
wei.lld meve from point b te peint d, meving te a larger heuse (Sz square feet ef living space). If
moving is cestly,the heusehold stayin its present heuse, cheesing e instead ef d. If inceme
increases again and the househeld dees net meve, it cheeses peint e. The heuseheld meves te a
larger dwelling (point f) if the resulting increase in utilicy is large enough to offset moving cost.
458
Part 5 Housin o
to !2 shifts the budget line upward, giving the household two choices. One option
is to stay in its original house and spend al!the extra income on nonhousing goods
(point e). Altematively, the household could move toa larger house (Sz square feet
instead of S1 square feet), choosing point d. In the absence of moving cost, utility
would be higher at point d (point d lies on a higher indifference curve), but the
questionis whether the increase in utility is large enough to justify the cost of moving.
If the moving cost is relatively high, the household stays in its original house and
chooses point e.If income increases to fs,the budget line shifts upward again, and the
household faces the same choice:Stay put (point e), or pay the moving cost and move
to a larger hou.se (point f). Because the gap between the initial housing consumption
(S1) and the desired housing consumption (S3 ) is relatively large, the household is
more likely to move than it was when the gap was relatively small (Sz - S1).
There are two lessons from Figure 17-8. First, a small change in income (or
price) is unlikely to change housing consumption. Most households tolerate sorne
dissatisfaction with their housing circumstances because moving is costly. A household moves to a different dwelling only if the change in income or price is large
relative to moving cost. Second, when a _household changes houses, it is likely to
make a large change in its housing consumption. A household moves when its dissatisfaction with its current dwelling is large enough to justify the cost of moving.
459
$10,000 prize in a bank account yielding 1O percent interest, her annual income from
the prize would be $1,000, meaning that the bingo prize increases her permanent
income by 5 percent. The income elasticity would then be computed as 1.0 (the
5 percent increase in housing consumption divided by the 5 percent increase in
permanent income).
What is the income elasticity of demand for housing? There have been dozens
of studies of housing demand, and there is a consensus on three points. First, the
overall income elasticity is about 0.75 (Ellwood and Polinski, 1979); a 10 percent
increase in income increases housing consumption by about 7.5 percent. Second, the
income elasticity for renters is less than the income elasticity for owner-occupants.
Third, the income elasticity increases with income. According to Thlanfeldt
(1982), the elasticity for low-income households is between 0.14 and 0.62, and the
elasticity for high-income households is between 0.72 and 1.10.
The results of hedonic studies can be used to estmate the income elasticities of
demand for individual components of the housing bundle. Follain and Jiminez (1985)
summarize the results from several hedonic studies and cometo three conclusions:
l. The demand forliving space is inelastic with respect to income; in seven ofnine
cases, the elasticity is below 0.46.
2. The demand for structural quality is highy elastic with respect to income; in
four of five cases, the elasticity exceeds 1.64.
3. The demand for neighborhood amenities such as public safety is highly elastic
with respect to income.
The hedonic studies suggest that, compared to the demand for housing in general,
the demand fo.: living space is less income-elastic and the demands for structural
quality and-neighborhood amenities are more income-elastic.
Price Elasticities
What is the price elasticity of demand for housing? Most estimates of the price elasticity fall between -.75 and -1.20 (Ellwood and Polinski, 1979). The consensus is
that demand is slightly price-ine!astic (an elasticity slightly less than l.O in absolute
value). This means that an increase in price increases total expenditures on housing
by a small amount; an increase in price decreases the quantity demanded by a slightly
smaller percentage amount, so total expenditure (price times quantity) increases by
a small amount.
THE HOMELESS
The McKinney Homeless Assistance Act of 1987 defines a homeless person as
someone who sleeps (a) outside, (b) inside in places not intended for sleeping (e.g.,
the lobby of a public building), or (e) in housing shelters (places providing temporary
housing). A consensus estmate of t.lJ.e number of people who are homeless on a given
rught is between 250,000 and 350,000 (Honig and Filer, 1993).
460
Part 5 Housing
The rest of the homeless population was adults traveling together (with relatives or
nonrelatives). Children make up about 15 percent of the homeless population.
Figure 17-9 shows the percentages of the homeless population that experience
various difficulties. About half are high school dropouts, and about one in :five has
FIGURE 17-9 Elasticities for Homelessness
!.50
1.25.'
1.00
0.50
Employment
growthrate
Z;>
:g
Size of
welf.rre
payment
Institutionalization
of mentally ill
0.22
0.00 J--.L....L...---,-,-----r--r--::::..::::.;.::.:::::.-=---L-L-_
Rent for
City size
" _ _
0 50
!ow-quality
housing
-0.15
-0.32
-1.00
-1.50
-1.54
-2.00
Source: Author's calculations b ed on data in Marjorie Honig and Randall K...Filer. "Causes of Imercicy
Variation in Homelessness." Amencan Economic Review 83 (1993), pp. 248-55.
461
spent sorne time in a mental hospital. About one in tbree homeless people has
spent sorne time as a patient in a chemical-dependency program. Over half of the
homeless has spent time in jail (a short period for misdemeanor violations), and
about a quarter has spent time in prison (longer periods for felony violations). About
one in five homeless people hasattempted suicide.
What causes homelessness? From an ecor10mic perspective, a person will be
homeless if bis or her income is low enough relative to the price of housing tbat it
is not sensible-or not even possible-to purchase housing services. A study of
the causes of homelessness bears out this simple theory (Honig and Filer, 1993).
Looking at data for different metropolitan areas, we see tbat the homeless rate varies
across metropolitan areas, and so do several variables tbat that affect a household's
ability to purchase housing services. The question is, do variations in these other
variables explain the observed variation in homeless rates?
Figure 17-10 shows the sensitivity of homeless rates to changes in several
variables. The numbers shown are the elasticities of the homeless rate with respect
to selected variables. For example, the elasticity for rent on low-quality housing is
1.25, meaning that a 10 percent increase in rent on low-quality housing increases tbe
homeless rate by 12.5 percent. the sample of cities, the average rent on low-quality
housing is $135, so a rent hike to $148 would increase the average city's homeless
rate from 186 per 100,000 population to 209.
One factor in the growth of the homeless population in the 1970s and 1980s
was a reduction in the stock of low-quality housing in large cities (Burt, 1992).
Historically, many of the poor lived in boarding houses, rooming houses, and residential hotels (also known as single-room occupancy unj.ts or SROs). This housing
In
r--:
49
50 r-
40-
...
.....
"'a 30 -
.C
21
20-
....
10 -
e-
19
r-
1):
dropouts
in mental
hospital
--
:. .
'--
,,
-- L-L-,Spent time
in prison
Attempted
suicide
Source: Author's calculations based on data in Marcha R. Burt, Over the Edge: The Growrh of Homelessness
in 1M 1980s (New York: Sage, !992).
462
Part 5 H01.ising
had small living quarters, shared bathrooms, and !imited kitchen facilities, but it
was affordable for most pensioners and day laborers. During the 1970s and 1980s,
much of the SRO stock was eliminated, a result of urban renewal and conversion
to other uses. Many of the hotels and large houses used for boarding or rooming
houses had been "rejuvenated" for higher-income households. The percentage loss
in SRO units was 19 percent in Chicago, 50 percent in Los Angeles, 38 percent in
New York, 42 percent in Cincinnati, and 25 percent in San Diego. Nationwide, a
total of 1.1 million SRO units were lost during the 1970s and 1980s. The loss of
SROs contributed to the homeless problem by decreasing the supply of low-quality
housing and increasing its price.
The numbers in Figure 17-1O indicate that changes in income affect the number
of homeless people. The elasticity for employment growth suggests that a 1O percent
increase in a city's employment growth rate will decrease the homeless rate by
1.5 percent. This is sensible because employment growth results in more jobsand higher wages-for the working poor, so they are more likely to be able to
pay for housing. The elasticity for welfare payments suggests that cities with more
generous welfare programs have lower homeless rates; a 1O percent increase in
welfare payments decreases the homeless rate by 15.4 percent. This is sensible
because welfare recipients can use the additional money to pay for housing.
The elasticity for institutionalization of the meritally ill indicates a negative relationship between institutionalization and homelessness. A 10 percent increase in
the institutionalization rate (the number of patients in mental hospitals per 100,000
population) leads toa 3.2 percent decrease in the homeless rate. This result is consisten!with the idea that policies concerning the mental!y ill affect the number of
homeless people. In the 1970s, many mentally ill people were released from state
mental hospitals with the-idea that they would live in the community and be treated in
community centers. A common perception is that many of the people relesed from
mental hospitals ended up homeless. Figure 17-10 suggests that this perception is
accurate.
SUMMARY
l. Housing is different from other products: it is heterogeneous, immobile, and
durable. Housing is costly in two senses: it is a large part of the household
budget, and changes in housing consumption are costly.
2. The hedonic approach is based on the notion that a dwelling is composed of a
bundle of individual components, each of which has an implicit price.
3. Housing is durable, and the landlord can control the rate of physical deterioration
by spending money on repair and maintenance.
4. The supply of housing is relatively inelastic for relatively long periods of time
because the bulk of the housing stock is used housing.
5. The filtering model explains how the quantity of housing services generated by
a dwelling decreases over time, so the dwelling is occupied by households with
progressively lower inconies.
463
6. The costs of rental housing and owner-occupied housing are determined by the
cost of capital (the opportunity cost of money invested in housing), depreciation,
and maintenance cost.
7. Three factors push many households toward homeownership: the renter externality (tenants have little incentive to inform landlords of maintenance problems), higher interest rates for rental property loans, and the hobby aspect of
homeowners' repair and maintenance.
8. The federal tax code has special tax breaks for housing. Rapid depreciation
of renta! property defers taxes, increasing the profitability of renta! property
and decreasing monthly rent. The deductibility of mortgage interest decreases
the cost of homeownership, increasing ownership rates and increasing housing
consumption.
9. The cost of moving is relatively la:rge, so households change their housing
consumption infrequently and make large changes when they move.
10. The factors that cause homelessness include high prices for low-quality housing,
low welfare payments, low rates of institutionalization for the mentally ill, and
slow growth in employme t.
\ .
l. Consideran apartment building that does not generate enough rental revenue to
cover the landlord's costs.
a. Use a graph to show this situation (see Figure 17,.-3).
b. Suppose that the govemment provides a 50 percent subsidy for operating, maintenance, and repair costs. Show the effects of the subsidy progratn on your graph. What happens to the optimum Q and the likelihood of
retirement?
2. Suppose that you own an apartment building that does not generate enough renta!
revenue to cover your costs. You have two options: First, you could demolish
the building and set up a parking lot; second, you could abandon the building.
Suppose that the cost of demolition is $20,000 and the annual profit from the
parking lot would be $!,500. Should you demolish or abandon? If you don't
have enough information to answer this question, list the information you need
and explain how you would use it.
3. Dwellings are eventually retired from the housing stock as a result of either
decreases in market rent (a downward tilt of the total-revenue curve in Figure
17-3) or rising maintenance cost(an upward tilt of the total-cost curve).
a. Use a graph like Figure 17-3 to prove or disprove the following statement:
"A dwelling that leaves the market as a result of decreases in market rent
leaves the market aa higher quantity of housing services (a larger Q) than
one that leaves as a result of increases in maintenance cost."
b. Explain why this statement is true or false.
4. In city E, all householdshave 1:he .same income and .real income is constant over
time. Will there be filtering in the city?
464
Part 5 Housing
465
e. What is the consumer surp!us at the new price, assuming that the household
moves to a new residence?
f. If the cost of moving is $25, will the typical household move to a new
residence?
12. In city D, landlords charge a security deposit of $900 and do not refund the
money when the tenant leaves. Suppose that a new deposit law is passed. The
law requires that alllandlords refund 100 percent of the security deposit unless
the landlord can prove that the tenant damaged or dirtied the dwelling. Comment
on the following statement: "The new law will save the typical tenant $900."
13. Comment on the following statement: "I live in an apartrnent complex built in
ayear when interest rates were re!atively low. My rent is lower than the rent on
other apartments because the cost of financing the complex is relatively low."
REFERENCES
l. Burt, MarthaR. Over the Elige: The Growth ofHomelessness in the 1980s. (New
466
Part 5
Housing
12. Kohlhase, Jane E., The Impact ofToxic Waste Sites on Housing Values. Journal
of Urban Economics 30 (1991), pp. 1-26.
13. Olsen, Edgar O. "A Competitive Theory of the Housing Market." American
Economic Review 59 (1969), pp. 612-22.
14. Ozanne, L., and Raymond Struyk. "The Price Elasticity of Supply of Housing
Services."In Urban Housing Markets: Recent Directions inResearch and Policy,
ed. L.S. Bourne and J. R. Hitchcock. Toronto: University ofToronto Press, 1987.
15. Palrnquist, Raymond. "Valuing Localized Externalities." Journal of Urban
Economics 31 (1992), pp. 59-68.
16. Quigley, John M. "What Have We Learned about Housing Markets?" In Current
Issues in Urban Economics, ed. Peter Mieszkowski and Mahlon Straszheim.
Baltimore: Johns Hopkins University Press, 1979.
17. Rosen, Harvey S. "Housing Decisions and the U.S. Income Tax: An Econometric
Analysis." Journal of Public Economics 67 (1979), pp. 1-23.
18. Struyk, Raymond J.; Neil Mayer; and John A. Tuccillo. Federal Housing Policy
at President Reagan's Midterm. Washington, DC: Urban Institute, 1983.
19. White, Michelle. "Property Taxes and Urban Housing Abandonment." Journal
of Urban Economics 20 (1986), pp. 312-30.
CHAPTER 18
Housing Policies
Tru.s is the second of two chapters on the urban housing market. It discusses three
types ofhousing policies. The fustis housing assistance; the federal government uses
various policies to improve the housing conditions and decrease the housing cost
of poor households. Second, the federal governrnent uses a number of community
development prograrns to support local efforts to improve housing conditions and
revitalize neighborhoods. Third, under a rent-control policy, the local govemment
sets a maximum prce for renta! housing.
The federal government provides housing assistance through a number of prograrns administered by the Department ofHousing and Urban Development (HUD).
In 1997, HUD's prograrns for renta! housing provided subsidies for about 4.8 million low-iil.corrie households. About two-thirds of the households were served by
policies that operate on the supply side of the market; the government either builds
new housing for the poor (public housing) or subsidizes prvate developers to build
and manage low-income housing. About 1.4 rnillion households occupied dwellings
in conventional public housing projects, and another 1.8 rnillion households lived in
prvatey! owned housng projects that were subsidized by the federal govemment. In
contrast, under a demand-side policy, the government hands out housing vouchers
(coupons) that low-income households can use to help pay for prvatey! owned renta!
housing. In 1997, about 1.6 rnillion households received housing vouchers. In recent
years, the emphasis of housing policy has shifted away from supplying housingeither directly or indirectly-in favor of housing vouchers. Most vouchers are now
portable; they can be used anywhere in a particular metropolitan area.
Why do governments around the world intervene in housing markets, providing
low-income households with public housing or housing vouchers? As we'll see, a
dollar spent on housing policies generales a benefit to the recipient, but the benefit is
much less than a dollar. Why don't govemments simply give money to low-income
households and let them spend it themselves? Why notjust give cash?
One defense for housing policy is a practica! one. Although a cash transfer to a
low-income household may increase its welfare by a larger arnount, most taxpayers
467
468
Part 5 Housing
do not support a policy of giving cash, preferring instead to provide specific goods
such as housing and food to ensure that the money is spent on goods deemed worthwhile by the donor. If cash transfers are impossible for political reasons, it may be
better to give housing rather than nothing at al!.
The second defense for housing policy concerns the externalities associated with
housing consumption. In the previous chapter we discussed neighborhood effects:
the property values of individual properties depend on the exterior appearance of
nearby properties. Low-income households may economize on housing to free up
money to spend on other necessities, generating neighborhood externalities. In sorne
neighborhoods, the externalities associated with dilapidated and abandoned buildings are severe. In addition, most communities have building codes and zoning that
outlaw the sort of high-density, low-quality housing that low-income households
might otherwise demand. The absence of low-quality dwellings raises housing costs
for low-income househo!ds, and a system of housing subsidies can malee housing
more affordable.
---'--
469
TABLE 18-1 The Condition of the Housing
Stock and Housing Cost Burdens, 1991
Percent of Households
Al/ households
Inadequate
Crowded
Cost-burdened
Total with problems
Poor households
Inadequate
Crowded
Cost-burdened
Total with problems
7.9
2.7
28.0
38.7
17.4
7.5
62.1
87.0
Metropolitan. residents
Inadequate
Crowded
Cost-burdened
Total with problems
7.3
2.9
29.6
39.8
Black households
Inadequate
Crowded
Cost-burdened
Total with problems
17.4
5.2
38.0
60.6
In the last several decades, there has been considerable progress in improving
the condltion of the housing stock. The percentage of households in overcrowded
housing dropped from 13.5 percent in 1950 to 2.9 percent in 1991. The percentage
ofurban households living in "substandard" housing (defined differently from inadequate housing) was 21.9 percent in 1950, but the percentage in inadequate housing
was only 7.3 percent in 1991.
Neighborhood Conditions
Table 18-2 provides evidence concerning the leve!of neighborhood dissatisfaction
for various types of households. The first column of numbers shows the percentage
of households of each type that gave their neighborhood a score of 3 or less on
a 10-point scale. The greatest dissatisfaction is among households living in multifamily dwellings in central cities. For example; 16.8 percent of low-income black
households living in multifarnily housing in the central city said that their neighborhood was of "poor" quality and 36.1 percent of these households said that the
neighborhood was so bad that they wanted to move. Among similar households
living in single-farnily housing in the suburbs, only 3.3 percent considered their.
neighborhood to be of poor quality and only 13.6percent said that they wanted
to move. The figures for Jow-income white households are 11.2 percent (poor
4
470
Pan 5 Housing
Household type
Black, low, multi, city
Black, high, multi, city
Black, high, multi, suburbs
Black, low, multi, suburbs
White, low, multi, city
White, high, multi, city
B!ack, low, single, city
Black, high, single, city
White, higb, multi, suburbs
Wbite, Jow, single, city
White, higb, single, city
Black, high, single, city
White, low, multi, suburbs
Black, low, single, suburbs
White, high, single, suburbs
Whi te, low, single, suburbs
Neighborhood
Quality Is Poor (%)
Want to
Move (%)
16.8
9.7
6.3
10.5
11.2
36.1
32.2
5.7
10.3
5.4
3.9
6.5
3.3
3.3
5.2
3.3
1.8
3.4
25.7
25.6
22.3
2!.8
19.7
17.4
16.3
15.9
15.4
14.9
14.1
13.6
10.4
10.0
Definitions
Poor quality: score 3 or less on a 10-point quality scale.
.quality) and 22.3 percent (want to move) for multifamily housing in the central city,
and 3.4 percent (poor quality) and 10.0 percent (want to move) for single-farnily
housing in the suburbs. Neighborhood dissatisfai::tion varies with race and income:
blacks are genera!ly more dissatisfied than whites, and low-income households are
more dissatisfied than high-income households.
Boehm and Thlanfeldt (1991) explore the effects ofvarious neighborhood characteristics on the perceived quality of the neighborhood. Their results suggest that
dissatisfaction is most prevalen!in neighborhoods with high crime rates, high noise
levels, run-down and abandoned buildings, and large quantities of trash, litter, and
junk. One implication from Table 18-2 is that the inciJence of neighborhood dissatisfaction is higher among black households. Boehm and Thlanfeldt show that the
most irnportant factors in explaining the difference.in .satisfaction are differencs in
(1) crime rates, (2) the amount oflitter, trash, andjunk, (3) the quality ofbuildings,
and (4) the number of abandoned buildings.
471
.5
12
"" 10
6
Source: Figure from ''The Revelati.on ofNeighborhood Preferences: An N-Chotomous Multivariate Probit
Approach," by Thomas P. Boehm and KeithR.lhlanfe1dt inJoumal ofHousing Economics 1, pp. 35-59,
@ 1991 Elsevier Science (USA), reprnduced by permission of the publisher.
Figure 18-1 shows the results of the following thought experiment. Suppose
that the crime rate in the typical black neighborhood changes to the crime rate that
prevails in the typical white neighborhood. By how much would the gap between
black and white dissatisfaction decrease? The answer is that the gap would decrease
by 12 percent. Figure 18-1 shows the results of repeating this thought experiment
for the other factors that contribute to neighborhood dissatisfaction: Abandoned
buildings are responsible for 14 percent of the gap; and litter, trash, and junk are
responsible for another 17 percent.
472
Part 5 Housing
housing athorities to give priority to househo1ds with "very 1ow" incomes (1ess than
half the median income of the area).
500
Uz
400
..
o
u ---Kr .
B
350
"O
-5
o
;z
1
1
1
L
1
1
1
1
1
1
1
1
1
1
1
1
1
600 800
2,000
Housing consumption (units of housing service). _
Income is $500 and the price of housing is 25 cents per unit of housing service. With
.. the original budget line, the house old chooses point B. publicbousing (apartments
generating 800 units of service at acost of $100) adds point C to the budget set. Public
housing allows the household to increase its consumption ofbothhousing and other goods.
473
. Suppose that the typical public dwelling is an apartment that generates 800 units
of housing service and is provided to the poor for $1OO. Public housing adds point e
to the household's budget set; the household can consume an 800-unit apartment and
have $400 left to spend on other goods. Given the household's indifference curves,
utility is maximized at point e, so the household moves into public housing. The
public housing program increases housing consumption (from 600 to 800 units of
housing service) and decreases housing costs (from $150 to $100).
Will public housing always increase housing consumption? Figure 18-3 shows
the ehoices of a household with the sa.,-ne income as the household in Figure 18-2, but
different tastes for housing. In the absence of public housing, the household lives in a
1,000-unit apartment (point D). The household spends $250 onhousing (1,000 units
times 25 cents per unit), leaving $250 for other goods. If the household moves into
public housing, it occupies a dwelling with fewer units ofhousing ervice (800 units)
and has an additional $150 to spend on other goods. Public housing increases utility
FIGURE 18-3 Public Housing Decreases Housing Consumption
$
500
Additiona! budget point
with $100 public housing
400
-----
< 250
l1
1
1
1
1
1
1
1 .
1 .
1
1
1
1
1
--
800 1,000
Housing consumption (units of housing service)
2,000
The household starts at point D (1.000 square feet of housing and $250 of other goods).
Public housing adds point C tb the budget se t. If the household occupies public bousing,
it decreases its housing consumption fiom 1,000 to 800 and increases its consumption of
other goOds from $250 to $400.
'
1<
474
Part 5 Housing
if the additional $150 worth of nonhousing goods is worth the sacrifice of 200 units
ofhousing service. Given the indifference curves in Figure 18-3, utility is higher at
point C than at point D, so the household accepts the offer of public housing and
decreases its housing consumption.
500
400
"O
o
o
350
00
o
-"
600 800
2,000
Housing consumption (units of housing service)
2,400
The alternativo to public housing (800 units of housing service ata price of $100 instead of
$200) is a $100 cash paymenL Under public housing, the household chooses point C. The
$100 cash payment shifts the budget line up by $100, and the household chooses point E.
. Utility is higher tii:tder the casb payment because th household candecide how to spend the
money.
-
475
575
U1
1
1
1
1.
1
1
1
1
1
1
....'
600 800
Housing consumption (units ofhousing service)
Public bousing ($100 for an 800-unit apartment) is equivalen!te a $75 cash payment. A $75
payment allows the household te choose point F..;. which is en e ame indifference curve as
.. point C (the point cbosenwith publichousmg).be household !S tndifferent between $] 00
public housing anda cash payment of $75.
476
Part 5 HouSfug
maxin:llzes utility at point F. Since the household reaches the indifference curve U2 ,
it is indifferent between $75 in cash and the $100 housing subsidy. Studies by Kraft
and Olsen (1977), Barton and O!sen (1976), andMurray (1975), suggestthat tenants
receive a benefit of about 75 cents per dallar spent on public housing. More recently,
Smeeding (1982) estimated that the benefit per dollar is 80 cents.
How does the cost of new public housing compare to the cost of private housing?
Public housing is more expensive for two reasons. First, the private sector can build
new low-income housing more efficiently than the public sector: Weicher (1979)
cites a number of studies showing that new public housing costs more than new
private housing. Second, there is a plentiful supply of used low-quality housing, so
even the least costly new housing costs more than used housing. According to the
National Housing Review (1974), the cost ofnew public housing is about 33 percent
higher than the cost of used private housing. If new public housing costs one-third
more than used private housing, the government spends $266 (not $200) to provide
a new 800-unit apartment. Therefore, the government's cost of getting to point C in
Figure 18-4 (the public housing option) is $166 ($266 !ess the $100 rent paid by
tenants). In contras!,the cost of getting to point E (the cash option) is only $100, so
the cash option generates a larger increase in utility for about three-fifths of the cost.
What is the bang per buck of public housing? In other words, what is the recipient
benefit per dallar spent on public housing? As shown in Figure 18-5, recipients
receive a benefit of $75 from public housing, so the bang from public housing is
$75. If the government spends $266 to build the apartment and collects $100 in
rent from the tenant, the government's cost is $166. Therefore, the bang per buck is
45 cents ($75 divided by $166).
477
FIGURE 18-6 Crirne Rates in Chicago Housing Projects
90
84
70
"
f:1
60
50
41
""
a
: 40
" 30
;>
22
20
10
"""
...
"
"8
o"'
:
"'
o,.,
"'
::
"'
"f:1
"
;::
.e
u"'
<>
-g
""
o
"'
:"2
cii
.""
::">
Source: Newsweek, January 2, 1989, p. 25.@ 1989 Newsweek, Inc. All rights reserved. Reprinted
by permission.
the price is $250 per dwelling per month. If the local housing authority builds 200
new public housing dwellings, the demand curve for private housing shifts to the left
by 200 dwellings: the 200 households in public housing leave the private housing
market. As the poor move into public housing, they vacate private dwellings, causing
excess supply that decreases the price of housing from $250 to $150 (point C). In
the short run, al! housing consumers gain from the public housing program: sorne
households occupy subsidized public housing, and others pay lower prices for
private housing.
What happens to the private market over time? In Figure 18-7, the interrnediaterun supply curve (the supply curve over a five-year period) is positively sloped. The
decrease in the price of housing caused by public housing decreases the profitability of private dwellings, so fewer low-quality dwellings are supplied. The supply
response takes two forms:
M re retirement. As explauied in the previous chapter, a d crease in the price
of low-quality dwellings makes economic profits negative for sorne dwellings,
478
Part 5 Housino
Private supply
after 5 years
1?
o -----------
g_
208
O Jl
'1
'1
.1
1
1
;;;
_"g
"o 150 -------
"
1
1
1
-------\---- e
1
1
1
1
1
1
1
1
1
1
i'
', !
; '
Original
demand
Demand after public
housing
500
300 380
Number ofprivare low-income dwellings
Public housing decreases the demand for privare low-income housing by 200 dwellings.
In the short run, the supply of privare housing is fixed, so the price of privare dwellings
drops from $0 to $!50. Over a five-year period, the decrease in the price ofprivate
housing decreases the quantiry of housing supplied from 500 to 380, so the price rises ro
$208. The longer the period of time, the greater the displacement of privare housing and
the smaller the ntdecrease in price.
so more dwellings are retired from the housing market. The retired dwellings
are either converted to another use or abandoned.
Slower downward filtering. As the price of low-quality housing decreases
relative to the price of medium-quality housing, landlords hold their dwellings
in the medium-quality submarket for a longer period of time.
In other words, more dwellings drop out of the low-quality submarket and fewer
dwellings filter down into the low-quality submarket. As the number of low-quality
dwellings decreases, the market price rises. The exit process continues until the
price is high enough to restare zero economic profits (normal accounting profits).
In Figure 18-7, a total of 120 private dwellings are withdrawn from the market, and
the price rises from $150 to $208. Because public housing displaces private housing,
there is a net increase in supply of only 80 dwellings.
Ohls (1975) used a computer model ofthe housing market to explore the effects
of public housing on the filtering process. He divided the housing stock into 60 quality
levels, and allocated housing in each leve!to the city's households: the highestquality dwellings were occupied by the wealthy; the lowest bythe poor. Under a
479
public housing policy, the bottom fifth of the population is removed from the private
housing market, causing the following changes:
l. The amount of new private construction decreased: the decrease in the demand
. for used low-quality housing lowered the incentive to build new housing.
2. Fewer dwellings filtered down to the low-quality submarket: the low-quality
market became less lucrative, so the filtering rate decreased.
3. The retirement rate increased: the decrease in the demand for low-quality
dwellings lowered the incentive to keep low-quality dwellings on the market,
so the retirement rate increased.
In general, the private market responded by building fewer dwellings, slowing the
filtering process, and increasing the retirement rate.
--480
Pan 5
Housing
,,
1'
Por examp!e, suppose that the fair market rent is $400 and the househo1d's income
is $600. The househo1d's contribution is $180, and the subsidy equals the difference
between the actual rent (up to a maximum of $400) and $180. If the actual rent is
only $350, the subsidy is $170.
Figure 18-8 shows the househo1d's choice between housing and other goods
under the nmt-cert(fi.cate program. Household income is $600 per month, and the
fair market rent is $400.-The original budget line AB shows the initial trade-off
between spending on.housing and other goods: every dollar of spending on housing
decreases spending on other goods by one dollar. The certificate program adds the
points between e and D to the budget set. The household receives a subsidy equal
to the gap between its actual spending on housing (its monthly rent) and 30 percent
of its income ($180). Therefore, it can consume $420 worth of nonhousing goods
and between $180 and $400 worth of housing. If the household rents an apartment
. for $181, it receives a subsidy of $1; if it rents an apartment for $400, it receives a
subsidy of $220. Because the government pays the di:fference between actual rent
and $180, an increase in housing consumption does not decrease the consumption of
other goods as long as rent is less than the fair market rent. If the household spends
more than the fair market rent ($400), it is ineligible for the certificate program, so
it is stuck with the original budget !ine.
How does the certificate program affect the household's budget choice? In the
absence of moving cost, a move from E to D increases utility.It makes sense to go all
the way to D because between points e and D the opportunity cost of more housing
is zero. The tenant contribution to housing is fixed at $180 and the government pays
any increase inrent. Ifmoving cost is relatively1arge, however, the household moves
straight up from E to F. In this case, the increase in utility is not large enough to
offset the moving cost, so the household stays in its original residence and spends
the entire subsidy on other goods.
481
FIGURE 18--8 Effects of Section 8 RentCertmcate on Housing Consumption
$
600 A
-----+F-----
1
1
1
---------\-----
1
1
1
1
1
1
1
1
1
1
1
180
300
400
Spending on housing (monthly rent)
B
600
With !he original budget hne AB, !he household spends $300 on rent and $300 on other
goods (point E). The rent-certificate program with a fair market rent of $400 adds points
between e and D to !he budget set: tbe government pays !he difference between actual
rent and 30 percent ofincome ($180), so tbe opportunity cost of spending beyond $180
is zero. In tbe absence of moving cost, tbe household switches to point D because it lies
on a higher indifference curve. lf tbe moving cost is relatively Iarge, the household
chooses point F.
One of the surprising results of the rent-certificate program was that relatively
few households moved to dwellings that rented for the fair market rent. In Figure 18-8, most households moved from point E to F, notE to D. This occurred
despite the fact that the entire in9rease in rent would have been covered by the
subsidy. Most households stayed in their existing dwellings, and used the subsidy to
cut their housing costs. This result suggests that the cost of moving is relatively large.
Housing Vouchers
The housing-voucher program was started on an experimental basis in 1983. The eligibilityrequirements fQrhousing vopchers are similar to those for rent certificates; the
bulk of payments go to the very poor. To qua!ify for a voucher, the household must occupy a dwelling that meets the minimum quality standards. A voucher differs from a
482
Part 5 Housing
rent certificate iri one irnportarit respect: The tecipient can use the voucher to rent any
dwelling that meets minirnum physical standards. In 1997, about400,000 households
received housing vouchers. The federal budget for fiscal year 1999 included 50,000
new vouchers (at a cost of $283 million) for households affected by the welfarereform package passed in 1996. These vouchers are "portable" in the sense that they
can be used anywhere in a particular metropolitan area. The idea behind making them
portable is to allow the recipients to move to residential areas accessible to jobs.
The face value of the voucher is based on household income and the fair market
rent. The formula is
Pace value = Fair market rent - 0.30 Income
Figure 18-9 shows the household's choice between housing and other goods under
the voucherprograrn. Household income is $600 and the fair market rent is $400, so
FIGURE 18-9 Effects ofHousing-VoucherProgram onHousing Consumption
$
A
F
600 --------
-g
\
.oS
=
o
""
=_
420
'6
"' 300
.1l
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
B
220 300
400
600
Spending on housing (montbly rent)
820
Witb the original budget Iine AB, tbe household spends $300 on rent and $300 on otber
goods (point E). The voucber program witb a fair market rent of $400 shifts tbe budget
Iine outward to AFG. The housebold receives a voucher wortb $220 {tbe difference
between tbe rent standard and 30 percent of income). In tbe absence of moving cost,
tbe housebold. cbooses point H, and is better off than it would be under a rent-certificate
prograrn (point D). Ifthe cost ofmoving is relatively Jarge, tbe householdcbooses point J.
483
the face value of the voucher is $220 ($400 -$180). The voucher program shifts the
budget line from A B to AF G. Point F is in the new budget set because the household
could use the voucher to get $220 worth of housing and spend all of its own income
($600) on other goods. As spending on housing rises above $220, there is a dollarfor-dollar trade-off between housing and other goods. The voucher program gives
the household more options than the rent-certificate program because the voucher is
equivalent to a cash transfer. In the absence of moving cost, the household with the
choices shown in Figure 18-9 increases spending on housing from $300 (point E)
to about $345 (point H), leaving $175 of the subsidy to spend on other goods. If
the cost of moving is relatively large, the household ehooses point J and spends the
entire subsidy on other goods.
The voucher household also has the option of spending more than the fair market
rent. In other words, the household has access to points along the budget segment
DG. In contrast to the rent-certificate program, the voucher program allows the
household to spend as much as it wants on housing.
,.
l. Downward filtering. Dwellings filter down more rapidly from the mediumquality submarket to the more Jucrative moderate-quality submarket.
2. Increased maintenance. The owners of moderate-quality dwellings spend
more on rriaintenance and repair to keep their dwellings in the market for a
longer period of time: downward filtering is slowed.
4
484
Pan 5 Housinero
350
292
.:;
o
250
R
";;"
6
Demand with
housing vouchers
"o'
u
"
Original demand
500 620
700
Number ofmoderate-qua!ity dwellings
The housing-voucher program increases the demand for moderate-quality dwellings. In the short
run, the supply of moderate-quality housing is fixed, so the price increases from $250 to $350. Over
a five-year period, the increase in the price of housing increases the quaniity supplied from 500 to
620, and the price drops to. $292. The longer the period oftime, the greater the increase in the
quantity suppliedan.thesmaller the net increase in price.
485
Ohls (1975) used.a computer model of the housing market to explore .the effects
of consumer subsidies on the filtering process. As explained earlier, his model divides
the housing stock into 60 quality levels, with the wealthy occupying the highest
quality dwellings and the poor occupying the lowest-quality dwe!lings. Under his
. consumer subsidy program, the government provides a 50 percent housing subsidy
for the poorest fifth of the population. The subsidy program causes the following
changes in the market:
l. The poorest households upgrade to higher-quality housing, moving from quality
leve!5 to 8. The typical recipient moves from quality leve!8 to 11.
2. The increase in the demand for moderate-quality housing stirnulates new construction (at the high-quality leve!) because the market for used housing becomes
more lucrative.
3. Filtering down to the moderate-quality range speeds up, and filtering down from
the moderate-quality range slows down.
L.Low. participation rates.]3etween 40 percent and 45 percent of the eligible households participated in the allowance programs. Participation rates were
e
486
Part 5 Housina
o
sensitive to the quality standards; the stricterthe standards, the lowerthe participation
rate.
2. Small changes in housing consumption. Between one-quarter and one-third of
the allowance payment was used to increase housing consumption, and the remainder
was spent on other goods. In other words, most of the allowance money was used
to decrease rent burdens. The measured income elasticity of demand of allowance
recipients was relatively low: between 0.20 and 0.44 for renters, and about 0.45 for
owner-occupants.
3. Mobility. Relatively few households changed dwellings. In Pittsburgh, 40 percent of allowancerecipients moved over a two-yearperiod, compared to 35 percent of
households in a control group. In Phoenix, 63 percent of recipients moved, compared
to 53 percent of the control households. The allowance prograrn did not encourage
integration with respect to race or income.
The general conclusion is that the housing allowances caused relatively small changes
in housing choices. Recipients spent most of the money on nonhousing goods, and
few moved from their original dwellings.
The allowance prograrn did not cause any measurable price increases in the two
cities involved in the supply experiment. Because participation rates were relatively
low and recipients spent a relatively small fraction of their allowances on housing,
the rightward sbift of the demand curve was relatively small. To the extent that
the demand for housing increased, the supply of housing was sufficiently elastic to
accommodate the increased demand without any measurable price effects.
What are the lessons from the supply experiment? As pointed out by Kain (1981),
the supply experiment tested the effects of one type of allowance prograrn, with
ffiinimal housing qualicy standards and relatively low participation rates. Because
the allowance prograrn increased the housing consumption of a small number of
households by relatively small arnounts, it is not surprising that housing prices did
not rise. If a national housing allowance prograrn had bigher participation rates,
larger allowance payments, and more stringent housing standards, it would sbift the
housing demand curve by a larger arnount and would probably increase housing
prices. Kain cites studies by the National Bureau of Economic Research and the
Urban Institute that suggest such an allowance prograrn would increase prices in
sorne housing submarkets.
487
The legislated mandate for federally supported community development cal!s for
"systematic and sustained action by the federal, state, and local governments to elirninate blight, to conserve and renew older areas, to improve the living environment of
low- and mciderate-income farnilies, and to develop new centers of population growth
and economic activity." The two principal purposes of community development policies are to (1) revitalize declining areas of the city, and (2) irnprove the housing of
poor households. Community development policies are neighborhood-oriented in
the sense that they concentrate spending in relatively small geographical areas.
Urban Renewal
Urban Renewal, the first community development program in the United States, was
established under the Housing Act of 1949 and eventually dropped in 1973. The
national government provided locaJ. governments with the power and the money to
demolish and rebuild parts of their cities. Local agencies acquired property under
the right of eminent domain, cleared the site of "undesirable" uses (like low-income
housing and small businesses), and then either built a public facility or sold the site
to a private developer. The local agencies charged the developer less than the cost
of acquiring and clearing the site, and the federal government covered two-thirds
of the local government losses. The private developer built housing (usually for
middle-income and high-income households), government buildings, or commercial
establishments.
488
Part 5
Housing
According to the Congressional Research Service, Urban Renewal had the following effects:
l. Demolition. A total of 600,000 dwellings were demolished, displacing about
two million people, most of whoin were poor.
2. New housing. A total of250,000 new dwellings were built, most ofwhich were
occupied by middle-income and high-income households.
3. Public facilities. The total fioor space of new public facilities was 120 mil!ion
square feet.
4. Commercial facilities. The total fioor space of new co=ercial facilities was
224 million square feet.
5. Property assessments. The assessed value of property on renewed sites increased by 360 percent.
In sum, Urban Renewal generated both costs and benefits. It displaced poor households, but also provided housing for middle-income and rch households, allowed
the construction of co=ercial and public facilities, and increased tax revenue.
Was Urban Renewal worthwhile? The crtics of the program focus on its demolition aspects, pointing out that two million poor people were displaced. The
defenders of the program focus on its rebuilding aspects, pointing out that the new
co=ercial developments provided jobs for the poor residents of the central city.
Recent Community Development Programs
More recentfederal co=unity development programs have avoided many of the
probleins of the Urj:>an Renewal program. The newer programs are executed on a
srrialler scale, so they displace fewer households, and they place a greater emphasis
on providing housing for the poor.
The bulk of funding for co=unity development is for a program called CornmunityDevelopmentBlock Grants (CDBG).In1997, the CDBG budgetwas $4.7 billion, with about 70 percent of the funds going to central cities and urban counties, and
the remainder going to smaller localities. The allocation formula used to distribute
CDBG funds gives large grants to cities withrelatively old and overcrowded housing,
high poverty rates, and slow economic growth. The funds are used to improve housing, support public servces, promote economic development, and clear land for new
development. Under the Urban Development Action Grants (UDAG), funds are used
to leverage prvate investment in co=unity development. The governrnent
provides small subsidies to transform slightly unprofitable prvate development
projects into profitable ones.
Cities spend their co=unity development grants in a wide variety of projects.
The projects can be divided into four types:
l. Improved housing. Sorne programs support local projects that renovate buildings, enforce building codes, and build new low-income housing. The Boston
Housing Partnership ilsed $8.5 million in CDBG funds to acquire and rehabilitate 700 rental units. The city of Cleveland joined with a nonprofit organization and
489
a private developer to build a 183-unit rental housing project, using $4.4 rnillion in
federal funds ($1.7 rnillion frorn CDBG and $2.7 rnillion frorn UDAG) to leverage
$9.9 rnillion frorn private and other sources.
2. Renewed infrastructure. Sorne programs support projects that improve streets,
. roads, and water and sewage facilities. In Fort Worth, Texas, the Main Street Project
used $2.6 rnillion in UDAG funds to improve the street's infrastructure (new curbs,
gutters, sidewalks, and street lighting). As part of the renovation program, private
developers agreed to build a $48 million hotel. The city of Oshkosh, Wisconsin,
used $600,000 of CDBG and UDAG funds to expand its airport to accommodate
increased traffic.
3. Job developrnent. Sorne programs support projects that encourage econornic
development and jb creation. In St. Louis, the Union Station was renovated with
$10.2 million in CDBG and UDAG money and $126 rnillion in other public and
private funds. The station was developed into a hotel-retail entertainrnent complex
that produced 2,000 new jobs, over half of which were filled by rninorities. The
project rejuvenated a part of the city that had been dormant for 17 years. Salt La.k:e
City is using a $1.1 rnillion UDAG to support the construction of a downtown auto
mall, using the grant to leverage $4.7 rnillion in private financing. Baltimore used
$10 million in UDAG funds to support the revitalization of its Inner Harbor. Prvate investrnent in of:fice space, retail space, and a hotel totaled $73.4 rnillion and
produced 1,600 jobs.
4. Responsive public services. Sorne programs support public services for the
elderly, the homeless, and children. Orange County, California, used $892,000 in
CDBG funds to support the construction of a $7.5 rnillion facility to provide emergency services for abused, abandoned, or neglected children. Phoenix used $361,000
in CDBG funds to build a recreation center for the handicapped. Fort Worth, Texas,
used $106,000 in CDBG funds to start a neighborhood youth project designed to
curtail gang activity.
These examples show that community development funds are spent on a wide variety
of projects.
.
What are the effects of community development programs? The programs that
support hciusing projects supplement federal housing programs (public housing, subsidized new construction, housing vouchers). The difference is that the city has more
control over the spending of community development funds. Other community development programs support econornic development projects and can be evaluated
as alternative means of promoting local econornic growth. The costs and benefits of
econornic development programs are d.iscussed in Chapter 6 (Urban Econornic
Growth).
RENT CONTROL
. During World War TI, the federal governrne Jt instituted a national system of rent
controls. Following the war, New York City was the only city to retain rent controls.
490
Part 5 Housing
During the 1970s, rent controls returned to dozens of cities, including Boston;
Cambridge; Los Angeles; Washington, D.C.; Albany; Berkeley; and Santa Monica.
According to the U.S. Department ofHousing and UrbanDevelopment (1991), more
than 200 localities had rent contro1s in 1991, and over 10 percent ofthe nation's private rental housing was subject to these controls.
Figure.18-11 shows the short-ron and long-ron effects of rent control. The short
ron is defined as the period of time over which the supply of housing is fixed at
FIGURE 18-11 Short-Run and Long-Run Effects of Rent Control
Sbort-iun
supply
Long-ron
supply
\
600
-------------------- B
e
Demand
Shortage
50
90
120
Number of apartments
In the short run, the supply of apartments is fixed, so rent control (maximum rent $400)
simply transfers income from landlords to tenants: the ritarket moves from point B to e.
In the long run, the decrease in price decreases the quantity suppliedt:he market moves
from point e to point D. The shortage (the gap between deinand and.supply) is 70
apartments.
l"
l..
:e
1
Jrt
at
491
90 dwellings.In the short run, rent control does not affect the supply ofhousing, but
simply transfers $200 per month from the landlord to the tenant.
In the long run, the supply of apartments depends on the price of renta! housing.
As rent decreases, sorne landlords convert their property to more lucrative uses;
sorne apartment buildings are c0nverted to condorniniums, and other buildings are
demolished to make way for commercialland uses. In Figure 18-11, rent control
decreases the number of apartments from 90 to 50. As the price of renta! housing
decreases, the market moves downward along the long-run supply curve. The supply
response to rent control depends on the long-run elasticity of supply. The more elastic
the supply, the larger the decrease in the quantity of housing supplied.
Rent control causes a shortage of housing. At the controlled price of $400, the
quantity of apartments demanded is 120, compared toa supplyof only 50 apartments.
There are too many tenants chasing too few apartrnents. In such a market, tenants
spend relatively large amounts of time and money searching for apartrnents. Por
tenants, rent control brings good news and bad news: although rent control decreases
the rent, it also increases search costs.
Rent control produces winners and losers. The city's population can be divided
into four groups:
l .. Nonmovers. Sorne households live in the city when rent control is implemented
and occupy the same apartment for life. These households are winners in the rentcontrol game: they pay a lower price for the same apartment.
2. Successful searchers. Sorne people move to the city after rent control is irnplemented. They spend severa! months searching for an apartrnent and eventually
find one. The full cost of a rent-controlled apartment is the sum of the rent ($400 per
month) and thopportunity cost of search time. The full cost of a rent-controlled
apartrnent may exceed the full cost of an uncontrolled apartrnent ($600 per month
and small search cost). In general, if the search cost is relatively large, the benefit
of rent control (decreased rent) is smaller than the cost (increased search cost), so
successful searchers are losers in the rent-control game.
3. Displaced households. Sorne tenants are displaced when apartrnents are converted or demolished, and are unsuccessful in their search for a new apartrnent.
In Figure 18-11, the number of dwellings decreases by 40 (from 90 to 50), so 40
households are displaced. The displaced households are obviously losers.
4. Landlords. Rent control decreases the annual renta!income per apartrnent from
$7,200 ($600 times 12 months) to $4,800. Because the market value of an apartment
building reflects its income-earning potential, the announcement of the rent-control
program decreases themarket value ofthe city's apartrnent buildings. IfBamey owns
an apartment building when rent control is announced, he pays for rent control in
the form of a decreased market value for the building. He cannot escape the effects
of rent control by selling the building to someone else: once prospective owners
bear about rent control, they want to pay less for the building. According to Srnith
and Tomlinson (1981), Toronto's rent-control program decreased the market value
of apartment buildings by about 40 percent over a five-year period (1975-1980).
Olsen (1972) estimated the effects f New York City's rent-control program
on tenants and landlords. Ignoring search cost, he estimated that the occupants of
492
Part 5 Housing
controlled units experienced a 3.4 percent increase in real income. On the other hand
landlords experienced losses in real income about twice as large as the total gain of
tenants.
Households in
controlled units
2,000
."!:i. 1,800
1,600
".:; 1 400
l"
(l '
All renter
households
!:i 1,200
j
"C
'"' 1,000
"'
All housebolds
800
>11
600
400
200
10
20
30
Annual income ($1,000)
40
50
Source: Fi.gure frorn ''quit)...ad Efficiy Aspecrs ofRt Control: .AJJ. Empirical Study of New York City," by
Joseph Gyourko and Peter Linnernan in Joumal of Urbcm Econ.offlics 26, pp. 54-74, @ 1989 Elsevier Science
(USA), reproduced by permission ofthe publisher.
493
The middle line in Figure 18-12 shows the benefit-income relationship for
all renting households. The average benefit among renting households is about
$1,311 per year.The average benefit among renting households is smaller than the average benefit among the occupants of controlled units because the benefits are spread
over more households: only 57 percent of renting households occupy controlled
units. The benefit per household decreases as income increases, refiecting the fact
that the average income of households in controlled units is lower than the average
income of renting households. Poor households are over-represented in controlled
dwellings, so poor households, on average, experience larger benefits from rent
control.
The lower line in Figure 18-12 shows the benefit-income relationship for all
households, including both renters and homeowners. The average benefit among all
households was about $856 per year. The average benefit among all households is
relatively small because less than a third of the households occupy controlled units.
The benefit decreases as income increases, refiecting the relatively low income of
households in controlled units.
Figure 18-12 suggests that the benefits of rent control are mildly progressive
in the sense that the average benefit per household decreases as income increases.
It's important to note, however, that the benefits experienced by households within
an income group vary widely across households. The variation in benefits across
households within an income group occurs for two reasons. First, only sorne of
the households of a given income leve!occupy controlled dwellings (and receive
benefits from rent control). The other households (renters of uncontrolled units and
homeowners) receive no benefits from rent control. Second, among households that
occupy controlled dwellings, there is a large variation in the benefits from rent
control. Given fue large variation in the benefits within each income group, rent
control is a very blunt too!for income redistribution.
'1r
494
Pan 5 Housina
o
Studies of rent-control cities show that landlords do indeed cut back on maintenance and repair. The Rand Corporation studied the effects of rent controls in
New York City (1970), and conc!uded that rent control decreased the quality and
quantity of renta! housing. Between 1960 and 1967, the inventory of "sound" housing increased by 2.4 percent, the inventory of "dilapidated" housing increased by
44 percent, and the inventory of "deteriorating" housing increased by 37 percent.
Between 1965 and 1967, 114,000 dwellings were retired from the housing stock.
A study of Cambridge, Massachusetts, found that maintenance cost per apartrnent
decreased by $50 per year (Navarro, 1987).
Gyourko and Linneman (1990) exarnined the effects of rent control on the quality
of renta!housing in New York City in 1968. They show that rent-controlled buildin os
are less likely than uncontrolled builclings to be in "sound" condition, meanin"o
that rent-controlled builclings are more likely to be deteriorating or dilapidated".
One measure of the effect of rent control is the decrease in the probability that a
building is sound, given that the building is subject to rent control. Por example,
if the probability of being sound is 0.90 for uncontrolled buildings (90 percem of
uncontrolled builclings are sound) but only 0.85 for controlled buildings, rent control
decreases the probability ofbeing sound by 0.05.
Figure 18-13 showsthe decrease in the probability that a building is sound for
severa! types oflow-rise buildings (less than seven stories) in Manhattan, Brooklyn,
FIGURE 18-13 Rent Control and the Probability that Building Is Sound
0.10
[1
Manhattan
O Brooklyn
"O
Queens
0.08
"g
.S
..1l 0.06
o
:-':=:'
.0.04
0.00
Pre-1947
19 7-59
1960-68
Year of coostructioo
Source: Figur from "Rene Controls and Rental. Housing Quality: A Note 00 the Effects of New
York City's Old Controls," by Joseph Gyourko and Peter Linneman in Joumal of Urban
Economics 27, pp. 398-409, 1990 ElsevierScience (USA), reproduced by permission ofthe
publisher.
'\;.
-. .'
- -...
-
495
and Queens. For buildings in Manhattan built before 1947, rent control decreases
the probability by about 0.09 (from 0.63 to 0.54). Newer buildings, which are more
likely to be sound, experience smaller decreases in the probability of being sound.
For buildings in Manhattan built between 1947 and 1959, rent control decreases the
probability of being sound by about 0.04 (from 0.89 to 0.85); for buildings built
between 1960 and 1968, rent control decreases the probability of being sound by
0.01 (from 0.98 to 0.97). The effects of rent control are smaller in the two other
boroughs of New York.
''
't,:
'
(!,,(
l'
1. -,
'
l.
l
l
Housing Submarkets
One of the assumptions of the rent-control model is that all households live in
identical apartrnents. If there are housing submarkets, the analysis changes in two
ways.
Consider first the possibility that there are two controlled submarkets. There are
large apartrnents (equilibrium rent of $600 and controlled rent of $400) and small
apartrnents (equilibrium rent of $300 and controlled rent of $200). Rent control
inhibits movement between the two markets because the search cost is large. Suppose
that a household would like to move from a large apartrnent to a small one. In the
rent-control city, there are two options. First, the household can stay in their large
apartrnent, paying $400 per month. Second, they can spend time and money searching
for a small apartrnent (which rents for $200). If the search cost is relatively high,
they are better off staying in their original apartrnent.
Considernext the possibility that sorne ofthe dwellings in the city are not subject
to rent control. Rent control decreases the supply ofhousing in the controlled sector.
The displaced..hoiiseholds move into the uncontrolled submarket, bidding up the
price of housing. Therefore, part of the cost of rent control is borne by households
living in uncontrolled dwellings.
.1
J
l
1
1
1'
(
J
1'
1
(
1.:
r'
l r.
'
ce.,
'
496
Pan 5 Housino
o
control. In fact, tenants displaced from the controlled market increase the demand
for new renta! housing, stimulating the supply of new renta! housing. It is possible
that new housing offsets the loss in housing stock caused by rent control. But tbere
are two reasons to suspect that new housing will not offset the effects of rent control.
First, new housing is typically a small part of the housing market, so it would take
a relatively large increase in the supply of new housing to counteract tbe effects of
rent control. Second, builders and owners are likely to be skeptical about the city's
promise to never impose rent control on tbis year's new renta! housing. If builders
suspect that today's new housing will become tomorrow's rent-control housing, they
may be reluctant to supply new housing.
3. Restrictions on conversions. Many cities restrict the conversion of renta!housing to condominiums. If successful, these restrictions limit the conversion of renta!
housing to other uses, diminishing the supply effects of rent control.
4. Subsidies for new construction. Public policies encourage the building of new
dwellings to replace the dwellings lost as a result of rent control. The city ofToronto
offered low-interest construction loans to housing firms. In the United States, one of
the purposes of public housing is to replace the dwellings lost because of rent control.
5. Vacancy decontrol. In sorne cities, rent is fixed during a tenancy, but can be
changed when a new tenant moves into the dwelling. This provision increases the
landlord's renta! income, weakening rent control. It also increases the incentive for
long-term occupancy.
Another way to decrease the supply effect of rent control is to cheat. Tenants
and landlords often find ways to violate the spirit of rent-controllaws. Landlords
charge tenants "key money" (e.g., $5,000 for akey to the apartment) or nomefundable
security deposits: To thextent that landlords use indirect payments to increase the net
price ofhousing, the negative supply response to rent control is reduced. In the city of
Cairo, which has had rent control since 1944, key money is an accepted practice. On
average, payments to property owners in the form ofkey money increase the effective
rent from about 38 percent ofthe market price to about 71 percent (Malpezzi, 1998).
Tenants also collect bribes for rent-controlled apartments; when a tenant vacates
a controlled apartment, he can charge the new tenant a finder's fee. The fee is often
implicit; in California, old tenants reportedly sold bean-bag chairs to new tenants,
charging hundreds of dollars for chairs that sold for justa few dollars in thrift stores.
Tenant bribery does not increase the supply ofhousing, but simply transfers income
from the new tenant to the old one.
Altematives to Rent Control
In many cities, rent control is viewed as an indirect form of income redistribution.
The idea is to take money from wealthy landlords and give it to poor tenants. At
best, rent control is a blunt instrument for redistribution. Rent control helps al! of
the occupants of controlled dwellings, rich and poor alike. Moreover, a poor person
who occupies a controlled dwelling is worse offunder rent control if the search cost
is larger tha:n tbe savings in rent.
497
Suppose that a city wants to redistribute income from the wealthy to the poor.
What are the altematives to rent control? The first option is to finance income
redistribution with a land tax. Because the supply of Iand is perfectly inelastic, the
Jand tax wi1l not affectthe supply ofland orthe supply ofhousing. Therefore, the land
tax will not cause the samesupply problems as rent control. The second approach is
to leave income redistribution to the federal govemment. Local attempts at income
redistribution are likely to be frustrated by the movement of households into and out
ofthe city.Ifthe city gives the proceeds of a Iand tax to its poorresidents, poor people
from other cities may rnigrate to the city, diluting the effects of the redistribution
program. Because intemational mobility is less than intercity mobility, a national
redistribution program would not suffer from this problem.
SUMMARY
l. Although the condition of the U.S. housing stock has improved over the last
. i
498
Pan 5 Housing
499
' ':
8. Suppose that the poor households of city H are emolled in a federal housingallowance prograrn. You are a middle-income renter in the city.
a. Will you ?e helped or harmed by the allowance prograrn?
b. If you Wlil be harmed, what variables determine how much you will be
harmed? In other words, what are the relevant elasticities, and how does the
cost you incur vary with them?
c. How would your answer to (a) change if you owned your home?
9. Contrast the effects of ren1; control in two cities. In city R, the elasticity of supply
of housinis 5.0 (i. ., a 10 percent increase in the price of housing increases
the quantity of housmg by 50 percent). In city S, the supply of housing is less
elastic; the supply e!asticity is 0.40.
10. Cheating on rent control is common. If you were a prospective tenant in a
rent-controlled city, would you encourage your prospective landlord to cheat?
.11. Figure 18-13 suggests that the effects of rent control increase with the age of
the building. Provide an explanation for this result.
12. Discuss the trade-offs associated with demolishing the high-rise clusters of public housing and replacing tl:em with dispersed low-rise buildings.
REFERENCES
l. Barton, David M., and Edgar O. Olsen. "The Benefits and Costs of Public
Housing in New York City." Institute for Research of Poverty Discussion Paper
373-76, Madison: University ofWisconsin, 1976.
2. Boehm, Thomas P., and Keith R. Iblanfeldt. "The Revelation of Neighborhood
Preferences: An N-Chotomous Multivariate Probit Approach." Journal of Housing Economics 1 (1991), pp. 33-59.
3. De Salvo, J.S. "Reforming Rent Controls in New York City." Regional Science
Association Papers 2 (1971), pp. 195-227.
4. Gyourko, Joseph, and Peter Linneman. "Equity and Efficiency Aspects ofRent
Control: An Empirical Study ofNew York City." Journal oj Urban Economics
26 (1989), pp. 54-74.
5. Gyourko, Joseph, and Peter Liruieman. "Rent Controls and Rental Housing
Qua_Jity: A Note on the Effects of New York City's Old Controls." Journal of
Urban Economics 27 (1990), pp. 398-409.
6. Kain, John F. "A Universal Housing Allowance Prograrn." In Do Housing Allowances Work? ed. Katherine Bradbury and Anthony Downs. Washington, DC:
Brookings Institution, 1981.
7. Kraft, John, and Edgar O. Olsen. "The Distribution of Benefits from Public
Housing." In The Distribution ojBenefitsfromPublic Housing. Vol. 41 of Studies
in Income and Wealth, ed. F. Thomas Juster. New York: National Bureau of
Economic Research, 1977.
8. Malpezzi; Stephen. "We!fare Analysis ofRent Control with Side Payments: A
. NaturalExperiment in C.airo, Egypt." Regional Science and Urban Economics
. 28 (1998), pp. 773-95.
500
::
:i:
.,
t
. :,!
'
.1
:1
1
.1
1
. '1
'
Part 5 Housing
PART SIX
Local Government
f
l
1
1
Tms
part of the book explores the spending and taxing programs of local governrnents. Chapter 19 examines the role of local governrnent in the federal systern
of governrnent and takes a closer look at local decision-rnaking through elections.
Chapter 20 eals with local revenue sources, examining the issue of who actually
pays the local property tax and exploring how local governrnents respond to intergovernrnental grants from higher levels of governrnent.
1
1
j
1
l
501
CHAPTER 19
Tru.s chapter provides an overview of the spending policies of the local public
sector. The first part of the chapter presents the facts on local government, including
inforrnation on spending programs and revenue sources. The second and third parts
discuss the economic rationale for local government, explaining why goods such
as public schoo!ing, public-safety services, parks, and transit systems are produced
by local governrnents rather than private firrns or higher levels of governrnent. The
fourth part explores local decision-making through voting.
.
:; -
503
'1
504
Units in 1992
Percent of Local
Expenditures
County
Municipal
Township and town
School district
Special d\strict
3,043
19,279
16,656
14,422.
31,555
23
33
3
30
11
Total
84,995
100
Sources: U.S. Bureau of the Census, Stariscical Abstraer ofthe United States 1997, Table
474 (Washington, OC: U.S. GovernmentPrinting Office, !998); U.S. Bureau of Census,
CensusofGovernm.erns, Govemment Finances 1991-92 (Washington, DC: U.S.
Al! Local
Governments
Municipa!ities and
Townships
Education services
Education
Libraries
Social services
Public welfare
Hospitals
Health
Transportation.
Highways
Other
'
Public safety
Police protection
Fire protection
Other
Erwironment and housing
Narural resources
$961
18
$81
10
!33
18
56
33
30
12
108
40
64
19
123
59
51
90
47
16
.2
:'54
64
. 81
45
1
36
35
51
31
125
General expenditure
120
58
73
Source: U.S. Bureau ofthe Census, Compendium ofGovernment Finances, no. 5 (Washington, DC: U.S.
Government Printing Office, 1997).
school districts (30 percent of local expenditures). As shown in Table 19-2, the
largest spending category for local government is education, with $961 per capita,
about 42 percent of total spending. The spending of municipalities is more evenly
divided, with the largest expenditures on police protection, education, highways,
sewage, and fue protection.
sos
THE ROLE OF LOCAL GOVERNMENT
What is the role of local government in the market economy? This question can be
answered by asking two separate questions. First, what is the role of government
in the market economy? Second, how does local government fit into the general
scheme of government involvement?
Musgrave and Musgrave (1980) distinguish between three types of government
policy. The first is stabilization policy: the government uses monetary and fiscal policy to control unemployment and inflation. Under the second type of policy, income
redistribution, the government uses taxes and transfers to alter the distributions of
income and wealth. The third policy is resource allocation: the government makes decisions about what to produce and how to produce it. When the government actually
produces a particular good or service, it makes these resource-allocation decisions
directly. When the government subsidizes or taxes private activities, it influences the
resource-allocation decisions of the private sector.
How do local governments fit into this three-part scheme of governmental activity? The responsibility for stabilization policy has been assi.lmed by the national
government for two reasons. First, although each local government could print its
own money and execute its own monetary policy, such a system would be chaotic.
Instead, the national government prints the money and manages a national monetary
policy. Second, because a large fraction of local income is spent on goods produced
outside the local area, local monetary and fiscal policies would be relatively weak
and ineffective. Considera city that wants to stimulate the local economy, and tries
to do so by cutting its taxes without changing its spending on local public goods.
Tbis deficit-spending policy will be ineffective because a large fraction ofthe tax cut
will be spent-on imports, so the tax cut will have a relatively smalllocal multiplier
effect. Fiscal policy is more effective at the nationallevel because a relatively small
fraction of national income is spent on imports, so a national tax cut has a relatively
large multiplier effect.
Consider next the distribution role of government. Local attempts to redistribute
income will be frustrated by the mobility of taxpayers and transfer recipients. Suppose that a city imposes a tax on its wealthy citizens and uses the tax revenue
to finance transfer payments to the poor. To escape the tax, sorne wealthy households willleave the city, causing a decrease in total tax revenue. At the same time,
sorne poor households will migrat to the relatively generous city, causing a decrease in the transfer payment per recipient. In combination, the fleeing of the
wealthy and the migration of the poor weaken the city's redistribution program:
there is less money to transfer to more poor households. A national redistribution program is more effective because there is less mobility between nations than
between cities.
The third role of government is resource allocation. As shown in Table 19-2,
local governments are responsible for the provision of severa! goods and services,
including education, highways, police and tire protection, parks, and sewers. Under what conditions can government provision of these go9ds be justified on ef-
ficiency grounds? There are thtee possibilities. First, the good may be produced
under conditions that generate a natural monopoly. Second, the good may generate
506
positive externalities. Final!y, the good may have the characteristics of a local public
good.
Natural Monopoly
A natural monopoly occurs if the production of a particular good is subject to relatively large scale economies. In Figure 19-1, thelong-run average total-cost curve
for sewage services is negatively sloped over a wide range of output, refl.ecting the
fact that (1) there is a substantial :fixed cost associated with setting up a sewage system and (2) the marginal cost of adding another customer is relatively small. Because
the average-cost curve is negatively sloped, the marginal cost is less than average
cost (the marginal cost pulls down the average cost as quantity increases). Sewage
service is a natural monopoly in the sense that a single sewage authority (providing
S* units of service) could serve the city ata lower average cost than several sewage
authorities (for example, five firms, each supplying S' units).
What is the optimum leve!of sewage service? The basic efficiency rule is that
an activity should be increased to the point at which the marginal social benefit of
FIGURE 19-1 Natural Monopoly in Sewage Services
$
Demand = marginal
social benefit
P' -------
S'
S"
Quantity of sewage services
Sewage service is a natural monopoly: scale economies generate a negatively sloped Iong-run
average-cost curve. The optimum output is where the demand curve.intersects the marginal-cost
curve. Since the optimum price is less !han the average total cost, the sewage system generales a
deficit. The govemmelt could provide sewge service itself, covering the deficit with general tax
revenue. AJtematively, the govemment could regulate a private sewage authority.
507
the activity equals the marginal social cost. In Figure 19-1, the demand curve, wbich
shows consumers' willingness to pay for different levels of sewage service, is also
the marginal-social-benefit curve. The optimum output (S*) is the output at wbich
the demand curve intersects the marginal-cost curve. One problem with an output
of S* is that a firm producing this output would lose money. To get consumers to
purchase S* units of sewage service, the :firm would charge a price of P*, wbich is
less than the average total cost of producing s. The deficit is shown by the shaded
area.
The government has severa!options.First, it can provide the service itself, charge
a price of P* and cover the deficit with general tax revenue. Second, it can get a
private :firm to provide sewage service. If the government forces the firm to charge a
price of P*, subsidies are required to cover the deficit. A :final option is to allow the
private :firm to charge P' instead of P*. At P', average total cost equals the price, so
the firm would cover all of its cost. Under this scheme, the output (S') is less than
the optimum output.
Local governments have the same options for other natural monopolies. Chapter 12 (Mass Transit) discussed_the pricing and regulation of mass transit. Por other
natural monopolies (water systems, cable television networks, telephone networks),
the government can either produce the service itself or regulate a private producer.
Whatlevel of government should operate orregulate anaturalmonopoly? Should
sewage and transit services be run by local, metropolitan, regional, state, or national
governments? The appropriate leve!of government is determined by the extent of
scale economies: the greater the scale economies, the larger the politicaljurisdiction
required to realize the scale economies associated with providing the service. Local
governments are responsible for the provision and regulation of services for wbich
scale economi"s are exhausted at the locallevel.
Externalities
.,
508
E'
E*
Quantity of education
There are extemalities in the consumption of education, so the marginal social benefit
exceeds tbe marginal private benefi.t In the absence of government involvement, the
student chooses E' (wbere the marginal private benefit equals the marginal cost). The
optirnurn leve!of education is E*, wbich the student will cboose ifthe government
provides a subsidy equal to the gap between the marginal private and marginal social
benefit.
'
civic extemalities, the marginal social benefit of education exceeds the marginal private benefit. In the absence of govemment intervention, the worker-citizen ignores
these extemal benefits, choosing E' units of education instead of the optimum (E*).
The govemment has two options to solve the extemality problem. The first is to
take responsibility for providing the good. A system of free compulsory education
could encourage citizens to consume more edi.Ication. Altematively, the govemment
could subsidize education, encouraging citizens to consume more education while
allowing theiii to choose their own schools. In Figure 19-2, the subsidy would be
equal to the gap between the private and social benefits of education. The subsidy
would increase consumption to E* units of education. As discussed in the chapter
on urban education, there is heated debate over whether the govemment should
supplement the public school system with subsidies for private education (tuition
tax credits and education vouchers).
What leve!of govemment should deal with education extemalities? Should the
public school system (or education subsidies) be managed by local, metropolitan,
regional, state, or national govemments? One factor in deterrnining the optimum leve!
of govemment is the geographical extent of the externalities. If people throughout
the country benefit from the education of chilc!Ten in a particular school, national
509
510
103 -
20 P'"'-+==------=.,.,_:::;--------"'";<:::
A,
A
Park acreage
The marginal social nefit of park acreage is the sum of the private benefits of three citizens.
The optimum park acreage is A (the arnount such that the marginal social benefit equals
the marginal social cost).
How do local govemments decide how much oflocal public goods to provide?
As we saw earlier in the book, one option is to let people vote on different quantities
of the local public good. The result is that the preferences of the median voter will be
decisive.In our exatnple, Marian is the median voter: she splits the voting population
into two equal halves, half ofwhom have a smaller preferred quantity and half of
whom have a larger preferred quantity. The altemative approach is to let people vote
with their feet, setting up homogeneous municipalities in which each voter gets his
or her preferred level of the local public good. This is the Tiebout process described
in the chapter on household sorting.
Benefit Taxation
There is an altemative to Tiebout sorting that will generate unanimous support for the
optimum level oh public good, eyen in a heterogeneous municipality. In a heterogeneous municipality, majority voting is inefficient because the costs of the public good
-----=-=- --- - -- - -
511
'-
35
E': =====
A,
Ab
Park acreage
Tbe optim'm park acreage is A' (the arnount wbere the marginal social benefit equals the marginal
social cost). If each bousebold pays $20 per acre, the preferred arnounts are A,Am, and Ab. Under
majority voting, the preferred size ofthe median voter (Am) is chosen. Under a benefit tax, each
bousehold pays a tax (per acre) equal to its marginal benefit at the optimum park size (A*), so there
will be unanimous support for the optimum park size.
are divided equally among the three voters. An alternative approach is to allocate park
costaccording to the willingness to pay forparks: the greaterthe citizen's willingness
to pay, the higher the tax liability. This is the Lindahl approach to local decisionmaking (named afterErik Lindahl), also known as benefit taxation: A household's
tax liability is based on the household's benefit from the local public good.
Figure 19--4 shows how benefit taxes work. Suppose that the government knows
the marginal-benefit curves of its citizens. It announces that Sam's tax liability per
acre is $9, Marian's is $16, and Bertha's is $35. Given these prvate costs, all three
citizens have the same preferred park size, A, which also happens to be the optimum
size.In this case, the government is clever enough to figure out how much individual
citizens are willing to pay for different levels of the local public good, and it sets
each citizen's tax equal to his or her marginal willingness to pay. Sam (who prefers
512
srnall parks) agrees to a largerpark because he faces a lower tax liability: he pays
$9 per acre instead of $20. Bertha (who prefers big parks) gets a largerpark ata cost:
she pays $35 per acre instead of $20. Under benefit taxation, citizens with larger
benefits pay higher taxes.
Is the benefit principie practical? One problern with this approach is that the
government doesn't know the shapes of the rnarginal-benefit curves of individual
citizens, so it can't precisey! estirnate the appropriate taxes for thern. Moreover, there
is little incentive for taxpayers with relatively high benefits to reveal their willing-
ness to pay for local public goods: in doing so, they would be volunteering to pay
higher taxes. If the benefits received are roughly proportional to property value, the
property tax could be used to approxirnate a benefit tax. Similary!, if the benefits are
related to incorne, an incorne tax could serve as a benefit tax, albeit an irnprecise one.
Chapter !9
513
514
60
1
1
1
1
1
1
1
20 1---!------!--------=>."
A'
:,."""""
',
'1
A*
If municipal spencling generares benefits for people outside the municipality, the marginal social
benefit exceeds the marginal local benefit. The optimum spending leve!is A'.Local voters, basing
tbeir decisions on local benefits and costs, choose less than the optimum spencling leve!(A'). A
subsidy equal to tbe gap between tbe local benefit and tbe social benefit would cause tbe
municipality to choose tbe optimum spencling leve!.
.)
'
_)
_)
We can use Figure 19-6 to explore the trade-offs associated with intemalizing
spillovers by consolidating decision-making. There are three municipalities that
._)
.
_)
515
Cornrnunlty S
Small parks
Cornmunity M
Medium parks
Cornmunlty B
Bigparks
MLB,
--
------------
----------
------ --Marginal
cost
s s
M' M*
B' B*
If extemalities are relatively small, municipal decision-making will be relatively efficient: tbe gaps
between the equilibrium and optimum spending amounts (S' - S'; M' - M'; B' - B') will be
re!atively small. If the diversity in demand is relatively large, metropolitan decision-making will be
relatively inefficient: the gaps between the equilibrium (M') and optimum spending amounts
(M'- S'; B' - M') will be relatively large. If externalities (spillovers) are small relative to
diversity in demand, the municipal systern will be more efficient than the metropolitan systern.
provide parks, S (small park), M (medium park), and B (big park). Under the municipal system, there are spillovers: In each community, the marginal local benefit of
spending is less than the marginal social benefit, so each municipality spends Iess than
the optimum amoun t. For example, in community S, the marginal local benefit equals
the marginal cost at a spending leve!S', which is less than the optimum leve!S*.
Similarly, M and B spend less than the optimum amounts (M'< M* and B' < B*).
Suppose that the three municipalities are consolidated into a single metropolitan
government. Tne metropolitan government is large enough to contain al!the citizens
affected by local spending, so citizens base their votes in the metropolitan election
on the social benefits of parks. For example, citizens in community S base their
votes on the marginal social benefits (MSBs) not the marginal local benefits (MLBs ).
Sirnilarly, citizens in communities M and B base their votes on MSBm and MSBb.
Given the marginal-cost curve, the preferred budgets for the three types of household
are S*, M*, and B*.
Suppose that the metropolitan government intends to build three parks, one in
each community. The three. parks will be the same size. If the metropolitan government holds an election to determine the size of the parks, the winning size is the
size preferred by the median voter, M'. Therefore, each community will get a park
with M* acres. The chosen size (M*) exceeds the optimum size for Iow-demand
households (S*) and falls short of the optimum size for the high-spending household (B*). Under the metropolitan goveiiunent, externalities are internalized, but
'
516
two of the tbree communities must compromise on their ideal park sizes. In other
words, there are trade-offs associated with dealing with externalities and diversity
in demand.
Is the metropolitan system superior to the municipal system? In Figure 19-6,
interjurisdictional externalities are relatively small, so the gaps between the localbenefit and social-benefit curves are relatively small. Therefore, under municipal decision-making, the gaps between the equilibrium and optimum amounts of
park acreage are relatively sinall. For example, the small-park municipality would
choose S', which is relatively close to the optimum spending S*. Similar!y, the gaps
between M' and M* and between B' and B* are relatively small.In contrast, because
diversity in demand is relatively large, the compromises associated with metropolitan government are relatively large. For example, the gap between M* (the leve!
chosen imder majority rule at the metropolitan leve!) and the optimum leve!for
community S is relatively large: (M* - S*) is relatively large. Sirnilarly, there is a
relatively large gap between B* and M*. In this case, it is likely that the municipal
system is more efficient than the metropolitan system.
In general, metropolitan government will be more efficient than municipal government if interjurisdictional spillovers are large relative to diversity in demand. In
this case, the advantages of a smalllocal government (the ability to accommodate
diverse demands for local public goods) are relatively small, and the disadvantages
. (the inefficiencies ass_ociated with externalities) are relatively large. Therefore, a
metropolitan system of government will be more efficient.
Among these networks are streets and highways, water systems, sewage-disposal
systems,school systems, airsheds, and.watersheds. Sorne ofthe problems that cross
jurisdictional boundaries are highway congestion, air pollution, crime, and low
517
educational achievement. In the current political system, the power to deal with
these problems is divided among many small jurisdictions, most of which contain
only a small fraction of the people affected by the problems. Two metropolitan
areas-Portland, Oregon and the Twin Cities in Minnesota-have governmental
bodies with the power to deal with problems that cross municipal boundaries.
Q.-.
518
budget because an $80 btidget is closer to his preferred budget ($84). Gis the only
voter who votes for the $90 budget, so the first budget proposalloses by a vote of 6
to l.The second proposal ($80) fails by a vote of 4 to 3: only citizens with preferred
budgets greater than $75 vote yes. By threatening to propase a smaller budget in
the next election, the school board convinces citizens with preferred budgets above
the median budget (E, F, and G) to vote for the $70 budget; combined with the
vote of the median voter (D), this generates a majority in favor of the proposed
budget.
In this exarnple, the school district decreased its proposed budget in $1O increments, from $90 down to the preferred budget of the median voter. The same
result would occur if the board decreased the proposed budget in increments of $1.
Similarly, the same result occurs if the school district reverses the direction of the
budget sequence. If it starts out with a low budget and works its way upward, the
median voter is still decisive.
Suppose that the candidates start out with. different proposed pollee budgets:
Penny proposs a budget of $4, and Buck propases $8. Citizens with preferred
budgets less than or equal to $5 vote for Penny (70 votes), and those with preferred
budgets greater than or equal to $7 vote for Buck (70 votes). The two candidates
split the 30 voters with a preferred budget of $6 (halfway between the two proposed
budgets), so each candidate gets a total of 85 votes. The e1ection resu1ts in a tie vote.
Penny could increase her chance of being e1ected by increasing her proposed
budget. Suppose that she propases $5 instead of $4. She picks up al! 30 votes of the
citizens with a $6 preferred budget (Penny's $5 proposal is closer than Buck's $8),
and she wins the e1ection by a vote of 100 (10 + 15 + 20 + 25 + 30) to 70 (25 +
20 + 15 + 10). If Buck responds by decreasing his proposed budget to $7, the
e1ection results in a tie vote again. Penny and Buck will continue to revise their
519
25
25
.s,'i-; .,
..
. l.,. ..
20
1
i_
1.(
:J
_:
ll.. '
l'
'
10
10
r;
..-:;.t .
.:;-'_
ji
.1
l
;I
l
J
J
J
e _
!.(
2
7
8
5
6
Preferred police budget ($)
!O
Citizens have different preferences for the pollee budget. If Penny proposes a
budget of $4 and Buck propos.S a budget of $8, the election will result in a tie
vote: the two candidatos split the votes of citizens with $6 preferred budgets,
while Penny gets the votes of !hose with lower preferred budgets, and Buck gets
the votes of !hose with higher budgets. By moving toward the median bdget,
Penny can increase her chances ofbeing elected. In equilibrium, both
.t
t
1
J
]
:J
f
1
.1
t
f
l
1
1
l
1
1
l
proposed budgets until each candidate's budget is close to the budget ofthe median
voter ($6).
This examplshows that the median-voter result occurs in a representative
democracy. In equilibrium, both candidates propose a budget equal to the preferred
budget of the median voter. Sinee both candidates have virtually the same proposed
budget; it doesn't matter which candidate actually wins the election. In either case,
the median voter determines the size of the police budget.
520
Public Goods
Public Good or Service
Total expenditures
Education
Price Elasticity
Incorne Elasticity
-0.23 to -0.56
-0.07 to -0.51
-0.19 to -0.92
-0.19 to -1.00
-0.92 to -1.00
0.34 to 0.89
0.24 to 0.85
0.99 to 1.32
0.52 to 0.71
0.79
Source: Robert .iaman, ''The Fiscal Performance of Local Govemm.tnts," in Currenr Issues in
Urban Ecorwmics, ed. Peter Mieszkowski and Mahlon Straszheim (Baltimore: Johns
Hopkins University Press, 1979).
income ($1,200). Assume that the "price" of police services (the opportunity cost of
money spent on police) is the same in the two cities. Theincome elasticity of demand
for police services is defined as the percentage change in the police budget divided
by the percentage change in income. City L, with 20 percent higher income, has a
25 percent larger police budget, so the income elasticity of demand is 1.25 (25 percent
divided by 20 percent). Inm.an (1979) summarizes the results of emprica! studies
based on the median-voter modeLAs shown in Table 19-4, the income elasticities
for most local goods are less than l.O.
If there is variation across comrnunities in the price of local public goods, we
can use the median-voter model to estmate the price elasticity of demand for public
goods. To plot the demand curve for local spending, one needs information on price
(the opportunity cost of local spending) and quantity (the local spending leve!). As
shown in Table 19-4, the demands for local public goods are price-inelastic; the
price elasticities aie. a!ess than or equal to l.O in absolute value. We'll discuss
sorne of the implications of these e!asticities in the next chapter.
521
3. All citizens vote. In the median-voter model, all citizens vote. In the typical
budget election, only a fraction of eligible voters actually cast ballots. A rational
citizen abstains from voting if the cost of voting (the time and monetary costs of
becoming informed and going to the polls) exceeds the benefit. The benefit of voting
is relatively small if (a) the two proposed budgets are so close to one another that
it makes little difference which person is elected (indifference), or (b) the proposed
budget of the best candidate is so far from the citizen's preferred budget that the
citizen is alienated from the election process (alienation). If sorne citizens abstain
from voting, the median-voter result will not necessarily occur.
SUMMARY
l. There are almost 85,000 local goverrunents in the United States, including municipalities, school districts, counties, and special districts.
2. In the federal system, the national government is responsible for stabilization
and redistribution policies, while local governments are involved in resource
allocation, Local goverrunents are responsible for sorne goods subject to relatively large scale econornies (water, sewage, transit), other goods that generate
externalities (education), and local public goods such as parks, public safety,
and education.
3. The inefficiencies resulting from majority rule could be elirninated by the use
of the Lindahl system of taxation (tax equal to the marginal benefit of the local
public good).
4- There are trade-offs associated with the public goods at the local leve!: local
provisipn acco=odates diversity in demand but may generate spil!overs and
not fully exploit scale econornies.
5. The median-voter model predicts that the government will adopt the preferred
budget of the median voter (the voter that splits voters into equal ha!ves, with
one-half preferring smaller budgets and the other half preferring larger budgets).
..i.
522
'
the same?
b. If you don't have enough information to answer the question, describe the
information you need and explain how you would use it.
5. In the example shown in Table 19-3, the school board started with a budget of
$90 and worked its way downward until a majority of voters voted yes. Show
that the same outcome would occur if the school board started with a budget of
$10 and worked upward.6. In the example shown in Table 19-3, the winning budget is $70. Explain the
effects of the following changes on the winning budget. In each case, use as the
starting point the distribution of voters shown in Table 19-3.
a. Four elderly voters move into the school district. The elderly voters have
preferred budgets of $20, $25, $30, and $35.
b. In response to "A Nation at Risk," a report critical of educational programs,
the desired spending of every household increases by $5.
7. Consider the effect of prvate schools on the local school election. Suppose that
households F and G (in Table 19-3) enroll their children in prvate schools.
Predict the effects gf prvate schooling on the public school budget under the
following two cases:
a. F and G abstain from voting in the school election.
b. F and G vote in the election.Their preferred budgets drop by $40 ($44 instead
of $84 and $51 instead of $91).
S. Co=ent on the following: "If the median-voter model is correct, it's silly to
mess around with elections. We should hire Mr. Gallup to identify the median
voter and find out his or her preferred budget. Such a system would be cheaper
than a series of budget elections."
9. You are the campaign manager for Dick Tracy, who is running for mayor. His
opponent has committed herself to spending $500 per capita on police services.
You must decide how much Dick will propose for the police budget. What
information would you collect, and how would you use it?
10. Consider Metro, a metropolitan area with three school districts, East, West,
and North. The optimum spending is $500 in East, $900 in West, and $700 in
North. If the districts make independent decisions, spending will be $450 in
East, $850 in West, and $650 in North. In other words, there are externalities
in the provision of education, causing each district to spend too little. Suppose that the three districts are consolidated into a single metropolitan school
district.
-------
'
te
:'
}(
lJ
523
a. Assume that there are no scale economies in the provision of schools. Will
the metropolitan school district be more or less efficient than in the original
arrangement?
b. How would your answer to (a) change ifthe expenditures under independent
decision making were $200 (East) and $600 (West)?
c. How would your answer to (a) change if there are scale economies in the
provision of education?
11. The example shown in Figure 19-7 assumes that all citizens vote in the mayoral
election. Suppose that sorne citizens refuse to vote if the proposed budgets of
the two candidates are not very far apart (they abstain because of indifference).
Specifically, suppose that if the difference between the candidates is 1ess !han
$1, only 50 percent of citizens (of all budget preferences) vote.
a. Will Buck increase his chances ofbeing e1ected ifhe increases his proposed
budget from $6 (!he median budget) to $7?
b. Will the median-voter result persist if there is indifference?
12. The example shown in Figure 19-7 assumes that all citizens vote in the mayoral
election. Suppose that sorne citizens refuse to vote if the proposed budget of the
closest candidate is too far from their preferred budget (they abstain because of
alienation). Specifically, a voter abstains if the difference between the closest
candidate and the citizen's preferred budget is greater !han $3.
a. Suppose that Penny decreases her proposed budget to $5. Does she increase
or decrease her chances of being elected?
b. Will the median-voter result persist if there is alienation?
REFERENCES
l. Downs, Athony. "The Devolution Revolution: Why Congress ls Shifting a Lot
of Power to the Wrong Leve1s." Brookings Policy Brief no. 18. Washington,
DC: Brookings Institution, 1998.
2. lnman, Robert P. "The Fiscal Performance ofLocal Governments: An lnterpretive Review." In Current lssues in Urban Economics, ed. Peter Mieszkowski
and Mahlon Straszheirn. Baltimore: Johns Hopkins University Press, 1979,
pp. 270-321.
3. Musgrave, Richard A., and Peggy B. Musgrave. Public Finance in Theory and
Practice. New York: McGraw-Hill, 1980.
4. Oates, Wallace E. Fiscal Federalism New York: Harcourt Brace Jovanovich,
1972.
5. Savas, E.S. Privatizing the Public Sector. Chatham, NJ: Chatham House, 1982.
6. Stevens, Barbara J. Delivering Municipal Services Efficiently: A Comparison
of Municipal and Private Sector Delivery: Summary. Report no. 3744, U.S.
Department ofHousing and Urban Development, Washington, DC, June 1984.
7. U.S. Bureau of the Census. Compendium of Government Finances, no. 5.
Washington, DC: U.S. Govemment Printing Office, 1997.
-CHAPTER 20
hi.s chapter explores the econornics of fue property tax, one of the most important
revenue sources for local government. When a govemment imposes a tax, the people
who pay the tax in a legal sensehave the incentive to change their behaviorto avoid the
tax. By doing so, they shiftitto someone else.The policy questionis: Aftertaxpayers
have fully adjusted to the new tax, who actually pays local taxes on property? This
chapter also explores the econornics of intergovernmental grants, another source of
local revenue.
'
Table 20-1 shows the distribution of revenue fordifferent types oflocal govemments.
For local governments as a whole, 38 percent of total revenue comes from intergovernmental grants, 39 percent comes from local taxes, and the remaining 23 percent
comes from charges and general revenue. School districts are heavily dependent
on intergovernmental grants, receiving about 54 percent of their revenue from the
federal and state governments. For municipal governments, 28 percent of revenue
comes from grants, 44 percent comes from local taxes, and 28 percent comes from
user charges and general revenue. In contrast, special districts are heavily dependent
on user charges, receiving 55 percent of their revenue from charges and general
revenue.
Figure 20-1 shows the distribution of tax revenue for municipalities and al!local
govemments. In 1992, the property tax generated about three-fourths of alllocal tax
revenue. Municipalities depend less on the prbperty tax: For municipalities as a
whole, the property tax generated about half of tax revenue; for large municipalities
(population greater than 300,000), the property tax generated only about 46 percent
of tax revenue. The second largest revenue source is the sales tax: In1992 it generated
524
525
Type of Govemment
County
Municipa!ity
Townshlp
School district
Special district
Alllocal governments
Intergovemmental
Grants
37%
28
24
Local
Taxes
38%
44
60
38
16
39
54
29
38
Charges and
General
Revenue
25%
28
16
8
55
23
Total
100%
100
100
100
100
100
Source: U.S. Bureau ofthe Census, Compendium ofGovemment Finances, on. 5 (Washington, DC: U.S.
,',
'
76
" 70
Al! local
govemments
" 60
!il Municipal
governments
50
"a
40
"-
"
27
lO
OL....--L"'""
Property
1
l
1
1
1
1
Sourr::e: U.S. Bureau ofthe Census, Compendiiun ojGovemme111 Finan.us, no. 5 (Washington, DC:
U.S. GovernmentPrinting Office, 1997).
15 percent of total tax revenue and 27 percent of municipal tax revenue. The third
largest source is the local income tax, which generated 5 percent of local taxes and
13 percent of municipal taxes.
Table 20--2 shows the effectiveproperty-tax rates for severa!cities. The effective
tax rate is defined as the tax liability divided by the market value, in contrast with
the nominal tax rate, defined as the tax liability divided by the assessed value.
Because many local governments assess property at less than its full market value,
the effective tax rate is typically less than the nominal tax rate.
--
526
Bridgeport, CT
Newark,NJ
Milwaukee, WI
Des Moines, IA
Houston, TX
Philadelphia, PA
Jacksonville, FL
Phoenix, AZ
Portland, OR
Chicago, TI...
Memphis, TN
Boston, MA
Minneapolis, MN
New York City, NY
Los Angeles, CA
3.68
3.58
2.98
2.67
2.64
2.20
1.82
1.80
1.79
!.53
1.39
1.34
0.86
0.74
Table 497.
It will be useful to divide rental housing into two components: the dwelling
itself (the structure) and the land that comes with the dwelling. The annual rent on
the residential property is the sum of dwelling rent and land rent. Por example, an
annual rent of $5,000 can be broken down into $4,000 of dwelling rent and $1,000 of
land rent. Similarly, the market value of residential property is the sum of the market
values of the dwelling and the land. As explained earlier in the book, the market
value of property equals the present value of the rental income from the property. If
the market interest rate is 10 percent, the market value of the dwelling is $40,000
($4,000/0.10), the marketvalue ofthe landis $10,000 ($1,000/0.10), and themarket
value of the residential property (both dwelling and land) is $50,000.
The residential property tax is an annual tax based on the market value of the
property. Por example, linder a 2 percent property tax, the owner of a $50,000
property would pay $1,000 per year in taxes. This can be broken down into an $800
tax on the dwelling itself (2 perc.ent of a market value of $40,000) anda $200 tax on
the land (2 percent of a market value of $10,000)..
"lllf'
:,J
;,::-
.. 1'4-
527
,{
t
.
1<
f.
f
(
l
1
1
1
1
1
1
To explain the market effects of the property tax, consider the city of Taxvil!e,
where allland is used for renta! housing. The city has the following characteristics:
l. ldentical dwellings. The city has 900 identical renta! housing units.
2. Housing inputs. There are two inputs to housing: land and improvements (the
dwelling). Every dwelling occupies a quarter-acre lot.
3. Capital mobility. Dwellings are irnri:J.obile in the short run, but can be moved
from one city to another in the long run.
4. Unit property tax. The annual property taxis paid in legal terms by the property
owner. The taxis $200 per quarter-acre lot and $800 per dwelling, regardless of
the market values of land and improvements.
The fourth assumption means that the property tax is not based on market value, but
is a fixed unit tax. This assumption simplifies the exposition without changing the
analysis in any substantive way.
The Land Portion of the Property Tax
Consider first the effect of the land portien of the property tax. Figure 20-2 shows
the city's land market. The supply of land is perfectly inelastic, with a fixed supply
of 900 quarter-acre lots. The demand curve intersects the supply curve at point B,
1
1
t
1
Supply
1,200
1
1
1
1
1
1
1
1
1
l
1
1
1
1
1
1
1
L1
-;::-
: 1,000
>-
&
800
J.
1
1
1
1
1
1
1
1
1
Demand
1
1
450
1,350
900
Quantity of land (number of quarter-acre 1ots)
Land rent is $1,000 in the pre-tax equi1ibrium. Since the supp1y of land is fixed, the land tax
decreases rent by the amount of the tax ($200). If the landowner tried to pass on the $200 annual
tax to consumers, tbe net cost ofland would rise to $1,200 (point C), and the quantity supplied
. (900) would exceed the quantity demanded (450). The surplus ofland wou1d cause rent to fa!]
lO $800.
528
generating an annual rent of $1,000 per 1ot. Ifthe market interest rate is 10 percent,
market value of the land is $10,000 per lot.
Can !andowners shift the land tax onto land consumers? The property tax increases the property owner's tax liability by $200 per year. Suppose that the property owner increased land rent from $1,000 per year (point B) to $1,200 (point C).
At point C, the quantity ofland demanded (450 lots) is less than the fixed supply
(900 lots), so sorne land is vacant. The excess supply ofland causes landowners to cut
prices to attract consumers. The price cutting continues until the gap between supply
and demand is e!iminated. In equilibrium, land rent falls to $1,000 and the net retum
toproperty owners drops to $800 ($1,000 less the $200 tax). Because the supply of
land is fixed, the land taxis ultirnately borne by!andowners. The decrease in net renta!
income decreases the market value ofland from $10,000 to $8,000 ($800/0.10)
..
8
""
11
.5' 3,600
-:Z;o
o
Demand
3,200
420
660
900
Quantity of!and (number of dwellings)
. 1.140
Befare the dwelling tax, B is the equilibrium point (rent = $4,000 per year). The property taxis
$800 per dwelling, so the supply curve shifts up by $800. Dwelling rent increases to $4,800 per
year, and the number of dwellings decreases to 420.
.'
529
assumed to be a small part of anational marketfordwellings, so the marketpriceis unaffected by events in the city. Given the supply and demand curves, the initial equilibrium (point B) generates a dwelling rent of $4,000 per year and 900 dwellings.If the
market interest rate is 10 percent, the market value of the renta! dwelling is $40,000.
Why is the long-run supply curveperfectly elastic? One way to picture the long
run is to imagine that property owners can put wheels on their dwellings and roll
them to other cities. Although property owners do not literally move their dwellings
from place to place, they can decrease the supply of housing in other ways. They
can (1) decrease their investment in repair and maintenance, (2) abandon dwellings,
(3) convert property to other uses, and (4) not respond to increases in the demand for
housing. Given these supply responses, property owners can decrease the effective
supply of housing without actually moving their dwellings to another location. The
house-on-wheels metaphor captures the idea that property owners can decrease the
supply of housing in the long run.
The property tax generates an annual tax of $800 per dwelling. Because the
property owner pays the improvement tax in a legal sense, the tax shifts the long-run
supply curve up by the amount of the tax. The supply curve is a marginal-cost curve,
and the tax increases the margmal cost of renta! property by $800 per dwelling per
year. The market moves from point B to F; the equilibriurn rent rises by the full
amount of the tax (from $4,000 to $4,800), and the quantity of dwellings drops from
900 to 420. Net renta!income remains at $4,000: the property owner collects $4,800
in rent and pays $800 in taxes, leaving $4,000 in net income. The tenant pays the
entire tax: the annual cost of renta! housing rises by $800.
The improvement taxis borne by consumers because the supply of dwellings is
perfectly elastic. The city is a small part of the national market for dwellings, so it
takes the national rent of $4,000 as given. If the city's property owners were to receive
anything less than $4,000 in net renta! income, they would roll their dwellings out
of town. In fact, the decrease in the supply of dwellings (from 900 to 420) increases
dwelling rent by an amount large enough to restore the original net renta! income of
$4,000.
530
4. Equilibrium price. Property owners move their dwellings to the city with
the highest return (the highest annual rent). In equilibrium, the market rent on
dwellings is the sarne in the two cities.
The regional dwelling market is shown in Figure 20-4. Each city has a negatively
sloped demand curve, and the initial equilibrium occurs with points B (in Taxville)
and K (in Notax). The equilibrium price of dwellings is $4,000, and each city has
900 dwellings.
In the short run, the property owner pays the irnprovement tax. In the short
run, the supply of dwellings in each city is fixed. The property tax ($800 per year)
decreases the netrental income in Taxville from $4,000 (point B) to $3,200 (point T).
The property tax is borne by property owners because the supply of dwellings is
perfectly inelastic. In the nontaxing city, rent stays at $4,000.
In the long run, property owners move their dwellings to the more lucrative
market in Notax. Suppose that property owners move 120 dwellings from Taxville
to Notax. In Figure 20-4, the Taxville market moves from point B (900 dwellings) to
point e (780 dwellings), causing therent on Taxville dwellings to rise from $4,000 to
$4,200. The net income ofTaxville's property owners increases from $3,200 ($4,000
rent less $800 tax) to $3,400 ($4,200 less $800). Taxville's loss is Notax's gain: The
Notax market moves from point K to point L (an increase of 120 dwellings, from
900 to 1,020), and rent drops from $4,000 to $3,800.
Is the dwelling market in equilibrium with points e and L? Equilibrium occurs
when two conditions are satisfied. First, the total number of dwellings in the two cities
must equal 1,800 (the :fixed supply). Second, property owners have no incentive to
move their dwellings from one city to another; the net rental income must be the sarne
in the two cities. Forpoints e and L, the first condition is satisfied (780 + 1,020 =
1,800), but the secori.d condition is violated. The net income in Taxville ($3,400)
is stillless than net income in Notax ($3,800). Therefore, the market will continue
to move up the Taxville demand curve and down the Notax demand curve. The
movements along the demand curves will continue until both equilibrium conditions
are satisfied. Equilibrium occurs with points D and M; there are 660 dwellings in
Taxville and 1,140 in Notax, and the net income is $3,600 per rental dwelling.
There are two lessons from this general-equilibrium analysis. First, property
owners pay the tax in an econornic sense. The tax decreases net rental income in
both cities to $3,600. The movement of dwel.l.i.:Iigs to the untaxed city decreases rent
in the untaxed city (to $3,600) and increases rent in the taxing city (to $4,400).
The tax is paid by property owners throughout the region, not just property owners
in the taxing city. Second, the average rent in the region is unaffected by the tax:
Consumers in Notax gain (they save $400) at the expense of consumers in Taxville
(they pay $400 more). The changes in housing costs offset one another, so the tax
does not change the average rent.
The general-equilibrium analysis suggests that the irnprovement tax is paid by
property owners (the owners of dwellings). Because the total (regional) supply of
dwellings is perfectly inelastic, property owners are unable to escape the tax. On the
other side of the market, as a group, housing consumers escape the tax: the losses of
Notax
$
4,800
4,600
4,400
4,200
4,000
$
Demand
E
--------1-- D
--------1--o
-------- 1--11
J S
3,600
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
540 660 780 900
Number of dwellings
Demand
--
3,200
420
"...'.
4,000
3,800
3,600
1
1
1
1
1
1
1
1
1
1
1
1
1
1
900 1,020 1,140
Number of dwellings
The initial (pretax) equilibrium is shown by points B and K. The improvement taxis $800 per dwelling per year. In the short run, dwellings are
immobile, so dwelling rent is fixed at $4,000 and property owners pay the tax. In the long nm, property owners move their dwellings to the
nontaxing city, increasing dwelling rent in the taxing city (movement upward along the demand curve) and decreasing rent in the nontaxing city
(movement down the demand curve). In Jong-run equilibrium, property owners receive the same net (posttax) retum, so Taxville rent exceeds
Notax rent by the amount of the tax: with points D and M, Taxville rent is 3.4,400 (net return = $3,600) and Notax's rent is $3,600. The tax is
shifted to property owners in lhe nontaxing city.
,,'
532
Taxville households are offset by the gains ofNotax households. The improvement
tax causes zero-sum changes in rent, so housing consumers are no worse off as a
result of the tax.
533
impose the tax, there is no incentive for property owners to move their dwellings
to the other city. Therefore, the short-run effect is the same as the long-run effect:
the tax decreases net income, but does not a:ffect the market rent. Property owners pay the improvement tax, just as they d.id in the single-tax case. Because the
tax does not change the equilibrium rent, it does not a:ffect housing consumers.
Housing consumers escape the ta.X beause dwellings do not move from one city to
another.
..
rt
] ti
fi
t
ri
r:,
'
. 1;
.l..
(,i,
l
;
l,:
J.
l. Dwellings are withdrawn. Only some ofthe dwellings thatleave Taxville arrive
in the untaxed city. Sorne of the dwellings are withdrawn from the regional housing
market. Suppose that one-third f the dwellings that leave Taxville are transferred
to the untaxed city, and two-thirds are withdrawn from the regional market. In Figure 20-4, equilibrium occurs with points E and L (compared to D and M ifthe total
supply of dwellings is fixed). The quantity of dwellings in Taxville decreases to 540
(a reduction of360), while the quantity of dwellings in Notax increases to 1,020 (an
increase of only 120), and 240 dwellings are retired from the market.
. 2. Property owners lose less. The withdrawal of dwellings increases net renta!
income. At points L and E, dwelling rent is $3,800 in Notax and $4,600 in Taxvi!le.
The net i.come-for property owners in both cities is $3,800 (compared to $3,600
with a fixed supply of dwellings). The tax causes a smaller decrease in renta!income
because property owners respond to the tax by decreasing the supply of dwellings.
3. Tenants pay part of the tax. The withdrawal of dwellings shifts part of the
tax to housing consumers. The tax increases dwelling rent to $4,600 in Taxville and
decreases rent in Notax to $3,800. In other words, the loss ofTaxville tenants (rent
increase of $600) exceeds the gain of Notax tenants (rent reduction of $200). The
tax harms Taxville tenants more than it helps Notax tenants, so the tenant class is
worse off. Because property owners decrease the supply of dwellings, they can shift
part of the improvement tax to housing consumers.
Mobile Consumers One ofthe assumptions of this simple model is that tenants are
imrnobile. When dwelling rent increases in the taxing city, none of the households
leave the city. What happens if households can move to the nontaxing city?
If households can move at no cost between the two cities, they willleave the
taxing city as long as it has a higher housing rent. If the market is at points D and
M in Figure 20-4, housing rent (the sum of dwel!ing rent and land rent) is $5,400
in the taxing city ($4,400 for the dwelling plus $1,000 for the land) and $4,600
in the untaxed city ($3,600 plus $1,000). When households leave the taxing city,
the demand for.Taxville land decreases, decreasing land rent and housing rent. Por
example, ifland rent drops to $900, housing rent decreases to $5,300.
534
At the same time, the rnigration of households to the untaxed city bids up land
rent, increasing housing rent: if !and rent increases to $1,100, housing rent increases
to $4,700. In other words, rnigration from Taxville to Notax narrows the difference
between housing rents in the two cities. In locational equilibrium, the two cities have
the same housing rent. For example, if land rent drops to $600 in Taxville and rises
to $1,400 in Notax, housing rent is $5,000 in both cities ($4,400 plus $600 in the
taxing city and $3,600 plus $1,400 in the untaxed city).
The rnigration of households causes capital gains and losses for landowners. If
the annual rent on Taxville land drops from $1,000 to $600, the market value ofthe
land drops from $10,000 to $6,000. In contrast, ifrent on Notax land rises to $1,400,
its market value increases to $14,000. The rnigration of households elirninates the
intercity differences in housing cost. Instead of differences in housing cost, there are
differences in land rent and differences in the market values of land. In other words,
the rnigration of households allows consumers to shift the benefits and the costs of
the improvement tax to landowners.
535
'
0-5,000
5,000-10,000
10,000-15,000
15,000-20,000
20,000-25,000
25,000-30,000
30,000-50,000
50,000-100,000
100,000-500,000
500,000-1 million
Over 1 million
Local Perspective
(partial equilibrium)
National Perspective
(general equilibrium)
7.9%
3.0
2.4
2.1
2.1
2.1
2.2
2.3
2.2
2.2
2.3
1.0%
0.6
0.9
0.9
1.0
1.2
1.4
2.2
3.9
5.2
5.8
Source: Joseph A. Pechman, Who Paid rhe Taxes, 1966-1985? (Washington, DC: Brookings
Instirution,!985), Table 3-1.
l
l
1
(
l
1
l
(
{
(
(
1,
1;
aL
Bertha, the president of a nation of 100 cities, wants to estimate the national effects
of a ptoperty tax in a single city. She should use the general-equilibrium model.
Under the assumption of a fixed national supply of dwellings, the improvement tax
is paid by capitalists in the form of a lciwer return on capital (lower net income from
rental property). The tax causes zero sum changes in the cost of housing: consumers
outside the taxing city gain at the expense of consumers in the taxing city.
Tab1e 20-3 also shows the distribution of the property tax burden from the
national perspective (the new view}. In computing the numbers, Pechman (1985)
assumes that (1) the land taxis borne entirey! by landowners and (2) the improvement
taxis borne entirely by property owners. In other words, the computationalresults are
based on general-equilibriurn analysis. Because capital and 1and ownership increase
rapidly with income, the property tax inthis case is progressive: the tax burden as a
fraction of income increases with income.
What if Bertha assurnes that the supply of capital is not fixed? If the supply is
not perfectly inelastic, part of the taxis shiftedto housingconsurners. The price of
.. housing increases by a relatively large amount {n the taxing citY and decreases by a
536
relatively small amount in the other 99 cities. As a result, the average cost of housing
increases.
2. Immobile households
a. Fixed supply of dwellings: zero-sum changes in. dwelling rent.
b. Variable supply of dwellings: positive-sum changes in dwelling rent.
3. Mobile households
. a. Fixed supply of dwellings: zero-sum changes in land rent.
b. Variable supply of dwellings: negative-sum changes in land rent.
Tax Imposed by Ail Cities (A National Property Tax)
Weifare effecrs in the taxing citie.s
l. Fixed supply of dwellings: Housing consumers and landowners unaffected by the tax.
2. Variable supply of dwellings: Dwelling rent increases and land rent decreases.
National weifare effects
l. Property owners receive lower net renta! income.
2. Fixed supply of dwellings: Property owners pay the entire tax.
3. Variable supply of dwellings: Tax shared by property owners, consumers, and landowners.
537
Differentil.TaX: Rates
l
L
(
J
l
Most jurisdictions irnpose the same nominal tax tate on residential and business
property. Because assessment ratios differ across property types, however, effective
tax rates often differ. Considera citywith the following characteristics:
l.
2.
3.
4.
5.
Suppose that the city cuts the property tax rate on pizza parlors and increases
the tax rate on housing. The change in tax policy is revenue-neutral: the loss of
revenue from pizza parlors equals the gain in tax revenue from residences. The new
tax policy causes the following changes in the city.
l. Transfer of capital. Property owners move capital from the residential sector
538
3. Decreased pizza prices. The movement of capital into the pizza sector decreases the price of pizza.
Who gains and who loses? The change in tax policy is revenue-neutral, so property
owners are not affected by the change in tax policy. Housing consumers lose because
the price of housing increases, and pizza consumers gain because the price of pizza
decreases. This general-equilibrium analysis suggests that differences in effective
tax rates are refl.ected in differences in consumer prices. If a city taxes residential
property at a higher rate than business property, its citizens pay more for housing,
but less for other goods.
What if nonresidential goods are consumed by people outside the city? The
taxation of property used in the production of exports increases the export price, so
part ofthe city's tax burden is borne by consumers outside the city. This is known as
tax exporting; the city uses tax revenue from outsiders to pay for local government.
Suppose that a large fraction of the city's pizza is consumed by outsiders. If the
city wants to decrease the local tax burden, it could cut the residential tax rate and
increase the tax rate on co=ercial property (including pizza parlors). Although
pizza prices will increase, partly offsetting the decrease in the cost of housing, local
. consumers are better offbecause a larger fraction oflocal taxes is shifted to outsiders.
What are the limits to tax exporting? The taxation of exports decreases the city's
export volume. Consumers respond to the increase in price by decreasing the quantity
of exports demanded, so the city's tax base decreases. In addition, export firms have
an incentive to move to cities with lower taxes. Because taxation decreases the city's
tax base, there are limits to tax exporting. Tax exporting is more likely in a city that
has a unique production advantage that makes it the dominant locational choice for
. export firms. .
539
f:
t
('
l
1
(
l
f
1
t
l
1
1
1
1
.L
$2,000
200
2,000
200
2.0%
0.20
4.0
0.40
As shown in Table 20-5, households sort themselves with respect to park demands
and property values, forming four municipalities. There are two wealthy communities, one with a large park budget (tax rate = 2.0 percent) and one with a small park
budget (tax rate = 0.20 percent). Similarly, there are two poor municipalities.
Because the households in Metro sort themselves into homogeneous communities, the property tax is a user fee, not a conventional tax. The tax liability of a
particular household depends on its park consumption, not on its housing consumption (property value). Every household pays a tax equal to its desired park budget,
regardless of its.property value. The poor pay a higher tax rate to generate the same
park budget as their counterparts in the wealthy community; the low spenders pay
a lower tax rate than their counterparts in the high-spending municipalities. In the
Tiebout world; there is no link between property value and property tax liability.
Instead, the tax liability depends strictly on park consumption.
In the Tiebout world, an increase in housing consumption does not affect property taxes. Suppose that a few of the wealthy, high-spending households decide to
buy $200,000 homes. They will forma new community of high-spending households with $200,000 houses, and finance their $2,000 park program with a 1 percent
property taJL The doubling of housing consumption cuts the property tax rate in half,
so property taxes remain at $2,000. In general, a cha:nge in housing consumption
causes the household to move to a community where other households have (1) the
same demand for public goods and (2) the same housing consumption.
In the Tiebout odel, the discussion of the incidence of the property tax is
meaningless. Ifhouseholds are sorted into homogeneous communities, the property
taxis a user fee, and every household gets exactly what it pays for. The "incidence"
of a $200 tax liability is the same as the ''incidence" of $1 paid for a hot dog: You
pay the $1 to get the hot dog; you pay the $200 tax to get $200 worth of parks. Since
the consumer's tax liability is independent ofhis housing choice, the property taxis
a user fee, not a conventional tax.
How realistic is the Tiebout model? The conditions required for Tiebout equilibrium are unlikely to be satisfied. There are sirnply not enough local govemments to
allow households to sort themselves into perfectly homogeneous communities. The
Tiebout model is clearly inapplicable to central cities, where a single municipality
serves a large and diverse population. Therefore, the central-city property taxis not
a user fee, but a conventional tax. In the suburbs, where there is more sorting of
households with respect to demands for public services and housing consumption,
the property ta.X is eloser to a user fee.
540
Limit on Tax
Rate for Specific
Type of Local
Government
Limit on Tax
Rate for Overall
Local Spending
Limit on
Property Tax
Revenue
Growth
Limit on
Assessment
Growth
32
12
28
28
11
21
15
541
half the states with levy limits have a maximum growth rates in the range 4 percent
to 6 percent. In two states, the maximum growth rate equals the infiation rate as
determined by the CPI, and in three others, the !imit is the maximum of 5 percent
and the inflation rate. Outside this middle range, three states allow no growth in
property tax revenue, and two states allow property tax revenue to grow by up to
1O percent per year. Over half the states have override options for levy limits, with
two requiring a supermajority.
The fourth column shows the number of states with limits on the growth rate
of assessed value. Nine states have reassessment caps, and in six states, the caps
apply to individual properties. In all nine cases, the reassessment caps were either
enacted or amended since 1978, with the caps used in combination with a limit on
tax rates. In most states, the maximum growth rate of assessed value is in the range
4 percent to 6 percent. Outside this range, California and Arizona have limits of
2 percent and Florida has a limit of 3 percent. Most states exempt new construction
and improvements in calculating the aggregate growth in assessed value. Only one
state-Arizona-has an override option for assessment limits.
...
10
"
o<.>E
....o 6
-"'
O )"
."
e
4
1
1
1
1
1.
1932
1945 1950
1975 1980
1960
Year
1995
542
doubled between 1929 and 1932, reaching 11.3 percent in 1932. During this threeyear period, personal income was cut in half while property tax revenue decreased
by only 9 percent. At this time, most of the money to support local government
carne from the property tax, so changes in the share of income absorbed by local
government were similar; the share of income spent on local government increased
from 5.4 percent in 1929 to 11.7 percent in 1932.
The doubling of the share of income absorbed by the property tax led to tax
delinquencies and the formation of taxpayers' leagues. Delinquency rates for the
property tax rosefrom 10.1 percentin 1930to 26.3 percentin 1933.In 1933, between
3,000 and 4,000 localleagues were agitating for tax reform (Beito, 1989), with the
demand for less local government. The message to government was that it should
scale back its operations to reflect lower income during the Great Depression (Beito,
1989, page 18): "!buy less food, less tobacco, less recreation, and I'd like to buy
less government."In mass meetings organized by the tax leagues, citizens demanded
the elimination of local services, including weed inspectors and county nurses. It
appears that taxpayers expected simultaneous cuts in taxes and local services.
Local officials tried to counteract d linquencies and tax resistance with campaigns urging citizens to pay their property taxes. Fearing massive defaults on municipal bonds, the business community supported these campaigns by paying for
lapel buttOns, mass inailings, and parades. The parades included the descendants
of canine war heroes, who carried signboards urging people to pay their taxes. The
National Pay Your Taxes Campaign published a handbook for local governments
with tips on how to get citizens to pay their taxes. Among the suggestions was a
tax-receipts scoreboard to be placed in front of city halls.
The tax revolt resulted in the passage of tax limit measures and a decrease in
the taX burden. In 1932-and 1933, a total of 16 states passed tax limits, with most
of the measures setting a maximum overall rate for local property taxes. Between
1932 and 1934, income increased by 42 percent, but property tax revenue decreased
by 5 percent. By 1940, the share ofincome absorbed by property taxes had dropped
to 5.8 percent. The decrease in the property tax share resulted from a combination
of income growth and the tax limits. By 1940, personal income had almost reached
the leve!observed in 1930, while the share of income absorbed by the property tax
was 5.8 percent, corripared to 6.3 percent in 1930.
543
believed that state and local govemments could absorb a 40 percent cut in tax revenue
without cutting services (Citrin, 1979).In Massachusetts, 82 percent of the supporters
of Proposition 2 1/2 believed that the Proposition would cut taxes without reducing
the quality oflocal public services. On!y 10 percent of supporters identified service
cutbacks as the most important result of the Proposition. In Michigan, only one in
four of the supporters of the Headlee Amendment expected simultaneous cuts in
taxes and services, while three of four supporters expected a free lunch, or at least a
more effi.cient government (Courant, Gramlich, and Rubinfeld, 1980).
In the period 1978 to 1981, a total of 16 states enacted property tax limits, with
California, Michigan, and Massachusetts leading the way. In California, Proposition
13 rolled backassessed values to the 1975 values, setamaximum taxrate of 1 percent
of assessed value, and capped the reassessment rate at 2 percent per year. In the
sarne year, Michigan limited the growth rate in property tax revenue from existing
properties to the inflation rate, as deterrnined by the Consumer Price Index. In 1980,
Massachusetts capped the property tax rate for municipalities at 2.5 percent and
capped the growth rate of property tax revenue at 2.5 percent per year. Most of the
other 13 states that imposed limits during this period established new lirnits on the
growth rate of property tax revenues (levy limits).
In 1991, Illinois established limits on the growth rate of property tax revenues in
the "collar" counties of the Chicago metropolitan area. Local governments in Cook
County were not subject to the limit. The maximum growth rate is the maximum of
the inflation rate (as measured by the CPI) or 5 percent per year. The levy limit was
extended to local governments in Cook County in 1995. In 1996, other counties in
the state were given the option ofimposing the levy limit.
Although Michigan adopted a tax-limit prograrn in 1978, it was ineffective in
controlling property taxes (Courant and Loeb, 1997). In 1995, the state reformed its
entire education finance system, cutting property tax revenue in half. The revenue
loss from property tax reform was offset by increases in sales taxes, tobacco taxes,
and real estate transfer taxes. The state also established a reassessment cap equal to
the lesser of 5 percent or the inflation rate.
Most of the modem property tax limits involve direct or indirect lirnits on the
growth rate of property tax revenues. In the case of levy limits, there is an explicit
limit on the growth of tax revenueThe altemative approach is to combine a limit
on the property tax rate with a limit on the growth rate of assessed values, an approach adopted in nine states. In most cases, new construction and improvements
are exempt from the levy limitar reassessment lirnit. Most of these modem limits restrict the growth rate of tax revenue from existing property to between 4 percent and
6 percent.
544
The evidence from states that enacted property tax limits suggests that the limits
decrease property tax revenue. In California, Proposition 13 decreased property tax
revenuefrom$10.3 billionin 1977to $5.6 billionin 1978. By 1995, nominalproperty
tax revenue had risen to $19.5 billion, but in real terms, property tax revenue was stil!
25 percent be1ow the leve!in 1977, and real property tax revenue percapita was 46
percent Jower (Sexton, Sheffrin, and O'Sullivan, 1999). In Massachusetts, the initial
impact of Proposition 2 112 was an 18 percent decrease in property tax revenue
(Cutler, Elmendorf, and Zeckhauser, 1999) and Jater significantly constrained local
spending (Bradbury, Mayer, and Case, 1997). Using national cross-sectional data,
the Advisory Committee on Intergovernrnental Relations (1995) and Shadbegian
(1998) suggest that tax Jimits reduce property tax revenue, with ACIR reporting a
reduction of 6 percent to 8 percent, and Shadbegian reporting a reduction in per
capita property taxes of about 3 percent.
Shires (1999) examined the effects of Proposition 13 and subsequent Jegis!ation
on the revenue structure of state and local govemments in California. Between 1978
and 1995, the share of state and local revenue from taxes decreased, while the share
of revenue from intergovernrnental grants and other revenues (fees, charges and
other revenue) increased. At the city Jevei, the tax share decreased s!ightly, while the
Jarge reduction in the intergovemmental share was nearly matched by the increase
in other-revenue share. The Joss of intergovernmental transfers to cities was not a
direct result of Proposition 13, but a consequence of the state's decision to shift state
funds from cities to school districts. At the county leve!, the tax share decreased
while the intergovernmental share and other-revenue share increased. For school
districts, the reduction in the tax share was nearly matched by an increase in the
intergovernmental. share.
Preston and Ichniowski (1991) examined the revenue effects of property tax
limits on municipalities over the period 1976 to 1986. The average limit on the tax
rate alone decreased the growth rate of property tax revenue by 7.4 percent, while the
average rate limit combined with a limit on assessment growth decreased the growth
rate of property tax revenue by 44 percent. The average Jevy limit decreased the
growth rate of property tax revenue by 9 percent. In terms of revenue-substitution
effects, Preston and Ichniowski suggest that losses in property tax revenue were
part!y offsetby increases in intergovernmental grants. They fi.nd no evidence that
limits on the property tax increased revenue from other local sources such as other
taxes or charges.
Shadbegian (1999) examined the effects on tax and expenditure limits on county
govemments over the period 1962 to 1987. In states with limits, the per capita property taxes collected by counties are $37 Jower (a 12 percent reduction), while other
per capita taxes are $6 1ower (a 17 percent reduction) and per capita miscellaneous
revenues are $12 higher (a 7 percent increase). In other words, counties substitute
miscellaneous revenue (including charges and fees) for property taxes. For each
$1 reduction in tax revenue, there is a $0.27 increase in miscellaneous revenue. In
the states with the Jeast stringent limits, miscellaneous revenue increases by $0.51
_for each $1 reduction oftaxes, In the states with the moststringent limits, tax limits
appear to decrease misce!laneous revenue along with taxes.
545
Dye and McGuire (1997) studied the effects of property tax limits in Chicago
collar counties, showing that the limits decreased the growth rate of property tax revenue. The limits were imposed on jurisdictions with above-average revenue growth
rates, and the limits narrowed the gap between the growth rates of capped and uncapped jurisdictions. Por example, the tax growth rate in capped municipalities was
14.57 percent before the limits, compared to 9.03 percent in uncapped municipalities. A few years after the limits had been irnplemented, the growth rate in capped
municipalities had dropped to 7.14 percent, compared to 7.55 percent in uncapped
municipalities. Controlling for other factors, the lirnits decreased the growth rate of
property tax revenue by 4.63. percentage points for municipalities and by 3.61 percentage points for school districts.
INTERGOVERNMENTAL GRANTS
This part of the chapter explores the econornics of intergovernmental grants, exarnining how local governments respond to transfers of funds from higher Jevels
of government. As we saw earlier in the chapter, intergovernmental grants provide
about two-fiftbs of the revenue of local government and about one-fourth of the revenue of municipalities. Intergovernmental grants support dozens of local programs.
The programs receiving the most grant money are education (about 57 percent of
all intergovernmental grants), public welfare (13 percent), housing and community
(4 percent), highways (4 percent), and health and hospitals (4 percent).
Figure 20-6 shows how the intergovernmental revenue for municipalities is
divided among various spending programs. About one-fiftb of grant money supports
the general ,operations of local government, and another fiftb supports education.
. Tbe two redistributional programs-public welfare and housing and community
development-together receive just over a quarter of municipal grant money.
Public welfare
17%
General support
22%
Housing and communiry
development
Highways
8%
9%
. . SoUTCe: U.S. Bureau of che Census, Compendium of Government Finances, no. 5 (Washington, DC: U.S.
Govemmenc Printing Office, 1997).
546
Why don't local govern.ments pay their own way, supporting their spending
programs with local taxes? There are two principal reasons (Aronson and Hilley,
1986). First, intergovern.mental grants can be used to internalize interjurisdictional
spillovers, as discussed in the previous chapter. Second, local tax instruments are
relatively inflexible, causing mismatches between local tax revenue and local expenditures.
The mismatch problem occurs if the desired spending on local public goods
rises faster than the tax base. Suppose that local public goods are financed with the
property tax, and property values rise half as fast as income. If the desired spending
on local public goods is roughly proportional to income, the desired spending will
increase faster than property values, and there will be a mismatch between desired
spending and local tax revenue. The mismatch occurs because property values are
relatively sluggish and the tax rate is fixed.
There are two possible responses to the mismatch between taxes and desired
spending. The obvious response is to increase the tax rate. The alternative approach
is to transfer surplus federal tax revenue to the local govern.ment. The federal income
tax will generate a surplus if (1) the income tax has a progressive rate structure and
(2) federal spending is roughly proportional to income.
.-
'\
Unconditional Grants
An unconditional grant is a lump-sum grant with no strings attached. The local
Suppose that the state gives the city an unconditional grant of $20 per household.
The city makes the following announcement to its citizens: "The state has just given
us $20 per household. Tell us how rnuch to spend on public services, and we'll cut
taxes to distribute the rest of the grant to you." The grant shifts Marian's budget
line upward, from AB to CD, increasing her desired spending on public goods to
$30 (point F). The city, adhering to the preferences of the median voter, increases
its budget by $5 and cuts taxes, thus distributing the rest of the grant ($15) to its
citizens. The tax cut allows Marian to increase prvate consumption to $60.
One of the predictions of the median-voter model is that an unconditional grant
has the same effect on local spending as an equal increase in per capita income.
In fact, there is conclusive evidence that cities spend a relatively large fraction of
, ,
'-- /
547
90
70 A
"o
g,
60 f-------'"'---4.:
g:
".
e
o
.., 45
e
'6
e
30
15
25 30
70
D
90
An unconditional grant of $20 per capita shifts the budget lioe of the median voter frorr
AB to eD. The utility-rraximizing combination of public and private goods moves from
point E to point F. Spencling on local public goods increases by $5, and spending on
private goods increases by $15. Tbe local goverrunent responds ro tbe grant by cutting
local taxes, allov;ing citizens to spend tbree-fourths of the grant on private goods.
548
program-specific grants within each expenditure group. For example, the $1.94 billion in education grants that went to local governrnents in 1992 included specific grants for remedia] reading, school libraries, special education, and otber
programs.
Suppose tbat tbe state gives the city a lump-sum grant of $20 per capita for
special-education programs. Figure 20-8 shows Marian's preferences for special
education and all otber goods. "Otber goods" includes private goods and other local
public goods. The grant shifts Marian's budget line from AB toAGD. Point Gis
in tbe new budget set because Marian could spend all of her own money on otber
goods and use tbe $20 grant to support special education. As spending on special
education increases above $20, there is a dollar-for-dollar trade-offbetween spending
on special education and otber goods, so tbe slope of the budget line is l.O.
90
''
''
'
'
'', G
70 A
"o:, 60
1
1
---_1-.
"
.. 45
.5
"=
"o.
1
1
.------,
"'
1 1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
10
B
2025 30
70
Spending on special education ($)
90
A conditional grant of $20 per capita shifts !he budget line of !he median voter from AB to
AGD. The utility,maximizing combinationmoves from point E to point.F: one-fourth of
!he grant is spent on special education. The city uses the grant to decrease its contribution
to special education, providing more money to spend on other goods. Because !he desired
spending on special education exceeds !he grant, !he conditional grani is equivalent to an
unconditional gran!.
1
1
549
The grant increases Mruian's desired spending on both special education and
other goods. If the city adheres to the preferences of the median voter, spending on
special education increases from $25 to $30 and spending on other goods increases
from $45 to $60. One-fourth of the grant is spent on special education, and threefourths is spent on other goods.
The city can spend part of the grant on other goods because it decreases its own
contribution to special education. Before the grant, $25 oflocal tax money was spent
on special education. After the grant, spending on special education is $30, and the
city can pay for its special-education program with the $20 grant and $10 of local
tax money. The grant frees up $15 worth of local tax money, which can be spent on
other local public goods and private goods.
In Figure 20-8, the conditional grant is equivalent to an unconditional grant.
Because the desired spending exceeds the grant, the community can replace its own
spending on special education with the grant: in effect, the grant has no strings
attached. In graphical terms, the budget lines of the two grants are identical as long
as the desired spending exceeds the grant (as long as the city chooses a point to the
right of G).
The conditional grant is not equivalent to the unconditional grant if the desired
spending on special education is less than $20. Suppose that the city starts with
spending $5 on special education, and the desired spending increases to $1O under
the unconditional grant. In other words, the city moves to a point midway along the
segment C G. In contrast, the city is likely to choose point G under the conditional
grant. Since the city wishes to spend less than $20 on special education, it cannot
substitute the conditional grant for its own spending on special education, so a
conditional grant is not equivalen!to an unconditional grant. The unconditional
grant generates a higher utility leve!for voters in the city. The conditional grant
causes a larger i:ncrease in spending on special education.
\
1
'
--
550
70 A
,,
,1'
,,
;
'1
!,
:1
25
40
70
Spending on special education ($)
140
An open-ended matching gtant shifts the budget Jine ofthe median voter from AB to AC. The
utility-maximizing combination moves from point E to point H: three-fourths of the grant is spent
on special education. The matching grant decreases the opportunity cost of spending on special
education, providing a greater stimulus to special education than an equivalen!lump-sum grant.
551
tr:
(point E). The matching grant decreases the price to $0.50, and increases the quantity
demanded to $40 (point H). Since the grant does not affect the other determinants
of demand (income and other prices), points E and H identify two points on the
demand curve: ($1, $25) and ($0.50, $40).
1
if
.,
70 A
e
ti.'.
5o
.ll
l'
,;
l.
(
1'
::' 50
oe
45
0/l
40
..,
5
',
Cl)
\.<
l
t
',,
1
1
1
1
1''
1
1
t
f
''
',
1
1
1
1
1
25
40
1
1
1.
1
B
60
70
90 100
Spending on special education ($)
140
l.
1
f
t
(..
e,
.,_;
w.l.
A closed-ended matching grant shifts the budget line of the median voter from A B to A DI.The
utility-maxirnizing combination moves from point E ro point H. Since the desired spending leve!
($40) is Jess tban the limit ($60), the grant lirnit is irrelevant, and the closed-ended grant is equiva!ent
to .the open-ended grant If desired spending exceeded the limit, the city would spend Jess under tbe
closed-ended grant tban under the open-ended grant
-552
In Figure 20-10, the city's response to the closed-ended grant program is the
same as its response to the open-ended program. Since the desired spending leve!
($40), is less than the limit, the grant limit is irrelevant. Therefore, the closed-ended
grant is equivalen!to the open-ended grant.
The two grants are not equivalen!if the desired spending under the open-ended
grant exceeds $60. In Figure 20-:-10, the open-ended budget line is the line ADC,
and the closed-ended budget line is AD1. If the desired spending under the openended grant exceeds $60, the city will choose a point along the segment DC. In
this region, the open-ended budget line (segment DC) Jies above the closed-ended
grant (segment D1).Spending is higher under the open-ended grant for two reasons.
First, real income is higher under the open-ended grant, producing an income effect
that increases spending on all normal goods including special education. Second, the
opportunity cost of special education is lower under the open-ended grant, producing
a substitution effect that increases spending on special education.
553
k
....
)::
..,.
(:
r
r(
''"
'
.,
federal grant no longer depends on how much the state spends on welfare. There
are no matching funds or rebates for state welfare spending, so the price of a dollar
spent on welfare is $1.00.
Figure 20-11 shows the effects of welfare reform on the budget choices of a
poor state. The original budget line (shown by points A, m, and E) is relatively
fiat, refiecting the low price of welfare spending under a matching gran t. The state's
original choice is shown as point m (for "matching"): the state spends $21O million
on welfare and $260 million on other goods.The new block grant is $140 million, so
the new budget line is shown by points A, G, and D: there is a kink in the new budget
line at point G. Given the new dollar-for-dollar trade-offbetween welfare spending
and other spending, the block grant is large enough that the state has the option of
picking point m. If the original point is possible, does that mean that the state will
choose it?
Under the block grant, the state will actually spend less on welfare programs. To
maxirnize utility of the median voter, the slope of the indifference curve will be equal
to the slope of the budget line{the marginal rate of substitution will equal the price
ratio). For the original choice (point m), the slope of the indifference curve equals .
----==== ------------------....
554
the slope of the original budget line (-0.22). Under the block grant, the slope of the
budget line becomes -1.0, so the indifference curve is now fl.atter than the budget
line, and point m no longer maximizes utility. The new utility-maximizing point, b
(for "block") is the point at which an indifference curve is tangent to the new, steeper
budget line. In this case, when the price of welfare spending increases, the state's
spending on welfare prograrns decreases by $40 million, from $210 million to $170
million. Consistent with the new dollar-for-dollar trade-offbetween the two goods,
spending on other goods increases by $40, from $260 to $300.
We can explain the cut in welfare spending using the concepts of income and
substitution effects. Normally, when the price of a particular good increases, there
are two effects:
l. Substitution effect. The relative price of the good increases, and the consumer
substitutes other goods for the good that has become relatively more expensive.
2. Income effect. The price hike means that the consumer has less purchasing
power or real income, refiected in the fact that the consumer can no longer
afford the original bundle. A decrease in real income could either increase or
decrease consumption.
In the case of a switch from a matching grant to a block grant that is large enough to
make the original point affordable, there is no income effect. Because there is only
a substitution effect-which always decreases the consumption of a good that has
become more expensive-we can be sure that spending on welfare programs will
drop.
By how much will the state's welfare spending decrease? Por a poor state, the
price hike from $0.22 to $1.00 will decrease welfare spending by between 40 and
66 percent (Inman and Rubinfeld, 1997). Por a rich state, the price hike is smaller
($0.50 to $1.00), and the switch to block grants will decrease welfare spending
by between 1 and 18 percent. Congress was apparently aviare that welfare reform
would cause states to cut their welfare spending. The law requires states to continue
to spend at least 80 percent ofthe amount spent under the old matching-grant policy.
SUMMARY
l. About two-fifths of local government revenue comes from taxes on property,
income;and sales, and another two-fifths comes from intergovernmental grants.
The rest of local money comes from user fees and charges.
2. The new view of the residential property tax is based on general-equilibrium
analysis, and shows that property owners throughout a region bear the burden
of one city's property tax, and consumers outside the taxing city gain at the
expense of consumers in the taxing city.
3. The simple model of the property tax assumes that the national supply of capital
is fixed. If the supply of capital is responsive to changes in net returns, capitalists
shift part ofthe tax onto housing consumers in the form ofhigher housing prices.
.
1
555
4. In the Tiebout world, a household's property tax bill is independent of its housing
consumption: Households sort themselves into homogeneous communities, so
a household's tax bill depends only on its consumption of public goods.
5. Most modero property-tax limits restrict the growth rate of property taxrevenues.
The limits decreased property-tax revenues and were partly offset by increases
in other revenue sources.
2.
3.
4.
5.
Taxville example, compute the long-run effects ofTaxville's property tax on the
following individuals. How much do they gain or lose?
a. Ms. Partial rents a house in Taxville.
b. Mr:Wash is a traveling salesman who rents one dwelling in Taxville and one
inNotax.
c. Ms. Capitalist owns five wellings in Taxville.
d. Mr. Capitalist owns five dwellings in Notax.
In the Taxville example, a 2 percent property tax decreased net rental income
by $400 in each city. Suppose that the region has four cities instead of just two.
a. By how much will the property tax increase dwelling rent in the taxing city?
b. By how much will the tax decrease net rental income?
Choose the word in parentheses that makes the following statement correct, and
then explain your choice: As the moving cost of households decreases, the taxinduced increase in Taxville's dwelling rent
(increases, decreases).
The city of"Tinytown is one of 10 municipalities in a large metropolitan area;
Compare the incidence of a Tinytown income tax to the incidence of a metropolitan income tax.
Suppose Collegetown eliminates its property tax on pizza parlors and increases
its tax on residential property. Under what conditions will the change in tax
policy decrease the tax liability of Joe College?
REFERENCES
l. Aaron, Henry J. Who Pays the Property Tax? A New View. Washington, DC:
Brookingsinstitution, 1975.
2. Advisory Commission on Intergovemmental Relations. Tax and Expenditure
Limitations on Local Govemmnt. Washington DC: U.S. Govemment Printing
Office, 1995.
3. Aronson, J. Richard, and John L. Hilley. Financing State and Local Govemments. Washington, DC: Brookings Institution, 1986.
4. Beito, David T. Taxpayers in Revo/t. Chapel Hill, NC: University of North
Carolina Press, 1989.
556
5. Bradbury, Katharine L., Christopher J. Mayer, and Karl E. Case. "Property Tax.
Limits and Local Fiscal Behavior: Did Massachusetts Cities and Towns Spend
Too Little on Town Services under Proposition 2 1/2?" Federal Reserve Bank
ofBoston Working Paper, February 1997.
6. Citrin, Jack. "Do People Want Something for Nothing: Public Opinion on Taxes
and Spending." National Tax Jouma/32 (1979), pp. 113-30.
7. Courant, Paul N., and Susanna Loeb. "Centralization of School Finance in
Michigan." Joumal of Po/icy Analysis and Management 16 (1997), pp. 114-35.
8. Courant, Paul, Edward Gramlich, and Daniel Rubinfeld."Public Employee Market Power and the Leve!of Government Spending." American Economic Review
69 (1979), pp. 806-17.
9. Courant, Paul, Edward Gramlich, and Daniel Rubinfeld. "Why Vters Support
Tax Limitations: The Michigan Case." National Tax Joumal 38 No. 1 (1980),
pp. 1-20.
10. Cutler, David M., Douglas W. Elmendorf, Richard Zeckhauser. "Restraining
the Leviathan: Property Tax Limitation in Massachusetts." Joumal of Public
Economics 71 (1999), pp. 313-34. .
11. Dye, Richard, and Therese McGuire. "The Effect of Property Tax Limitation
Measures on Local Government Fiscal Behavior." Journa/ of Public Economics
66 (1997), pp. 469-87.
12. Filirnon, R.; T. Romer; and H. Rosenthal."Asymmetric Information and Agenda
Control: The Bases ofMonopoly Power and Public Spending." Journa/ ofPublic
Economics 17 (1982), pp. 51-70.
13. Inman, Robert P. "The Fiscal Performance ofLocal Governments: An Interpretive Review." In Current Issues in Urban Economics, ed. Peter Mieszkowski
and Mahlon Straszh"eim. Baltimore: Johns Hopkins University Press, 1979,
pp. 270-321.
..
14. Inman, Robert P., and Daniel L. Rubinfeld. "Rethinking Federalism." Jouma/
ofEconomic Perspectives 11:4 (Falll997), pp. 43-65.
15. Pechman, Joseph A. Who Paid the Taxes, 1966-1985? Washington, DC: Brookings Institution, 1985.
16. Preston, Anne E., and Casey Ichniowski. "A National Perspective on the Nature
and Effects oftheLocalPropertyTax Revolt 1976-1987." National Taxloumal
44 No. 2 (1991), pp. 123-45.
17. Sexton, Terri, Steven M. Sheffrin, and Arthur O'Sullivan. "Proposition 13:
Unintended Effects and Feasible Reforms." National Tax Jouma/ 52 (1999),
pp. 99-112.
18. Shadbegian, Ronald J. "Do Tax and Expenditure Limitations Affect Local
Government Budgets? Evidence from Panel Data." Pub/ic Finance Review 26
(1998), pp. 218-36.
19. Shadbegian, Ronald J. "The Effect ofTax and Expenditure Limitations on the
Revenue Structure of Local Government, 1962-87." Nationl Tax Joumal 52
(1999), pp. 221-38.
20. Shires, Michael A. Pattems in California Govemment Revenues since Proposition 13. San Francisco: PublicPolicy Institute of California, 1999.
._)
APPENDIX
Tools of Microeconomics
..
This appendix provides a review of sorne of the microeconomic tools that are used throughout the
text.The first part reviews the model of supply and demand. The second and third parts show how
to use indifference-curve analysis to explore the decisions of consumers and workers. The fourth
part explalns the model of producer input choice, using isoquants and budget lines. The fifth part
reviews the notion of marginal decision making.
Appendix
558
FIGURE A-1 Market Demand Curve
--------------------
e
1
1
1
1
".o": 20 ---------------------r----- e
1
";"E
1
1
1
1
1
1
1
Demand
1
1
1
1
1
1
200
Quantity of sbirts per minute
260
The demand curve shows the relationship between price and quantity demanded,
ceteris paribus (everything else equal).
559
------------------ e
200
250
2. Price of a substitute. An increase in the price of a substitute good will shift the demand curve
to the right. Suppose that the curve shown in Figure A-2 is the demand curve for cotton shirts. An
increase in the price of polyester shirts would shift the demand curve for cotton shirts to the right;
at every price; more cotton shirts would be demanded. This is sensible because an increase in the
price of polyester shirts will cause sorne polyester consumers to switch to cotton shirts.
3. Price of a complement. An increase in the price of a coniplementary good will shift the
demand curve to the left. If the curve shown in Figure A-2 is tlie demand curve for cotton shirts,
an increase in the price ofpants will shift the demand curve to the left (from Du to D'); at every
price, fewer cotton shirts will be consumed. This is sensible because an increase in the price of
pants increases the cost of a dress outfit (shirt and pants), so consumers will demand fewer outfits
and thus fewer shirts.
4. Number ofpeople. One oft.lJ.e variables held fixedin drawing the market demand curve is the
number of people in the relevant market area'. In Figure A-2, an increase in the number of people
will shift the demand curve to the right, from D 1 to Du; at every price, more shirts will be demanded.
r'
(
l
..
(
(
l.
l:
The market supply curve shows the relationship between the price of a particular good and the
quantity of the good supplied by firms. To draw the market supply curve, we change the price
and observe how producers respond, assuming that everything else that affects the production of
the good remains fixed. In Figure A-3, the market supply for shirts shows that an increase in
the price of shirts (from $20 to $30) increases the quantity supplied from 100 (point b)
to 200
:..
(point
e).
Accor
ding
to the
law
of
suppl
y, an
increa
se. in
price
increa
ses
the
quanti
ty
suppli
ed,
ceteris
paribu
s.
There
fore,
the
supply
curve
is
positiv
ely
slope
d.
560
Appendix
30
e
-------------------
,:!
20 ---------
....o
b
1
1
1
1
1
1
1
1
1
1
1
1
200
Quantity of shirts per minute
100
The supply curve shows the relationship between price and quanticy supplied,
ceteris paribus.
What sort of changes would cause the market supply curve to shift? A change in any one of
the variables held fixed in drawing the supply curve would cause the entire curve to shift.
'
L Input prices. An increase in the price of an input used to produce a particular good will
increase the cost of production, causing firms to supply a srnaller quantity at each price. In Fig
ure A-4, the supply curve for cotton shirts would shift to the left (from S" toS') if (a) the price of
labor increased, (b) the cost of cotton increased, or (e) the cost of shirt-making machinery increased.
2. Production technology. A tecllnological innovation that decreases the cost of production
will shift the rnarket supply curve to the right; at each price, firms will supply a larger quantity because it is less costly to produce the good. In Figure A-4, a cost-saving innovation in the
production of shirts would shift the supply curve from S', to S".
561
S"
.
-
100
200
Quantity of shirts per minute
250
The supply curve shifts as a result of a change in input prices ora change in production
technology.
Market Equilibrium
\
An equilibrium is a situation in which there is no tendency for things to change over time. In
the market for shirts, equilibrium means that there is no tendency for the price of shirts to change
over time; tomorrow's price will be the sarne as today's price. At the equilibrium price, firms
produce the quantity of sbirts that consumers want to buy; the quantity demanded (by consumers)
equals the quantity supplied (by producers). Once the market reaches the ec.juilibrium price, there
is no tendency for the price to change because there is neither a shortage nor a surplus of shirts.
We can use demand and supply curves to depict the market equilibrium. In Figure A-5,
the demand curve- shows the negative relationship between the prie of shirts and the quantity of
sbirts demanded; an increase in price decreases the quantity of shirts demanded. The supply curve
shows the positive relationship between the price of sbirts and the quantity of sbirts supplied; an
increase in price increases the quantity of shirts supplied. The market equilibrium is shown by
the intersection of the supply and demand curves. At a price of $30, the quantity supplied (from
the supply curve) equals the quantity demanded (from the demand curve), so there is neither a
shortage nora surplus of sbirts. Consumers want to buy 200 shirts, which is exactly the quantity
that producers want to sell.
Shortage or Surplus
_ If the price is less than_ the equilibrium price, therewill ]:e a shortage of shirts. For example, if
the price is $20, the inarket will be-at p-u'rntc on the demand curve (consumers would !ike to buy
260 sbirts) and point b on the supply curve (producers are wil!ing to sell only 100 shirts). In this
562
Appendix
-----------------------j
1
1
1
1
30
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
----------- ----
..,
"e
20
--------
--,-----,----
'"'
1
1
1
1
1
1
1
1
1
1
Demand
100
Quantity of shirts per minute
At the market equilibrium (point e, with price = $30 and quantity = 200), the quantity
supplied equals the quantity demanded. At a price less than the equilibrium price
(e.g., $20), there is ashortage: the quantity supplied (100 at point b) is less than the
quantity demanded (260 at point e). Ata price exceeding the equilibrium price
(e.g., $40), there is a Sllf>lus: the quantiry supplied exceeds the quantity demanded.
case, there is a shortage oran excess demand of 160 shirts; the quantity demanded (260) exceeds
the quantity supplied (100) by 160 shirts. Because consumers want to buy more shirts !han firms
are willing to supply, we would exp,ect the price of shirts to increase over time. Tomorrow's price
will be higher !han today's price.
An increase in the price of shirts will cause the market to move upward along both the supply
and demand curves. An increase in the price of shirts allows firrns in the shirt industry to outbid
other industries for the inputs required to produce shirts (machines, labor, land, raw materials),
so the output of the shirt industry will increase. In other words, the market moves 'upward along
the supply curve; as the price increases, the quantity supplied increases. On the demand side, the
increase in price decreases the quantity of shirts demanded; the market moves upward along
the demand curve. In combination, the movements along the supply curve and the demand curve
narrow the gap between the quantity supplied and the quantity demanded. The market continues to
move upward along the two curves unti!the shortage is e!irillnated. This occurs at point e, where
the quantity supplied equals the quantity demanded at a price of $30.
If the initial price exceeds the equilibrium price, there will be a surplus of shirts. For example,
if the price is $40, the market will be at point j on the demand curve (consumers would like to
buy 140 shirts) and point k on the supply curve (producers are willing to sell300 shirts). In this
case, there is a surplus oran excess supply of 160 shirts; the quantity supplied (300) exceeds the
quantity demanded (140)-by 160 shirts.Because producers wano sell more shirts than consumers
"""'
,.
,:w:
563
lf
lr
are willing to buy, we would expect the price of shirts to decrease over time; tomorrow's price will
be lower !han today's price.
As the price decreases, the market moves downward along both the supply and demand curves.
A decrease in the price of shirts means that the shirt industry will be outbid by other industries for
sorne of the resources used to produce shirts (machines, labor,land, raw materials), so the output
of the shirt industry will decrease. In other words, the market moves downward along the supply
curve; as the price decreases, the quantity supplied decreases. On the demand side, the decrease in
price increases the quantity of shirts demanded; the market moves downward along the demand
curve. In combination, the movements along the supply curve and the demand curve narrow the
gap between the quantity supplied and the quantity demanded. The market will continue to move
downward along the two curves until the surplus is elirninated. This occurs at point e, where the
quantity supplied equals the quantity demanded ata price of $30.
35 ---------------1
30------------------- ------1-1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
200
Quantity of shirts per minute
250
270
An increase in the demand for sbirts sbifts the demand curve to tlie righL At the initial price
($30), there is a shortage: the quantity demanded (point f) exceeds the quantity supplied
(point e). The resu!ting increase in price causes the market to move upward along the
supply urve and the new demand curve until equilibrium is restored at point g.
564
Appendix
Original supply
New supply
30-------------------
26--------------
"
.o...
-u"
"""
200 230
Quantity of shins per minute
An increase in the supply of shins sbifts the supply curve to the right. At the initial price ($30),
there is a surplus: tbe quantity supplied (point f) exceeds the quantity demanded (point e). The
resulting decrease in price causes tbe market to move down ward along tbe demand curve and tbe
new supply curve until equilibrium is restored at point g.
'.
price is $30 and quantity is 200 shirts per minute. Suppose that the demand for shirts increases,
shifting the market demandcurve to the right. As explained earlier, this increase in demand could
be caused by an increase in income, a. change in the price of a related good, or an increase in
population. At the initial price, there is a shortage of shirts, so the equilibrium price will increase.
We could also use the supply and demand curves to show the market effects of a change in
supply. In Figure A-7, a rightward shift of the supply curve would generate a surplus of shirts at
. the original price: the quantity supplied will exceed the quantity demanded. The new equilibrium
(shown by the intersection ofthe demand curve and the new supply curve) has a lower price and
a larger quantity..
'
, .J
'--'
,_j'
;--.,
565
goods). The consumer's problem is to find the affordable consumption bundle that maximizes hls
utility.The analysis ofthe consumer-choice problem can be divided into two parts. An indifference
curve summarizes the consumer's subjective preferences about alternative consumer goods. The
budget line shows the lirnits (constraints) imposed on the consumer.
00
o
. .<:
..o.
----1
-----r -
"'
"
o
"C
1
1
1
1
1
1
1
1
1
1
1
1
3
Quantity of harnburgers
The indifference curve shows the different combinations of hamburgers and hot dogs
that generate the same utility leve!.
.,
566
Appendix
combinations, we have an indifference curve. Mary will be indifferent among al!the combinations
along the indifference curve.
The indifference curve that passes through point C divides the combinations of hamburgers
and hot dogs into three groups.
1. Superior combinations. Al! combinations above the indifference curve generate more satisfaction (higher utility) than combination C. For example, Mary would prefer point E or F
to point C.
2. Inferior combinations. Al! combinations below the indifference curve generate less satisfaction (lower utility) than combination C. For example, Mary would prefer point C to point
H or point l.
3. Equivalen!combinations. Al!combinations along the indifference curve generate the same
satisfaction (the same utility) as combination C. For example, Mary would be indifferent
between combinations B, C, and D.
Because indifference curves sbow the subjective preferences of consumers, the shapes of
indifference curves vary from one consumer to another. Nonetheless, the indifference curves of
al! consumers share two characteristics: They are negatively sloped and convex.
Why are indifference curves negatively sloped? If the two commodities are "goods" (more
ofthe commodity is better), not "bads" (more ofthe commodity is worse), the indifference curve
will be negatively sloped. In other words, if we give Mary an additional hamburger, we must take
away sorne hot dogs to restare her original utility leve!. The slope of the curve is the marginal
rate of substitution (N.!RS) between the two goods. It shows the rate at which Mary is willing to
substitute one good for another. In Figure A-8, if Mary starts at point C and we give her one more
hamburger, we musttake away one hot dog to keep her on the same indifference curve. Therefore,
the marginal rate of substitution near point C is 1.0. Mary is willing to trade one hot dog for one
hamburger, so her subjective trade-off between the two goods is 1.0.
The slope ofthe indiffer<:_lce curve decreases (in absolute value) as the quantity ofhamburgers
increases. In other words,_the indifference curve becomes flatter as Mary's hamburger consumption
increases. In Figure A-8; the slope (the rise over the run) near point B is 2.5; if we give Mary
one more hamburger, we mst take away 2.5 hot dogs to keep her on the same indifference curve
(to keep her utility constant). Therefore, the MRS near point B is more than twice the MRS near
point C. The MRS nearpoint D is 0.30; ifwe give Mary one more hamburger, we must take away
only0.30 hot dog to keep her utility.constant.
Because the slope of the indifference curve decreases as hamburger consumption increases,
the indifference curve is convex to the origin. The convexity results from the assumption of
dirninishing marginal utility. The marginal utility ofhamburgers is defined as the increase in Mary's
utility per additional hamburger. Suppose that as Mary's hamburger consumption increases, the
marginalutility ofhamburgers decreases. As we move down Mary's indifference curve (increasing
hamburgersand decreasing hot dogs to keep her utility constant), we must take away progressively
fewer hot dogs to keep her at the same utility leve!because she gains progressively less utility per
additional hamburger. At the same time, as Mary's hot-dog consumption decreases, the marginal
utility of hot dogs increases, so a given decrease in the number of hot dogs has a larger effect
on her utility leve!. Therefore, it takes progressively fewer hot dogs to offset a given increase
in utility from an additional hamburger. The combined effect of decreasing marginal utility of
hamburgers (as hamburger consumption increases) and increasing margina!utility.ofhot dogs (as
hot-dog consumption increases) means that the subjective trade-off between hamburgers and hot
dogs decreases. In other words, the indifference curve becomes flatter and the marginal rate of
substitution decreases.
'
k
K
J
---Iz
l
...
t
,.
r
e
r
1.
Quantity of hamburgers
An indifference map shows a set of inclifference curves, with utility
An Indifference Map
An indifference map is a set of indifference curves. To construct an indifference map, we pick
severa! starting points and then draw indifference curves through each point In Figure A-9, the
starting points J, Kand L generateindifference curves 1 1, [z, and iJ. As Mary moves from any
point on the indifference curve 11 to any point on [z, her utility increases. To show that her utility
is higher along [z, al!we need to show is that she prefers one point on [z to one point on [.
Point K (on 12) is obviously better than point J (on [) because K provides more hot dogs and
more hamburgers. Under the assumption that both commodities are "goods," she will prefer the
combination with larger quantities of both cornmodities (shown by point K). In general, Mary's
utility increases as she moves in the northeasterly direction to higher indifference curves (from 11
to [z to !3 , and so on).
Appendix
568
FIGURE A-10 Budget Set and Budget Line
60 M
"O
_g
'O
l:- 30
o" 20
1
1
1
1
1
1
1
20
Quantity of harnburgers
40
The budget Iine shows aJl the combinations ofhamburgers and hot dogs that exhaust Mary's
budget. In Figure A-10, the budget line is the Iine connecting points M and N. Point Mis the
combination that occurs if Mary spends her entire budget on hot dogs; if the price is $2 and she
spends her entire $i20 budget on hot dogs, she will get 60 hot dogs. Point N is the combination
that occurs if Mary spen"ds her entire budget on hamburgers (40 hamburgers from spenc!ing $120
on $3 hamburgers).lf Mary spends sorne money on each good, she can reach the points between
M and N. Forexamp1e, she could reach point Q (20 hamburgers and 30 hot dogs)by splitting her
budget equally between the two goods.
The slope of .the budget line is the market trade-off between hamburgers and hot dogs. It
shows the rate at which the consumer can trade hamburgers for hot dogs, given the market prices
ofthe two goods. Starting frompoint M, ifMary gives up one hamburger, she frees up $3 that can
be spent on hot dogs.lf the price of hot dogs is $2, the $3 will buy 1.5 hot dogs, so the slope of the
budget line is -1.5; the "rise" is 1.5 hot dogs and the "run" is 1 hamburger. In general, the slope
is equal to the price of hamburgers (the good on the horizontal axis) c!ivided by the price of hot
dogs (thegood on the vertical axis).!fthe prices ofthe two goods do not change as Mary changes
her consumption pattern, the budget line will be a straight Iine (the s1ope will be constan!).
Changes in the prices ofthe two goods will change the budget Iine. In Figure A-11, a decrease
in the price of hamburgers from $3 to $2 will shift the Iine from MN to MT. The original vertical
intercept (point M) is still in the budget set: if Mary spends her entire budget on hot dogs, she wiii
be unaffected by the decrease in the price of hamburgers. The horizontal intercept moves from
N (40 hamburgers) to T (60 hamburgers) because at the lower price, a given budget buys more
hamburgers.lf Mary buys sorne hamburgers and sorne hot dogs (she chooses sorne point between
the horizontal and vertical intercepts), she can afford more combinations because she spends less
per hamburger. The same argument applies to a decrease in the price of hot dogs, shown in Figure A-11 as a shift from MN to RN (the price ofhot dogs decreases from $2 to $1.50).
569
FIGURE A-11 Changes in Prices Shift the Budgel Line
80 R
60
o"'
"o'
'"O" 40
-o
="'
Cl
20
.,
'
20
40
60
Quantity of hamburgers
A cbange in the price of either good changos the slope of rhe budget line.
(
1
How does an increase in income affecl the budgel !ine? For a given sel of prices, an increase
in income increases-the size oflhe budgel sel and shifts the budget line oulward. In: Figure A-12,
an increase in Mary's income from $120 lo $180 shifls the budgel line from MN lo ST. She
can now afford lo buy 90 hol dogs if she buys only hol dogs (poinl S, given a price of $2) or 60
hamburgers if she buys only hamburgers (point T, given a price of $3). If she spends half of her
income on each good, she can buy 30 hamburgers and 45 hol dogs (poinl U).
Maximizing Utility
The consumer's objective is lo maximize ulility, subject lo the budget constraint. In other words,
Mary will choose the affordable cornbination of hamburgers and hol dogs thal generales the
highesl utility. In graphical 1erms, Marywill choose a point on the highest feasible indifference
curve.
In Figure A-13, Mary will choose point V(22 han1burgers and 27 hol dogs) and will achieve
the utility leve!associated with indifference curve 1 5 Why does she choose point V inslead of
poinl W, X, or Y?
1.
Point W: Mary doesn't choose this poinl for two reasons. Firsl, il is nol on the budgel line, so
it does nol exhausl her budget. Second, il is on a lower indifference curve (and thus
generates less utility) than point V.
Point X: A!though point X exhausts Mary's budgel, X Iies on a Iower indifference curve than
V, so il generales less utility than V (another poinl on the budgel line). Starting from
Appendix
570
FIGURE A-12 Changes in Income Shift the Budget Line
90 S
60 M
"C
-='O
-E
45
"
&
20
40
30
Quantity of harnburgers
60
.point X, Mary could reallocate her budget and buy more hamburgers and fewer hot dogs.
As she movedoM! her budget !ine, she moves to progressively higher indifference
curves, ultimately reaching point V and indifference curve !5
Point Y: Although point Y generales a higher utility leve!(it's on ahigher indifference curve), it
!ies outside Mary's budget set, so she cannot afford it.
At point V, Mary reaches the highest indifference curve possible, given her budget set.
The indifference curve touches-but does not pass through-the budget line. In other words, the
indifference curve is tangent to the budget line. If the indifference curve cut through the budget
!ine at sorne point (e.g., at point X), Mary could increase her uti!ity by rea!locating her budget
between the two. goods. A tangency occurs at the utility-maximizing combination.
What is the economic interpretation of the tangency condition? At the tangency, the slope of
the indifference curve (the marginal rate of substitution) equa!s the slope of the budget line (the
priceratio). Therefore, the consumer's subjective trade-offbetween the two goods (the consumer's
l'vfRS) equa!s the rate at which the two goods can be traded off in the consumer's budget (the price
ratio). At any other combination, the l'vfRS would be unequa! to the price ratio, so the consumer
could rea!locate her budget and increase her utility. For example, starting from point X, Mary's
indifference curve suggests that she is willing to give up about six hot dogs to get one hamburger
(the slope of the indifference curve (the l'vfRS) is about 6.0). Given the prices of the two goods,
the price ratio is 1.5: Mary must give up only 1.5 hot dogs to get one hamburger. Therefore, she
will increase her utility by reallocating her budget to c0nsume more hamburgers and fewer hot
dogs. She will continue to reallocate her budget until the l'vfRS equa!s the price ratio (until the
'i,
Appendix Tools of Microecooomics
'
571
:
\
r/
t
d't
r.:
f
f,
i:
)
t
(
(
l
1'
15
.<::
'O
.iO'
0"1
27
20
---h
'
----5
'1
1
l
1
1
(
1
22
40
Quantity of hamburgers
Te maximizo utility, the consumer finds the combination of hamburgers and hot dogs at
which an indifference curve is tangent te the budget line, meaning that the marginal rate
of substitution equals the price ratio.
slope of the indifference curve equals the slope of the budget line). This occurs at point V, where
MRS = 1.5.
f
l
i
572
Appendix
1.
160 --r,-1., F
1
. 1
1
1
1
1
..
1
1
1
1
g_
11
] 80
1W
--L--4-12
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
55
60
40 --T-- ------
------------
70
Leisure time per week
1
1
1
1
1
90
100
To maximize utility, the wrker finds the combination of 1eisure time and income at
which an indifference curve is tangent to the budget line. meaning that the marginal rate
of substitution equals the wage.
573
income by the hourly wage. In Figure A-14, we assume that the worker has 100 hours per week to
allocate between leisure time and income-eaming time, and the wage is $4 per hour. The horizontal
intercept (point C) is the maximum leisure time (100 hours). As leisure time decreases, income
increases by $4 per hour, from $40 at point D (90 hours of leisure and 10 hours oflabor) to $120
at point E (70 hours of leisure and 30 hours of labor), and so on.
How would a change in the wage affect the budget line? A decrease i the wage would
decrease the slope of the budget line, causing it to pivot downward. At point C, labor time and
income are zero, so a decrease in the wage has no effect. As leisure time decreases (work time
increases), a given reduction ofleisure time (a given increase in labor time) generales less labor
income. Therefore, the decrease in the wage decreases the slope of the budget line. Por example, if
the wage drops to $3 per hour, the income earned from 1O hours ofwork (90 hours ofleisure) would
decreasefrom $40 to $30, so the budgetpoint associated with 90 hours ofleisure would drop by $10.
Maximizing Utility
'\
'
The worker will maximize utility at point E because it is the combination ofleisure and income that
lies on the highest feasible indifference curve. The indifference curve is tangent to the budget !ine,
so the worker's subjective trade-offbetween leisure and income (the marginal rate of substitution)
equals the wage. At the optimum com)lination, the marginal rate of substitution will equal4.0 (the
slope of the budget line). Por any other point along the budget line, the worker could reallocate his
time and increase his utility. Por example, starting with point F, the marginal rate of substitution
is about 9.0, meaning that the worker is willing to give up $9 of income to get one more hour
of leisure time. Because he gives up only $4 for every additional hour of leisure time, he will
move down the budget line toward point E. He will continue to trade income for leisure unti! his
subjective trade-off (the MRS) equals the wage, meaning that be will stop at point E.
How would an increase in the wage affect the amount ofleisure time? An increase in the wage
has two effects on the worker's choice between leisure and labor time. The firstis the income effect:
an increase in the wage means that the worker has a higher real income. The worker will consume
more of all "no al" goods, including leisure. Therefore, the income effect provides upward
pressure on leisure time and downward pressure on work time. The second effect of an increase in
the wage is the substitution effect, or the opportunity-cost effect. An increase in the wage increases
the opportunity cost of leisure time: given the higher wage, each hour of leisure time causes a
largerreduction ofincome. As the opportunity cost ofleisure time increases (as the benefit ofwork
time increases), the worker will tend to spend less time on leisure and more time working. In other
words, the substitution effect provides downward pressure on leisure time and upward pressure
on labor time. It is not possible to predict, on theoretical grounds, which effect will dorninate, so
it's impossible to predict whether an increase in the wage will increase or decrease labor time.
1
f
1
1
t
t.
574
Appendix
60
40
Quantity of capital
To maximize output, the fum finds the combination of capital and labor at
which an isoquant is tangent to the budget line, meaning that the marginal
rate of technical substitution equals the price ratio.
producticii:J. budget. Lt'.s assum theprice oflaboris $4, the price of capitalis $6, and the production
budget is $240. Figure A-15 shows the finn's budget line: the slope is 1.5 (the price of capital
divided bythe price oflabor); the verticalinterceptis 60 units oflabor ($240/$4); and the horizontal
intercept is 40 units of capital ($240/$6).
The finn's objective is to" maxirnize output, subject to its budget constraint. In other words,
the firm wants to reach the highestisoquant possible, given its budget set. In Figure A-15, output
is maximized at point E, where the isoquant is tangent to the budget line. The slope of the budget
line is the market trade-off between the two inputs (the ratio of the price of capital to the price of
labor). The slope of the isoquant is the marginal rate of technical substitution (MRTS) between
the two inputs, the rate at which one input can be substituted for another without changing total
output. The tangency condition means that the MRTS equals the price ratio, that is, the technical
trade-off between the two inputs equals the market trade-off.
The Marginal Principie: Pickthe level of an activity at which the marginal benefit
of the activity equals its marginal cost.
If the marginal benefit of sorne activity exceeds the marginal cost, you will be better off if you do
more ofthe activity. You should continue to increase the leve!of an activity as long as the marginal
benefit exceeds the marginal cost.
',..
(
.
575
To explain the marginal principie, consider a problem facing Betty the beautician. Suppose
that on a particular day, Betty must decide whether to keep her beauty shop open for an extra hour,
meaning that her shop would be open for 10 hours instead of 9. To use the marginal principie,
Betty must compare the marginal benefit of staying open to the marginal cost: if the marginal
benefit exceeds the marginal cost, it will be sensible to remain open for the extra hour.
;'
l. Betty ignores the fixed costs of the beauty shop, defined as costs that do not vary with the
time she remains open. For example, her monthly rent is fixed, so she ignores this cost in
calcu1ating the marginal cost. A common error in econornic analysis is to include fixed costs
in the calculation of marginal cost.
2. Betty includes the opportunity cost of her time. Although Betty does not receive an hour1y
wage, .the labor cost associated with remaining open equals the opportunity cost of her time.
A common error in econornic analysis isto include only explicit monetary c_osts in the
computation of coses.
--'""
Index
r
1
l
[
r
1.
r
577
578
Bay Area Rapid Transit; see
San Francisco
Becker, Charles, 86, 90
Becker, Gary S., 363, 369
Beesley, Michael E., 294,317
Bebrrnan, Richard, 387, 396
Beito, David T., 542, 555
Benefit-cost rule, 248
Benefit spillovers, 512
Benefittaxation,510-512
Berry, Brian J. L., 109, 118
Bid-rent function
business-residential intersect, 193
and congestion tax, 277-278
convex, 170-173
flexible/inflexible, 172-173
and income, 332
income mixing-vs. sorting, 337
and land use, 184
local public goods and, 330-333
of manufacturers, 168-170, 209-210
ofofficefirms, 173-174
peer-group sorting, 336-337
residential, 181-182, 190-192
Black, Duncan, 78, 90
Black-capitalism approach, 362-363
Block grants, 488--489
Blumenstein, Alfred, 422, 428, 429
Board.ing up, 441
388-389
Brueckner, Jan K., 188, 201
Budget elections, 517-519
Budget line, 553-554, 565,567-569,
572-573,574
Budget set, 567-569
Buenos Aires, 113
Building design, 237
Building permits, 231-232
lndex
Burglary, 402-403
Burt, Martha R., 460,461, 465
Bush, John, 86, 90
Business taxes, !24
Bus system, 301-302
Busway, 298
Button, Kenneth J., 290
Byzantine Empire, 32
e
California, 240
education ruling, 380-,381
Proposition 13, 542-543, 544
three-strikes law, 417
Capital cost, 450
light rail system, 303-304
Capital gains, 450
Capitalized project, 159-161
Carbon monoxide, 280-281
Card, David, 394
Carelli, Richard, 417, 429
Carlina, Gerald A., 51, 63,216, 226
Car!ton, D. W., 50, 63, 87, 90
Carpet manufacruring, 85-86
Carroll, Thomas, 435, 465
Case,Alne,378,394
Case, Karl E., 544, 556.
Castells, Manuel, 84, 90
Categorical grants, 547-549
Central business district, 212
bid-ret of manufacturers, 168-170
bld,rent of office firms, 173-174
convex bid-rent function, 170-173
financeindustry, 174-175
land use, 174-!76, 183-!85
!essened attractiveness, 214-215
in Los Angeles, 217-218
and subcenters, 221-'-222
Central cities, 12
b!ack-capitalism approach, .362 363
commuting patterns, 257-258
education in, 389-392
education problems, 372-375
enterprise zones, 363
increased scbool spending, 390
loss oflow-incorne housing,
461-462
poverry rate, 323, 354
problems of, 2, 189
redistribution policies, 357
school finance reform, 387
school quality, 360
schoolsegregation,387-388
lndex
L
.,.
.:-
f:
'
1
(
L
homelessness, 459-462
Clauretie, Terrence, 435, 465
job gains and losses, 65-66
Clean Air Act, 280-281
knowledge spillovers, 3
Clearance rate, 412
labor demand curve, 122-124
Cleveland
learning opportunities, 79-80
crime rate, 400, 401
location incentive programs, 81-82
educational attainment, 338, 340
market forces, 14
median income, 338, 339
mercantile,33-34
neighborhood transforrnations,
military, 30, 31
345-347
modem, 202-205
poverty rate, 323,351,353
monocentric, 167
racial composition of neighborhoods,
rnostinnovative,3
342,343
neighborbood transitions, 344--347
school dropout rates, 374, 375
nurturing environment, 52-53
Clotfelter, Charles T., 388-389, 395
outsiders and economies of, 140-141
Clustering
population increase 1800-1970, 34
of office space, 212-214
positivo and negativo aspects, 1
in subcenters, 216-217
problems of central cities, 207-208
CMSA; see Consolidated metropolitan
promoting innovation, 48
stati.stical area
reasons for existence, 19
Coa! resources, 75
regions without, 19
Coase, Ronald H., 245
religious, 30-31
Coase solution, 245
of Reman Empire, 31-32
Cognitivo acbievement. 375-376
size distribution of, 105-107, 109-112 Coinage, 31
spatial variations, 186
Co!lection phase oftrips, 295
sports stadiums, 82-84
Co!lection time cost, 296
telecornmunications and future of,
College graduates, 4
58-60
dissimilarity lndex, 79-80
trading cities, 20-23
Colorado, 240
trafficcongestion,258-262
Cornmercial development, 241
unzoned,244--245
_ Cornmercial property tax, 537-538
urban giants, 112-115
Cornmunications technology, 2!4-215
urban renewal, 487-488
Cornmunity development
effects of urban renewal, 488
varying in size and scope, 39-40
wage levels, 4
purposes, 487
world's largest, 12
recentprograms,488-489
Citrin, Jack, 556
urban renewal, 487-488
City size
Cornmunity Development Block Grants,
and crirne rates, 421-422
488-489
Cornmunity service system,412
effect of local employment, 56
and labor productivity, 51
Com.muting
by automobile, 257
Iimits to cornmuting costs, 27-29
choosing a travel mode, 294-300
limits to freight cost, 26-27
limits to production economies of
co!lection phase, 295
scale, 27
distribution phase, 295
and localization economies, 53
exarnple of modal choice,295-297
multiplier process and, 139
Iine-haul time, 295
and urbanization economies, 53-55
mass transit vs. cars, 297
variation in, 54
by working peor, 357-359
and variety of goods, 55-56
Cornmuting costs, 259
and wage increases, 28
automobiles, 300-301
City-states, 31
Bastan experience, 28-29
Clark, Williarn A., 342, 349
and city size, 27-29
579
lower, 206-207
in primary cities, 113
private vs. social, 260-261
time cost, 185
trade-offbetween land and, 186-189
and wage increases, 28
and wages, 183
Comrnuting pattems, 204
Comparativo advantage, 20-22
Comparison shopping, 57-58
Compensation rules, 248-249
Competition
effect on land rent, !57
efficiency trade-offs, 95-97
for monopolist, 94-95
Competitivo markets, 168
Complement, 559
Complementary goods, 58
Computer industry, 52-53
Conditional grants, 547-549
Congestible public good, 324
Congestion, 258-262
land-use pattems, 277-280
Congestion extemality, 260-261
Congestion tax, 262-267
alternativos to, 267-273
benefits and costs, 262-264
to decrease traffic volume, 268
estimates of, 265-266
for bighway financing, 275-277
implementing, 266
peak vs. off-peak travel, 264-265
pricing experiments, 266-267
and residentialland, 277-279
and urban labor market, 280
Consolidated metropolitan statistical
area. 8-10, 204
Constan!multipliers, 138-139
Constant returns to scale, 23-26
Constan!wage, 130
Consumer choice, 564-571
Consumer goods, 55-56
Consumer subsidies; see Demand-side
policy
Consumer substitution
convex housing price function,
179-181
linear housing-price function, 177-179
and residential density, 182
Consumer surplus, 264
Consumer variety, 5
Consumption, 5
nonrivalrous, 324
-lndex
580
Consumption patterns, 133-134
Contracting for transit services, 309
Conversion of housing, 441, 496
Convex bid-rent function, 170-173
Convex housing-price function, 179-181
Cook, Thomas D., 430
Corn Law debate, 159-161
Corporate headquarters, 42
Cost-benefit analysis of labor pooling,
44-46
Cost of living elasticity, 125-126
Cotton baling, 69
Coulson, N. Ed, 313,317
Courant, Paul N., 387,395, 543,547,556
Court system, 412-413
Crack cocaine, 422
Crandall, Robert W., 283, 285,289
Creation, 2-3
Crime, 397-425; see also Crime
prevention; Prison system
anguish cost, 404
arrest ratio, 411
burglary example, 402-403
in central cities, 208
city size and rates, 421-422
costs of, 401-402
and court system, 412-413
disposition offelony cases, 413
drop in rates in 1990s, 422-423
and drugs, 424-425
economic approacb to, 397
effects of equalizing penalties,
419-421
facts and statistics, 397-402
and housing prices, 399-401
in bousing projects, 476, 477
index crimes, 397-398
juvenile vs. adult penalties, 417-418
marginal cost, 407-408
moral-bazard problem, 410-411
morality and aversion to, 403-404
opportunity cost of prison, 403
optimum arnount of, 405-408
parole system, 415
and police services, 509
prevention cost, 405-407
principie of marginal deterrence,
418-421
rational criminal model, 402-405
recidivism rate, 414
supply curve for, 404-405
total cost of, 407
and unemployment, 409
"9;
lndex
581
and real per capita income, 143-144
and urban diversity, 51
Employment multiplier, 134
Employment rate
effects of employment growth,
141-143
increases in, 144
Employment ratios, 217
Energy orientation, 75
England
Corn Law debate, !59-161
growth of cities, 34
mass transit deregulation, 312
woolen industry, 23
Engle, Roben F., 313,317
Enterprise zones, 363
Entry; see Market entry
Environmental Protection
Agency, 281
Environmental quality, 126, 130-132
zoning for, 232-236
Epple, Dennis, 391, 395
Equal productivity, 20
Equal protection clause, 247-248
and education, 381
Equilibrium, 94-95, 96, 97, 561
Equilibrium employment, 140-141
Equilibrium land rents, !57
Equilibrium market areas, 97
Equilibrium price ofland, 238
Equilibrium wage, 126
Escudero, Emilio, 313,317
Euclid v. Ambler, 246
Evans, William N., 386, 395
Everett, Caro!T., 269,290
Exclusionary zoning, 238-240
coun rulings, 247-248
and propeny tax, 334-335
and racial segregation, 344, 362
Expected crash rate, 285
Experimental Housing-A!lowance
Program, 485-486
Expon industries, 134-135
Expon-oriented growth theories,
139-140
Expon production, 55-56
Exports, demand for, 124
Express Toll Road, 266-267
Externa! economies
evidence of localization econornies,
49-51
eviderice of urbanization economies,
51-52
582
Externa] economies of scale, 39
in urbanization economies, 49
Externalities
automobiles, 281-282
Coase solution, 245
of education, 507-509
from fire protection, 509
of high-density housing, 237
housing abandonment, 443
housing consumption, 468
of industry, 232-236
intejurisdictional, 514, 516
of local government, 507-509
public safety, 509
renter, 451-452,455-456
of retailing, 236
shopping, 56
of traffic congestion, 260--261
vs. demand diversity, 514
Extemality zoning, 233-236
Externa] trip coses, 261
F
Face-to-face communication, 58-60
Faccors ofproduction, fixed, 168
Factor specialization, 24
Factor substitution, 139, 170--173,
181-182
and residentia] density, 182-183
Factory cities
limits to city size, 2 27
scale economies, 23-26
Fair market rent, 479
Fare-box ratios, 305, 307
Fare increases, 294
Federal Bureau of!nvestigation,
397-398, 410-411
Federal housing policies, 344
Federalism, 512-517
Federal safety legislation, 283
Federal Transit Adrninistration, 309
Fe!drnan, Maryann P., 3, 16, 52
Female-headed households, 354, 368
Fertility analysis, !53-159
Feudal cities, 32-33
Field, David Dudley, 161
Fielding, Gordon J., 308, 318
Fifth Amendrnent, 248
Filer, Randall K., 459, 460,
461, 465
Filimon, R., 547, 556
Filtering effect, 449
downward, 483
lndex
of public housing, 478-479
upward, 484
Filtering model of housing market,
444-450
Financeindustry, 174-175
Fire protection externalities, 509
Firms
energy-intensive, 75
flexible and inflexible, 170--173
in Iocalization economies, 76
location decisions, 65--88
market areas of, 92
market-oriented, 70--71
potential profit, 65
resource-oriented, 67-70
transfer-oriented, 6 75
in urbanizati.on economies, 76
weight-!osing, 69
Fiscal neutrality principie, 380
Fiscal policy, local, 505
Fiscal problerns, central cities, 208
Fiscal zoning, 233,238-240
commercial and industrial
deve!opment, 241
developrnent fees, 240
and equal protection, 247-248
fringe Iand use, 240
high-density housing, 237
impact fees, 240
Fischel, William, 246, 247, 24S, 253
Fisher, P., 200,318
Fisher, Ronald c., 333, 349
Fishmim, Robert, 202
Fixed costs, 71, 575
Fixed-factorproportions, 67
Fixed faccors of production, 168
Fixed prices, 67, 155, 168
Fixed-rail systerns; 313-314
Fixed-rate mortgage, 456
F1eisher, Belton, 409,429
Florida, Richard, 3, 16,91
Flour milis, 75
Flypaper effect, 547
Follain, James R., 459,465
Forecastiog techniques
economic base study, 132-135
input-output analysis, 135-138
limitations, 138-141
Forst, Brian, 412, 413,429
Fort Worth, Mainstreet Project, 489
Foundation grants, 381-384
Fourteenth Amendment, 247
Fox, James AJan, 429
583
Index
Greeawood, Michael, 91
Griesiager, Harriet, 411,430
Grimes, O. F., 435, 465
Grogger,Jeff!e 409,415,422,429
Groenspecht, Howard K., 289
Guaraateed tax base plan, 384--386
Gyourko, Joseph, 492, 494, 499
H
!
'
'\
584
Housing submarkets, 495
Housing vouchers, 467,481-483
market effects, 483--485
vs. public housing, 486-487
Houston
city without zoning, 245
HOV !anes, 299
HOV system, 300
office space, 214
subcenters, 219
Hub-and-spoke streetcar system, 168
Huckster skills, 102
Hughes, James, 90
Human capital, uneven distribution of,
79-80
Hutchinson, Broce G., 64
1
lbsen, Henrik, 503
Ichniowski, Casey, 544, 556
Ideo1ogy, 520
lblanfe1dt, Keith R., 313, 317, 358, 359,
370,459,465,470,471,499
lllinois, property tax caps, 543
Impact fees, 240,241
Imperfect substitutes, 56-58
Imprisonment probabi1ity, 406
Improvement tax; see Property tax
lmpure pub!ic good, 324
lmputed renta! income, 454-455
Incapacitation, 417
Incapacitation function, 413,416
Incentive prograrns, 81-82
Incidence oftaxation, 526-532
Inclusionary zoning, 238-240
!ncome, 558
and bid-rent function, 332
black households, 342
changes in, 57!
and commuting costs, 206-207
and crime, 399
of expon workers, 120-121
household sorting by, 338-340
and marginal propensity to consume,
121-122
market area, 100-!02
peer-group sorting by, 336-338
Income effect, 554
Income elasticity of demand, 458-459
for land, 1O1
Income-location pattems
land-commuting cost trade-off,
186-!89
lndex
in other countries, 190
public policy irnplications, 192
and residential bid-rent function,
190-192
Income red.istribution, 357
rent control as, 496-497
taxation for, 505
Income segregation, 186-192
alternativo explanations, 189-190
outside United States, 190
public policy irnplications, 192
and residential bid-rent function,
190-!92
Income sorting, 360; see also Household
sorting
Incubation process, 52-53
lndex crimes, 397-398
Indifference curve, constructing.
565-566
Ind.ifference curve analysis
consumer choice, 564-571
housing choice, 437-438
local govemment revenue, 553-554
worker's choice, 572-573
lnd.ifference map, 567
Indivisible inputs, 24
Industrial bonds, 81, 130
Industrial development, 241
Industrial nuisance zoning, 233-236
Industrial property tax, 537-538
Industrial-Public services, 124
Industrial Revolution rail
transport, 26-27
urbanization during, 34-35
Industries
in localization econornies, 53
location quotie t, 133
in urbartization econornies, 53-55
Industry clusters
based on input sharing, 43
effect on employment growth, 50-51
information sbaring, 48
!nfrastructure, 114-115
Inman, Roben P., 523, 552, 554, 556
Innovation
centered in cities, 2-3
from econornic d.iversity, 52
in Industrial Revolution, 34-35
prometed in cities, 48
in telecommunications, 58-60
in war making, 34
Input coefficients, 136-137
Input-oriented fums, 108
!nput-output analysis, 135-138
input coefficients, 136-137
limitations of, 139-141
multiplier process, 137-138
transactions, 136-137
Inputs
price changes, 139
price of, 560
resource-oriented firms, 67-70
shared suppliers of, 41-43
transferable, 66
ubiquitous, 103
Inputsubstitution, 170-173
Insurance against crime, 410-411
lntegration, attitudes toward, 344-345
Intel Corporation, 84
Intercity location decision, '80-81
lntercity transportation, 25
Intercity truck, 209
lntergovemmental contracts, 513
Intergovernmental grants, 545-554
block grants, 552-554
categorical, 547-549
closed-ended matching grants,
551-552
for community development, 488-489
cond.itional, 547-549
flypaper effect, 547
rnismatch problem, 546
open-ended matching grants, 549-551
prograrn shares, 545
stimulative effects, 552
uncond.itional, 546-547
and welfare reform, 552-554
Inteijurisd.ictional externalities, 514, 516
!nteijurisdictional mobility, 328-330
Inteijurisd.ictional spillovers, 329
Intermed.iate goods, !28
Intermed.iate inputs, 49
lnternational trade, 113-114
Interstate highway system, 209
lntracity location decision, 80-81
lndex
585
Intracity truck, 208-209
Ishii, Ryoichi, 52, 63
Isoquant, 573-574
Ita!ian city-states, 4-5
l
1
J
Jackson, Jerry, 201
Jackson, Kenneth T., 29!
Jacobs,Jane,39,64
Jaffe, A B., 3, 16
Japanese automobile finns, 84-85
Jargowsky, Pau!A., 351, 370
Jensen, Jeff, 435, 465
Jiminez, Emmanuel, 459,465
Jitneys, 310-311
Job development prograrns, 489
Job dispersa!, 2
Job gains and losses, 65-66
Jobs
distance from city center, 202-205
entry-!eve!, 357
following population, 215-2!6
and sports stadiums, 82-84
suburbanization of, 363
Joelson, Mitchell, 399, 429
Joboson,Bruce,422,429
Jardan, Stacy, 206, 207, 226
Just compensation, 248-249
Juster, F. Thomas, 499
Just-in-time systems, 84
Juvenile pena!ties, 417-418
K
Kaho, Matthew E., 2,16, 202,216,226,
356,357,370
Kain, Jobo F., 204, 342, 350, 357, 370,
377,378,392,396,435,486,499
Kalla!, Hedi D., 51, 52, 63
Katz, Bruce, 357, 366, 370
Katz, Jeffrey, 380, 395
Kazimi, Camilla, 280, 290
Keeler, Theodore E., 265, 289, 290, 300,
301-302,303,3!6,318
Kelley, Maryellen R., 52, 63
Kemp, Michael A., 294, 31.7
Kentucky, school spending, 380
Key money, 496
King, Mervyn A., 456, 465
Kish, 30
Kleiman, Mark A R., 424, 428
Knowledge spillovers, 3, 48, 52, 128
Kohlhase, Jane E., 437, 466
Ko!ko, Jed, 5, 16, 79, 90
Kraft, Gera!d, 294, 317,318
586
Land-use patterns (continue)
fringeland,240
and local public goods, 330-333
ofmass transit, 312-314
monocentric city, 167-196
and poverty, 186-188
residential, 177-183
restrictive covenants, 245
and traffic congestion, 277-280
Land-use policies, 124
Lang, Robert E., 2!2, 213, 226
Large-lot zoning, 237-238,239
Lave, Charles A., 309,317,318,319
Lave, Lester B., 289
Law of demand, 99, 100, 558, 57!
Law of supply, 559
Learning, 3-4
in !ocalization economies, 127-!29
Learning opportunities, 79-80
Lees, Lynn H., 32, 33, 38, 190, 201
Leftover principie, 157, 169, 181
Legal environment
equal protection, 247-248
justcompensation,248-249
land-use controls, 245-249
substantive due process, 246
Leicester, England,- 33
..eisure-income trade,off, 572-573
LeRoy,Stephen,201
Letellier, Johnson, 397
Levin, Henry, 39!, 395
Levitt, D., 226
Levitt, Steven D., 208, 397, 411, 4!4,
417,418,429,430
Levy, Daniel C., 395
Levy, John M., SI, 91, 129, 149
Light rai1 system, 303-304
Lillydahl, Jane H., 240, 253
Lindahl, Erik, 5!1
Lindahl model, 511-512
Linear housingprice
function, 177-179
Line-haul econornies, 71
Line-haul time, 294, 295
Line-haul time cost, 296
Linneman, Peter, 492, 494, 499
Lipsey, Mark, 414,430
Loan guarantees, 82
Local employment, 55-56
Local government, 501-555
altemative to spillovers, 516--517
benefit spillovers, 512
benefittaxation,510-512
Index
-l
'
l
'
1
'
t
'
1'
'
t
{
,.
1
(:-
,.
!
1
:
Index
t:
587
Manhattan dressmaking industry, 41-42
Manufacturers
bid-rent, 168-170
convex bid-rent function, 170-173
Manufacturing
in Industrial Revolution, 35
innovations in, 35
location decisions, 210
and nuisance zoning, 232-236
single-story plants, 211
in subcenters, 217
suburbanization of, 208-211
Marginal benefit, 326-327, 575
Marginal-benefit curve, 260
Marginal cost, 326-327,575
of crime, 407-408
ofspeed,285
Marginal cost curve, 155,276.
Marginal decision making, 574-575
Marginal deterrence principie, 418-421
Marginal principie, 574, 575
Marginal private cost, 453-454
Marginal propensicy te consume,
121-122
Marginal rate of substitution,
565-566,570
Marginal rate of tecbnical substirution,
574
Marginal revenue product, 363-364
Marginal social benefit
ofeducation,507-508
and marginal cost, 514
of natural monopoly, 505-507
Marginal social cost, 453-454
Marginal travel cost, 261
Market area, 92
Market-area analysis
algebraic model of market areas, 98
changes in demand andpopulation,
98-100
demise of small stores, 101-102
determinants ofmarket'areas, 97-103
of clifferent industries, 102-103
efficiency trade-offs, 95-97
entry and competition, 94-95
market area income, 100-102
paree] post and general stores, 100
pricing with a monopolist, 93-94
scale economies, 99
travel costs, 99-100
Market cities, 22-23
medieval, 33
Market demand curve, 558-559
Market entry
efficiency trade-offs, 95-97
monopolists, 94-95
Market equilibrium, 561
Market forces, 19-36
in development of cities, 14
hierarchical system of cities, 108
and wage discrimination, 363-365
and welfare reform, 367-368
Marketing
agglomeration economies, 56-58
tecbniques, 102
Market-oriented firms, 70-71
in central place theory, 104-107
objective, 108
Market supply curve, 559-560
Market trade-off, 568
Market value ofland, 153-154
M hall,Alfred,48,64
Marvel, Thomas B., 416,430
Masen, PatrickL, 365,370
Massachusetts, Proposition 2 , 543, 544
Mass transit
British deregulation, 312
choosing a system, 300-304
choosing a travel mode, 294-300
contracting for services, 308-311
cost ofbus system, 301-302
and costs ofauto system, 300-30!
cross subsidization, 308
deregulation,309-311
effect offare increases, 294
facts and statistics, 291-294
failure te reach suburbs, 357-359
fixed-rail systems, 313-314
HOV lanes, 298-300
land-use pattems, 312-314
light rail system, 303-304
mainline vs_ integrated, 298
natural monopoly, 308-309
noncommuting trips, 294
paratransit, 310-312
and poverty, 356-357
price elasticity of demand, 294
public ridership, 292-293
reasons for deficits, 307-308
restricting entry into market, 308-309
in San Francisco, 295-298, 302
scale economies, 308-309
service and model choice, 297-298
spacing between stations, 298
subsidies, 271-273,282, 296,
305-308
,,
r
588
.;
'
')'
'
l
'f
r
J
oi
.,
,'.
,,;
1
'
'i
Index
1
Index
589
l.
N
Nakarnura, R., 50, 51, 64
National Bureau of Economic Research,
360,409
National Housing Review, 476
National Pay Your Taxes Carnpaign, 542
National self-sufficiency, 134
National Semiconductor, 84
Natural monopo1y, 506-507
Natural arnenities, 78-79
Natural monopoly, 308-309
Navarro, Peter, 499
Neal, Derek, 379, 396
Near East, 30
Negroponte, Nico!as, 58, 64
Neighborhood conditions, 469--471
Neighborhood crime rates, 399
Neighborhood effect, 435
Neighborhood homogeneity, 323
Neighborhood-level poverty rates,
351-353
Neighborhood transitions, 344-347
Ne!son, Arthur C., 253
New Jersey, school spending, 380
New York City
dressmaking industry, 41-42
educational achievement, 372, 374
housi.ng abandonment, 443
jitneys, 310
mass transit use, 293-294
rent control, 491-495
Nitrogen oxides, 280-281
Noll, Roger G., 82, 90, 91
Nominal tax rate, 525
Nomina! wage, 113
Noncommuting travel, 185-186
Nonexcludable public good,
324-325
Nonrivalrous public good, 324
Normal good, 558
Narman Conquest, 23
Nuisance zoning, 233-237
Parker, Theodore, 1
Parking cost, 296
Parking tax, 268-269
Parody, Thomas E., 269,290
Parole system, 415
Partial-equilibrium analysis, 528-529
Partialland tax, 163
Patents, 3
Peak travel times, 264-265
Pechman, Joseph A., 535, 556
Peer group effec376, 378
Peer-group sorting, 335-338
bidding for 1and, 336-337
for schools, 335-336
vs. income mixing, 337
Peloponnesian War, 24
Peltzman, Sarn, 283, 285, 290
Pennsylvania Coal v. Mahon, 248
Per capita demand, 102-103
systematic variation, 107
Per capita income, 143-144
Performance, light rai1 system, 303-304
Performance zoning, 236-237
residential areas, 237
retailing, 236
Perloff, H., 166, 201
Permanent income, 458-459
Personal crime, 398
reductionin,422-423
Personal Responsibility and Work
Opportunity Reconciliation
Act, 365
Peskin, R., 309,318
Petersilia, Joan, 429
Phelps, Edmund, 364, 371
Philadelphia, 358
Phillips, Liad, 409, 419, 430 .
Phoenician cities, 4-5
Phoenix, community deve1opment, 489
Physical invasion rule, 248
Pickerell, Don H., 307, 3!8
Pittsburgh, 163-164, 299
Pivo, Gary, 212-213, 214, 227
Plant re!ocations, 51
Plessy v. Ferguson, 388
PMSA; see Primary metropolitan
statistica1 area
Police, 411-412
arrest ratio, 411
clearance rate, 412
community service systern, 412
new crime control techniques, 422
Policeinvestigation,412
590
Pollee patrol, 411-412
Police services, 509
Polinski, Mitchell, 459,465
Politics, and development of primary
cities, 114-115
Pollution
air,280-282
effiuentfees,232-236
and mass transit subsidies, 282
nuisance zoning, 232-236
Pollution abatement program, 130-132
Pollution-control policies, 267
Pollution rights, 281
Pollution tax, 281
Pool, I. De Sola, 63
Population
density function, 205-206
density gradient, 206, 207
followingjobs, 215-216
global trends, 60
increase 1800-1970, 34
marker area changes, 98-100
and number of factory workers, 26
in primary cities, 112-113
spatial distribution in modem cities,
202-205
suburbanization of, 205-208
urban areas, 6-7
of urbanized areas, 93
world's largest cities, 12
Population density, 103
and demise of small stores, 1O1-102
Population growth, 228-232
Porell, Frank W., 79, 91
Port cities, 73-75
Poterba, James, 91, 371
Poverry
basic questions about, 352
concentration of, .351'-353, 356-359
consequences of income sorting,
359-360
definitions of, 354-356
and educational attalnment, 360
female-headed households, 354, 368
high school dropout rates, 372-375
home!essness, 459-462
and housing, 445-447
labor market discrirnination,
363-365
and land use, !86-188
and mass transit, 356-357
in racial segregation, 361-362
scope of, in 1999,355
lndex
Production
centered in cities, 4-5
consrant retums to scale, 23-26
!ack of scale economies, 20
output effect, 123
total value, 119-120
travel-cost trade-off, 96
Production activities, 120-121
Production budget, 573-574
Production facilities, 50
Production function, 119-120
approach to education, 375-379
Production technology, 77, 560
Productivity; see Labor'productivity
Products, 56-58
Professional sports, 82-83
Profit maximization
in location decisions, 65
minimizing transport costs, 67
Promotions, 143-144
Properry crime, 398
and drng use, 424-425
reductionin,422-423
Properry tax, 237-238, 524-545
assessment caps, 541
changing assumptions, 532-534
commercial properry, 537-538
and decreased housing supply, 529
differential rates, 537-538
effective rate, 525
effects of limits on, 543-545
exclusionary zoning, 334-335
general, 537
general-equilibrium analysis,
529-532
history oflimits on, 541-542
and household sorting, 334
housing abandonment and, 442-443
and housing demand, 333-335
incidence of, 526-532, 534-536
industrial property, 537-538
landed portian of, 527-528
local effects, 534-535
market effects, 527
mobile consumers, 533-534
modem limits on, 542-543
more than two municipalities, 532
national effects, 535-536
number of jurisdictions, 334-335
partial-equilitirium analysis, 528:..529
residential, 526-536
schoolspending,379-387
and single tax propasa!, 163
i1
591
lndex
taX in
'
'
Public services, 81
industrial, 124
residential, 126
Puga, D., 51, 63
Punishment, 406
Pure local public good, 324
Q
Race
annual earnings by, 364-365
and crime, 399
and poverty rate, 354
Racial confiict, 208
Racial discrirnination
and desegregation, 388
in labor market, 363-365
in real estate market, 362
Racial segregation
affirmative marketing policy and,
346-347
and commuting distances, 358-359
consequences of, 359-360
facrs and statistics, 341-342
federal housing policies, 344
and increase in poverty, 361-362
neighborhood transitions,
344-347
and public policy, 362-363
reasons for, 342-344
schoolsegregation,388
spatial mismatch, 357-359
Racial steering, 342-344
Radial highway, 258
Radio industry, 52
Rafael, Steven, 357, 371
Railroads, 26-27
Rarnis, Timothy V., 253
Rand Corporation, 494
Rank-size rule, 109-112
Rapid depreciation, 452-453
Rappaport, Jordan, 356,357,370
Rational criminal model burglary
example, 402-403 morality and
aversion to crime,
403-404
supply curve for crime, 404-405
Rauch, J. E., 79, 91
Real estate market; see also Housino
market
592
Residential bid-rent function, !81-182,
190-192
Residential density, 182-183
effect of congestion tax, 279
Residentialland use, 177-183,
277-279
convex housing-price function,
179-181
linear housing-price function,
177-179
Residentia! nuisance zoning, 237
Residential property tax, 526-536
Residentia!public services, 126
Residential segregation, zoning for,
232-233
Residential taxes, 126
Resnick, Mitchell, 112, 118
Resource allocation, 505-506
Resource-odented firms, 67-70
in central place theory, JOS
Restrictivo covenants, 245
Retail clusters, 56, 58
Retailers
nuisance zoning, 236
in unzoned cities, 245
Retirement decision
housing, 441-442
Retribution function, 419
Reynolds, Margan, 412,430
Ricardo, David, 155, 166
Richardson, Harry, 303,318
Richmond, Jonathan E., 298, 300,
304,318
Risk compensation
evidence of, 285
theory of, 283
Rivasplata, Antera, 253
Rivkin, Steven G., 377, 378, 392, 396
Robak, Jennifer, 125-126, 149
Rogers, Will, 153
Reman cities, 31-32
Romano, Richard E., 391, 395
Rome, 114-115
Romer, T., 547, 556
Roof age, 439
Rosen, Harvey S., 454, 466
Rosen, Kenneth T., 112,118
Rosenbloom, Sandra, 309,319
Rosenthal, H., 547,556
Rosenthal, Stuart S., 43, 48, 50-51, 64,
342, 349, 370
Ross, John, 206, 207, 226
Ross, Stephen, 330, 350
Index
S
Sacerdote, B. 421, 422, 429
St. Louis
community development, 489
Housing market, 435
S:riz, Albert, 5, 16. 79, 90
San Antonio Independent School
Disrricr v. Rodriquez. 381
Sander, William, 396
Sanderson, Allen R., 83, 90
San Diego
HOT lanes, 267
light r:ril system, 303-304
San Francisco
Bay Area Rapid Transit, 295-298,
301, 309, 313
building heights, 241
impact fees, 240
land values, 221
office space, 214
peak travel times, 264
. race and emp1oyment, 357-358
zoning..ordinances, 232-233
Sat!lrn plant, 86-87
Savas; E. S., 509, 523
Sawhill; Isabel, 366, 370, 371
Sawmills, 75
Saxenian, Annalee, 3, 16
Scale economies,
advertising carnp:rigns, 42
central place theory, 105-106
and city size, 27
externa!, 39
from factor specialization, 24
factory cities, 23-26
high-tech firms, 42
from indivisible inputs, 24
in intermediare goods, 128
lack of, 20
of local public goods, 512
in market areas, 95-96, 99
in mass transit, 308-309
and per capita demand, 102-103
sewing machine, 27
in transportation, 22-,23, 71
urbanization economies, 49
vs. demand diversity, 512-514
Scheinkman, Joe A. 51, 63
Schell, Steven R.. 253
Schiller, Bradley, 360,365,371
School vouchers, 390-392
Schwab, Robert M.. 125, 149, !53, 166,
386,395
Seattle, office space, 214
Section 8 rent certicates, 480-481
Segregation, in schools, 387-389
Semiconductor industry, 84
Semidvalrous public good, 324
Serrano, John, 380-381
Se"ano, v. Priest, 380-381
Sewage services, 506-507
Sewing machine, 27
Sexton, Terd, 544, 556
Shadbegian, Ronald J., 544, 556
Shared-ride taxis, 310
Sheffrin, Steven M., 544, 556
Sheinkman, Jose A. 52
Sherrnan, Lawrence, 411,430
Shilling, James, 454,465
Shires, Michael A. 544, 556
Shirt factory, 23-26
Shleifer, Andrei, 51, 52, 63
Shoe faetones, 27
Shopping extemalities, 103
imperfect substitutes, 56-58
Shopping paths, 106-107
Shortage,561-563
Shoup, Donald C., 269, 290
Siegan, Bemard, 244, 253
Silcock, D. T., 3!2, 318
Silicon Valley, 42, 48, 84
Silverman, L. P., 425, 430
Singapore, 266
Single-farni!y homes, 245
Single-issue politics, 520
Single-room occupancy units,
461-462
Single-story plants, 210-211
Single tax, 161-162
altematives to, 163
Single transferable input, 66
Single transferable output, 66
Sirmans, C., 465
Site development, 82, 130
Sjoquist, David L., 358,359,370
Skunk deodorizing, 69
Skyscraper, first, 35
Sa!t Lake City, 489
Index
..1
593
Strip development, 245
Struyk, Raymond J., 444,454,466
Subcenters, 216-222
in Chicago, 219, 220
and city centers, 221-222
clustering of firms in, 216-217
employrnent ratios, 217
in Houston, 219
in Los Angeles, 217-218
types of, 217
Subscription commuter services, 31O,
311
Subsidies, 81-82
fiscal imp1ications, 130
government loans and
guarantees, 130
industrial bonds, 130
for local public goods, 516
mass transit, 271-273,282,305-308
for new housing, 447-449
for new housing construction, 496
for private housin.g, 479
site development, 130
tax abatement, 129
in United Kingdom, 312
Subsidized housing, 240, 357, 449-450
Substantive due process, 246
Substituto, 559
Substirution effect
definition, 554
in housing market, 449
Suburbanization,205-222
airports, 210-211
automobiles, 210-211
and central-city problems, 189
Chicago area, 219
clustering of offices, 212-214
communications techno1ogy, 214-215
coinmuting costs and higher incomes,
206-207
commuting patterns, 257-258
and crime rates, 397
edge cities, 219-221
Houston area, 219
and intercity truck, 209
and intracity truck, 208-209
joblocation,204-205
ofjobs,363
jobs following people, 215-216
Los Angeles area, 217-218
of manufacturing, 208--211
new housing, 189
of office employment, 212-215
lndex
594
Taxation (continue)
of land rent, 161-164
land tax, 497
and location decisions, 80-81
mortgage deduction, 453-456
national, 536
parking tax, 268
partialland tax, 163
principie of fiscal neutrality, 380
renta! tax break, 452-453
residential taxes, 126
single tax, 161-163
splittax, 163
transit subsidies, 304-306
two-rate tax, 163
user fees, 277
Tax-free indusnial bonds, 81
Tax Reform Act of 1986, 81, !30
Tax revolts, 540-543
Taylor, Brian D., 309, 318
Teacher ability, 377
Tea!, RogerF., 309,318
Teleconnunications, 58-60
Teleconferencing, 215
Telephone, 60
Television industry concentration, 43
Temporary Aid to Needy Families,
365-366
Tenant bribery, 496
Terman, Donna, 387, 396
Texas, school spending, 380
Thaler, Richard, 401, 430
Theory of risk compensation,
283,285
Theory of statistical
discrirnination, 364
Thisse, Jacques-Fran9ois, 188, 201
Three-snikes Jaw, 417
Tidewater, Va., 309
Tiebout, Charles M., 134, 149,
328,350
Tiebout model, 328-330
altematives to, 5!0-512
and property tax, 333--335, 538-539
Time cost of connuting, 185, 259
Time elasticities, 294
Toffier, A!vin, 58,64
Tolley, G., 465
Tomlinson, Peter, 491,495,499
Toronto, 266-267
office sector, 50
Total costs vs. labor cost,
76-78
Trade
centered in cities, 4-5
with coinage, 31
generated by comparative advantage,
20-22
in Middle Ages, 32-33
in primary cities, 112-113
regions without, 20
Trade resnictions, 113--114
Trading cities
comparativo advantage, 20-22
English woolen industry, 23
scale economies in transportation,
22-23
Traffic design, 269-271
Traffic flow, 271
Traffic volume, 260-261
average road cost, 273-275
congestion tolls, 275-277
long-ron average cost curve, 275
optimum volume and road widrh, 275
Trajtenberg, M., 3, 16
Transactions table, !36-137
Transferable development rights,
242-243
Transfer-oriented firms, 66-75
basic assumptions, 66-67
resource-oriented, 67-70
Transit gap, 273
Transit subsidies, 271-273, 282
)ustificagon for, 305-306
Trarisponation, 14; see also
Commuting; Mass transit;
Urban transportation
freight cost, 26-27
in Indusnial Revolution, 35
intercity, 25
lack of scale ec<lnornies, 20
scale econornies, 22-23,71
Transportation costs, 22
and location decisions, 66
market-oriented firms, 70-71
resource-oriented firms, 67-70
vs.labor costs, 76-78
zero, 155
Transportation technology, 77
Transshipment points, 73--75
Travel costs
in market areas, 99-100
production trade-off, 96
Travel route, 268
Traveltime,260-261,268,357
Treyz, George, 125-126, 149 .
Trucks
intercity, 209
intracity, 208-209
Tuccillo, John A., 454, 466
Turner, M., 50, 51,64
1\vo-earner households, !86
1\vo-rate tax, 163
Tye, Williarn B., 290
u
Unemployment, and crime, 409
Unemployment rate, 142
Uniform consumption pattems,
133--!34
Uniform Crime Repons, 397-398
Uniform demand, 103
Unions, effect on location decisions,
87-88
United States
size disnibution of urban areas, 40
size disnibution of urbanized areas, 93
urbanization of, 35-36
United States Census Bureau, 6-13
Upward filtering, 484
Ur,30
Urban area
census definition, 6-13
defined by population, 6-7
definition, 5-6
metropolitan areas, 8-13
municipality, 7-8
rank-size rule, 109-112
size disnibution, 40, 110-112
world comparisons, 8
Urban Development Action Grants,
488-489
Urban econornic growrh, 119-145
benefits and costs, 141-145
econornic base study, 132-135
effects of employment growth,
141-!43
and environmental quality, 130-132
and export indusnies, 134-135
export-oriented theories, 139-140
input-output analysis, 135-138
localization econornies and learning,
127-129
Jocation decisions, 65-88
location-quotientapproach, 133-134
multiplier process, 120-122
production activities, 120-121
production function approach,
119-120
Index
595
r
I
1
l
I
r
{
f
f
t
f
l
1
'
(
.t
w
Wachs, Martin, 309,318
Wage discrirnination.-363
and market forces, 363-365
by race and education, 364-365
Wage levels, 4
and city size, 27-28
and com.muting costs, 183
constan!wage, 130
and cost ofliving, 125-126
and crime rate, 409
effect of congestion tax, 280, effect
ofwelfare reforrn, 367-368 effect
on location decisions, 87-88_ and
labor demand curve, 123
and labor pooling, 44
and labor supply curve, 124-126
and mass transit costs, 297
migration effect, 124-125
of peor fantilies, 354-356
by race and education, 364-365
real wage, 143
and schooling, 378
of transit workers, 307
Walking cost, 297
Wallman, Joel, 422, 428, 429
Walters, A. A., 308, 319
War making, 34
Warner, Charles Dudley, 524
Warth v. Seldem, 247
Washington, D.C., Metro system,
302-303
Wassmer, Robert W., 82, 90, 91,
333,349
-
596
Wasylenko, Michael J., 201
Waterwheel, 75
. Weaver, Kent, 370
Webber, Melvin W., 302,313,319
Weicher, John, 476, 499
Weight-losing finns, 69
Welfare reform, 365-368
and block grants, 552-554
effects of, 366-367
legislation, 365-366
market effects, 367-368
Wells, H. G., 372
Wheaton, William, 188, 191,201
Wbite, Michelle, '442, 443, 466
Williamson, Harold, 208, 227
lndex
--1'
y
Ybarra v. Town of Los Alros Hills, 247
Yinger, John, 330, 344, 350,387, 395
z
Zeckhauser, Richard, 544, 556
Zenou, Yves, 188, 201
Zero economic profit,
155, 157
Zero transport costs, 155
Zimbalist, Andrew, 82, 90, 91
Zoning; see Land-use controls
and zoning
Zoning map, 236
'.
J
'