Financial Constraints and Exchange Rate Flexibility. in Emerging Market Economies * Michael B. Devereux

Transcription

1 Financial Consrains and Exchange Rae Flexibiliy in Emerging Marke Economies * Michael B. Devereux Hong Kong Insiue for Moneary Research Universiy of Briish Columbia CEPR 26 July 2001 JEL Classificaion F40 F41 Absrac Empirical evidence from he Asian financial crisis of suggess ha exchange rae depreciaion may have had a conracionary effec on he raded good secor of he wors-hi economies. Many wriers have suggesed ha his was caused by exchange rae sensiive credi consrains affecing he producion secor. This paper documens some of his evidence, and uses i o develop a srucural model ha feaures an imporan role for credi consrains in he financing of raded goods producion. We show ha his model can explain why emerging marke governmens would be more concerned wih variaions in exchange raes han would be implied by sandard `opimal currency area crieria. Moreover, he model implies ha moneary policy may be a very ineffecive ool in emerging marke economies. * I hank SSHRC for research funding, and Desmond Hou for excellen research assisance. 1

2 Secion 1. Inroducion Since he financial crises of he lae 1990 s, here has been an acive debae abou he appropriae exchange rae policies for emerging marke economies 1. While here is some degree of consensus ha inermediae exchange rae regimes such as sof pegs or arge zones are no longer feasible, being oo vulnerable o speculaive pressures, here is far less agreemen on wheher emerging marke economies should move owards freely floaing exchange raes (for insance, wihin a moneary regime governed by an inflaion argeing objecive), or adop a hard peg, currency board, or fully dollarize. This paper make a conribuion o his debae by firs providing some evidence on he response of he Eas Asian economies o he crisis of , and hen by developing a model ha is consisen wih his evidence. The model can be used o evaluae he appropriae moneary policy sance for emerging marke economies ha are vulnerable o rapid movemens in flows of capial. The essenial message of he paper is ha he sandard opimal currency area (OCA) analysis of exchange rae regimes, used heavily in he debae on he coss and benefis of he European single currency, for insance, is no a fully appropriae framework for evaluaing exchange rae regimes in emerging markes. In order o adequaely undersand he coss and benefis of exchange rae movemens for emerging markes, we have o acknowledge he imporance of financial marke consrains. In paricular, in he presence of disorions or consrains in financial markes, (discussed, for insance, by Krugman 1999, Aghion, Bachea, and Banerjee, 2000), exchange rae adjusmen will no longer necessarily play a sabilizing role in he economy, as in OCA analysis, bu may in generae endogenous supply shocks by ighening consrains in financial markes. 1 Among many oher papers, see Change and Velasco (2000), Calvo and Reinhar (2000), Fischer (2001) Frankel (1999), Schmi and Uribe (2000), Mendoza (2000). 2

3 When we develop his feaure in an analyical framework we find ha i is always desirable o penalize exchange rae volailiy more heavily han would be he case if we based our judgemen purely on OCA ype consideraions, and in fac, a hard exchange rae peg migh dominae a policy of pure floaing exchange raes. There is an exensive lieraure based on he noion ha exchange rae movemens have differen effecs in emerging markes han in more developed economies (see Agenor and Moniel 1998, for references). Recen analyses of he Asian crisis have added o his lieraure by emphasizing credi consrains and financial marke disorions in emerging markes ha may depend on he exchange rae (Krugman 1999, Chang and Velasco 2000, Aghion Bachea and Banerjee 2000). Our model can be seen as a simple applicaion of his lieraure o he discussion over he appropriae exchange rae policy for emerging markes. A key macroeconomic feaure of recen emerging marke experience has been he insabiliy of capial flows. Sachs (1998), Calvo (1999), and McKinnon and Pill (1998) provide a macroeconomic inerpreaion of capial marke reversals in small open economies. They emphasize he join `ransfer-problem aspec of capial marke reversals; a sudden reversal of capial flows requires a shif of resources from he domesic economy o he res of he world, represening an exernal ransfer, and also a shif of resources from he shelered non-radables secor, o he radables or expor secor, represening an inernal ransfer. In combinaion wih a sharp real exchange rae depreciaion, we would anicipae ha a capial marke reversal would cause a slump in he non-radables secor and a boom in he radables secor. Bu he evidence shown of Secion 2 below reveals ha for mos Eas Asian economies, he crisis was associaed wih a very sharp decline in oupu of boh radable and nonradable goods. To he exen he response of radable goods producion in he crisis 3

4 moved he `wrong way, a leas for a considerable ime, his gives prima facie evidence of he role of financial marke consrains in producion ha are sensiive o exchange rae movemens. We hen develop a model of nominal and real exchange rae deerminaion which emphasizes he imporance of financial marke consrains. The key consrain ha we idenify is in he purchase of impored inermediae inpus. Much anecdoal evidence a he ime of he crisis suggesed ha expor firms could no ake advanage of he improved compeiiveness associaed wih exchange rae depreciaion because of difficulies in obaining credi for impored raw maerials purchases. In our model, imporing inermediae firms are subjec o a credi consrain, which limis he purchases of inermediae impors o be no greaer han heir iniial nominal asses. This is a varian of he collaeral consrains inroduced by Aghion e al. ( 2000). We show ha a capial marke reversal may lead o a nominal exchange rae depreciaion of such a magniude ha his consrain becomes binding. In ha case, he fall in he availabiliy of inermediae impors may generae a negaive `supply shock in he raded goods secor. This can accoun for a fall in raded goods oupu following a capial marke crisis. We go on o show ha his financial consrain alers he balance of he argumen beween `fixed and floaing exchange raes oward greaer exchange rae sabiliy. While he effec of financial consrains migh no offer a case for fully fixed exchange raes, i will always have he effec of reducing he desired volailiy of he exchange rae ha a policy maker would choose. Moreover, in ligh of he difficuly of achieving `inermediae regimes, i may il he argumen in favour of full dollarizaion for some emerging marke economies. Moreover, i provides a raionale for he empirical observaion, made by Levy Yeyai and Surzenegger 4

5 (1999), and Calvo and Reinhar (2000), ha many emerging marke economies display a `fear of floaing. The paper is organized as follows. The nex secion gives some empirical evidence from Asian counries. Secion 3 develops a model of real and nominal exchange rae deerminaion. Secion 4 illusraes he resuls of he model. Secion 2: Evidence from he Asian Crisis The sandard model of a capial marke shock, as discussed in he las secion, implies ha a sudden reversal of capial inflows o an emerging marke economy should be followed by a combinaion of boh an exernal and inernal ransfer (see Krugman 1999, and McKinnon and Pill 2000). The exernal ransfer occurs because he sudden reversal of capial inflows mus involve a ne ransfer of funds from he economy o he res of he world. This requires he economy o generae a curren accoun surplus, eiher by reducing absorpion, increasing GDP, or boh. The inernal ransfer is necessary because in almos all episodes of capial inflows, much invesmen goes ino he non-radable secor (e.g. consrucion). Bu on ne, his invesmen mus be repaid hrough rade surpluses. As a resul, a capial marke reversal requires a shif of resources ou of he non-radable goods secor owards he radable goods secor, which occurs simulaneously wih a real exchange rae depreciaion. These predicions follow almos direcly from he macroeconomic accouning of open economies subjec o capial marke shocks. They are no sensiive o he paricular heory of crises ha is used for analysis (see e.g. Tornell and Schneider 2000, Aghion, Bachea, and Banerjee, 2000). Figures 1-3 documen he experience of aggregae GDP, he real exchange rae, and he curren accoun for he eigh major Eas Asian economies. In erms of he response o he crisis, he paern is essenially he same for all 5

6 counries. There was a subsanial real depreciaion, large increase in he curren accoun balance, and a fall in GDP. In paricular, he real depreciaion and curren accoun urnaround is seen in all economies. In erms of GDP, he wors hi economies were Indonesia, Korea, Thailand, and Malaysia. Afer his, Hong Kong experienced a sharp fall in oupu, whereas Singapore, Taiwan and he Philippines had only a relaively mild a slowdown in GDP (especially in Taiwan), bu also experienced a subsanial real depreciaion and curren accoun increase. One predicion of he basic model of capial marke shocks (e.g. Tornell and Schneider 2000), is ha here should be a sharp increase in he size of he radable goods (or expor ) secor. In face of a large real exchange rae depreciaion, domesic expor goods will become more compeiive, eiher hrough a fall in he domesic real wage, a fall in he world relaive price, or a combinaion of he wo. Bu he evidence from he Asian crisis does no suppor his paern of response in he raded goods secor. Figure 4 documens he breakdown of secoral oupu dynamics for each counry ino radable and non-radable secors (Hong Kong figures were no available) 2. For Thailand, Malaysia, he Philippines, Indonesia, and Korea, here is evidence ha he non-radable goods secor grew faser han he radable good secor in he early 1990 s, as capial inflows were aking place. This is wha would be expeced, as capial inflows are used for invesmen in infrasrucure, ec. However, as regards he componens of GDP, he impac of he crisis differs from he sandard model. Following he capial marke reversal for he same five counries, we see no a decline in non-radables and a rise in radables, bu a decline in boh secors. Thus, while he exernal ransfer ook place very quickly, here is lile immediae evidence 2 In each case, Traded goods are defined as Agriculure, Mining and Manufacuring. Non-raded goods, are Consrucion,, Communicaions and Transpor, and oher Services indusries. 6

7 of he inernal ransfer. The Figures provide no evidence a all ha facors of producion were ransferred from he non-radable secor owards he radable secor. Raher, producion fell dramaically in boh secors, in all counries. A he same ime, he Figures also show ha he fall in raded goods producion was more emporary han ha in he non-radable secor. In Malaysia, Korea, Thailand, and Indonesia, radable producion begins o increase in 1999 and generally grows much faser han non-radable producion. One objecion o he argumen made above is ha he real exchange rae depreciaion was a common phenomenon across he region. There is subsanial inerregional rade in all Eas Asian economies. This means ha he fall in he real exchange raes agains Europe or Norh America did no necessarily improve effecive compeiiveness for a subsanial componen of regional expors. Figure 5 avoids his problem by focusing only on expors o he US, for Malaysia, Thailand, and Korea. Given he large increase in compeiiveness vis a vis he US economy, one would anicipae a subsanial burs of expors following he crisis. For all hree economies, he figures seem o indicae a subsanial lag in US expor growh. Korean expors were essenially fla unil Thailand and Malaysia had relaively small US expor growh during 1998, followed by subsanial growh in The sharp fall in producion of radable goods in face of a rise in compeiiveness, followed by a subsanial growh afer abou a delay of one year, suggess ha radable good secors in hese economies were subjec o producion consrains generaed by he dynamics of he crisis iself. Much anecdoal discussion a he ime poined o he difficuly of obaining rade credi financing for exporers in he crisis-hi counries. The key aspec of producion in many emerging markes is ha a large componen of gross oupu (in he raded goods secor) is accouned for by 7

8 impors of inermediae goods, or parly finished goods, which are hen combined wih local facors of producion o be expored. Given ha inermediae impors are purchased from he non-crisis counries, a sharp real depreciaion raises he domesic cos of purchasing hese goods. Since local labour coss are unlikely o be indexed (a leas fully) o he exchange rae, he depreciaion sill improves he effecive compeiiveness of he expor secor. Bu wih weak domesic financial insiuions, he higher cos of inermediae impors may increase he difficuly of obaining rade credi. In paricular, rade credi may be resriced by he domesic balance shee posiion. A real exchange rae depreciaion is likely o make rade credi harder o ge in hese circumsances. In he model below, we develop his idea in a simple manner. Secion 3. The srucural model We ouline a model of a small economy which has wo secors; radable and non-radable. I is small in he sense ha i akes boh world ineres raes and is domesic erms of rade as given. In he non-radable secor, here are nominal rigidiies; prices of non-radable goods are se in advance. Household uiliy depends upon composie good ha combines radable and non-radable goods. 3.1 Household preferences and consrains The represenaive agen ges uiliy from consumpion C, and disuiliy from ime spen working in he radables and non-radables, T H and NT H. Work in each secor is perfecly subsiuable, so ha oal work ime is H = H + H. We may NT T wrie he iner-emporal uiliy funcion as: E β ( U( C ) V( H )) 0 = 0 8

9 where U is increasing and concave, and V is increasing and convex. Aggregae consumpion is a linear homogenous funcion of consumpion of radable and nonradable goods: φ 1 φ 1 1 φ φ = + NT φ T ( ) (1 ) ( ) C a C a C. Where φ is he elasiciy of subsiuion beween radable and non-radable goods in preferences. The non-radable good is in urn defined over he consumpion of a coninuum of differeniaed goods, so ha ρ ρ () 1 NT C = x i di, wih ρ > 1. 0 I herefore follows ha he consumer price index and he price index for non-radable goods are wrien respecively as: NT T ( ) (1 ) ( ) P = a P + a P ρ NT 1 ρ () 0 NT P = P i di where NT P is he common price for all radable goods, and P ( i ) is he price of ype T i non-radable good. Each household faces he choice of purchasing curren consumpion goods, eiher radable or non-radable, working, accumulaing domesic or foreign nominal bonds and allocaing financing o inermediae imporing firms. The household earns income from wages, profis from he non-radable goods firms, dividend paymens from he inermediae impored firm (see below),and reurns on domesic and foreign bond holdings. The household s budge consrain is herefore wrien as: (1) where PC + B + S B = W H +Π + (1 + i ) B + (1 + i ) S B + P d T, * * * T I 1 * B ( B ) represens he holding of domesic (foreign) bonds, i ( i * ) is he nominal ineres rae on domesic (foreign) bonds, W is he nominal wage, Π is 9

10 profi income, S is he nominal exchange rae, I d is dividend income received from inermediae imporing firms, and T is a nominal governmen ax. 3.2 Household opimal choices The household s opimal choice of bond holdings and labor supply resuls in he following firs order condiions: * S 1 P + U'( C) = Eβ 1 + i+ 1 U'( C+ 1) S P+ 1 (2) ( ) P U'( C) = Eβ 1 + i+ 1 U'( C+ 1) P + 1 (3) ( ) W (4) U'( C) V '( H) P =. Finally, he individual demands for non-radable and radable goods are given as: (5) φ T NT T P NT P () i C = (1 a) C, C ( i) = a C. P P φ 3.3 Firms Tradable goods firms funcion: A perfecly compeiive indusry produces radable goods using he producion T V I (6) Y = Min[, ] ω 1 ω, Here ω is he share of value added ( V ) in producion, and 1 ω is he share of impored inermediaes ( I ). The assumpion ha value added and inermediae impors are combined in fixed proporions seems a reasonable one, especially in he shor erm, where echnological srucure is fixed. Value added is defined by he producion funcion: 10

11 T T V = F ( H ) T where F (.) is increasing and concave. Profis for he non-raded goods firm are given by Π = PY W H Q I, where Q is he price of he inermediae impor. T T The raded good firm will choose employmen and inermediae inpus o maximize profis. The following condiions follow from profi maximizaion: (7) T T F'( H ) ( P (1 ω) Q) = W ω (1 ω) (8) I = V. ω Inermediae imporers Compeiive firms purchase inermediae impors from abroad, and sell hem o domesic raded goods firms. These inermediae imporers are owned by domesic households. The flow budge consrain of a ypical imporer is I T I I I B + 1 P d =Π + + (1 + i ) B, where Π = QI SP I represens he profi earned by he inermediae imporer, I * I and I B represens bond holdings of he imporer. We assume ha purchases of inermediae impored goods are subjec o he credi consrain given by: (9) SPI B. * I I Thus, inermediae impors mus no be greaer han he iniial bond holdings of he imporing firms. This consrain is no derived endogenously, bu simply assumed as par of he operaing environmen of he inermediae imporer. Bu i capures he noion ha movemens in exchange raes may impose raioning on he purchases of inermediae goods. If he consrain is no binding, hen imporers simply sell inermediaes o raded goods producers a cos and hus we have Q = S P. If he * I 11

12 I B consrain is binding, hen I =, and he domesic price of inermediae impors * SP I paid by raded good firms may be higher. The problem of he imporer is o choose inermediae impors o maximize is profis, ensuring ha he consrain (9) is saisfied. Our analysis below looks a he case where here is an unanicipaed shock in he firs period, which leads he consrain o bind in ha period. Afer his he consrain will no longer bind 3. Non raded goods firms The non-radable firm i has he increasing and concave producion funcion given by: NT N N Y () i = F ( H ()) i. In he non-radable secor, each producion firm has marke power, and ses he price as a markup over marginal cos. If non-radable goods prices were perfecly flexible, hen he profi maximizing decision for firm i would imply: NT N NT ρ (10) P () i F '( H ()) i = W, ρ 1 where ρ ρ 1 represens he monopoly markup. We will also look a he impac of foreign ineres rae and moneary policy shocks under he assumpion ha he nonradable goods price canno adjus in he period of he shock. For his experimen, we make he assumpion ha he non-radable price is se a he value peraining o he iniial seady sae. 3.4 Moneary and fiscal policy We follow he recen lieraure (Woodford (2001), Gali e al (2000)) in absracing from he deails of he moneary mechanism, and simply assume ha he 3 This follows because in a perfec foresigh equilibrium households will lend o he inermediae imporer so ha (9) will never bind. In a more general sochasic model i is possible ha credi consrains would be occasionally binding over ime. 12

13 moneary auhoriy follows a domesic ineres rae argeing rule. We may define he domesic nominal ineres rae implicily by he condiion (1 i ) U'( C ) P β + + =. U'( C) P+ 1 Then he moneary auhoriy is assumed o follow he rule: 1 S (11) (1 + i+ 1) = exp( v), ω > 0. β S0 ω The parameer ω represens he coefficien of exchange rae inervenion. So long as ω>0, here is a deerminae equilibrium value for he nominal exchange rae. The higher, is ω, he closer he moneary rule approximaes a pegged exchange rae, where he arge for he exchange rae peg is S 0. The variable v represens an independen moneary policy shock ha could be hough of as a discreionary money expansion. One aspec of his srucural model is ha here monopoly power in he nonradable good secor. This means ha here is a welfare loss due o monopoly pricing of non-radable goods in a seady sae oupu and employmen of non-radable goods will be oo small. We assume ha he fiscal auhoriy offers an employmen subsidy o eliminae his inefficiency. This subsidy is financed wih lump-sum axes. In addiion, because we are no concerned wih governmen deficis, we assume ha he sock of governmen bonds ousanding is mainained consan, wih ineres paymens on hese bonds being also financed by lump-sum axes Equilibrium Given he equaions of household opimaliy, profi maximizaion, saisfacion of he moneary auhoriies budge consrain, and marke clearing condiions, we may consruc an equilibrium of his economy. Since non-radable goods mus be boh produced and consumed only in he domesic economy, he marke clearing condiion 13

14 in he non-radable secor is wrien as (since all non-radable firms are alike, we drop he firm specific subscrip hereafer): (11) Y NT = C. NT From he combinaion of (1) and (11), his implies ha he economies exernal balance of paymens relaionship is (13) P B = 1 (1 + i ) B + + ( C Y ). T * * * T T S Finally, we assume ha he radable goods price obeys he `law of one price condiion (14) P T = S, where he foreign currency price of he radable good is normalized o uniy. Secion 4. Soluion and calibraion of he model 4.1 Seady sae equilibrium We focus on a perfec foresigh equilibrium where agens have full knowledge of he pah of he money supply and foreign ineres raes from dae zero. In addiion, assume ha he foreign ineres rae is consan, and equal o he domesic rae of ime preference, so ha * β (1 + i ) = 1. Then if he economy sars ou wih a zero ne foreign asse posiion, i is easy o esablish ha a seady sae equilibrium is aained a dae zero, where he curren accoun is zero, and consumpion and oupu are consan over ime. This equilibrium may be described by he following condiions, NT T N which give he four equilibrium values of H, H, p, C N NT NT p (15) F ( H ) = a C N Γ( p ) φ 14

15 (16) (17) (18) (19) T T F ( H ) 1 (1 (1 ω) q) = (1 a) C N ω Γ( p ) (1 ω) T T I = F ( H ) ω ρ F ρ 1 T T '( H ) (1 (1 ) ) N N '( NT ω q = p F H ) ω T ' T F ( H ) NT T U'( C) = V '( H + H ) N Γ( p ) where 1 N(1 φ ) 1 φ Γ ( p N ) = ap + (1 a), p N NT P Q =, and q S =. From he soluions T P given by his sysem, we may recover all oher variables. 4.2 Moneary policy The equilibrium depends on he sance of moneary policy. If all prices are perfecly flexible, hen moneary policy deermines only he level of domesic prices and he domesic nominal ineres rae. If, however, he non-raded goods prices is pre-se, hen he moneary policy rule is imporan for he pah of real variables in he economy, following an unanicipaed shock. We look a wo aspecs of moneary policy. Firs, when he economy is subjec o an ineres rae or capial marke shock (as described below), he degree o which moneary policy responds o he exchange rae is a criical facor in he impacs of his shock. This feaure of moneary policy is governed by he inervenion parameer ω. Bu we also allow for he possibiliy of independen, discreionary moneary policy shock, capured by he parameer v Ineres rae shocks We wish o compare he effecs of differen exchange rae rules on he response of he economy o exernal shocks. In paricular we wan o compare he rade-off beween he need for exchange rae flexibiliy o deal wih he macro effec 15

16 of exernal shocks agains he cos of exchange rae variaion coming from he `balance shee impac on he inermediae imporers. To illusrae he rade-off we focus on he economy s response o foreign ineres rae shocks. The foreign ineres rae deermines he cos of borrowing for a small economy, and simulaneously deermines he scale of capial inflows or ouflows. A capial marke shock of he ype seen in recen emerging marke crises may be hough of as a sudden and unanicipaed rise in he ineres rae. We can hink of his shock as being an exogenous rise in he res of world risk premium for he foreign counry (McKibbon 2000), or more srucurally, i can be hough of as he resul of a collapse in confidence in an emerging marke ha precipiaes a self-fulfilling `bank-run ype ouflow of capial. In eiher case, he proximae effec is for he economy o face a sudden rise in he cos of borrowing. To model his as simply as possible, we conduc he following experimen. Assume ha he iniial foreign ineres rae is such ha he curren accoun is in balance if he level of ne foreign asses is zero; i.e. * β (1 + i ) = 1. Bu in he firs period, here may be a posiive or negaive shock o he foreign ineres * * rae; β (1 + i ) > 1 ( β (1 + i ) < 1 ). Following his, however, he foreign ineres rae 1 1 reurns back o he domesic rae of ime preference, so ha β + = for all > 1. * (1 i ) 1 Assuming ha he economy sars ou in he seady sae posiion described in he previous paragraph, hen his perurbaion in he foreign ineres rae will generae a period zero curren accoun surplus (defici) if he foreign ineres rae rises above (falls below) he domesic rae of ime preference. Following his however, in period 1 16

17 and for all fuure periods, he curren accoun will be zero, and consumpion, labor supply, and he relaive price of non-radable goods will be consan Calibraion I is no possible o obain an analyical soluion for his model, due o he naure of he imporing firms credi consrain, and o he presence of curren accoun dynamics. Insead, we calibrae he model, and obain he soluion numerically. We make he following assumpions wih respec o funcional forms. Le he uiliy funcion be: (1 σ ) C η H 1 σ 1+ ψ 1+ ψ. The parameer σ represens he iner-emporal elasiciy of subsiuion, and ψ represens he inverse of he elasiciy of labour supply. In addiion, he producion funcions for raded and non-raded goods respecively are: T T NT γ NT NT NT α F = A ( H ), F = A ( H ) α < 1, γ < 1. Table 1 repors he baseline calibraion assumpions. We follow he open economy macro lieraure in picking parameer values. The iner-emporal elasiciy of subsiuion is se a 2, following Backus, Kydland and Kehoe (1995). The rae of ime preference is se a 0.05, so he subjecive discoun facor is Iniially, he foreign ineres rae is hen se a 0.05 also. The value of η is unimporan, jus deermining scale, so we se i arbirarily o uniy. The share of non-raded goods in he consumer price index is se a 0.5, following he evidence cied in Schmi and Uribe (2000) for Mexico and Cook and Devereux (2001) for Malaysia and Thailand. The elasiciy of labour supply is se o 0.5, so ha ψ = 2. This is roughly in he 4 Alhough his is quie a simple approach o modeling a capial marke shock, i has some degree of generaliy. An self-fulfilling panic of he ype described by Chang and Velasco (2000) can be hough of as equivalen o a emporary rise in he exernal real ineres rae (Cook and Devereux 2001). 17

18 middle of he various esimaes of labor supply in he lieraure, based on micro evidence and aggregae macro daa. The elasiciy of subsiuion beween non-raded and raded goods is se a 1.5. This accords wih he assumpions made in Backus Kydland and Kehoe (1995) over he elasiciy of subsiuion beween home and foreign goods. The elasiciy of subsiuion beween varieies of non-raded goods is ρ, and his governs he equilibrium markup of price over cos in he non-raded good secor. We assume ha his markup is 20 percen. This is slighly higher han he common value of 10 percen (Basu and Fernald 1999) used for macro sudies of he indusrial economies, bu i is likely ha markups are higher in emerging markes. We assume ha non-raded goods producion is relaively labour inensive, wih α = 0.7, and raded goods is relaively non-labour inensive, wih γ = 0.3. Evidence of his from Eas Asian daa is presened in Cook and Devereux (2001). We also need o choose he iniial sock of asses held by inermediae imporers. We will firs describe he resuls when his consrain never binds. Then, relaively arbirarily, we will assume ha iniial asses are 2 percen higher han he seady sae inermediae impor requiremen. This means, for insance, ha any shock which generaes a greaer han wo percen nominal exchange rae depreciaion will, ceeris paribus, lead he credi consrain (9) o bind. Finally, we vary he ineres rae adjusmen parameer ω, beween 0.01, which represens a case where he moneary auhoriy is conen o allow significan adjusmen in he nominal exchange rae, and 9000, represening a case where he moneary auhoriy essenially keeps he exchange rae pegged. In addiion, in he analysis below we examine an opimal moneary rule ha lies beween hese wo values. 18

19 Table 1: Baseline Calibraion Variable Value Variable Value σ 2 φ 1.5 β ρ 6 η 1 α 0.7 a 0.5 γ 0.3 ψ 2 ω 0.01, 9000 I D * 1.02SPI I 4.4 Capial Marke Shocks We now examine he impac of changes in he res of he world ineres rae (as defined in secion (3.2) above) on he curren accoun, he real exchange rae, oupu and welfare in he domesic economy. As noed above, he saring poin is assumed o be he economy in iniial seady sae wih zero ne deb, defined by he sysem (15)-(19). Given he calibraion se ou in Table 1, a value of he world ineres rae of 0.05 produces his seady sae. We define he firs period curren accoun as * B 1, he real exchange rae as P P NT 1 1 (he relaive price of non-raded goods in erms of he domesic CPI), and oupu is defined as he CPI deflaed sum of producion in he non-raded and raded goods secor: p F ( H ) + F ( H ). N NT NT T T N Γ( p1 ) We examine he impac of he world ineres rae shock in wo cases; firs in he case where he credi consrains are non-binding, and hen he case where he credi 19

20 consrains do bind for posiive shocks o he ineres rae. In each case, we compare he impac of he shock under flexible and fixed exchange raes. 4.5 Ineres rae shocks wihou financing consrains Figure 6 illusraes he impac of variaions in he ineres rae on he period 1 levels of GDP, raded secor oupu, he real exchange rae, and overall lifeime welfare. This Figure perains compares he case ω = 0.5 (flexible exchange raes) wih ω = 9000 so he exchange rae is held fixed. The solid lines represens he flexible exchange case, while he doed lines represen he fixed exchange rae case. Figure 1a shows ha beginning a curren accoun balance, a emporary ineres rae increase will raise GDP in boh he fixed and flexible exchange rae economy. The rise in he ineres rae will cause a fall in curren consumpion and a rise in fuure consumpion, causing a rise in curren labour supply and a rise in oupu. Oupu will rise by more under he flexible exchange rae case however, because under flexible exchange raes he ineres rae shock can generae a emporary real exchange rae depreciaion. Under a flexible exchange rae he real depreciaion is faciliaed by a nominal depreciaion, while under a fixed exchange rae, he real exchange rae is unaffeced by he ineres rae shock, in he iniial ime period. Tha depreciaion under flexible exchange raes simulaneously increases demand in he non-raded goods secor, as well as cushioning he overall fall in curren consumpion, since he effecive real ineres rae rises by less han he rise in he foreign ineres rae. Traded good oupu acually rises by more under fixed exchange raes, because he greaer fall in curren consumpion generaes a higher shif up in labour supply, and a greaer fall in he real wage. Figure 6d illusraes he welfare impac of he ineres rae shock. Welfare will rise under eiher exchange rae regime, wheher he ineres rae rises or falls, simply 20

21 because of he gains from inernaional borrowing or lending, relaive o auarky. This resul is sensiive o he assumpion of zero iniial ne foreign asses. If he economy was in an iniial ne debor posiion, hen a rise in he foreign ineres rae would have end o reduce welfare direcly hrough negaive wealh effecs. Bu despie his, i would sill be he case ha welfare under flexible exchange raes is higher han under fixed exchange raes, even if overall welfare falls in each case. Flexible exchange raes sricly welfare dominae. Welfare is higher everywhere wih flexible exchange raes excep in he special case where here is no borrowing or lending. Essenially, his illusraes he indicaes he sandard OCA benefis of a flexible exchange rae in he presence of asymmeric shocks, since a movemen in world ineres raes may be hough of as reflecing he presence of asymmeric macro shocks. In face of exernal shocks ha require real exchange rae adjusmen, i is beer o allow he nominal exchange rae o faciliae his adjusmen. 4.6 Ineres rae shocks wih financing consrains someimes binding Now le us look a he case where financing consrains someimes bind, and his depends on he movemen of he exchange rae. Specifically, we ake a case where he finance consrain binds under flexible exchange raes if he ineres rae shock is posiive and high enough, bu he consrain never binds under fixed exchange raes. Graned, his is a special case, and i is easy o calibrae so he consrain never binds (for he range of ineres rae shocks we look a in he Figures) in eiher case, or where he consrain someimes binds in boh fixed and flexible exchange rae regimes. Bu wha is rue is ha he finance consrain will always be more binding under flexible exchange raes, due o he impac of nominal depreciaion on he consrain. 21

22 In Figure 7, he finance consrain binds under flexible exchange raes when he foreign ineres rae is higher han 6 percen. The Figure illusraes ha he finance consrain has dramaic effecs on he effecs of posiive ineres rae shocks under flexible exchange raes (negaive ineres rae shocks work in he same way as before, since hey do no make he finance consrain more binding). A posiive ineres rae shock will lead o nominal exchange rae depreciaion. Wihou finance consrains, his would lead o a srong rise in GDP, more han ha seen under fixed exchange raes. Bu when he finance consrain binds, an ineres rae shock ha is high enough can generae a fall in GDP, as he nominal depreciaion leads o a raioning of inermediae impors, and a fall in raded goods oupu. In effec, he nominal exchange rae depreciaion arising from he ineres rae increase has a conracionary effec on he economy. Now we find ha GDP would be greaer even under fixed exchange raes, when he finance consrain is no binding, han under he flexible exchange rae. (The behaviour of he real exchange rae is idenical wheher he finance consrain is binding or no, and is omied from Figure 7). The presence of finance consrains also has a direc effec on he welfare comparison of fixed and flexible exchange raes. Figure 7 shows ha lifeime uiliy as a funcion of he period 1 ineres rae. We see ha when he finance consrain does no bind, uiliy is everywhere higher under he flexible exchange rae case. Bu as he ineres rae reaches he hreshold where he finance consrain binds, welfare is higher under a fixed exchange rae. How would we compare overall welfare beween fixed and flexible exchange raes? A simple approach is o assume ha he disribuion of foreign ineres raes is such ha each ineres rae beween 0.01 and 0.09 is equally likely. In his case we may compue expeced uiliy by inegraing across he welfare loci in Figures 6 and 7, 22

23 weighing by probabiliies. A compuaion for he calibraion under sudy reveals he following. As is clear, in he unconsrained environmen, a flexible exchange rae moneary rule will dominae a fixed exchange rae rule. Bu wih he finance consrain binding in he manner described in Figure 7, we find ha a fixed exchange rae dominaes. In his sense, we may conclude ha he presence of financial marke disorions may generae a preference for greaer nominal exchange rae sabiliy. While we have only described he rade-off beween a fixed exchange rae rule and a rule in which he moneary auhoriy pus almos no weigh on he exchange rae, we could also sudy inermediae rules. I will be he case ha a rule which pus more weigh on he exchange rae does beer han eiher a fixed exchange rae or a flexible exchange rae rule as described above. Bu he essenial message is unchanged. The presence of exchange rae-sensiive financing consrains in rade increases he desirabiliy of exchange rae sabiliy. 4.7 Perverse effecs of moneary policy shocks The presence of credi consrains in he raded goods secor alers he calculaion of he ne benefis of exchange rae flexibiliy. The exchange rae no longer necessarily acs as a mechanism for sabilizing GDP, bu may be de-sabilizing, when financing consrains are sensiive o exchange rae movemens. There is an immediae corollary of his conclusion ha perains o he effeciveness of moneary policy in an emerging marke economy operaing under his ype of credi consrain. Moneary policy works in his economy only by influencing he nominal and real exchange rae. For insance, in Chang and Velasco (2000), moneary expansion can reduce real ineres raes by generaing a curren real exchange rae depreciaion, simulaing aggregae demand and oupu. Bu when an exchange rae depreciaion pushes he economy owards a binding credi consrains, moneary policy expansion 23

24 may have indirec negaive effecs on he oupu of raded goods. In fac, moneay policy may be conracionary for he overall economy. Figure 8 illusraes he impac of an unanicipaed moneary policy expansion (a rise in v ) in he model, boh wih and wihou binding finance consrains. In he absence of he finance consrain, moneary policy works in a familiar way. The moneary expansion generaes a nominal depreciaion, and a rise in curren consumpion. Demand and oupu rise in he non-radable secor, and overall GDP expands in he firs period. When he exchange rae effecs of moneary policy expansion lead he credi consrains o bind, however, Figure 7 shows ha he ne negaive effec on he producion of radable goods may be large enough o offse he increased demand in he non-radable secor. Overall, GDP may fall. This example provides an illusraion of he viewpoin of Calvo and Reinhar (2000), who emphasize he limiaions on he use of moneary policy for macroeconomic purposes in emerging marke economies. I should be emphasized ha his perverse effec of moneary policy is no relaed o he sandard criicisms of independen moneary policy in emerging markes. We have purposely absraced from he convenional problems relaed o poliical consrains or he credibiliy of moneary policy. In his economy, he usefulness of moneary policy is consrained purely by financial marke disorions. Secion 5. Conclusions This paper uses empirical evidence from he response of some Asian economies o he financial crisis of o suppor a heoreical model ha feaures an imporan role for exchange rae sensiive financial marke disorions. We show ha his model can explain why emerging marke counries would be more 24

25 concerned wih variaions in exchange raes han would be implied by sandard OCA crieria. Moreover, he model implies ha moneary policy may be a very ineffecive ool in emerging marke economies. All his suggess ha a currency board or an explici dollarizaion of he moneary sysem may be less cosly for emerging markes han suggesed by convenional analysis. Neverheless, he issue is more complicaed han ha. As emphasized by Eichengreen and Hausmann (1999) and Krugman (1998), emerging markes and ` maure developed economies are reaed quie differenly by inernaional capial markes. Maure developed economies do no face consrains on issuing deb in heir own currencies, and are no forced o follow pro-cyclical policies in response o inernaional crises. From he perspecive of he policy-maker, i would clearly be more desirable o gain saus as a `maure developed economy, han o face he consrains imposed on emerging markes, whaever he form of he exchange rae sysem used in he laer. If an emerging marke economy could cross his `hreshold in saus, i migh wish hen o allow more exchange rae flexibiliy. 25

26 References Agenor Pierre Richard and Peer Moniel 1998 Developmen Macroeconomics, Princeon Universiy Press Aghion, Philippe, Philippe Bacchea, and Abhiji Banerjee 2000 Currency Crises and Moneary Policy in an Economy wih Credi Consrains, mimeo. Backus, David, Finn E. Kydland, and Parick J. Kehoe (1992) Inernaional Real Business Cycles, Journal of Poliical Economy Bernanke, Ben, Mark Gerler and Simon Gilchris (1999), "The Financial Acceleraor in a Quaniaive Business Cycle Model", in John Taylor and Michael Woodford, eds, Handbook of Macroeconomics, Volume 1c. Norh Holland. Calvo, Guillermo (1999a), "On Dollarizaion", mimeo, Universiy of Maryland. Calvo, Guillermo (1999b), "Fixed versus Flexible Exchange Raes", mimeo, Universiy of Maryland. Calvo, Guillermo and Carmen Reinhar (2000) ``Fear of Floaing, NBER d.p Cook, David, and Michael B. Devereux (2000) The Macroeconomics of Capial Marke Crises mimeo, HKUST. Caramazza and Aziz, 1998 Fixed versus flexible: geing he exchange rae righ in he 1990 s, IMF Economic issues, no. 13. Cespedes, Luis Felipe, Robero. Chang, and Andres Velasco (2000) Balance Shees and Exchange Rae Policy, mimeo, Federal Reserve Bank of Alana Chang, Robero and Andres Velasco (1998), "Financial Fragiliy and he Exchange Rae Regime", NBER Working Paper No Chang, Robero and Andres Velasco (1999), "Liquidiy Crises in Emerging Markes: Theory and Policy", Economic Research Repor RR#99-14, CV Sarr Cener for Applied Economics, New York Universiy, June. Cook, David (1999) Moneary Policy in Emerging Markes Hong Kong Universiy of Science and Technology De Gregorio, José, Albero Giovannini and Holger Wolf (1994), Inernaional Evidence on Tradables and Nonradables Inflaion, European Economic Review 38, Eichengreen, Barry and Ricardo Hausmann (1999), "Exchange Raes and Financial Fragiliy", NBER w.p Fischer, Sanley (2001) Exchange Rae Regimes: Is he Bi-polar view correc? Finance and Developmen,

27 Frankel, Jeffrey (1999), "No Single Currency Regime is Righ for All Counries or a All Times", NBER Working Paper #7338. Gray, F. Dale and Mark R. Sone (1999) Corporae Balance Shees and Macroeconomic Policy, Finance and Developmen Krugman, Paul (1998) The Confidence Game The New Republic. Krugman, Paul (1999), "Balance Shees, The Transfer Problem and Financial Crises," Inernaional Tax and Public Finance, November. Levy Yeyai, and Surzenegger (1999) Classifying Exchange Rae Regimes, Words Versus Deeds, mimeo, Universidad Di Tella. McKinnon, Ronald and Huw Pill (1998) Inernaional Overborrowing: A Decomposiion of Credi and Currency Risks, w.p , Sanford Universiy. Mendoza, Enrique (2000) The Benefis of Dollarizaion When Sabilizaion Lacks Credibiliy and Financial Markes are Imperfec, mimeo, Duke Universiy. Mishkin Frederic and Miguel A. Savasano (2000) Moneary Policy Sraegies for Lain American NBER w.p Sachs, Jeffrey (1998) ``The Causes of he Asian Crisis, mimeo. Schmi Sefanie and Marin Uribe (2000) Sabilizaion Policy and he Coss of Dollarizaion, mimeo Rugers Universiy. Schneider Marin and Aron Tornell (2000) Balance Shee Effecs, Bailou Guaranees, and Financial Crises, NBER d.p Woodford, Michael (2000) Ineres and Prices mimeo. 27

28 Figure 1: GDP Malaysia Thailand Feb-96 Nov-98 Korea Singapore Taiwan Indonesia Feb-96 Nov-98 28

29 Figure 1: GDP Philippines Hong Kong GDP

30 Figure 2: Real Effecive Exchange Raes Malaysia: REER Thailand REER Jul Korea REER Singapore REER Taiwan: REER Indonesia REER

31 Figure 2: Real Effecive Exchange Raes Philippines REER Hong Kong REER

32 Figure 3: Curren Accoun o GDP raio Malaysia CA/GDP Thailand CA/GDP Feb Jun-97 Nov-98 Mar Feb Jun-97 Nov-98 Mar Korea CA/GDP Singapore CA/GDP Feb Jun-97 Nov-98 Mar-00 Taiwan CA/GDP Indonesia CA/GDP

33 Figure 3: Curren Accoun o GDP raio Philippines CA/GDP Hong Kong CA/GDP

34 Figure 4: Tradable and Non-Tradable Secor Oupu Malaysia Thailand NT T Feb-96 Nov-98 Korea Singapore Taiwan Indonesia Feb-96 Nov-98 34

35 Figure 4: Tradable and Non-Tradable Secor Oupu Philippines

36 Figure 5: Expors o US Malaysia, Expors o US Jan-93 Jun-94 Oc-95 Mar-97 Jul-98 Dec-99 Apr-01 Thailand Expors o US Jan-93 Jun-94 Oc-95 Mar-97 Jul-98 Dec-99 Apr-01 Korea Expors o US May- 90 Sep-91 Jan-93 Jun-94 Oc-95 Mar-97 Jul-98 Dec-99 Apr-01 36

37 Figure 6: Capial Marke Shocks wihou Finance Consrains Figure 6a: GDP Figure 6b: Real Exchange Rae Fixed Flexible ROW ineres rae ROW ineres rae Figure 6c: Traded Goods Oupu Figure 6d: Welfare ROW ineres rae ROW ineres rae 37

38 Figure 7: Capial Marke Shocks wih Finance Consrains Figure 7a: GDP Figure 7b: Traded Goods Oupu Fixed Flexible ROW ineres rae ROW ineres rae Figure 7c: Welfare ROW ineres rae Figure 8: GDP, Expansionary Moneary Shock No Credi Consrains Credi Consrains