Sei sulla pagina 1di 15

American Economic Association

Exchange Rates and Prices Author(s): Rudiger Dornbusch Reviewed work(s): Source: The American Economic Review, Vol. 77, No. 1 (Mar., 1987), pp. 93-106 Published by: American Economic Association Stable URL: http://www.jstor.org/stable/1806731 . Accessed: 30/01/2012 16:43
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The American Economic Review.

http://www.jstor.org

Exchange Rates and Prices


By RUDIGER DORNBUSCH* The adjustment relative of pricesto exchange rate movements explainedin an is industrialorganization approach,using various models. The extentof price is adjustment, givenlaborcostsin therespective currencies, shownto dependon number domestic productsubstitutability, relative the of and foreign firms, and market structure. Some empirical evidence offered support theory. is to the The large appreciationof the U.S. dollar over the 1980-85 periodand the subsequent depreciation open importantareas of rereal search.The factof a largeand persistent appreciation poses a challenge for equilibrium theorists to uncover the change in fundamentals.For those who explain in medium-term macroeconomics termsof the wage contracts, Fischer-Taylor long-term example of the episode provides a striking of differential speeds of adjustment wages, goods, and assets prices. I adopt this perof spectivehere to explain the determinants groupsof relativeprice changesof different goods. Specifically,I advance hypotheses about those sectorswherean exchangerate change should lead to large relativeprice changes,and otherswherethe relativeprice shouldbe negligible. effects The approach is to draw on models of industrialorganizationto explain price adjustmentsin termsof the degreeof market the concentration, extentof producthomoand geneityand substitutability, the relative and foreign firms. marketsharesof domestic Models of industrialorganizationhave, of applied in trade thecourse,been fruitfully isory; their application to macro-pricing slow.1 sues, however,has been surprisingly There is a long-standingquestioning of purchasingpower parity(PPP), especially in the work of Irving Kravis and Robert Lipsey (1978, 1984).2 But so farthereseems to exist littleformalanalysisof price-setting behaviorin thiscontext.3 This paper adopts a partial-equilibrium a approach in that it assumes throughout in given,exogeneousmovement thenominal exchangerate. The exchangerate movement flexible moneywage and the less-than-fully interactto produce a cost shock for some firmsin the firmsin an industry-foreign abroad-and home marketand home firms thus bring about the need for an industrywide adjustment in prices. Although the assumption of exogeneous exchange rate movements and sticky wages is open to it criticism, is a usefulworkinghypothesis relative price forthepurposeof investigating issues. Section I reviewssome facts. Section II offersa stylizedview of the link between exchange rates and prices, and the third
See Avinash Dixit (1984) and Elhanan Helpman work on and and Paul Krugman (1985) for extensive to references trade applications.In the macro context, see Olivier Blanchard (1985), Oliver Hart (1982), and N. GregoryMankiw (1985). see 2For a reviewof the PPP literature, my papers (1985, 1986). 3JoshuaAizenman(1985, 1986) and AlbertoGiovanbehaviorin theconprice-setting nini (1985) investigate Theirfocus,however, textof exchangerate movements. costs and uncerissues of transactions is on short-term in movements ratherthan on thelarge,persistent tainty the real exchangerate. See, too, themorerecentpapers (1986), and by CatherineMann (1986), RobertFeinberg Eugene Flood (1986).
1

*MassachusettsInstitute Technology, of Cambridge, MA 02139. I am indebtedto OlivierBlanchard, Stanley Fischer, Paul Krugman, Michael Rothschild, Julio Rotemberg,Sergio Sanchez, Lawrence Summers,and Jean Tirole. Avinash Dixit providedespeciallyvaluable An earlierversionof this paper was presuggestions. sented at the 1985 NBER SummerWorkshopand the NBER Meeting on Business Fluctuationsand I acknowledgehelpful commentsreceivedon those occasions. David Wilcox providedvaluable researchassistance. 93

94

THE AMERICAN ECONOMIC REVIEW

MARCH 1987

section studies the behaviorof equilibrium prices. I. SomeFacts The large dollar movements reflected are both in absolute and relative prices.Table 1 shows two measuresof the change in U.S. relativecosts and prices: relativeunit labor in costs and therelative value-addeddeflator In manufacturing. each case, the U.S. series is deflatedby the corresponding time-series for the trade-weighted averagein dollars of of our trading The largemagnitude partners. costs and pricesarises the change in relative fromthefactthatunitlabor costsand prices abroad (in nationalcurrencies) wererising at a lowerratethanin theUnitedStates.But at thanoffsetthe same time,the dollar,rather trendby a depreciation, ting the divergent further reinforced divergence a strong that by appreciation. The depreciationsince mid1985 has not been sufficient eliminate to the changein competitiveness. Figure 1 shows absolute prices measured and the deflators by the U.S. GNP deflator for imports. Prior to 1980, import prices increase more rapidlythanthe deflator and, to a lesser extent,so do exportprices (not shown). During this period, the dollar was After1980,however, dollar the depreciating. and appreciation getsunderway, import price increases slow down and ultimately import prices fall in absolute terms.Export prices moreclosely,though trackthe GDP deflator is the patternof divergences similarto that for imports.At this broad level, it is clear then that importprices fell relativeto the and relative export deflator to prices. In the absence of comprehensive price series,Table 2 showsunitvalues fordifferent exportand importgroups.The table brings out thattheabsolutedeclinein import prices mustbe primarily attributed thefirst to three groups, and not to finishedmanufactures. Oil priceincreasesin 1979 easilyexplainthe divergent patternof exportand importunit values for crude materials.The interesting is comparison,therefore, betweenthe relatively homogeneous commoditygroupsfood and semimanufactures-andfinished and prodmanufactures whereprice setting

TABLE 1-RELATIVE COSTS AND PRICES IN MANUFACTURING

Change) (CumulativePercentage

1976-80 1980-85:1 1980-85:IV RelativeValueAdded Deflator Relative Unit Labor Cost -14.7 -12.6 49.3 59.8 27.0 22.0

Source: International FinanceStatistics, Yearbook 1985 and August1986.

Import Prices 140-

148

7~~~~
136 + GNP Deflator 0/1,
//'

120 -

+0

100

1979
FIGURE

1980

1981

I I I l 1982 1983 1984 1985 1986

1. THE IMPACT OF DOLLAR APPRECIATION

uct differentiation likelyto be important. are For the formergroup, export and import unit values move roughly line, while for in finished manufactures, exports follow the domestic price trend and importsshow a much smallerincrease. II. Standard Models There are two extrememodels of price in literature. relationships theopen economy One assumes that the "law of one price" holds. Prices of goods are geographically arbitraged and, adjusted for tariffsand transport costs, theyare equalized in different locations.Homogeneity, and information perfectcompetitionassure this result.4Let
4For a reviewof PPP, see mypaper (1985).

VOL. 77 NO. I

DORNBUSCH: EXCHANGE RATES AND PRICES


TABLE 2-UNIT VALUES OF IMPORTSAND EXPORTS

95

= (Index 1980:J 100) Materials

Foods E 1979:11 1980:1 1985:1 87 100 94 M 82 100 87

Semimanufactures E 71 100 86 M 77 100 82

Finished manufactures E 95 100 139 M 91 100 106

E 92 100 91

M 60 100 97

and Note: E = Exports, M = Imports;fordefinition source,see the Appendix.

Pi, p,, and e denote the price of good i in the the home countryand currency, foreign price of forprice, and the home-currency thenimplies eign exchange.Arbitrage (1) pi=ep*.

veror In thisform, in the first-difference sion of Gustav Cassel, thelaw of one priceis The law of asserted in the PPP literature. one price has been applied in the monetary approach to exchangerates in combination of with the quantitytheory moneyand the to flexibility obtain assumptionof full-price a theory the exchangerate.An important of not of spatialarbitrage, implication complete but onlyforcommodities forall goods,is the idea that relativenational price levels in a common currencyare independentof the exchange rate, since exchange rate movenadivergent reflect, passively, mentsmerely tional price trends.That is, of course, an postulate application of the homogeneity whichholds whenmoneyis fully neutral. The alternativemodel might be called "Keynesian." Here it is assumed that each is country fullyspecializedin theproduction of "its own" good. Domestic and foreign goods are less than fullyhomogeneousor substitutable.Wages are fixed in national currencies at least sticky. or Letting P and P * be the national GDP the deflators, relativeprice of domesticand foreign goods or the real exchange rate then,is (2) A = P/eP *

is constant,then forgivenunit labor costs, priceswillalso be given.Hence in thismodel, exchange rate movementschange relative prices one-for-one.Exchange rate-induced the priceaffect world changesin the relative distributionof demand and employment. This approach tends to be used in open economyversionsof theIS-LM modelin the Meade-Mundell tradition. Equation (1) would be a usefulmodel of price relationsfor materialsinternational say sisal, copper, tea-whereas (2) more nearly describeswhat happens with manufactures.But the assumptionof a constant as justified we shall now markupis no longer have see when domestic and foreignfirms in pricing. interactions their strategic Models Pricing III. Equilibrium Table 1 gave evidenceof large,persistent relain fluctuations exchangerate-adjusted In tive prices in manufacturing. thissection models that would exI explore theoretical as plain theseprice movements theresultof changesin relativeunitlabor costs. I The basic assumption make is thatfirms in any industryhave a linear technology, with labor as the only input. Unit labor costs, w and w*, are given in home and This assumprespectively. foreigncurrency, witha modelof tion about costsis combined about thebehavpricingto yieldpredictions is ior of relativeprices.The experiment simply this: the exchange rate change, say a unitlabor lowersforeign dollar appreciation, the equicosts in dollars.As a result, market librium is disturbedin each industryand must occur. price and output adjustments

If the markupof pricesoverunitlabor costs

96

THE AMERICAN ECONOMIC REVIEW


*

MARCH 1987
J

look like depends What these adjustments on threefactors:5 or Market integration separation.Is a particular commodity traded in an inor tegratedworld market, are theresignifispatial arbitrage? to cant barriers restrict betweendomesticand forSubstitution eign variants of a product.The extent of and the price setting influences substitution of outputeffects cost and pricechanges. Is perMarketorganization. the market are in fectlycompetitive, which case firms price takers, or is the marketimperfectly in competitiveor oligopolistic, which case firmsare price settersand may interactin ways? strategic in Two models lend themselves a straightfashionto formulating pricerethe forward sponse to cost shocksof partof theindustry. The Cournot model assumes perfectsubsuppliersand stitutionbetween alternative places more emphasis on the extent of more variaoligopoly.It allows in principle in to tionin themarkup response costshocks, patand thus has the potentialfora richer tern of response to cost shocks.The Dixitemphasizes Stiglitz(1977) modelby contrast imperfectsubstitutionbetween alternative looks very suppliers and in its predictions much like the Keynesian model discussed above. An alternativeto the Dixit-Stiglitz productdifferentimodel,again emphasizing ation,is the Salop (1979) model of competition on a circle. Model A. The Cournot

J*

EQUILIBRIUM FIGURE2. THE COURNOT

are capwhere all nonprice determinants turedin the constant.There are n domestic firms withrespecsuppliersand n * foreign Aggregate tive sales of q and q * per firm. Q, sales of thesefirms, have to sumto market demand: (4) Q = nq + n*q*.

Each firmmaximizes profitstaking the of as sales of other firms given.Profits the in firm domesticand foreign representative the home marketare

- w) In the Cournotformulation, analytical (5) q'=(p the sold commodity focus is on a homogeneous x [a - bp - (n -l1)q - n*q*], in an oligopolisticmarket.Each seller assumes that other sellers defend theirsales nj= (p/e - w*) volume. I assume that thereis an effective spatial separationbetweenthehome market X [a - bp - nq - (n* - I)q*] and discussthepricing and foreign markets, in the U.S. market. Maximization gives rise to the reaction For expository purposes,I startwith the functions shown in Figure 2. The home case of a lineardemandfunction: is reactionfunction JJ,whileJ *J* country's representsthe foreigncountry.They yield a - bp, (3) Qd= the Cournot-Nash equilibrium shown at quantity point A whichgivestheequilibrium 5A fourth form of is itemof relevance the functional domestic allocation between representative the demand curve. See Jesus Seade (1983) and Dixit firms. and foreign The commonequilibrium (1986).

VOL. 77 NO. I

DORNB USCH: EXCHANGE RATES AND PRICES

97

is pricein theindustry givenby (6) p = (nw + n*ew*)/N+ a/bN; N n+ +1.

A dollar appreciation shifts the J *J* thusleadingto scheduleout and to theright, sales and reduceddomestic increasedforeign sales. At the initial level of sales for every other supplier, the individual foreignfirm faces a given marginalrevenueschedulein dollars but experiencesa reductionin its dollar marginal cost, and hence wishes to at increase output. In the new equilibrium firms increasetheiroutput point A', foreign The whilehomefirms contract. industry price (6). declines,as seen from in We are now interested the extentto which exchange rate movements move(or mentsin relativeunit labor costs) affect the of equilibrium price. The elasticity the equilibriumprice with respectto the exchange rate is (7)
=

(n */N)(ew */p).

The other extremecase where exchange on rate has no influence home pricesresults when there are few firmsin the industry, most of which are domestic.In that case, absorb the dollar appreciation foreignfirms rather primarily the formof extraprofits in than increasedsales. exThe Cournot model thus potentially pricesand steepprice plains bothunchanging share structure-import declines.The market and concentration-are the key parameters that explain the outcome. competConsider next U.S. exportfirms A ing in a foreignmarket. dollar appreciation will lower their marginal revenue in dollarcost, dollars.Withunchanged marginal In will contract. termsof Figure these firms our market, schedule 2 applied to the foreign JJ shifts down and to theleft.The common currency pricerises,but in dollarsit foreign to declines,thoughless thanproportionately the appreciation.Using the same model for theforeign we market, findthattheelasticity to of foreign pricewithrespect theexchange rate is (8) *=-(nt/N *)(w/ep*),

The elasticityformulahas two determifirms nants: the relativenumberof foreign withwages (or the relativenumberof firms not fixedin dollars), and the ratio of marSince suppliers. ginal cost to priceof foreign it both termsare fractions, is immediately will clear from(7) thata dollar appreciation The lowerprice less thanproportionally. decline in the dollar price is largerthe more the (i.e., the smallerthe competitive industry cost) and the markupof price overmarginal larger the share of importsin total sales. This lattertermis represented n*/N on by and initially the assumption of symmetry equal wages betweencountries. because it Equation (7) is interesting all stretches the way fromthe "small country" case to the case whereexchangerates have virtuallyno impact on home pricing. The small countrycase, in the trade literais ture,is the case wherea country a price In takerin world markets. thatcase, a currency depreciationwill raise prices in the This is, of course,thelimitsame proportion. and a large competition ing case forperfect firms relative thenumto numberof foreign ber of home firms.

in of where n' is the number domesticfirms and N * thetotalnumber the foreign market the of firms.With p* a negativefraction, dollar priceof exports, *e, has an elasticity p 1 + (p,and hence mustdeclinein responseto a dollar appreciation. Rememberingthat the marketsare separated, let us look at the price of U.S. exp/ep*. portsrelativeto thepriceof imports dollar exIn case of a dollar appreciation, port prices rise relativeto importprices if holds: the following condition (9) sP>Iq+c*.

In principle,the conditioncan go either case, exportand way. In the small country import prices in dollars fall in the same as appreciates = (q proportion the currency 1, * = - 1), so thattherelative pricep/ep * remains constant. In general,the outcome depends on the relativeoligopolisticstructure of the two markets. Exportprices will pricesin theappreciarise relativeto import is tion case, if at-homeimportcompetition

98

THE AMERICAN ECONOMIC REVIEW

MARCH 1987

pervasiveand foreign marketsare strongly affected U.S. suppliersas well as highly by competitive. Jesus Seade (1983) and Avinash Dixit (1986) have shown that the price effects of disturbances oligopolymodels are highly in sensitiveto the functional form.It is thereforeinteresting ask what happens witha to more generalfunctional form.For example, with a constantelasticity demand function D = Ap exp(- /), the elasticity of equilibriumprice with respectto the exchange = rate becomes qp (n */N)(ew */W), where N= n + n* and W= (n/N)w + (n */N ) ew*. The exchangerateimpactthus dependsno longeron cost-price markup and, when costs are initially equal betweencounof tries,is only a function the relative number of firms. In general the elasticity price withreof spect to theexchangerateis' (9'-) p = [n*/(N- 0)] (ew *1p) > O, of where 0 is the elasticity the slope of the inversedemand curvewhich,forstability, is less than N. In the linearcase, the formula reduces to (9) since 0 = 0. But, since 0 is positive (or zero), thereis no upper bound it on q. From public finance applications, is known that tax disturbances may lead to magnified price adjustment underoligopoly. I find the same possibility here for the exon change rate effect prices. B. The Dixit-Stiglitz Model The representative consumer thismodel in maximizes a utilityfunctionV with conz sumptionof two commodities and x as

arguments:
(10) V= U(z, x); X =(EXa )la

0 < a <1. x I focuson commodity whichis an indexof of brandsof thesame consumption different are good. It is assumedthatthere n domestic firmssupplyingsome varianteach, and n* firms foreign doing the same. Maximizationyieldsthe demand foreach individualbrand,as well as theutility-based x: price index forcommodity (11) (12) xi=x(P/pi)c; p =[(Epih+ c=1/(1-a), Ep h)l; h= -a/(1-a). In equation (12) pi denotes the price of a p1 brand produced in the home country, is thepriceof an imported brand,and P is the industry price. in We are now interested the responseof prices to cost shocks.The individualimperfirm facesa demandcurve fectly competitive as in (11) withthe relative priceof its prodThe firmasuct pi/P as the determinant. small so thatits own sumes it is sufficiently price changes leave the industry price, P, firm's unchanged.The representative profits are
(13)
7i=(Pi-W)Xi.

Maximization yields the familiarconstant markuppricingequation: (14) pi= aw; a=1/(1-1/c),

6Let p = F(Q) be theinverse market demandcurve. The elasticity the slope of theinversedemand curve of The derivation (9') is of thenis 0 = - QF"(Q)/F'(Q). maximizesprofitsr = as follows: The individualfirm F(Q)q - wq and s = (F(Q)/e)q* - w*q*. The firstorder conditionsare F(Q) + qF'(Q) = w and F(Q) + the conditions q * F' (Q) = ew*. Multiplying first-order respectively by n and n* and adding them yields: (n + n*)F(Q)+ QF'(Q) = nw+ n*ew*. Differentiating this expression,using dp = F'(Q)dQ, yields the expressionfor 0 in (9').

where a depends inversely the elasticity on of substitutionamong variants. Since the is each industry structure symmetric, domeswill followthesame pricing tic firm rulewith an equal markup. Now assume again markets separated are and we can thus meaningfully discuss the firm our market. for price set by a foreign in Foreign firms the home marketface the

VOL. 77 NO. 1

DORNBUSCH: EXCHANGE RATES AND PRICES

99

same formof demand curve as home firms and hence theyalso followthe same pricing rule,withthe same markup, withforeign but wages in dollars, ew*, as the base of their pricing: (15) pj = aew *.

stay constantrelativeto domesticprices of the same variant.This is, of course,theexact Keynesian specificationof the fixed-price model which is derivedhere as an implication of given labor costs and an invariant markup. Model Dixit-Stiglitz C. An Extended model assumes ChamThe Dixit-Stiglitz competitionand hence berlinian imperfect each supplierassumesthathe does not affect industry price. Strategic interactionwith excluded. But the other firmsis therefore can of products same structure differentiated ineasily be adapted to introducestrategic variation. teractionby way of a conjectural Assume, contraryto the precedingsection, large is thattheindividualfirm sufficiently to industry price.Assume,too, thatfirms affect price,and respondto changesin theindustry be variation theparameter let theconjectural betweenzero and one. Thus a a, a fraction rise in the industry one-percentage-point to price is assumed to cause each firm raise theirprice by a percent.Assuminga given it conjecturalvariationratherthan deriving is framework froma dynamicgame-theoretic Nor is thereany conobviously a shortcut. conjecturalvariacern here with consistent tions. With this adaptation,the demand curve price policy no facing the individualfirm's longeris a constantmarkupover unit labor costs but ratherbecomes (18) Pi =a'w; a'=1/[1-1/c(I-efl,

From (14) and (15), we have two strong predictions: First the relativeprice of domestic and foreignvariants in the home marketdepends just on relativeunit labor costs in a commoncurrency: (16) pi/pj= w/ew*.

Second, it is readilyshown that the relative price of a domesticvariantin termsof the industryprice index (pi1P) is just a functionof the relativewage, w/ew*. The of elasticity the relative pricewill be (17) nzl/(n + n*z);
z=(w/ew*)l/h.

If wages are initiallyequal between counof tries,the effect an exchangerate change on the industryprice and on the relative pricedependsmerely thefraction firms on of in thathas wages fixed foreign and currency, hence experiencesa reduction theircosts of in dollars when the dollarappreciates. Given the wages in home and foreign currency, the Dixit-Stiglitzmodel provides about theimpactof dollar strong predictions appreciation: The prices of imported variantsfall in proportion thedeclineof dollarunitlabor to costs of foreign firmsand the prices of domesticvariantswould remainunchanged. Exportingfirms home,althoughthey at have to compete in foreignmarkets,still followtheirmarkuppricing dollarwages. on Accordingly, changein thedollar does not a affect theirdollar exportprice.Of course,it does affect A theirsales and profits. dollar appreciationwill raise theirforeign currency price in the same proportion henceraise and theirrelativepricein the foreign market. The strongprediction the model is to of look fora sharpfallin import pricesrelative to domesticprices and to see exportprices

where the term e=(dP/P)/(dPJ1Pi) capbetweenfirms interaction tures the strategic as perceived by the individualprice-setting of prices The termis a function relative firm. variation:7 and theconjectural
of is 7The derivation as follows:maximization profits - w)c for a domestic firm k in (13) yields '+(Pk (E- l)/pk = O which yields the markup equation in price level with (18). The elasticityE of the aggregate pricePk domestic respectto a variationin an individual 9h + is derived as follows: from(12) we have ph Now assuming that a=d lnpi/dlnP h +ypjh. for all firms i and j other than k we obtain the price withrespectto a variation of elasticity aggregate

in Pk.

100

THEAMERICAN ECONOMIC REVIEW

MARCH 1987

/ H~~~~H

p? I
pi HH

/
,,18~~~~~~~~~,
oo~ /~~~~~d H*

0~~~~

more nearly the Cournot model. Foreign firmsreduce theirprice proportionally less than the reductionin dollar unitlabor costs and home firms cut theirprice. But at A', the relativeprice of domesticproductshas increased compared to A as can be seen by the slope of a ray through compared A' to OR. D. Competition theCircle on

H1
,,

I conclude the discussionof manufactures prices with a sketch of a third model of pricing for differentiated products. In the O p p. p. Dixit-Stiglitz model,consumers buy some of each brand of a product.Applied to toothFIGuRE 3. THE EXTENDED DIXIT-STIGLITZ MODEL paste, that is an implausible model; we should thereforelook for an alternative model whereconsumers buyonlyone brand. A particularlymanageable version is the Salop model where consumers'tastes (defined by preferences for the attributes or (19) ?< e(G,Pj/Pj)7'=of characteristics goods) are uniformly spread over the unit circle.Since domesticand for[ +i n){n + <1. have potentially eign firms different costs,a jpi) symmetric equilibriumdoes not necessarily exist. From (18) and (19), it is clear thatpricing I simplify matters assumingthatthere by decisions are now interdependent. can We is an even numberof firms, that domestic represent each firm's pricing policyin terms and foreignfirms alternate along the circle, of a price reactionfunction: and that each consumerbuys a unit from one or the otherof the firms adjacent to his location.This is a verystrong (her) preferred (20) Pi = F( p/pi, a, c) w simplifying assumptionbecause it impliesa very different competitionpattern from a P.= F *(pi/pj,a, c)ew* circumstancewhere two foreignfirmsare adjacent. Figure 3 shows the impact of a dollar Producershave constantunit labor costs appreciation in this setting.The schedules and, other than entrycosts, there are no HH and H*H* are the reactionfunctions fixedcosts. With theseassumptions, can we and A is theinitialequilibrium. derive equilibrium prices and studythe imAn appreciation willshift foreign reacthe pact of dollar cost changesforforeign suption function and to theleftwhileleaving pliers. Each consumeris located at a point up the home country's reaction functionin on thecircle.The significance thelocation of in place. The magnitude the shift H*H* of on the circle is that firms may not supply at given relativeprices (for example,along precisely the most preferred product. Acthe OR ray) is proportional the apprecia- cordingly, to the consumeris forcedto chose tion. Thus AB/AO representsthe peroffered themost betweenthe alternatives by is centageappreciation. The new equilibrium adjacentlylocated brands. FollowingSalop, therefore A'. at consumers' surplus derived frombuying a Note that this equilibriumat A' differs good that is a distance T from the best from the Dixit-Stiglitz one and resembles location (on the circle)dependson theprice

/~~~~~~ Hw

VOL. 77 NO. 1

DORNBUSCH: EXCHANGE RATES AND PRICES

101

and on thedistance and therelationship is assumed linear: (21)


h= v- c--p,

firms:
(26) p=c/n+(ew* +2w)/3; * p * = c/n + (2ew + w)/3.

wherev is a constant, denotes utility c the From(26), we can calculate elasticity the costperunitdistance from bestlocation, ofprices the with respect theexchange in to rate and p is the price of a particular firm. thismodel: Consumers will be indifferent between the brandsoffered two competing by = firms on p9 (1/3) (ew */p); (27) either side of their preferred location the if
consumers'surplusis thesame, hi= hi. Tak-

ing the case of n firms that are equally spacedon thecircle, condition indif- Note thattheseelasticities the for once again are ference a between domestic and a foreign fractions. wagesand hence If prices iniare
supplieris

* = (2/3)(ew*/p*).

tiallyequal, w = ep*, theelasticities simplify

to thefollowing expressions: (27') q=4/3;

(22) v-CT-p*=V-C(Iln-T)-P.

Hence thedistance served a foreign by firm = 24'/3; 4--1/(1+ c/nw). is an increasing function theprice of charged by domestic firms a declining and function The elasticities showthattherelative price ofitsownprice: of imported goods declinesand that the changein therelative price4/3 is smaller, the smaller number firms theinthe of in (22') T=(p+c/ln-p*)/2c. and dustry, thelower substitutability the as Profits theforeign areequal to 2Lx for firm measuredby the termc. Along withthe times excessofprice the overmarginal cost: change relative in will there be a shift prices, in demand from homefirms foreign to firms (23) as consumers trade thereduction price off in for a largerdistance from their most-preIT* (p*-ew*)2L( p + c/n-p*)/2c, ferred brandlocation. At thispoint, is worth it commenting on of there whereL denotesthetotalnumber con- theproperties theequilibrium when of pattern between domessumers henceL also represents den- is notan alternating and the firms. Specifically, suppose sityper unit distance servedby the firm. tic and foreign are two and Sincethefirm serves bothsidesof its loca- thatthere fivefirms, domestic It firms. is apparent 2LLTis thetotal tion, number units of sold. threeadjacentforeign firm only Maximization taking domestic price given thatthemiddleforeign competes as withforeign firms, hencewill cut its and yields foreign the reaction function: pricemorethanthe outlying foreign firms thatcompete withhomefirms whichhave p* = (p+c/n+ew*)/2. (24) notexperiencedcostreduction. a Hencethere The typical domestic firm's reaction function willbe three prices. is derived thesamemanner: in The same modelcan be appliedto the foreign market.In termsof foreign exp = (p* + c/n + w)/2. the change, priceswillriseand therelative (25) priceof our export brands abroadwillrise. From(24) and (25), we obtainthesolution But,becauseit rises proportionally than less forthepricescharged homeand foreign thecurrency by appreciates, export the price in

102

THE AMERICAN ECONOMIC REVIEW

MARCH 1987

conThe empiricallytestablehypotheses cost markups and the cern price-marginal prices.For differentiated behaviorof relative )*=1/(1+c/nn*w), (28) qV=1-24*/3; products,it is always the case that export serv- and domestic prices will stay closer in line where n* is the total numberof firms ing the foreignmarket.We can therefore than import and domestic prices. In the Dixit-Stiglitzmodel, importsfall in terms find the change in the relative price of of and in domesticexportsin terms imports of domesticgoods and the relativeprice of termsof home brands: exports goods stays unchangedin termsof home goods. In other models, the export T -q* =1-2(4 + 4*)/3; (29) price can in principleeven decline in terms of imports. 9)' 9=1-24*/3-+ /3. prices The first point to noteis thatexport pricesas a of may riseor fallin terms import result of appreciation. But the fewer the in the numberof firms each country, more leads to a fall in likelythat an appreciation as the relativeprice of exports.By contrast, increases(and hence 4 the numberof firms and 4* tend to unity),the relativeprice of the exports must rise, reflecting increasein the relativeunit labor cost at home which relative prices. sets competitive pricesmay The second pointis thatexport decline relativeto domestic pricesas a result This mustbe the case if of an appreciation. in the numberof firms the two marketsis the same (4)= 4*). As the numberof firms increases,the relativeprice tends to remain unchanged.This resultarises because price cost which gets competeddown to marginal is the same forhome and exportproduction. It is apparent from(29) that the change in will the relative of pricein terms importables always be larger than that in terms of domesticgoods. IV. SomeEvidence

dollars changesin theproportion:

is Econometrictestingof the hypotheses unfortunately precludedby the absence of a matcheddata set of export, comprehensive import,and domesticprices.The BLS now and pricesforexports publishestransactions to importsthat are disaggregated the 4-digit on level and classified theSIC basis. But few of theseriesgo back beyondtheearly1980's. Wherethey therevisions theSIC-based of do, U.S. producerpricesin mostcases are either not all available yet,or only go back very few years. A completeoverlap betweenexport, import,and domesticprices for more than two years apparentlyonly exists for fewerthan a handfulof cases, and overlap to betweenany two seriesis limited less than a dozen. levelthere interestare At a moreinformal ing patterns to observe. First, consider a comparisonof U.S. exportpricesin dollars with those of Germanyand Japan. Table 3 shows the percentage loss in U.S. competitiveness over the period 1980:IV to 1984:IV, using as a sampleall available data E. Summary at a highly disaggregatedlevel. In the of thereare 36 difWe have now seen commonfeatures a U.S.-Germany comparison, in numberof models: theyall predictthatapferent matchedtime-series, theU.S.-Japan preciation should lead to a decline in the comparison,thereare 20. Typical items in In are priceof imports. thecase of homogeneous the list of commodities "gears and gear match of goods, domesticfirms, course,fully units," or "household electricalspace heatthe declinein price.If productsare differen- ing." The data do not allow us to tell whether tiated, it will always be the case that the relative price of the importedbrands detheseare pricesof thesame productssold in the same third market (say France), or clines in response to an appreciation.The exportsto different extentof the decline depends on a measure whetherthey represent of competitionand on the relativenumber markets(say U.S. sales to France and German sales in the United States). Accordfirms. of home and foreign

VOL. 77 NO. 1
TABLE 3-CHANGES

DORNBUSCH: EXCHANGE RATES AND PRICES


IN RELATIVE PRICES: 151

103

U.S. VS. GERMANY AND JAPAN (Percentage changein relative export prices: 1980:IV-1984:IV) U.S.-Germany Mean StandardDeviation Source: See the Appendix.
100 TABLE 4-CUMULATIVE
0-

U.S.-Japan 24.9 8.3

-e134
130 120 Dol lar
X /

39.3 6.1

Telecommunications Equipment

1980:IV-1985:1

PRICE CHANGE:

90
90 I 77 78

ExportPrices ImportPrices Non-Electrical Machinery Scientific Instruments Source: See the Appendix. 18.0 18.0 - 10.1 - 13.4

79

80

8 1

82

83

84

151

ingly, cannottellfrom we thesedata whether marketsegmentation imperor they reflect fect substitution. with Theyareconsistent markets being segmented, goods being perbut fectsubstitutes and havinga commonprice in the same marketindependent supply of source. But they are also consistentwith markets being integrated-a commonworld market-but goods being imperfectsubstitutes that the relative so priceof different supplierscan change. Consider next a comparisonof the transactions prices of U.S. exportsand U.S. imports in the same commodity group. There is overwhelmingevidence that, virtually without the exception, dollarappreciation of 1980-85 has been accompanied by an increase in the price of exports relative to imports. Evidence in this directioncomes fromexport-import pricecomparisons the at more narrow2- and 4-digitlevel. An example is providedin Table 4 whichshowsdata fortwo2-digit industries. Figure 4 shows the ratio of exportprices to import prices for telecommunications equipmentand fornonelectrical machinery. The figure also show the index of the nominal dollar exchange rate index. The dollar appreciationsince 1980 gives rise to an increase in the relative price of exports in termsof imports.Table 4 shows indices of the relativeexport-import priceforall series

140_

Dollar

120 0-0-.O'O

, 100
0-o, 0'

o o' o

0,0?'

116

Non -E lectrical Machinery

I
77 78

II
79

I
80 8 1

I
82

I
83 84

FIGURE 4

wherecomparableSIC data exist.The same pattern would be obtained by comparing U.S. to German and Japaneseexportprices in these individualcommodity groups.The firstfindingthen is, that across industries, virtuallywithout exception,export prices have increasedrelative import to prices.This is true at the level of individualcommodities,but also, as shownat the outsetof the exportand importunit paper, foraggregate values. This result would obtain strictly only in the Dixit-Stiglitz model. In theotherformuthoughit need not lations,it is a possibility occur. Tables 5-7 look at the price of exto ports and importsrelative each otherand

104

THE AMERICAN ECONOMIC REVIEW


TABLE 5-THE RATIO EXPORT TO IMPORT PRICES

MARCH 1987

(Index 1980:I = 100) 353 100 118 135

2011 1979:IV 1981:IV 1985:1 105 108 126

301 103 105 104

35 112 131

356 95 119 152

3569 96 121 143

357 106 110

3643 91 115 152

38 92 108 136

Note: The headingsare SIC codes. For definitions source,see theAppendix. and

TABLE 6-THE

RATIO OF EXPORT TO DOMESTIC PRICES

(Index 1980:IV = 100) 3533 99 100 100

3546 1979:IV 1981:IV 1985:1 Note: See Table 5. 101 100 95

3555 101 104 107

3674 109 91 93

3523 99 100 102

3519 103 103 105

3494 99 103 108

2011 110 93 108

3537 97 99 100

TABLE 7-THE

RATIO OF IMPORT TO DoMESTIc PRICES

(Index 1980:IV = 100) 3143 96 95 88

2311 1979:IV 1981:IV 1985:1


Note: See Table 5.

2033 108 92 90

3651 100 92

3531 98 90 76

2435 120 114 102

2011 105 92 85

3312 101 98 84

3313 88 88 74

100 101 110

relativeto domesticproducerprices in the commodity group. Exportpriceschange little relativeto domesticprices,even though there is no clear patternof decline in all industries.By contrast, most importprices decline in termsof domesticgoods. But the order of magnitudeof the decline remains relatively small compared to the change in relativeunit labor costs. With a change in relative unit labor costs of more than 40 the percent, declinein therelative priceis in most cases less than 20 percent. That is not once some at all out of line withthe theory degreeof "pricingto theAmericanmarket" is taken into account,just as the price-settingmodels above suggest. It is worthnotingthat,at the retaillevel, this effect would obtain even morestrongly. The reason is that here distribution costs come into play, so that even with the full of on pass-through cost reductions imported goods, the proportional decline in the price

of importedgoods would be muchless than the exchangerate appreciation. V. Concluding Remarks The models reviewedin this paper focus on a relativelyshorttime perspective. The wage rateis assumednot to reactto changes in output and profitability, the number and in and location of firms an industry unis affected. These assumptions plausiblein are the shortterm, it is clearthata sustained but real appreciation will ultimatelyshow its in effects wage cuts in thoseindustries where the loss in competitiveness causes unemployment and wage increasesin the expanding sectors.Firms will close in high-wage areas and entry will into an industry take place in areas where labor costs are low. These longer-term adjustments also part of the are macroeconomicsof adjustment exchange to rate movements.They imply that the ab-

VOL. 77 NO. 1

DORNBUSCH: EXCHANGE RATES AND PRICES

105

n *) will be endogeneousas the location of

of solute and relativenumbers firms and (n

firms the productcircle. on It is clear fromthe analysis offered here that for these issues, a microeconomic perIn it spectivewill also be helpful. particular, will be interesting see how pricingdecito sions are affectedby entryand relocation at possibilities an international level,and by of the anticipatedpersistence disequilibrium exchangerates. The analysis developed here has application not only to the exchangerate queseffectof tion, but also to the short-term The commonargument trade liberalization. is that a small country openingup can by take advantageof theworldmarkets, enjoying price reductionsin proportionto the tariff reduction. That clearly assumesperfect competition.If marketsare less than fully competitive,the analysis offeredhere becomes relevant to the trade liberalization issue.
DATA APPENDIX The data in Table 2 are unit values obtained from Data Resources,Inc., U.S. CentralData Bank. The data in Table 3 are compiledby the Bureau of Labor Statistics, Division of International Prices. The unpublished data are from a compilation entitled "Comparisons of U.S., German and Japanese Export Price Indices," dated May 1985. The data forexportand import pricesin Tables 4-7 are SIC-based price data compiled by the Branch of Export and Import Price Indexes of the Bureau of Labor Statistics.The data are froma printout, dated July 1985. The domestic price index for SIC-based producerprices is obtained fromData Resources,Inc. Table 4: The data refer to SIC codes 34 and 38, In respectively. Tables 5-7, thefollowing thedefiniare tionsof the SIC codes shownin the table heading: Table 5: 2011-Meat and Meat Packing Products; 301-Tiresand Inner Tubes; 35-Machinery, exceptelectrical;353-Construction, MiningEquipment;3569-General Industrial Machines; 357-Office, Computingand AccountingMachinery;3643-Current-Carrying Wiring Devices; 38-ScientificInstruments, Optical Goods, Clocks. Table 6: 3546-Power-Driven Hand Tools; 3555PrintingTrades Machines, Parts; 3674-Semiconductor Devices; 3533-Oiland Gas Field Equipment;3523-Farm Machineryand Equipment; 3519-Internal Combustion Engines; 3494-Metal Valves, Pipe Fittings;2011-Meat and Meat-Packing Products; 3537-Industrial Trucks, PortableElevators. Table 7: 2311-Men's or Boy's Suits & Coats, except Raincoats; 2033-Canned and Preserved Fruits,

Vegetables,Jams,Juices;3651-Videoand Audio Equipment; 3143-Men's FootwearexceptAthletic;3531-Construction Machinery; 2435-Hardwood Veneer and Plywood; 2011-Meat and Meat-Packing Products; 3311-Rolling Mill Products; 3313-Electrometallurgical Products. The index of the nominal dollar exchangerate in index Figure 4 is the Morgan Guarantytradeweighted of the dollar. The serieswas obtained fromData Reequipsources, Inc. The series for telecommunications ment and non-electricalmachineryare unpublished, SITC-based prices of exports and imports obtained Prices, Bureau of from the Division of International Labor Statistics.

REFERENCES
Aizenman, Joshua, " Monopolistic Competition and Deviations fromPPP," unpub-

lished manuscript, of University Chicago, 1985. ,"Testing Deviations fromPurchasing PowerParity (PPP)," Journal Interof nationalMoneyand Finance,March 1986, 5, 25-35. Blanchard, Olivier,"Monopolistic Competition, Small Menu Costs, and Real Effects of Nominal Money," unpublishedmanuscript,MIT, 1985. Dixit, Avinash, "InternationalTrade Policy for Oligopolistic Industries," Economic Journal, Suppl. 1984, 94, 1-16. , " Comparative Statics Oligopoly," for International EconomicReview, February 1986, 27, 107-122. and Stiglitz, Joseph, " Monopolistic Competition and Optimum Product Diversity," American Economic Review,June 1977, 67, 297-308. Dornbusch, Rudiger, "Flexible ExchangeRates and Interdependence," IMF StaffPapers, March 1983, reprinted Dollars, Debts in and Deficits, Cambridge:MIT Press,1986. "PurchasingPower Parity,"NBER Working Paper No. 1591, 1985, in Palgrave's Dictionary Economics, of London: Macmillan,forthcoming. ) "Inflation, Exchange Rates and Stabilization,"Essays in International Finance, International Finance Section, PrincetonUniversity, 1986. Giovannini, Alberto, "Exchange Rates and Traded Goods Prices,"unpublished manuscript, Columbia University, 1985.
_

106

THE AMERICAN ECONOMIC REVIEW

MARCH 1987

Feinberg,Robert M., " The Interaction of Foreign Exchange and Market Power Effects on German Domestic Prices," Journalof IndustrialEconomics,September 1986, 61-70. Flood,Eugene, "An EmpiricalAnalysisof Jr., the Effects ExchangeRate Changes on of Goods Prices," unpublishedmanuscript, StanfordUniversity, 1986. "A Hart,Oliver, Model of Imperfect CompetitionwithKeynesianFeatures,"Quarterly Journalof Economics, February1982, 97, 109-38. and Helpman, Ephanan, Krugman, Paul,Market Structure and ForeignTrade, Cambridge: MIT Press,1985. Kravis, Irving Lipsey, and Robert, "Price Behavior in the Light of Balance of Payments EcoTheories," Journalof International nomics, May 1978, 8, 193-246.
and
_,

"Prices and Terms of

Trade for Developed CountryExportsof Goods," in B. Csikos-Nagy Manufactured et al., eds., The Economics of Relative Prices, New York: Saint Martin's Press, 1984. Krugman, Paul, "Pricingto Marketwhen the ExchangeRate Changes,"NBER Working Paper No. 1926, May 1986. N. "Small Menu Costs and Mankiw, Gregory, Large Business Cycles: A Macroeconomic Journal of Model of Monopoly,"Quarterly Economics,May 1985, 100, 529-37. Marginsand "Prices, Profit Mann,Catherine, Exchange Rates." Federal ReserveBulletin,June1986, 72, 366-79. Seade, Jesus,"Prices, Profitsand Taxes in of Paper, University Oligopoly,"Working Warwick,1983. Salop, Steven, "Monopolistic Competition of withOutside Goods," Bell Journal Economics,Spring1979, 10, 141-56.

Potrebbero piacerti anche