ECOLOGICAL

ECONOMICS
E L S E VI E R Ecological Economics 11 (1994) 213-226
An experi ment al analysis of the ef f ect i veness of an envi ronment al
assurance bondi ng system on pl ayer behavi or in a si mul at ed firm
Laur a Cornwel l * Rober t Cost anza
9
Maryland International Institute for Ecological Economics, Center for Encironmental and Estuarine Studies, Unicersity (~f Maryland,
Solomons, MD 20688, USA
Received 30 March 1993; accepted 4 January 1994
Abstract
It has l ong been r ecogni zed t hat t he pr esent command and cont rol met hods for pol l ut i on abat ement ar e
inefficient. Usi ng mar ket mechani sms for envi r onment al management is a promi si ng al t er nat i ve to t he di rect
r egul at or y appr oach. Mar ket mechani sms are j ust begi nni ng to appear in U.S. envi r onment al policy. For exampl e,
t r adabl e per mi t schemes ar e bei ng devel oped and i mpl ement ed for some ai r pol l ut ant s under t he amended Cl ean
Ai r Act . Var i ous ot her forms of taxes and t r adabl e per mi t s have been pr oposed; however, t hese systems do not
addr ess t he l arge uncer t ai nt y i nher ent in most envi ronment al pr obl ems. One mechani sm cur r ent l y bei ng st udi ed to
addr ess uncer t ai nt y mor e effectively is a flexible envi r onment al assurance bondi ng system, desi gned to i ncor por at e
envi r onment al cri t eri a and uncer t ai nt y i nt o mar ket incentives. This st udy uses an exper i ment al appr oach, empl oyi ng
an i nt eract i ve comput er game/ s i mul at i on model with human pl ayers to exami ne t he effect i veness of t he assurance
bondi ng system under varying degr ees of uncert ai nt y. An envi r onment al cost efficiency index ( ECEI ) , or pr of i t per
unit waste, is used to measur e pl ayer per f or mance. Prel i mi nary resul t s i ndi cat e t hat pl ayers are mor e successful
under t he assurance bondi ng system, in t erms of ECEI , t han under t he si mul at ed cur r ent system.
Keywords: Assur ance bondi ng system; Si mul at i on model
1. I n t r o d u c t i o n u n d e r p r e s e r v a t i o n p r o t e c t i o n by t he g o v e r n me n t
we r e o p e n acces s , f r e e a n d a va i l a bl e t o al l . Th e
1.1. Bac k gr ound r e s ul t was ove r e x p l o i t a t i o n o f t he e n v i r o n me n t
a n d t he " t r a g e d y of t he c o mmo n s " ( Ha r d i n , 1968).
Pr i o r t o t h e 1960s e n v i r o n me n t a l pol i c y i n t he Wi t h t h e e n v i r o n me n t ' s d e c l i n i n g c o n d i t i o n
Un i t e d St a t e s was b a s e d u p o n Ri p a r i a n Ri g h t s t h r o u g h t he 1960s a n d t he i n c r e a s e of p u b l i c
a n d En g l i s h c o mmo n l aw. Th o s e r e s o u r c e s not a wa r e n e s s t h r o u g h p u b l i c a t i o n s s uch as Ra c h e l
Ca r s o n ' s Si l ent Spri ng ( 1962) a n d Ke n n e t h Bo u l d -
i ng' s The Ec onomi c s o f t he Comi ng Spaceshi p
Ear t h (1966), t h e e n v i r o n me n t ' s s t a t us c h a n g e d t o
* C o r r e s p o n d i n g a u t h o r , t ha t of a s c a r c e a n d v a l u a b l e r e s o u r c e . Le gi s l a -
0921-8009/94/$07.00 1994 Elsevier Science B.V. All rights reserved
SSDI 0921- 8009( 94) 00006- H
214 L. Cornwell, R. Costanza / Ecol ogi cal Economics 11 (1994) 213-226
tion through the 1970s set ambitious goals for of damages, liability, clean-up costs, enforcement,
environmental protection with fixed emission lim- potential legal action, and amount of time the
its for individual sources based largely on speci- polluter is responsible for their emissions (cf.
fled abatement technologies (command and con- Peles and Stein, 1976; Roberts and Spence, 1976;
trol strategies) to impose immediate controls on Harford, 1987). These uncertainty factors can
relatively few, widespread, problematic poilu- translate into incentives for resource exploitation.
tants. Since this landmark legislation, much has In addition, legislative language is often so vague
been learned about the effectiveness of this type that a firm can present a legally convincing case
of environmental regulation. Economic analysis that the regulatory agency's abatement require-
indicates that present methods of pollution con- ments are unfeasible. The polluter can further
trol are inefficient and provide disincentives for expect to negotiate a new compliance schedule
directing resources toward abatement. For policy with a regulatory agency which is resource limited
to be both effective and efficient, steps need to and often more willing to gain minimal compli-
be taken to make it privately optimal to manage ance than enter into lengthy and expensive legal
waste in a socially optimal manner, battles (cf. Ackerman and Stewart, 1985).
In the past two decades, there has been ex-
1.2. Direct regulation vs. market alternatives haustive discussion in the literature of the effi-
ciency that can theoretically be achieved in poilu-
The traditional command and control regula- tion abatement through the use of market mecha-
tory system is inefficient because it treats firms nisms (cf. Tietenberg, 1973, 1985; White, 1976;
homogeneously, both in their production process Common, 1977; Kohn, 1977; Beavis and Walker,
and geographiclocation (cf. Atkinson and Tieten- 1979; Burrows, 1979; Bohm, 1981; Marquand,
berg, 1982; Seskin et al., 1983). It also places a 1981; Endres, 1983; Krupnick et al., 1983; Koenig,
tremendous information burden on the appropri- 1984; Lee, 1984; Haas, 1985; McHugh, 1985;
ate regulatory agency (Hahn and Noll, 1983) . Stollery, 1985; Webber and Webber, 1985; Brooks
Under command and control, the agency is re- and Heijdra, 1987; Conrad, 1987; Costanza, 1987a;
sponsible for determining the best available tech- Katzman, 1987; Perrings 1987, 1989; Shaw et al.,
nology for each industry, setting emissions levels 1987; Baumol and Oates, 1988; Hahn and Hester,
for polluting substances, and monitoring and im- 1989; Hamilton et al., 1989; Milliman and Prince,
posing penalties in cases of noncompliance. Di- 1989; Pethig and Fiedler, 1989). Some market
rect regulation guides polluters towards irrespon- alternatives that have been suggested include pol-
sible pollution abatement in several ways. Be- lution taxes, tradable pollution discharge permits,
cause it is based on a particular abatement tech- financial responsibility requirements, and de-
nology, it provides no incentives for development posit-refund systems. Tradable permits have been
of innovative technology; it promotes avoiding included in the most recent amendments to the
regulatory detection rather than regulatory com- Clean Air Act.
pliance; and, because it places stricter controls on The efficiency that market mechanisms can
new plants and processes, it provides disincen- achieve for pollution abatement has been well
tives for growth and facility upgrade (cf. Acker- studied both theoretically and empirically (cf.
man and Stewart, 1985). Krupnick 1983; Seskin et al., 1983). For example,
Optimization of pollution abatement requires studies in the St. Louis metropolitan area indi-
that all costs and benefits associated with the cate that the existing command and control sys-
activity be identified (Freeman, 1990). Polluters tems for air pollution abatement cost three to five
respond to direct regulation by comparing the times as much as an optimal incentive based
(often) substantial costs of installing and operat- approach that would yield the same ambient air
ing abatement equipment with the uncertain costs quality (Atkinson and Lewis, 1974, 1976; Atkin-
associated with penalties for noncompliance. Un- son and Tietenberg, 1982). In reality, policy struc-
certainty arises with respect to detection, amount ture is more complex than some analysis would
L. Cornwell, R. Costanza / Ecol ogi cal Economics 11 (1994)213-226 215
suggest. The i nst r ument t hat we are suggesting t he bur den of pr oof ( and t he cost of t hat bur den)
shoul d be consi der ed al ong with ot her r egul at or y is shi ft ed f r om t he public to t he r esour ce user,
tools t hat of f er fexi bi l i t y, and a st rong economi c i ncent i ve is provi ded to
r esear ch t he t r ue costs of envi ronment al l y damag-
1.3. Assurance bonding ing activities and to devel op innovative, cost-ef-
fective pol l ut i on cont rol t echnol ogi es. Assur ance
Uncer t ai nt y is not t hor oughl y addr essed in en- bondi ng is an ext ensi on of t he " pol l ut er pays"
vi r onment al policy despi t e its i mpor t ance and pri nci pl e to " t he pol l ut er pays for uncer t ai nt y as
pervasi veness in envi r onment al probl ems. An in- well" or t he " pr ecaut i onar y pol l ut er pays princi-
novat i ve policy cur r ent l y bei ng st udi ed is a flexi- pl e" (4P) (Cost anza and Cornwell, 1992).
ble envi r onment al assurance bondi ng system de- Nei t her t he pri nci pl e nor t he i nst r ument pro-
signed to i ncor por at e envi r onment al cri t eri a and posed are new. The envi r onment al bond has its
uncer t ai nt y i nt o t he mar ket system, and to i nduce root s in t he "mat er i al use f ees" first pr oposed by
innovative envi r onment al t echnol ogy. Thi s varia- Mill (1972) and Solow (1971), t he simplest work-
tion of t he deposi t r ef und system is desi gned t o ing exampl e of which is t he r ef undabl e deposi t on
i ncor por at e bot h known and uncer t ai n envi ron- glass bottles. The aim of t he deposi t is to encour -
ment al costs into t he i ncent i ve system and to age t he users to di spose of t he commodi t y in t he
i nduce positive envi r onment al t echnol ogi cal inno- most desi rabl e way (by recycling) and to avoid its
vat i on (Cost anza and Cornwell, 1992). The assur- disposal in t he least desi rabl e way (as litter). The
ance bondi ng system has been discussed else- deposi t may not be sufficient to cover t he cost of
wher e in t he l i t er at ur e (Cost anza, 1987a; Per- t he worst possible met hod of disposal, but it is
rings, 1989; Cost anza and Perri ngs, 1990; Far ber , general l y set at a level high enough to make
1991; Cost anza and Cornwel l , 1992) and is not r et ur ni ng t he bot t l e pri vat el y profi t abl e. The im-
t he t opi c of this paper; however, a br i ef overview por t ant f eat ur e of t he fee is t hat by insisting t hat
of t he pri nci pl es involved is war r ant ed, consumer s pay in adt,ance for t he costs t hey
The envi r onment al assurance bondi ng system might inflict on society if t hey adopt ed t he most
requi res t hose seeki ng to use society' s resources harmful met hod of disposal, it reverses t he usual
to post a bond equal to t he worst -case damages pr esumpt i on of " i nnocence" over "gui l t " as ap-
t hey coul d inflict on t he envi r onment , in advance plied to envi r onment al damages. The i nnocent -
of any activity. Worst -case damage scenari os unt i l -proven-gui l t y ar gument is not appl i cabl e in
woul d be est abl i shed by t he r egul at or y aut hor i t y t he case of firms using societal resources as re-
with t he best i nf or mat i on available and with t he ceivers of pri vat el y gener at ed wast e since t her e is
advice of i ndependent scientists. If r esour ce users no quest i on t hat t he act is bei ng commi t t ed. It is
coul d demons t r at e t hat damages to t he envi ron- t he amount of damage t hat is uncert ai n, and it is
ment wer e less t han t he amount of t he bond our cont ent i on t hat society shoul d not bear this
(over a pr edet er mi ned l engt h of time, speci fi ed in risk.
t he bond), this di f f er ence and a por t i on of ear ned In t he U.S., exampl es of bond use for envi ron-
i nt erest woul d be r ef unded. Thus, t he envi ron- ment al policy are few. Wher e t hey do appear in
ment al assurance bondi ng system i nsures t hat t he legislative l anguage, t hey are general l y one of
funds available for pr ot ect i ng t he envi r onment several financial mechani sms available to firms
are equal to t he pot ent i al harm facing its re- t hat are r equi r ed to demonst r at e financial re-
sources, sponsibility. Ther e are t hr ee exampl es wher e
If damages did occur, t he bond woul d be used bonds are requi red: Owner s and oper at or s of
to rehabi l i t at e or r epai r t he envi r onment , and under gr ound i nj ect i on wells are r equi r ed to post
possibly t o compensat e i nj ured parties. By requi r- bonds to ensur e pr oper plugging and abandon-
ing t he users of envi r onment al r esour ces to post a ment of wells. Bonds are cur r ent l y r equi r ed by
bond adequat e to cover pot ent i al f ut ur e envi ron- compani es leasing public land for oil and gas
ment al damages (with t he possibility f or refunds), expl or at i on/ ext r act i on to ensur e pr oper cappi ng
216 L. Cornwell, R. Costanza /Ecological Economics 11 (1994) 213-226
of wells and restoration of lands or surface waters lieve that the experimental approach is a neces-
after the cessation of the lease operations. Fi- sary component of the analysis because it can
nally, bonds must be posted by logging companies better address the incentives of the regulatory
to use existing roads on public lands for the system without dependence on (potentially
transport of timber, flawed) theoretical models of individual response.
Costanza and Perrings (1990) categorize the This experimental approach consisted of four
current command and control system using the stages: model development, game programming,
"social trap" paradigm (Platt, 1973; Cross and data generation by human volunteers, and finally,
Guyer, 1980; Costanza, 1987a). Several experi- data analysis.
mental games have been designed to study behav-
ior in social traps. The already-mentioned 2.1. The mode l
"tragedy of the commons" is a trap used to study
resource exploitation (Edney and Harper, 1978). The model was designed using STELLA
The well-known "prisoner's dilemma" game is an (Structured Thinking Experimental Learning
externality trap that has been used to study the Laboratory with Animation), a software package
evolution of cooperation (Axelrod, 1984), and developed by High Performance Inc. for the Ap-
more recently, the dollar auction game, an invest- pie Macintosh (cf. Costanza, 1987b). Smith and
ment trap, was used to study the effect of taxa- Williams (1992) recently used simulated markets
tion on the conflict escalation process (Costanza to analyze the principles that govern trading deci-
and Shrum, 1988). For this study, an experimen- sions; however, as far as we know, our study is
tal firm management game was developed that the first dy nami c , interactive simulation model
measures player behavior and performance while examining pollution abatement systems. Model
making management decisions in a theoretical development occurred in stages, with model be-
firm, under both the command and control regu- havior observed at each stage. Values and units
latory system and the assurance bonding system, in the model are arbitrary. The model is not
intended to simulate any particular industry, but
instead examines relative behavior and the im-
2. Methods pact of the human player's decisions.
As with all models, this one is a trade-off
The purpose of this experiment was to deter- between realism, precision, and generality. Upon
mine the effect of two charge systems on individ- initial inspection of the model diagram (Fig. 1), it
ual behavior, under varying degrees of uncer- may appear quite complex. However, upon fur-
tainty. This was accomplished by developing a ther examination of the model symbol definitions
game designed for one player who manages a (Appendix 1), it can be seen that the model is
computer-simulated company by controlling pro- really quite basic. Two of the seven state vari-
duction level and amount of resources devoted to ables, Cum Env Damages and Acc Env Fnd, are
waste reduction. As previously mentioned, sev- merely accounting components that have no out-
eral models of the profit-maximizing firm have flows and no external variables. The model was
been constructed to examine the effectiveness of made as simple as possible while including those
pollution abatement using market mechanisms, components the authors felt were minimally nec-
and several have considered the importance of essary to simulate the workings of a firm under
uncertainty (cf. Averch and Johnson, 1962; two environmental policy regimes. The final ver-
Weitzman, 1974; Adar and Griffin, 1976; Fishel- sion includes 7 state variables (boxes in Fig. 1), 8
son, 1976; Roberts and Spence, 1976; Baron and input flows, 7 output flows (double-lined arrows
Taggart, 1977; Magat, 1978; Mendelsohn, 1984; with valves in Fig. 1), and 25 auxiliary variables
Harford, 1987; Plourde and Yeung, 1989; Farber, (circles in Fig. 1). Flows of money, products, and
1991). None, however, have included the "experi- wastes are shown by double-lined arrows while
mental" aspect addressed in this paper. We be- flows of information are indicated by single-lined
L. Cornwell, R. Costanza / Ecological Economics 11 (1994) 213-226 217
a . A s s e t s S u b m o d e l
" ~ a l e s M . t P c e 0 ~ o s t s / ~ : ~ s i t s / D i s ~ a r 0 e ' . , t _ ~
T T ( I%~'~ Charge System
Shipments ~_~ I T ~ }, ~ ~ I , ~ L / C h a r g e S :
~ Monet ar y\ Inputs ~ ~ Total Cost s ~ . Uncert
Refunds .......... ' ~ a l Profits Env D~isastePOStPaid Damages
~ - "r f Usr P'rod - -
Waste F r a c t ~ Inventory
Waste Reduc Tech Unit Waste Red Cost Forcast Shipmts Product Demand
c . Wastes Submodel ~E [
Unit Waste ~ Waste Fraction
. . . . . . ~ Cure I Di s c h a ~ 1 " ~
Waste Reduc k j / ~ / ~ Diaster Threshold
Tech Production Env Disaster
~ }~ ~ Rand ~ - ~...J
Postp-ay ~ T of Change to Pre f . ~ Discharged
Charge [ System , T [
~ ' ~ ' ~ ~ J ~ L "r 'nt Frac~t Firm I~ X ~ s t e m
0~o~ ~ , ~ . A _ J C' ~ . ~. . ~, \ . . . . .
l " v~ ( ~ ~ - ~ ACCEnv Fnd
(Worst Case Est i mat e ~ Assur IIBnd Env Res Fnd ~ . ~
\ , , ~ ,nterest II ~ " /
, , . . . , f - . , . , . I~ -,.- I ~ 3 " - 1
Max Discharges
Interest
Charge
LP' ~ ...... ,;r.: ........ U Re,u~s
Unc--eert ~ j / Total Withdraw Discharged
Env Disaster
I I ~ ~ ~ , . - ' ~ 0
St at e Va r i a bl e Goods or Se r vi c e Fl ow Func t i ona l Au x i l i a r y or
De p e n d e n c e Ex o g e n o u s
Va r i a bl e
L E GE ND
Fi g. 1. Di a g r a m of t h e S T E L L A Mo d e l . i n c l u d i n g f our s u b mo d e l s .
218 L. CornweU, R. Costanza / Ecological Economics 1l (1994) 213-226
arrows. Symbol definitions are given in Appendix environmental disaster "events" as well as addi-
1. The model is composed of four connected tive effects over time, resulting in perceivable
submodels depicting firm assets, company inven- changes in ecosystem health. The environmental
tory, level of wastes, and the assurance bond. threshold variable determines the level at which
The Assets stock is incremented by flow from an environmental disaster will occur. During the
monetary inputs and decremented by outflow to postpay system of regulation, environmental dam-
total costs (Fig. la). Monetary inputs include age payments are deducted from firm assets. Dur-
sales based on market price and number of ship- ing the prepay system, they are deducted from
ments as well as refunds from the bond. Total the assurance bond (Fig. la, Assets Submodel).
costs are the sum of total input costs, assurance The assurance bond is established with a one-
bond deposits made during the prepay charge time deposit from firm assets (Fig. ld). This oc-
system, and environmental damages paid during curs when the regulation system switches to the
the postpay charge system, including a randomly prepayment system at some randomly selected
generated uncertainty component. Uncertainty time between time steps 20 and 40 during the
enters into the model through payments for envi- simulation. The firm management simulation runs
ronmental damage. Uncertainty is a random for a total of 60 time steps (simulated years).
number between 0 and 1 and affects the amount Once the bond is established, it accumulates in-
a player must "pay" for a given amount of envi- terest, which increases the bond's value over time.
ronmental wastes. Paying variable amounts for Refunds to the firm are granted on a yearly basis
the same level of environmental wastes repre- when costs of environmental damages are less
sents the aforementioned uncertainty that arises than the assessed estimate. A portion of the
with respect to detection, amount of damages, interest also accumulates in an environmental
liability, clean-up costs, enforcement, potential research fund.
legal action, and amount of time the polluter is
responsible for their emissions. 2.2. The game
The stock of inventory is based on inputs from
production and outputs to shipments (Fig. lb). The model was made into an interactive game
Shipments are based on a randomly generated using STELLAStack (High Performance Inc.), a
product demand that forecasts five years into the two-way interface between STELLA and Hyper-
future. The cost of inputs for production is the card software. Hypercard makes the simulation
sum of unit costs for inputs and waste reduction model "playable" as a game with only minimal
technology multiplied by the amount of produc- training required for the player. It enables the
tion. As the level of production increases, so do novice to interact indirectly with the STELLA
the costs associated with inputs to attain that model and, thus, not only make decisions while
level, the simulation is unfolding, but also examine the
Stocks of waste are based on by-products of consequence of those decisions. Production level
production and discharge (Fig. lc). Waste pro- and the amount of resources devoted to waste
duction is simply a function of the level of firm reduction are the only variables players can ma-
production and the amount of resources devoted nipulate. The game begins with these variables
to waste reduction technology. Waste reduction set at a default level (see Appendix 1). Because
technology can improve efficiency by decreasing the play screen is linked directly to the model
the waste fraction (i.e., by recycling or by using through StellaStack, when players make changes
new technology). As wastes are released into the in production level or waste reduction technol-
environment, they accumulate as a stock of dis- ogy, these translate into parameter changes in the
charge. This discharge, in turn, determines the model. Changes can be made at any time during
stock of cumulative environmental damage. Envi- the running of the game. If the player does not
ronmental disasters are considered to be any choose to make any manipulations, these vari-
adverse effects on the environment. They include ables will be determined by the model.
L. Cornwell, R. Costanza / Ecol ogi cal Economics 11 (1994)213-226 219
The game begi ns by pr ompt i ng t he pl ayer to pant s wer e given some prel i mi nary i nst ruct i ons
log in t hei r name, occupat i on, address, gender , about using t he comput er and t ol d t hat t he obj ect
and age as well as some simple i nst ruct i ons on of t he game was to maxi mi ze profits. Because of
how t o play t he game and what t he objectives are. t he real i t i es of i ncreased envi r onment al aware-
Each t i me a pl ayer begi ns a new game, a r andom hess, moral obl i gat i ons to society and t he influ-
level of uncer t ai nt y is established. Model - gener - ence negat i ve envi r onment al press coul d have on
at ed dat a on assets, annual profits, t ot al costs, sales, pl ayers wer e also t ol d t hat t hey mi ght want
cumul at i ve envi r onment al damage and t he assur- to t hi nk about minimizing envi r onment al damage.
ance bond are di spl ayed for pl ayers on a yearl y They wer e also i nf or med t hat t he game was de-
basis (Fig. 2). Besi des t he i nf or mat i on on t he play signed to st udy two types of pol l ut i on abat ement
screen, part i ci pant s can view t hei r progress over strategies. All responses ent er ed by t he study
t i me by exami ni ng t he envi r onment al graphs, eco- subjects were r ecor ded by t he comput er in an
nomi c graphs, bond graphs, and profi t graphs out put file.
(Fig. 3). When t he charge system changes f r om As previously ment i oned, t he game is desi gned
post pay t o prepay, t he pl ayer is not i fi ed t hat t he to test pl ayer behavi or under a pr epayment (as-
assurance bondi ng system is in effect and t hat it surance bondi ng) versus a post payment (corn-
gener at es i nt erest , a por t i on of which will be mand and cont rol ) system, with varying degr ees
r ef unded to t he firm. of uncert ai nt y. Our measur e of pl ayer per f or -
mance, an envi r onment al cost effi ci ency i ndex
2.3. Players ( ECEI ) , was cal cul at ed by dividing t ot al cumul a-
tive profi t by t ot al cumul at i ve waste. We did not
Vol unt eer s for this st udy consi st ed of Envi ron- use t he st andar d i ndex of effi ci ency used in eco-
ment al Pr ot ect i on Agency per sonnel , gr aduat e nomi c exper i ment s (t he exper i ment al l y observed
st udent s in a publ i c policy cour se at t he Uni ver- net social benefi t s di vi ded by t he maxi mum possi-
sity of Maryl and, st udent s and faculty at t he ble expect ed net social benefi t s) because we are
Chesapeake Biological Labor at or y and, for an argui ng t hat t he social cost of pol l ut i on is of t en
i nt er nat i onal perspect i ve, facul t y and st udent s at uncert ai n or unknown. Because of this, we did
t he Uni versi t y of St ockhol m in Sweden. Partici- not want to assign an arbi t rary val ue to pol l ut i on.
Y e a r 0 A s s e t s 0 I Waste Reduction Fraction Scale
A n n u a l P r o f i t s 0 ] O f the fraction o f resources that a_re available for
S a l e s 0 w~ste reduchon, what f~uc~on ~ould y o u like
R e f u n d s o n B o n d 0 to d e v o ~ ~o this technology?
CProfits = Sales + eefund, - rot,, Costs) I ~ I O ~
T o t a l C o s t s 0 O 1 0 0 %
E n v i r o n C h a r g e s O
I n p u t C o s t s 0 Production
B o n d D e p o s i t s 0 P]~yer D e f n ~ d P~od~ction lev~l (if 0 then the budt
( T o t a l C o s t s = E n v i r o n Ch a r g e s + Input C o s t s i n ~ e r w i l l d e t e F m l ~ p r o d u c t i o n l e v e l b a s e d
+ B o n d D e p o s i t s ) o n a forecast o f sales)
Cumm Envir Damage 0 I [ ~ l o ~
Wastes O I L ~ H ~ . ~
E n v i r o n D i s a s t e r s O
C u r n m Envir D a m a g e is a function of wastes I Control B u t t o n s
,vet time. Environ Di sasters Occur when I ~ . . . ~ . . ~ . . . ~ . . . . . . . . .
Start a N e w G a m e ~ G o to H E L P i
gour C u m i n D a m a g e reaches critical levels) .
A s s u r a n c e B o n d 0 I P a ~ e O ~ m e ( ~ ShowEnvlronO~,ph.s ( ~
Interest on Bond O I R , , ~ O ~ ~ ~ho.Eco~O~,~
I
~toP Oame ( ~ ~ h o w Bond Omphs ( ~
Refunds o f the b o n d to the firm, are s h o w n
nder Annual P r o f i t s , a b o v e . ) Iv[ed~ M e n u ( ~ S h o w Profits G r a p h ( ~
Fig. 2. The "play" screen where management decisions are made by players.
220 L. Cornwell, R. Costanza / Ecol ogi cal Economics 11 (1994) 213-226
Instead, we decided to determine which system 3. Results
offered the greatest environmental efficiency. As
the ECEI value increases, management decisions Data were partitioned into two samples, values
are considered more efficient since annual profits generated by the model during the postpay sys-
were being made with the least environmental tern and those generated during the assurance
impact. We hypothesized that players would have bonding system. Thirty-three games were run with
the incentive to manage "their firm" in a more no players. Without player manipulation, all vari-
environmentally efficient way under the prepay able values are determined by the model. The
system and that this efficiency gain would be output files from these games constituted the
most pronounced when uncertainty was high. As control data. The efficiency index of the two
a result, ECEI values under the prepay system charge systems was compared using a paired t-
will be greater than under the postpay system, test. The relationship between ECEI and uncer-
Year ~]e~ Refund~ Total Co~t$ t O 0 Sales I . Msur=nce Assurance
/ Y e a r Refunds I ntere~t Bond 2 0 0 0 ~
1.00 70.61 0.00 22.8Z 7 5 111.00 0.00 0.00 0.00 I I 1500"J
3.00 b'7.59 0.00 18.Z4 r ~ S O " ~ " ~ A - - 3.00 000 O.OO 0.00 I 0 0 0
~ ~ . w 0.00 re.era r ~ , 114.00 0 oo 0 oo 0.00 r ~ , ~ j j
" R ' ~ u r n t ' ~ e e y C o ~ _ . - - S h ~ r e P " ~ 2 5 ~_ - t ( . _ , ' R e t u r n l o P l a y Card~(Slmm Eon 6 r a p l * s I S O 0 ~ J L...,......,_
S h e m E n u i r o n 6 r e p b s | / $ h o w B o n d 6 r n p l l s l 0 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . "", " h e w E n o l r o n B r a P h U ( ? h w P r o f l t s G r e p h ' j 0:~ J . . . . . . . . . . . . .
* " - - " - 15 30 45 60 15 30 45 60
5 0 ] n T o t a l Costs 250"~. Refunds 2 0 0 1 I n t e r e s t 2501 Refunds
, . . ' V L _ _ "i
I
6 2 . 5 ' 50 62.5~
/ ' %
O l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . o . . . . . . . . . . I,.~ . . . . . . . . . . . . . . . . . . . o . . . . . . . . . . . . . . . . . . . o . . . . . . . . . . I , ~ t . . . . . . . . . . .
0 , s 30 . 5 60 ~ ~ . ~ 60 0 , s 30 . 5 60 i s 30 . 5 60
a. Profit Graphs b. Bond Graphs
Year ~l~e~e Eiwlr'" ~ln~srtn=,mentalwa=,e!, 4 0 0 1 C u m u l e t i v e E , v l r o n m e n t a l Year Assets nnualProfits Production 1 0 0 0 0 ~ A s s e t s
I o o o00 , 0 0 0 ~ 3 o o t Damages O.CO 5000.00 56.00 10.00 ~.-~7500-~
I : : : ; o : N 2 o o ] : : : N s o o o
0 ~.~.. r . . r . . ~ . . . ~ . . . , . . . .
I ( , , . r n t o P l . , ~ . r X s , o - , o . , r n , , . ) . . _. o . . . . . i'5 ..... 3'~ ..... 4'5 ..... 6'o how[nuironGra--~($homProntsGraphs) 0]0~,.~.,~,~-~.-3,0.~-..4m 5 ..... 6,0
" S h o w E c o n 6 r a p h s ) I S h o w P r o f i t s G r a p h . ) Production 1 0 0 ] Annual P r o f i t s
Environmental I 0 ~ 0 ~ ~
15 ~ Wastes 1 O0 ~ OIsasters 0
; ' . 5 ~ 5 0 5
75 25 2.5 -50 1 ........ . , U ............
o = 5 ~o ~5 6o o t s ~0 45 60 [ 0 ~ 5 ~0 4'~'""20 0 15 ~0 45 60
c. Environmental Graphs d. Econ Graphs
Fig. 3. Exa mpl e s of pr ogr es s over t i me as vi ewed by pl ayer s of t he game. Onl y one set of graphs can be vi e we d at a t i me. a. Pr of i t
Gr a p h s i ncl ude out put over t i me of f i r m sal es, r e f unds f r om t he b o n d and total cost s, b. Bond Gr a p h s i ncl ude out put over t i me of
total bond a mount , i nt erest e ar ne d on t he bond and r e f unds t o t he f i r m of t he bond. c. Envi r onme nt a l Gr a p h s i ncl ude o u t p u t over
t i me of cumul at i ve e nvi r onme nt a l damages was t es produced by t he f i rm and e nvi r onme nt al di sasters caus ed as a r esul t of f i r m
wast es, d. Ec on Gr a p h s i ncl ude out put over t i me of f i r m as s et s , pr oduct i on level a nd a nnua l pr of i t s.
L. Cornwell, R. Costanza / Ecological Economics 11 (1994) 213-226 221
tainty was determined for both samples using fessionals, students, and engineers. Because the
linear regression. The resulting slopes were corn- last two categories have small sample sizes, they
pared using the F-test (Sokal and Rohlf, 1981) to were not included in all analyses (but were in-
determine if under high uncertainty conditions cluded in the 8-variable model). For all possible
players performed significantly better under the occupation pairs, variance homogeneity was tested
assurance bonding system than under the corn- using the Fma x test (Sokal and Rohlf, 1981). Anal-
mand and control system, yses were then performed using the unpaired
Player characteristics were analyzed with ECEI t-test for unequal sample sizes and, where appro-
to better understand the role that these charac- priate, unequal variances (Snedecor and Cochran,
teristics have on player performance. Two analy- 1980). ECEI scores of males and females were
ses were used to determine if player characteris- also compared using the t-test for unequal sam-
tics affect ECEI. Multiple regression models were ple size. Finally, regression models were con-
constructed with eight variables (age, game num- structed to examine the relationship between
ber, time of change to the prepay system, charge ECEI and various combinations of explanatory
system, gender, residence, occupation and uncer- variables for both the prepay system and the
tainty), five variables (age, game number, time of postpay system (Table 1).
change to the prepay system, charge system, and Thirty-six people played the game a total of
uncertainty) and three variables (age, game num- 101 times for an average of 2.89 games per player.
ber, and charge system). Unpaired t-tests were The average age of the study subjects was 35
used to examine ECEI between places of resi- (range 22-60). The model inherently performs
dence (U.S. or Sweden) and occupations. Player better (i.e., with no player input) under the pre-
occupation was divided into the following cate- pay system (mean ECEI = 9.706) than the post-
gories: economists, natural scientists, policy pro- pay system (mean ECEI = 4.552), due to initial
Table 1
Multiple regression of the eight-, five-, and three-variable models. The independent variable in all cases was ECEI
Dependent variables Coefficient St. error t-value Probability
8-variable model a
Game number 0.927 0.232 3.996 0.0001
Charge system 8.113 1.375 5.903 0.0001
Age - 0.227 0.090 2.532 0. 0124
Change to prepay system - 5. 312 2.372 2.239 0.0266
Uncertainty 107.278 47.181 2.274 0.0244
Occupation 0.1341 0.7227 0.186 0.8531
Resi dence 0.353 0. 46l 0.765 I).4453
Gender - 0.067 1.969 0.(134 0.973
5-variable model b
Game number 1.026 0.189 5.42 0.0001
Charge system 7.178 1.194 6.012 0.0001
Age - 0.226 0.064 3.516 0.0005
Change to prepay system - 3.902 2.068 1.887 0. 0607
Uncertainty 78.314 41.048 1.908 0.0579
3-variable model c
Game number 0.976 0.188 5.202 0.0001
Charge system 7.164 1.200 10.973 0.0001
Age - 0.226 0.063 3.612 0. 0004 ~'
Intercept = 109.438, R 2 = 0.359, F = 10.304, P = 0.0001.
b I nt er cept = 82.788, R 2 = 0.322, F = 18.159, P = 0.0001.
c I nt er cept = 7.055, R e = 0.309, F = 28.741. P = 0.0001.
222 L. Cornwell, R. Costanza / Ecological Economics 11 (1994) 213-226
condi t i ons and r et ur ns f r om i nt erest . Because of Resul t s of an unpai r ed t-test f or unequal sam-
this bias, t he mean ECEI f or cont rol games was pl e size showed t hat pl ayers f r om Sweden had
subt r act ed f r om exper i ment al ECEI values t o nearl y twice t he adj ust ed ECEI val ues of pl ayers
cr eat e adj ust ed scores whi ch wer e used f or subse- f r om t he U. S . (t0.05,199 = 2.356, P = 0.0195). ECEI
quent analysis. The command and cont r ol system was significantly hi gher f or nat ur al scientists t han
gener at ed mor e t han f our t i mes t he amount of for policy professi onal s (to.os,t03 = 2.19, P = 0.031)
wastes per unit profi t t han t he assurance bond and was t he onl y t -t est bet ween occupat i ons t hat
system (t0:05,199 = - 5. 14, P = 0.001). Ther e was was significant. However , adj ust ed ECEI scores
no significant rel at i onshi p bet ween adj ust ed ECEI i ncreased with game number , and when analyzing
and uncer t ai nt y duri ng t he pr epay system ( y = scores for first and second games only, no signifi-
3.95x + 7.38, F1,99 = 0.9422, P = 0.3341) nor dur- cant di f f er ence occur r ed (t0.05,61 = 0.978, P =
ing t he post pay system ( y = 0. 7413x + 2.099, F] , 9 8 0.332). The r e we r e consi derabl y mor e mal e play-
= 0.1622, P = 0.688). A compar i son of t he re- ers t han femal es and, in general , mal es pl ayed
gression lines r eveal ed t he rel at i onshi p bet ween mor e games. Adj ust ed ECEI mean val ues wer e
ECEI and uncer t ai nt y was not significantly differ- significantly hi gher f or mal es t han for femal es
ent bet ween t he two char ge systems (F1, 98 = 0 . 0 6 5 , (t0.05.199 = -- 2.162, P = 0.0318), but onl y when
P > 0.05). consi deri ng all games played. ECEI val ues f r om
The resul t s of t he regressi on model s are shown first games i ndi cat ed no significant di f f er ence be-
in Tabl e 1. The r e wer e highly significant rel at i on- t ween mal es and femal es (t0.05,70 = 1.268, P =
ships bet ween adj ust ed ECEI val ues and game 0.2088). Likewise, t her e was no significant differ-
number , char ge system, and age for all t hr ee ence bet ween mal es and femal es for second games
model s. The t i me of change t o t he pr epay system (t0.05,43 = 0.355, P = 0.725).
and uncer t ai nt y wer e onl y significant in t he 8-
vari abl e model while occupat i on, resi dence, and
gender wer e not significant in any of t he model s. 4. Di scussi on
The resul t s of t he 5-variable regressi on model f or
pr edi ct ed and actual ECEI val ues are pl ot t ed in As expect ed, t he assurance bondi ng pol l ut i on
Fig. 4. It is i nt erest i ng to not e t hat al t hough t her e abat ement system i mproved pl ayer per f or mance
is a significant r ange of ECEI val ues f or bot h when managi ng a si mul at ed firm. We suspect this
char ge systems, all of t he hi ghest ECEI val ues is due to t he i ncent i ves t hat this system provi des
( > 20) wer e gener at ed duri ng t he pr epay system, for t he "manager s . " Charge system is by far t he
most i mpor t ant expl anat or y vari abl e in all of t he
model s. When t he change t o t he pr epay system
i. Prepay System(withAssuranceBond) Occurs, pl ayers are i nf or med by a pause in t he
40 o Po,tpay Sy,ter, (w,,out Assur,nce Boo~) si mul at i on and an i nf or mat i on screen. Thi s allows
3s pl ayers t o eval uat e past per f or mance and r et hi nk
30 + ~ ..... ,,,~o~,~ strategy. However , cont r ar y to expect at i ons, t he
2s . %...~-"5~Prepaypo~n,soo,y level of uncer t ai nt y does not significantly affect
LI.I * ~ . 'i~',11, T ~ Postpay points only
20 ~ ~ " " ~ ] ~ U- pl ayer behavi or except in t he 8-variable model .
~1~ The r eason for this may be t hat pl ayers wer e not
4- 10 [ . ~ given di rect i nf or mat i on about uncer t ai nt y levels.
5 ..4
....... ~" ~ o ~ ~ o o They r eact ed t o uncer t ai nt y by assessing model
o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . out put . I f pl ayers wer e di rect l y i nf or med about
--10 lb 26 36 " 4~ uncer t ai nt y val ues (e.g., by showing a message at
ECEI t he begi nni ng of t he game t hat i ndi cat es " your
chances of get t i ng caught a n d / o r char ged for a
Fig. 4. Pl ot of ECEI for t he pr epay and post pay syst ems
agai nst t he pr edi ct ed val ues for ECEI f r om t he 5-variable pol l ut i on vi ol at i on in t he post pay system a r e
model . The overal l r z for t he model was 0.322. X%"), t hey woul d have t he oppor t uni t y to de-
L. Cornwell. R. Costanza / Ecol ogi cal Economics l l ( 1994) 213- 226 223
vel op st rat egi es and per haps woul d have engaged " l e a r ne d" as t hey pl ayed t he game). It would be
in mor e risky behavi or if t he probabi l i t y of get t i ng i nt erest i ng in f ut ur e exper i ment s t o provi de dif-
caught and char ged was low. In f ut ur e experi- f er ent i nst ruct i on sets to di f f er ent pl ayers to de-
ment s, we pl an to communi cat e mor e i nf or mat i on t er mi ne how i nf or mat i on affect s results. In gen-
about uncer t ai nt y levels duri ng t he game. eral, we t hi nk this prel i mi nary appl i cat i on of t he
The strongly significant rel at i onshi p bet ween si mul at i on mo d e l / g a me / e x p e r i me n t appr oach
ECEI and game number suggests t hat pl ayers was qui t e successful and subst ant i at ed our main
l earn as t hey pl ayed mor e games. The significant hypot hesi s t hat t he pr epay system provi des play-
di f f er ence bet ween males and femal es appear s to ers with t he i ncent i ve to i mprove t hei r economi c
be a di rect result of number of games played, and envi r onment al per f or mance in t he game when
Onl y one f emal e pl ayed mor e t han t hr ee games, compar ed with t he post pay system. Thi s experi-
Al t hough comput er literacy and familiarity with ment al evi dence is bol st er ed by r ecent t heoret i cal
t he Maci nt osh system wer e not surveyed in t he results ( Far ber , 1991). The exper i ment al ap-
game, t hose unfami l i ar with comput er s may have pr oach to under st andi ng economi c behavi or has
initially been i nt i mi dat ed by t he mani pul at i ons t r emendous pot ent i al (cf. Smith and Williams,
r equi r ed to play t he game. The fact t hat age is a 1992), bot h as an analytical tool and as a t eachi ng
significant expl anat or y vari abl e and exhibits a device. The results of t he exper i ment descri bed in
negative coeffi ci ent may be a result o f " c o mp u t e r this paper are prel i mi nary, but encouragi ng.
phobia.' " It is also i nt erest i ng to not e that, while Whi l e not conclusive, t hey i ndi cat e t hat t he pre-
pl ayers f r om Sweden have significantly hi gher pay (assurance bondi ng) system may be an effec-
ECEI val ues t han pl ayers from t he U.S., when tive i ncent i ve to i mprove a firm' s envi r onment al
taking ot her vari abl es into account (i.e., in t he behavi or. However, as in any heal t hy science, we
8-variable model ) count r y is not a significant ex- fully real i ze t hat we have rai sed mor e quest i ons
pl anat or y variable. In f ut ur e exper i ment s we pl an t han we have answered and have set t he stage for
to investigate f ur t her t he effect s of cul t ural back- addi t i onal exper i ment al studies in t he fut ure.
gr ound on per f or mance.
The fact t hat " t i me of change to t he pr epay Acknowledgments
syst em" is significant in t he 8-variable model and
almost significant in t he 5-variable model sug- Part i al Fundi ng for this proj ect was provi ded
gests t hat this vari abl e is pot ent i al l y i mpor t ant by t he U.S. EPA, cont r act #CR-815393-01-0, S.
and also requi res f ur t her investigation. In f ut ur e Far ber and R. Cost anza, Pri nci pal Investigators,
exper i ment s we pl an to allow mor e vari at i on in titled: " A Flexible Envi r onment al Cost Chargi ng
t he " t i me to change" vari abl e to avoid cl ust eri ng and Assurance Bondi ng System for I mpr oved En-
of values in t he mi ddl e of t he game. Increasi ng vi r onment al Management . " The aut hor s woul d
t he sampl e size of pl ayers with di f f er ent occupa- like to t hank game part i ci pant s as well as Paul
tions also seems like a val uabl e exercise. The Jivoff, Lisa Wai nger, Enr i que Reyes, Dan Mus-
amount of knowl edge or i nf or mat i on t hat a pl ayer satti, Cl em Tisdell, and an anonymous revi ewer
had pr i or to begi nni ng any given game seemed to for hel pful comment s on earl i er drafts. Of course,
make a di f f er ence in pl ayer per f or mance (pl ayers any remai ni ng errors are our own.
224 L. Cornwell, R. Costanza / Ecol ogi cal Economics 11 (1994) 213-226
Appendix
Appendi x 1
Model symbol definitions
Symbol Defi ni t i on
State L,ariables
Acc Env Fnd Accumul at ed Envi ronment al Fund = Acc Env Fnd + dt *(Env Res Fnd)
INIT(Acc Env Fnd) = 0
Assets Assets = Asset s + dt *(Monet ary inputs - Total Costs)
INIT(Asset s) = 50Q0
Assur Bond Assurance Bond = Assur Bond + dt *(Deposi t s - Damage Deduct i on - Refunds + Int erest - Env Res
Fnd)
INIT(Assur Bond) = 0
Cure Di schar Cumul at i ve Di scharge = Cumm Di schar + dt * (Di scharged - Di scharged Liabil)
I NI T( Cumm Discbar) = 0
Cum Env Damage Cumul at i ve Envi ronment al Damage = Cumm Env Damage + dt *(damages)
INIT(Cumm Env Damage) = 0
Inventory Invent ory = Inventory + dt *( - Shi pment s + Product i on)
INIT(Invent ory) = 100
Wast es Wast es = Wast es + dt * (Waste Product i on - Di scharged)
INIT(Wast es) = 10
Inputs
Damages Damages = Di scharged
Deposi t s Deposi t s into Bond = IF Del Charge > 0.1 THEN Worst Case Est i mat e ELSE 0
Di scharged Wast es Di scharged into t he Envi ronment = Wast es *0.5
Env Res Fnd Envi ronment al Research Fund = Int erest *(0.8 - Int frac to firm)
Int erest Int erest = Assur Bond * 0.08
Monet ary inputs Monet ary Input s into Fi rm Assets = Sales + Refunds
Production Level of Production = IF Usr Prod = 0 THEN Forcast Shipmts ELSE Usr Prod
Wast e Product i on Amount of Wast e Product i on = 1 + Product i on *((1 - Wast eReducTech) * Wast e Fract i on)
Outputs
Damage Deduct i on Damage Deduct i on = IF Charge System = 1 THEN (4 *Di scharged *(1 - Uncert)) + 14nv Di sast er *(1
- Uncert ) ELSE 0
Di scharged Wast es Di scharged into the Envi ronment = Wast es *0.5
Di scharged Liabil Di scharged Liability = 0.6 * Env Di sast er
Env Res Fnd Envi ronment al Research Fund = Int erest *(0.8 - Int frac to firm)
Refunds Refunds to t he Fi rm = IF Charge System = 1 THEN (Max Di s c ha r ge s - Di schar ged) + (Int frac to
firm * Int erest ) ELSE 0
Shi pment s Shi pment s of Invent ory = SMTHl ( Pr oduct Demand, 5, 10)
Total Costs Tot al Costs to the Firm = Input costs + Deposi t s + PostPaid Damages
External ~'ariables
Annual Profits Annual Profits = Monet ary inputs - Tot al Costs
Charge System Charge System = IF ( TI ME < T of Change to Pre) THEN Postpay ELSE Prepay
Del Charge Del ay Charge = Charge System - DELAY( Char ge System, l , I NI T( Char ge System))
Di ast er Threshol d Di sast er Threshol d = 70
Env Di sast er Envi ronment al Di sast er = IF (Cumin Di schar * RANDOM) > disaster t hreshol d THEN Cumm Di schar
ELSE 0
Forcast Shipmts Fut ure Shi pment s = FORCST(Shipments, 5, 5)
Input costs Input costs = 1 *(Uni t Inp Cost + Uni t Wast e Red Cost) *(Product i on)
Int frac to firm Fraction of Int erest Ret ur ned to the Firm = 0.25
Market price Market Price = 8
Max Discharges Maxi mum Di scharge in any given year = 7
PostPaid Damages Envi ronment al Damages Occurri ng Duri ng the Postpay System = IF Charge System = 0 THEN (4 * Dis-
charged *(1 - Uncert)) + 14 *Env Di sast er *(1 - Uncert ) ELSE 0
L. Cornwell, R. Costanza / Ecological Economics 11 (1994) 213-226 225
Post pay An I ndi cat or Var i abl e of t he Post pay syst em ( = 0)
Pr epay An I ndi cat or Var i abl e of t he Pr epay syst em ( = 1)
Pr oduct De ma nd Level of Pr oduct De ma nd = 3 + (4 * RANDOM)
Rand Ge ne r a t e s number s bet ween 0 and 1, r andoml y
Sal es Sales = Mar ket Pri ce *Shi pme nt s
Tot al Wi t hdr aw Tot al Wi t hdr aws f r om t he Bond = Ref unds + Damage Deduct i on
T of Change t o Pr e Ti me in t he Ga me when t he Char ge Syst em Changes to Pr epay =
INT(20 + 20 * I NI T( Rand) )
Unc e r t Uncer t ai nt y = I NTI ( r andom) or a number is sel ect ed r andoml y bet ween 0 and 1
Uni t I np Cost Uni t Cost of I nput s = gr aph( Wast e Fract i on). Gr aphi cal Poi nt s Incl ude
(0.0,0.910),(0.100,0.745),(0.200,0.650),(0.300,0.550),(0.400,0.485),
(0.500,0.440),(0.600,0.395),(0.700,0.355),(0.800,0.335),
(0.900,0.315 ),( 1.00,0.300)
Uni t Was t e Re d Cost Uni t Cost of Wast e Reduct i on = gr aph( Was t eReducTech) . Gr aph Poi nt s Incl ude
(0.0,0.0200),(0.100,0.0500),(0.200,0.0800),(0.300,0.120),
(0.400,0.180),(0.500,0.250),(0.600,0.310),(0.700,0.380),
(0.800,0.495),(0.900,0.6651,(1.00,0.9501
Us r Pr od Us e r Pr oduct i on = Thi s vari abl e allows pl ayers of t he game to over ri de t he level of fi rm pr oduct i on set
by t he model
Wa s t e Re duc Te c h Wast e Reduct i on Technol ogy is initially set at 0.5, but can be changed at any t i me by t he pl ayer of t he
game
Was t e Fr act i on The Per cent age of I nput s t hat end up as Wast e = 0.5
Wor st Case Es t i mat e Es t i mat e of t he Wor s t Case Damages = Max Di schar ges *200
Ref erences
cat i ons to Envi r onment al , Conser vat i on, and Cons umer
Acker man, B. and St ewart , R. , 1985. Ref or mi ng envi r onmen- Policy. J ohns Hopki ns Uni versi t y Press, Bal t i mor e, MD.
tal law: t he democr at i c case for mar ket i ncent i ves. Boul di ng, K., 1966. The economi cs of t he comi ng spaceshi p
Col umbi a J. Envi ron. Law, 13: 171-199. eart h. In: H. Dal y and K. Towns end (Edi t ors), Val ui ng t he
Adar , Z. and Gri ffi n, J., 1976. Uncer t ai nt y and t he choi ce of
Ear t h. MI T Press, Cambr i dge, MA, pp. 297-309 (1993).
pol l ut i on cont r ol i nst r ument s. J. Envi ron. Econ. Manage. , Brooks, M. and Hei j dr a, B., 1987. Rent - s eeki ng and pol l ut i on
3: 178-188. t axat i on: an ext ensi on. Sout h. Econ. J., 54: 335-342.
At ki nson, S. and Lewis, D. , 1974. A cost - ef f ect i veness analysis Burrows, P., 1979. Pi govi an taxes, pol l ut er subsi di es, regul a-
of al t ernat i ve air qual i t y cont rol st rat egi es. J. Envi ron.
tion, and t he size of a pol l ut i ng i ndust ry. Can. J. Econ. , 12:
Econ. Manage. , 1: 237-250. 494-501.
At ki nson, S. and Lewis, D. , 1976. Det er mi nat i on and i mpl e- Carson, R., 1962. Si l ent Spri ng. Fawcet t Publ i cat i ons Inc.,
ment at i on of opt i mal air quality st andar ds. J. Envi ron. Gr eenwi ch.
Econ. Manage. , 3: 363-380. Common, M., 1977. A not e on t he use of t axes to cont rol
At ki nson, S. and Ti et enber g, T., 1982. The empi ri cal pr oper - pol l ut i on. Scand. J. Econ. , 79: 346-349.
t i es of two cl asses of desi gns for t r ans f er abl e di schar ge Conr ad, K., 1987. An i ncent i ve s cheme for opt i mal pri ci ng
per mi t mar ket s. J. Envi ron. Econ. Manage. , 9: 101-121.
and envi r onment al pr ot ect i on. J. Inst. Theor . Econ. , 143:
Averch, H. and J ohns on, L., 1962. Behavi or of t he fi rm under 402-421.
r egul at or y const r ai nt . Am. Econ. Rev. , 52: 1053-1069. Cost anza, R., 1987a. Social t r aps and envi r onment al policy.
Axel r od, R., 1984. The Evol ut i on of Cooper at i on. Basic Books, Bi oSci ence, 37: 407-412.
New York, NY. Cost anza, R. , 1987b. Si mul at i on model i ng on t he Maci nt osh
Baron, D. and Taggart , R. , Jr., 1977. A model of regul at i on using STELLA. Bi oSci ence, 37: 129-132.
under uncer t ai nt y and a t est of r egul at or y bias. Bell J. Cost anza, R. and Cornwel l , L., 1992. The 4P appr oach t o
Econ. , 8: 151-167. deal i ng wi t h sci ent i fi c uncer t ai nt y. Envi r onment , 34(9):
Baumol , W. and Oat es, W., 1988. The Theor y of Envi r onmen- 12-2(I.
tal Policy. Cambr i dge Uni versi t y Press, New York, NY, Cost anza, R. and Per r i ngs, C., 1990. A flexible assur ance
299 pp. bondi ng syst em for i mpr oved envi r onment al management .
Beavis, B. and Wal ker, M., 1979. I nt er act i ve pol l ut ant s and Ecol. Econ. , 2: 57- 75.
j oi nt a ba t e me nt costs: achi evi ng wat er quality st andar ds
Cost anza, R. and Shr um, W., 1988. The effect s of t axat i on on
wi t h ef f l uent charges. J. Envi ron. Econ. Manage. , 6: 275-
moder at i ng t he confl i ct escal at i on process: an exper i ment
286. usi ng t he dol l ar auct i on game. Soc. Sci. Q., 69: 416-432.
Bohm, P., 1981. De pos i t - Re f und Systems: Theor y and Appl i -
226 L. Cornwell, R. Costanza / Ecol ogi cal Economics 11 (1994) 213-226
Cross, J. and Guyer, M., 1980. Social Traps. University of Mendel sohn, R., 1984. Endogenous technical change and
Michigan Press, Ann Arbor, MI. envi ronment al regulation. J. Environ. Econ. Manage., 11:
Edney, J. and Harper, C., 1978. The effects of information in 202-207.
a resource management problem: A social t rap analogy. Mill, E.S., 1972. Ur ban Economics. Scott Forseman, Glenville,
Human Ecol., 6: 387-395. IL.
Endres, A., 1983. Do effluent charges (always) reduce envi- Milliman, S. and Prince, R., 1989. Fi rm incentives to promot e
ronment al damage? Oxford Econ. Papers, 35: 254-261. technological change in pollution control. J. Environ. Econ.
Farber, S., 1991. Regulatory schemes and self-protective envi- Manage., 17: 247-265.
ronment al risk control: a comparison of insurance, liability Peles, Y. and Stein, J., 1976. The effect of rate of ret urn
and depos i t / r ef und system. Ecol. Econ., 3: 231-245. regulation is highly sensitive to t he nat ure of uncertainty.
Fishelson, G., 1976. Emission control policies under uncer- Am. Econ. Rev., 66: 278-289.
tainty. J. Environ. Econ. Manage., 3: 189-197. Perrings, C., 1987. Economy and Envi ronment : a Theoret i cal
Freeman, A.M., III., 1990. Economics, incentives and environ- Essay on t he I nt er dependence of Economic and Environ-
ment al regulation. In: N. Vig and M. Kraft (Editors), ment al Systems. Cambridge University Press, New York,
Envi ronment al Policy in t he 1990's. Congressional Quar- NY, 179 pp.
terly Inc., Washington, DC, pp. 145-166. Perrings, C., 1989. Envi ronment al bonds and t he incentive to
Haas, C., 1985. Incentive options for hazardous waste man- research in activities involving uncert ai n fut ure effects.
agement. J. Environ. Syst., 14: 373-393. Ecol. Econ., 1: 95-110.
Hahn, R. and Noll, R., 1983. Barriers to i mpl ement i ng trad- Pethig, R. and Fiedler, K., 1989. Effl uent charges on munici-
able air pollution permits: problems of regulatory interac- pal wastewater t r eat ment facilities: in search of t hei r theo-
tions. Yale J. Regulat., 1: 63-91. retical rationale. J. Econ., 49: 71-94.
Hahn, R. and Hester, G., 1989. Market abl e permits: lessons Platt, J., 1973. Social Traps. Am. Psychol., 28: 642-651.
for theory and practice. Ecol. Law Q., 16: 361-406. Plourde, C. and Yeung, D., 1989. A model of industrial
Hamilton, J., Sheshinski, E. and Slutsky, S., 1989. Product i on pollution in a stochastic envi ronment . J. Environ. Econ.
externalities and long-run equilibria: bargaining and Pigo- Manage., 16: 97-105.
vian taxation. Econ. Inquiry, 27: 453-471. Roberts, M. and Spence, M., 1976. Effluent charges and
Hardi n, G., 1968. The tragedy of t he commons. Science, 162: licenses under uncertainty. J. Publ. Econ., 5: 193-208.
1243-1248. Seskin, E., Anderson, R., Jr. and Reid, R., 1983. An empirical
Harford, J., 1987. Self-reporting of pollution and t he firm' s analysis of economic strategies for controlling air poilu-
behavi or under imperfectly enforceable regulation. J. Env- tion. J. Environ. Econ. Manage., 10: 112-124.
iron. Econ. Manage., 14: 293-303. Shaw, B., Winslett, B. and Cross, F., 1987. The global environ-
Katzman, M., 1987. Pollution liability insurance and catas- merit: a proposal to eliminate mari ne oil pollution. Nat.
trophic envi ronment al risk. J. Risk Insur., 34: 75-100. Resourc. J., 27: 157-185.
Koenig, E., 1984. Uncert ai nt y and pollution: t he role of indi- Smith, V. and Williams, A., 1992. Experi ment al market eco-
rect taxation. J. Publ. Econ., 24: 111-122. nomics. Sci. Am., 267: 116-121.
Kohn, R., 1977. Emission st andards and price distortion. J. Snedecor, G. and Cochran, W., 1980. Statistical Methods. The
Environ. Econ. Manage., 4: 200-208. Iowa State University Press, Ames, IA, 507 pp.
Krupnick, A., 1983. Costs of alternative policies for t he con- Sokal, R. and Rohlf, F., 1981. Biometry. W.H. Fr eeman and
trol of NO 2 in t he Baltimore region. Resources for t he Company, New York, NY, 859 pp.
Future, Washington, DC. Solow, R., 1971. The economist' s solution to pollution control.
Krupnick, A., Oates, W. and Van De Verg, E., 1983. On Science, 173: 498-503.
market abl e air pollution permits: t he case for a system of Stollery, K., 1985. Envi ronment al controls in extractive indus-
pollution offsets. In: E. Joeres and M. David (Editors), tries. Land Econ., 61: 136-144.
Buying a Bet t er Envi ronment : Cost Effective Regul at i on Tietenberg, T., 1973. Specific taxes and t he control of poilu-
t hrough Permi t Trading. University of Wisconsin Press, tion: a general equilibrium analysis. Q.J. Econ., 87: 503-
Madison, WI. 522.
Lee, D., 1984. The economics of enforcing pollution taxation. Ti et enberg, T., 1985. Emissions Trading: an Exercise in Re-
J. Environ. Econ. Manage., 11: 147-160. forming Pollution Policy. John Hopkins University Press
Magat, W., 1978. Pollution control and technological advance: for: Resources for t he Future, Baltimore, MD.
A dynamic model of t he firm. J. Environ. Econ. Manage., Webber, B. and Webber, D., 1985. Promoting economic in-
5: 1-25. centives for envi ronment al prot ect i on in t he surface min-
Marquand, J., 1981. An economist' s view of pollution charges ing control and reclamation act of 1977: an analysis of t he
as regulatory instruments. In: J. Butlin (Editor), Eco- design and i mpl ement at i on of reclamation performance
nomics of Envi ronment al and Nat ural Resources Policy. bonds. Nat. Resourc. J., 25: 390-414.
Westview Press, Boulder, CO, pp. 153-160. Weitzman, M., 1974. Prices vs. quantities. Rev. Econ. Stud.,
McHugh, R., 1985. The pot ent i al for private cost-increasing 68: 683-691.
technological innovation under a tax-based, economic in- White, L., 1976. Effl uent charges as a fast er means of achiev-
centive pollution control policy. Land Econ., 61: 58-64. ing pollution abat ement . Publ. Policy, 24: 111-125.