Exchange Rates and Monetary Policy. in Emerging Market Economies * Michael B. Devereux. University of British Columbia and CEPR. Philip R.

Transcription

1 Preliminary Exchange Raes and Moneary Policy in Emerging Marke Economies * Michael B. Devereux Universiy of Briish Columbia and CEPR Philip R. Lane Triniy College Dublin and CEPR This draf: June 2001 Absrac This paper invesigaes he effecs of exchange rae regimes and alernaive moneary policy rules for an emerging marke economy ha is subjec o a volaile exernal environmen in he form of shocks o world ineres raes and he erms of rade. In paricular, we highligh he impac of financial fricions and he degree of exchange rae pass hrough in deermining he relaive performance of alernaive regimes in sabilizing he economy in he face of exernal shocks. Our resuls are quie sharp. When exchange rae pass-hrough is high, a policy of non-raded goods inflaion argeing does bes in sabilizing he economy, and is beer in welfare erms. When exchange rae pass-hrough is low, however, a policy of sric CPI inflaion argeing is beer. In all cases, a fixed exchange rae is undesirable. In addiion, financial fricions have no implicaions for he ranking of alernaive policy rules. Keywords: moneary policy, exchange raes, emerging markes. JEL Codes: E50, F02, F41. * We hank seminar paricipans a he Hong Kong Insiue for Moneary Research, he Bank of England, Universia Pompeu Fabra and Universiy College London. We are graeful o Mahias Hoffman for research assisance and he Social Science Research Council of he Royal Irish Academy for financial suppor. This work is par of a research nework on The Analysis of Inernaional Capial Markes: Undersanding Europe s Role in he Global Economy, funded by he European Commission under he Research Training Nework Programme (Conrac No. HPRN CT ). Devereux hanks SSHRC for financial suppor and Lane also graefully acknowledges he suppor of a TCD Berkeley Fellowship.

2 Secion 1. Inroducion Since he crises of he lae 1990s, here has been remendous ineres in he design of appropriae moneary policies for emerging marke economies. Should hese economies aemp o peg heir exchange raes o he US dollar via currency boards or dollarizaion, or should hey allow he exchange raes o floa and follow insead a domesically oriened moneary policy geared owards inflaion conrol, similar o he inflaion argeing ha has been successfully applied in many wesern economies in he pas decade? This paper develops a simple modelling framework in order o evaluae alernaive moneary policy rules for emerging markes, and in paricular we ask how imporan are movemens in exchange rae raes for implemening such rules. The model is specialised owards he emerging marke environmen in wo ways. Firs, he emerging marke economy may have specific srucural characerisics ha make i more vulnerable o exernal shocks. Two such feaures are consrains on he financing of invesmen hrough exernal borrowing, and he speed by which exchange rae shocks feed hrough o he domesic price level. The second way in which he model is geared owards emerging markes is in our measuremen of shocks. We calibrae our model wih he observed shocks o ineres raes and exernal erms of rade for Asia. The cenral quesion addressed is he appropriae moneary policy for an emerging marke, given hese srucural characerisics and he paern of exernal shocks. Much of he lieraure on emerging marke crises has focused on inconsisencies in policy making, and problems of susainabiliy of moneary and fiscal policy. By conras, our paper does no invesigae he credibiliy of moneary policies, or he ineracion beween poliical consrains and macroeconomic policies. 2

3 Raher, we assume ha all moneary policies are equally credible, and simply invesigae he properies of alernaive rules in erms of economic sabilisaion and welfare. The presence of financial marke imperfecions in capial inflows o emerging markes has received widespread aenion in he las few years. An imporan heme in his lieraure is he moral hazard problems wih financing invesmen in emerging markes, where conracs may be less enforceable han in Wesern economies. Accordingly, we explore he role of collaeral consrains in invesmen financing for emerging markes, following he work of Bernanke, Gerler, and Gilchris (1999) [hereafer BGG]. In paricular, as emphasized by Krugman (1999), Aghion, Bacchea and Banerjee (2001) and ohers, emerging marke borrowers may find ha ineres rae and exchange rae flucuaions have large effecs on heir real ne worh posiion, and so, hrough balance shee consrains affecing invesmen spending, have much more serious macroeconomic consequences han for richer indusrial economies. Our ineres is in how hese feaures affec he choice of moneary rules. For insance, i is suggesed by Eichengreen and Hausmann (1999) and Calvo (1999) ha emerging marke economies may be much more relucan o allow freely floaing exchange raes due o he problem of liabiliy dollarizaion in he presence of balance shee consrains on exernal borrowing. A second imporan feaure of emerging markes is he degree o which heir price levels are sensiive o flucuaions in exchange raes. As emphasised by Calvo and Reinhar (2001), exchange rae shocks in emerging marke economies end o feed ino aggregae inflaion a a much faser rae han in indusrial economies. While his difference may be due o hisorical feaures relaed o he conduc of moneary policy, 3

4 we simply focus on wheher and how his difference affecs he choice of moneary policy. We compare hree differen ypes of moneary rules. While a fixed exchange rae is a well-defined rule for a small economy, here is an infinie variey of differen ypes of floaing exchange raes. We resric our aenion o wo imporan rules: a policy of CPI inflaion argeing, and a policy of argeing inflaion in a subse of he CPI consising of non-raded goods prices. As we will see, he laer is a robus and aracive moneary rule in a wide variey of circumsances. While we focus on wo differen ypes of shocks ha hi emerging markes --- ineres rae shocks and erms of rade shocks --- i urns ou ha our resuls regarding opimal moneary rules do no really depend on he source of shocks. In addiion, echoing Cespedes, Chang and Velasco (2001a, 2001b) and Gerler, Gilchris and Naalucci (2001) in quie differen seings, we find ha exernal financing consrains have no implicaions for he ranking of moneary rules. 1 While balance shee consrains in he presence of liabiliy dollarisaion is an imporan propagaion channel, i essenially generaes a magnificaion effec in response o all shocks, wihou alering he ranking of alernaive moneary policy rules in welfare erms. Our resuls show ha, in he presence of high exchange rae pass hrough, he bes policy for an emerging marke economy is o sricly arge inflaion in nonraded goods prices alone. This ends o sabilise oupu in response o exernal shocks and also maximises welfare, when compared o he alernaive rules. Eiher an exchange rae peg or a policy of sric CPI inflaion argeing, by conras, desabilises oupu, and is welfare inferior. Neverheless, here is an elemen of rade-off involved in using a non-raded goods price inflaion arge. The oupu sabilisaion can be 4

5 achieved only by very high volailiy in he nominal and real exchange rae, and consequenly, high inflaion volailiy. If a policy maker were more concerned wih inflaion han our represenaive individual welfare index suggess she should be (perhaps for reasons of credibiliy along he lines of Rogoff 1985), hen his migh reduce he araciveness of he non-raded goods price inflaion argeing rule. When exchange rae pass-hrough is low however, we find ha he moneary policy choice is alered considerably. While an exchange rae peg is sill quie undesirable, a policy of CPI inflaion argeing is now much more aracive in his siuaion. The reason is ha he policy maker can simulaneously arge inflaion sricly, bu sill allow high nominal exchange rae volailiy in order o sabilise he real economy in face of exernal shocks. The low rae of pass hrough ensures ha exchange rae shocks do no desabilise he price level. When pass hrough is very low, he exchange rae no longer acs as an expendiure-swiching device, alering he relaive price of home and foreign goods, bu is sill of criical imporance in sabilising demand by cushioning he effecive real ineres rae faced by consumers and firms in he emerging marke. The imporan feaure of low pass hrough is ha i eliminaes he rade-off beween oupu volailiy and inflaion volailiy in he open economy. For insance, a policy of non-raded goods inflaion argeing does beer han a pegged exchange rae on boh couns; oupu volailiy and inflaion volailiy under his rule are lower han hey are under a pegged exchange rae. 1 The former se of auhors derive some analyical resuls in a highly-sylised sripped-down model. The laer is closer o his paper in building a quaniaive model. However, neiher allow for nonradables or incomplee pass hrough or consider he same se of moneary policy rules. 5

6 Lowering he rae of pass hrough also alers he welfare rankings of alernaive rules. 2 Wih rapid pass-hrough, he non-raded goods inflaion argeing rule is welfare dominan, wih or wihou exernal financing consrains. Bu wih delayed pass hrough he CPI inflaion argeing rule is preferable. Inuiively, he CPI rule does a beer job of sabilising boh secors in he presence of nominal sluggishness in boh he non-raded and raded goods secor. These resuls sugges ha a low rae of exchange rae pass-hrough may be an imporan prerequisie for he success of inflaion argeing in emerging markes. The paper is organised as follows. Secion 2 ses ou he model. Secion 3 discusses calibraion and he soluion of he model. Secion 4 develops he main resuls. Some conclusions follow. Secion 2. Moneary Policy in a Small Open Economy Ouline of he model The srucure is a sandard wo-secor dependen economy model. Two goods are produced: a domesic non-raded good, and an expor good, he price of which is fixed on world markes. Four cenral aspecs of he model are a) he exisence of nominal rigidiies; b) he requiremen ha all foreign liabiliies be denominaed in foreign currency; c) he presence of lending consrains on invesmen financing; and d), he degree of exchange rae pass hrough in impor prices. The firs feaure is of course necessary o moivae a role for he exchange rae regime a all. The specific assumpion made is ha he prices of non-raded goods are se by individual firms, and adjus only over ime. The specificaion of price seing follows Calvo (1983) and Yun (1996). 2 Monacelli (1999) also examines he implicaions of low pass hrough for alernaive moneary policy 6

7 The second feaure of he model is based on he observaion by Eichengreen and Hausmann (1999), among ohers, ha emerging marke economies have lile abiliy o issue exernal deb denominaed in local currency. They also noe ha almos all he exernal deb issued by Eas Asian counries during he period of rapid inflows in he early 1990s was denominaed in US dollars. Wih respec o he borrowing consrain on invesmen, we follow he BGG model, which assumes ha enrepreneurs underake invesmen projecs wih reurns ha can be observed by lenders only a a cos. This leads enrepreneurs o face higher coss of exernal financing of invesmen relaive o inernal financing, and as a resul invesmen depends on enrepreneurial ne worh. Finally, i is well esablished from Engel (1999) ha deviaions from he law of one price are a major facor in deermining real exchange raes. Accordingly, we consider alernaive speeds of adjusmen of impor prices o exchange rae movemens. There are four ses of domesic acors in he model: consumers, producion firms, enrepreneurs, and he moneary auhoriy. In addiion, here is a res of world secor where foreign-currency prices of expor and impor goods are se, and where lending raes are deermined. 3 Consumers We assume ha he economy is populaed by a coninuum of consumer/households of measure uniy. We will describe he model in erms of he represenaive consumer. She has preferences given by M (1) U = E0 β u( C,, ) 0 H =, P rules bu his model only conains radables and here are no financial fricions. 7

8 where C is a composie consumpion index, H is labor supply, and M P represens real balances, wih M being nominal money balances, and P being he consumer price index. Le he funcional form of u be given by u 1 ε 1+ ψ 1 χ M H C η 1 σ 1 ε P 1+ ψ 1 σ = + Composie consumpion is a CES funcion of consumpion of non-raded goods and an impor good, where ρ ρ ρ ρ ρ ρ 1 N M C = ( a C + (1 a) C ), ρ > 0. The implied consumer price index is hen 1 1 ρ 1 ρ 1 ρ N M P = ( ap + (1 a) P ). Since we wish o inroduce nominal price seing in he non-raded goods secor, we need o allow for imperfec compeiion in ha secor. In order o do his, we assume ha he consumpion of non-radable goods is differeniaed as follows: N 1 ( ) 1/(1 λ ) 1 λ () 0 N C = C i di, where λ > 1. We assume ha consumers do no face any capial marke imperfecions. Therefore, he consumer can borrow direcly in erms of foreign currency a a given ineres rae i *. The assumpion of a fricionless consumer credi marke is an exreme one. Bu he criical aspec of he financial fricions in he model below revolves around heir impac on invesmen financing. Thus, he resuls of he model wih respec o he properies of alernaive moneary policies would be essenially unaffeced by he presence of credi consrains a he consumer level. 3 Since our focus is on emerging marke economies ha are plausibly price-akers on world markes, we assume ha he foreign-currency prices of expors and impors are exogenously deermined. 8

9 A consumer s revenue flow in any period comes from her supply of hours of work o firms for wages non raded secor (see below) W, ransfers T from governmen, profis from firms in he Π, domesic money M, less her deb repaymen from las period * SD (1 + i ), where S is he nominal exchange rae and D is he ousanding amoun of foreign-currency deb. 4 She hen obains new loans from he inernaional capial marke, and uses hese o consume and acquire new money balances. Her budge consrain is hus (2) PC = W L +Π + S D + M M + i S D + T * (1 ) The household will choose non-raded and raded goods o minimize expendiure condiional on oal composie demand goods is hen C. Demand for non-raded and impored P = ρ N CN a C P ρ P M CM = (1 a) C P The consumer opimum can be characerized by he following condiions. 1 σ S+ 1 P (3) C * = Eβ C (1 + i ) S P σ + 1 W (4) P = ηc H σ ψ ε M σ h (5) = χc (1 Ed + 1) P Equaion (3) represens he Euler equaion for opimal consumpion. Equaion (4) is he labour supply equaion, while equaion (5) gives he implici money demand funcion. Money demand depends on domesic nominal ineres raes. The domesic nominal discoun facor is defined as 4 Noe ha consumers do no receive any capial income, as all invesmen in his economy is done by 9

10 σ h P σ + 1 P C + 1 (6) d+ 1 β 1+ i C where i + 1 is he domesic nominal ineres rae. The combinaion of (3) and (6) gives he represenaion of uncovered ineres rae pariy for his model. Producion Firms Producion is carried ou by firms in each secor. Secors differ in heir producion echnologies. Boh ypes of goods are produced by combining labour and capial. Following BGG (1999), labour comes from boh consumer/households and from enrepreneurs. Thus, in he non-raded secor, effecive labour of firm i is defined as L = H H, Ω e(1 Ω) Ni Ni Ni where H Ni is employmen of household labour and e H Ni is employmen of enrepreneurial labour. The overall producion echnology for a firm in he non-raded goods secor is hen (7) Y = A K L, α 1 α Ni N Ni Ni where A N is a produciviy parameer. Similarly, exporers (all domesically-produced radables are expored) use he producion funcion (8) Y = A K L γ 1 γ X X X X Firms in each secor hire labour and capial from consumers and enrepreneurs, and sell heir oupu o consumers, enrepreneurs (for heir consumpion) and capial producing firms. Cos minimizing behaviour hen implies he following equaions YNi (9) W = MCN(1 α) Ω H Ni enrepreneurs. 10

11 e Y (10) W = MC (1 α)(1 Ω ) H Ni N N e Ni YNi (11) RN = MCNα K YX (12) W = PX(1 γ ) Ω H e Y (13) W = P (1 γ )(1 Ω ) H N X X X X e X YX (14) RX = PXγ, K where X MC N denoes he marginal producion cos for a firm in he non-raded secor (which is common across firms). Equaions (9), (10), (12) and (13) describe he opimal employmen choice for firms in each secor. I is assumed ha enrepreneurial labour supply o each secor is inelasic in supply and fixed across secors. Thus, he enrepreneurial wage differs across secors. Equaions (11) and (14) describe he opimal choice of capial. Noe ha he price of he raded expor good is P X. Movemens in his price, relaive o he impor price P M, represen erms of rade flucuaions for he small economy. Producion of capial goods is also carried ou by compeiive firms. These firms combine impors and non-raded goods o produce unfinished capial goods. There are adjusmen coss of invesmen, so ha he marginal reurn o invesmen in erms of capial goods is declining in he amoun of invesmen underaken, relaive o he curren capial sock. Capial socks in he expor and non-raded secors evolve according o I X (15) K + 1 = φ( ) K + (1 δ) K K X X X X 11

12 I N (16) K + 1 = φ( ) K + (1 δ) K K N N N N where he funcion φ saisfies φ '> 0, and φ '' < 0. This reflecs he presence of adjusmen coss of invesmen. Invesmen in new capial requires impors and non-raded goods in he same mix as he household s consumpion baske. Thus, he price of a uni of invesmen, in eiher secor, is P, he price of one uni of he consumpion index. Compeiive capial producing firms will hen ensure ha he price of capial sold o enrepreneurs is (17) (18) Q Q X N 1 = P I X φ '( ) K X 1 = P I N φ '( ) K N Price seing Firms in he non-raded secor se prices in advance. Following he mehod of Calvo (1983) and Yun (1996), assume ha firms face a probabiliy ( 1 κ ) in every period of alering heir price, independen of how long heir price has been fixed. Following sandard aggregaion resuls, he non-raded goods price follows he parial adjusmen rule (19) P = κ P + κp 1 λ ~ 1 λ 1 λ N ( 1 ) N N 1 where P ~ N represens he newly se price for a firm ha does adjus is price a ime. The evoluion of P ~ N is hen governed by (20) P! N = (1 βκβκ ) MCN + E P! N

13 Taking a linear approximaion of (19) and (20), assuming an iniial seady sae where he rae of change of P N is consan, we can derive he familiar forward-looking inflaion equaion: (21) πn = λmcn + Eπ N + 1 where MC N mcn represens he log deviaion of real marginal cos in he non-raded secor, / P from is seady sae level (of uniy). Equaion (21) is analogous o he N forward-looking inflaion equaion in Clarida, Gali and Gerler (1999). The key difference here is ha boh marginal coss and inflaion are specific o he nonradable secor. Local Currency Pricing We assume ha he law of one price mus hold for expor goods, so ha (22) P = S P. * X X For impor goods however, we allow for he possibiliy ha here is some delay beween movemens in he exchange rae and he adjusmen of impored goods prices. Wihou loss of generaliy, we may assume ha impored goods prices are adjused in he same manner as prices in he non-raded secor. Tha is, a measure 1 κ * of foreign firms adjus heir prices in every period. Thus, he impored good price index for domesic consumers moves as (23) P = (1 κ*) P! + κ * P 1 λ 1 λ 1 λ M M M 1 where P! M represens he newly se price for a foreign firm ha does adjus is price a ime. The evoluion of P! M is hen governed by (24) P! = βκβκ S P + E P! +. * M (1 *) M * M 1 13

14 The inerpreaion of (24) is ha he foreign firm wishes o achieve an idenical price in he home marke as in he world marke. Bu i may incur a lag in adjusing is price. The coefficien κ * deermines he delay in he pass-hrough of exchange raes o prices in he domesic marke. Using he same approach as wih equaions (20) and (21), we can derive he familiar inflaion equaion: (25) * ˆ ˆ ˆ M = ( s + pm pm ) + E M 1 π λ π + where π M is he domesic-currency inflaion rae for he impored good, and s ˆ and * p ˆ m represen he log deviaion of he exchange rae and he world price for he impor good from seady sae. Enrepreneurs Unfinished capial is ransformed by enrepreneurs and sold o he final goods secor. Following BGG, he broad noion is ha capial producers face idiosyncraic risk arising from unverifiable individual invesmen oucomes. The effecive resul of his is o make borrowing more cosly for enrepreneurs han financing invesmen ou of inernal resources. In borrowing from capial markes, enrepreneurs face an exernal risk premium. Here we incorporae a simplified version of he BGG se-up which, in he aggregae, preserves he essenial feaures of he credi channel effecs for invesmen dynamics. There are wo groups of enrepreneurs. One group provides capial o he non-raded secor, while he oher provides capial o he raded secor. 5 Enrepreneurs mus borrow in foreign currency. We se his as a consrain on he ypes of borrowing conracs raher han deriving i endogenously. An 5 Since capial is quasi-specific o each secor, separaing he provision of capial o he differen secors is he simples way o handle he enrepreneurial capial supply decision. 14

15 j enrepreneur j in he non-raded secor who wishes o inves K N + 1 unis of capial j mus pay nominal price KN + 1Q. Say ha he enrepreneur begins wih nominal ne j worh given by Z N + 1. Then she mus borrow in foreign currency an amoun given by ej 1 (26) 1 ( j j DN+ = KN+ 1Q ZN+ 1 ) S The fac ha an enrepreneur faces a cos of exernal capial ha is inversely relaed o her ne worh is capured by he assumpion ha, when borrowing, an enrepreneur in eiher secor is faced wih a risk premium given by Ψ ( z), where z is he raio of ne worh o he value of capial. We assume ha where Ψ (1) = 1, and Ψ '(.) < 0. Thus, a higher ne worh reduces he exernal finance risk premium. The cos of invesmen funds for an enrepreneur is increasing he amoun borrowed. If ne worh is large enough so ha here is no requiremen o borrow, hen he cos of borrowed funds equals he world opporuniy cos (1 + ). Bu he larger * i + 1 is he amoun borrowed relaive o ne worh, he higher he probabiliy of defaul, and he higher is he cos of funds. By his mechanism, a shock o ne worh, generaed say by a nominal exchange rae depreciaion, may direcly reduce invesmen by increasing he cos of capial. A risk-neural enrepreneur ha faces such a risk premium will choose a rae of invesmen such ha he expeced reurn, evaluaed in erms of foreign currency, equals he cos of capial. Thus, for he non-raded secor, we have (27) j j S Z N+ 1 * E( RKN+ 1 ) =Ψ (1 i 1) j + +. S+ 1 KN+ 1Q A he beginning of each period, an enrepreneur in he non-raded secor receives he amoun given by 15

16 R Q K i Z S D j j * N KN 1 N (1 + ) Ψ j Q 1KN N This represens he nominal reurn on capial purchased las period, less he ineres coss on deb incurred las period, evaluaed a he curren exchange rae. The enrepreneurs are assumed o die a any ime period wih probabiliy (1 ϑ). They consume only in he period in which hey die. Thus, a any given period, a fracion (1 ϑ) of enrepreneurial wealh is consumed. I is shown in BGG ha he funcional forms used here allow for aggregaion, so ha he mean capial sock in each secor is deermined by equaion (27) as a funcion of he mean raio of ne worh o he value of capial in ha secor. Furhermore, he mean value of ne worh of enrepreneurs in he non-raded secor evolves as (28) Z = ϑ R Q K + i Ψ z S D + W * e N + 1 ( KN 1 N (1 ) ( ) N ) N For an enrepreneur in he non-raded secor, ne worh is deermined by he unconsumed reurns o invesmen plus he wages earned working in he non-raded secor. Noe ha ne worh depends negaively on he curren exchange rae. The deails of he conrac srucure and ne worh dynamics in he expor secor are described in he idenical way. Finally, we may define he reurn o capial o enrepreneurs as depending on boh he price of capial and he renal rae offered by firms. Thus RN + 1+ (1 δ ) QN + 1 (29) RKN+ 1 = Q RX + 1+ (1 δ ) QX + 1 (30) RKX+ 1 = Q X N Moneary Policy Rules Assume ha he moneary auhoriy uses a shor-erm ineres rae as he moneary insrumen. Given he ineres rae, he money supply will be deermined 16

17 endogenously by equaion (5), he aggregae demand for money arising from he consumer secor. I is imporan however ha ineres rae rules are se so as o ensure a unique price level and exchange rae, and o avoid he issue of real indeerminacy ha can arise under some ineres rae rules in sicky price models. 6 Under all calibraions of he model, as discussed below, a unique equilibrium is obained. (31) ( i ) The general form of he ineres rae rule used may be wrien as + 1 µ π µ n π µ s PN 1 P 1 S PN 11+ πn P 11+ π S 1 + = (1 + i ) where i is assumed ha µ π 0, µ 0, 0, 0. n π µ y µ s The parameer µ π allows he n moneary auhoriy o conrol he inflaion rae in he non-raded goods secor around a arge rae of π 7 n. The parameer µ π governs he degree o which he CPI inflaion rae is argeed around he desired arge of π. Finally, µ s conrols he degree o which ineres raes aemp o conrol variaions in he exchange rae, around a arge level of S. Equilibrium In each period, he non-raded goods marke mus clear. Thus, we have P P ρ N Ne Te (32) YN = a ( C + IN + IT + C + C ) Equaion (32) indicaes ha demand for non-raded goods comes from household consumpion, invesmen and he consumpion of enrepreneurs. In he calibraion of. 6 See Woodford (1999) for condiions on ineres rae rules required for uniqueness in he price level. In addiion, see Clarida, Gali, and Gerler (1999). 7 When pass-hrough is full, a arge of π n = 0 will acually replicae he response of he economy wihou nominal rigidiies of any kind. 17

18 he model, he las caegory is assumed o be very small relaive o he size of he economy. The evoluion of he household secor s ne deb is given by MC SD+ = + i SD + PC WL P Y * N 1 (1 ) N N(1 ) PN The expression on he far righ hand side gives he value of profis in he non-raded good secor ha is earned by he households ownership of monopolisic compeiive firms. We may add his o he budge consrain of enrepreneurs o obain he curren accoun equaion for he overall economy as (33) SD SD = isd P Y A A * A + 1 X X ρ P M Ne Te + (1 a) ( C + IN + IT + C + C ) P Labour marke clearing for he household secor implies (34) HN + HX = H. Finally, he price of he non-raded good and he impor good may be recovered from he equaions (35) PN = (1 + π N ) PN 1 (36) PM = (1 + π M ) PM 1 18

19 Secion 3. Soluion and Calibraion We now derive a soluion for he model, by firs calibraing and hen simulaing using sandard linear approximaion echniques. The calibraion of he model is somewha more involved han he usual dynamic general equilibrium framework, since he model has wo producion secors and i involves parameers describing he enrepreneurial secor. The benchmark parameer choices for he model are described in Table 1. Some sandard parameer values are hose governing preferences. I is assumed ha he ineremporal elasiciy of subsiuion in boh consumpion and real balances is The consumpion ineremporal elasiciy is wihin he range of he lieraure, and he equaliy beween he wo elasiciies ensures ha he consumpion elasiciy of money demand equals uniy, as esimaed by Mankiw and Summers (1986). The elasiciy of subsiuion beween non-raded and impored goods in consumpion is an imporan parameer, on which here is lile direc evidence. Following Sockman and Tesar (1995), we se his o uniy. The elasiciy of labour supply is also se o uniy, following Chrisiano, Eichenbaum, and Evans (1997). In addiion, he elasiciy of subsiuion beween varieies of raded goods deermines he average price-cos markup in he non-raded secor. We follow sandard esimaes from he lieraure in seing a 10 percen mark-up, so ha λ = 11. Assuming ha he small economy sars ou in a seady sae wih zero consumpion growh, he world ineres rae mus equal he rae of ime preference. We se he world ineres rae equal o 6 percen annually, an approximae number used in he macro-rbc lieraure, so ha a he quarerly level, his implies a value of for he discoun facor β. 19

20 The facor inensiy parameers are quie imporan in deermining he dynamics of he model. In he shor run, only labour is mobile beween secors, so he impac of ineres rae and erms of rade shocks on oupu will depend on he labour inensiy of he differen secors. For wo Asian economies, Malaysia and Thailand, we found ha here is clear evidence ha he non-raded secor is more labour inensive han he raded secor 8. Boh counries esimaes of secoral wage shares are quie similar. Following hese esimaes, we se oal share of labour in GDP o 52 percen, he labour share of raded goods (i.e. expor) oupu o 30 percen, and he share of wages in non-raded oupu o 70 percen. In combinaion wih he oher parameers of he model, he parameer a, governing he share of non-raded goods in he CPI, deermines he share of nonraded goods in GDP. Following he classificaion followed by De Gregorio e al. (1994), we found ha he average share of non-raded goods in oal GDP in Thailand was 54 percen over he period Cook and Devereux (2001) find a similar figure for Malaysia. Given he oher parameers, a value of a equal o 0.55 produces his share. To deermine he degree of nominal rigidiy in he model, he value κ, governing he speed of price adjusmen in non-raded goods, mus be chosen. Again, in he absence of direc evidence on his, we follow he lieraure (e.g. Chari, Kehoe and McGraen 1998), and se κ =0.75, so ha prices compleely adjus afer approximaely four quarers. Likewise, wih no direc evidence on he speed of adjusmen in capial, we follow BGG in seing he φ funcion such ha he elasiciy of Tobin s q wih respec o he invesmen capial raio is For Malaysia, evidence is presened in Cook and Devereux (2001). For Thailand, an earlier draf of his paper provides esimaes: see Devereux and Lane (2000). 20

21 κ* We consider wo values for he impor pass-hrough coefficien: κ * = 0 and = κ. The former represens he complee pass-hrough case; he laer implies he same degree of price sickiness in non-raded and raded secors. Finally, we choose a seady sae risk spread of 200 basis poins, as in BGG, and an elasiciy of he exernal finance premium wih respec o he ne worh o capial raio equal o 0.025, and a savings rae of enrepreneurs of 0.9. Shocks We consider wo ypes of exernal shock: a) shocks o he world ineres rae, and b) erms of rade shocks. In he model, a) is represened by shocks o i * + 1, and b) is represened by shocks o P / P. * * X M Secion 4: Exernal shocks under alernaive moneary rules Figures 1-6 illusrae he impac of shocks o he counries borrowing rae and shocks o he erms of rade, under he hree alernaive moneary rules. In order o illusrae he workings of he model, a his sage, we simply assume ha each shock is an AR(1) process wih persisence 0.5. The Figures show alernaively how he collaeral consrains and he speed of exchange rae pass-hrough affecs he ransmission of shocks o he economy. Ineres Rae Shocks Figures 1-3 illusrae he effec of a persisen shock o he world ineres rae. Figures 1 and 2 show he impac of he shock wihou and wih he presence of financing consrains respecively, under complee pass-hrough in impor prices. 9 Traded goods were defined as Agriculure, Mining and Quarrying, and Manufacuring. Non-raded goods were Elecriciy Gas and Waer, Consrucion, Hoel and Reail Trade, Transpor Sorage and Communicaions, Educaion, and oher service secors. 21

22 The unanicipaed rise in he cos of exernal borrowing leads firs o a fall in oal absorpion; boh privae consumpion and invesmen fall. The fall in absorpion causes a fall in demand for non-raded goods, leading o a real depreciaion. Nonraded goods oupu falls, while oupu in he expor secor will rise, and he economy experiences a sharp increase in he rade surplus. In principle, he impac of he ineres rae spike on oupu is ambiguous, since oal oupu combines boh nonraded goods oupu and expor secor oupu. As Figure 1 shows, he oupu impac of he ineres rae shock depends sharply on he moneary rule. The non-raded inflaion argeing rule rule involves an expansionary moneary policy, since he fall in demand ends o generae a deflaion in he non-raded goods secor, so ha in order o preven non-raded goods prices from falling, moneary policy mus be expansionary. The moneary expansion in effec sabilises aggregae oupu. Figure 1 shows ha employmen acually expands 10. Noe also however ha he non-raded inflaion argeing rule requires a very large nominal exchange rae depreciaion, followed by an appreciaion. Due o insananeous exchange rae pass-hrough, his means a large iniial burs of inflaion. The mechanism by which his sabilises GDP is seen in Figure 1. The sharp, bu emporary rise in he nominal exchange rae leads o cushioning of he nominal and real ineres rae from he full effecs of he rise in foreign borrowing. The domesic real ineres rae rises by only abou half of he rise in he foreign ineres rae. This cushions he impac of he shock on absorpion, demand, and aggregae GDP. Under he oher wo policy rules, however, he ineres rae shock ends o be sharply conracionary. Moreover, he exchange rae peg and he inflaion arge have almos idenical implicaions; o all inens and purposes hey become he same policy 10 Remember ha he non-raded secor is labour inensive, so an expansion in overall employmen is 22

23 rule in his case. Boh rules mus ac so as o preven a nominal exchange rae depreciaion, and preven any increase in aggregae inflaion. While he non-raded inflaion arge rule is expansionary, he exchange rae peg and he inflaion argeing rule have a conracionary sance, because he ineres rae shock generaes pressure for a nominal depreciaion and a jump in CPI inflaion. By sabilising he nominal exchange rae and inflaion, he wo rules ensure ha he full impac of he foreign real ineres rae shock is passed hrough o he domesic economy, and here is a much larger fall in absorpion, oupu in he non-raded secor, and overall GDP. Now we see ha oal employmen falls. On he oher hand, he rade surplus is larger, because oal absorpion is less. These resuls indicae ha a foreign ineres rae shock is cushioned much more effecively by a non-raded inflaion argeing rule han a policy of sric inflaion argeing or a pegged exchange rae. How do he presence of collaeral consrains in invesmen affec his conclusion? Figure 2 illusraes he impac of he same foreign ineres rae shock when invesmen in each secor is subjec o financing consrains, and exernal deb is foreign-currency denominaed. The key effec of he financing consrains is o increase he downward shif of invesmen, and so overall absorpion, o he ineres rae shock. This occurs because he higher borrowing coss reduce he value of exising capial for enrepreneurs in each secor, and also because he unanicipaed real exchange rae depreciaion raises he deb burden for enrepreneurs. Boh channels reduce ne worh, raising he effecive cos of borrowing, and reducing invesmen by more han we see in he model wihou financing consrains. no inconsisen wih aggregae GDP falling slighly. 23

24 In he aggregae, he impac of he financing consrains is herefore o is o magnify he impac of he ineres rae shock; oupu and employmen fall by more, and he rade balance increases by more, as he greaer fall in absorpion causes a sharper collapse in non-raded oupu. How do he differen moneary policy sances affec he response of he economy o an ineres rae shock in he presence of financing consrains? From he figures, he answer clearly is ha financing consrains have lile or no affec on he policy choice problem. The non-raded inflaion argeing rule sill acs so as o cushion oupu from he ineres shock, alhough oupu falls by more han wihou he financing consrains. Neverheless, he ranking of alernaive policies remains he same as in he economy wihou financing consrains. A second feaure of he resuls wih financing consrains is ha he response of prices is essenially he same, for each ype of policy rule. The rise in he real ineres rae, and he real exchange rae deprecaion, is essenially idenical o he case wihou he consrains. We conclude from his ha he presence of collaeral consrains in exernal invesmen for an emerging marke economy has essenially no implicaions for he choice of moneary policy. As we will confirm below, from a welfare perspecive, he non-raded inflaion argeing rule is sill superior o eiher a nominal exchange rae peg or a sric inflaion arge. 11 The resuls so far are based on he assumpion ha exchange rae pass-hrough o impored goods prices is immediae. How does he presence of delayed passhrough, following he specificaion se ou in secion 2 above, affec he resuls? 11 The resuls remain unchanged when we allow for he possibiliy ha exernal deb is denominaed in domesic currency. When enrepreneurs borrow in domesic currency, one of he channels by which an ineres rae shock impacs on ne worh (hrough he unanicipaed real exchange rae depreciaion 24

25 Figure 3 illusraes he response of he economy o an ineres rae shock under delayed pass-hrough 12. I is imporan o noe ha he response under he exchange rae peg is he same as before. When he exchange rae is pegged, he speed of impor price response o exchange rae shocks is obviously irrelevan. From a qualiaive poin of view, he slower exchange rae pass-hrough does no change he way in which he emerging marke economy responds o shocks. I is sill he case ha absorpion falls, he rade balance improves as resources are shifed ino he expor good secor, aggregae oupu falls, and here is a real exchange rae depreciaion. This indicaes ha closing off he `expendiure-swiching effec, by which exchange rae changes immediaely affec he relaive price of home o foreign goods, does no aler he qualiaive dynamics of he economy. Quaniaively, however, he presence of delayed pass-hrough has very significan affecs on he response o an ineres rae shock. Moreover, i has dramaic implicaions for he comparison of alernaive moneary policy rules. The mos significan feaure of Figure 3, when compared wih Figure 1, is ha here is now a very disinc difference beween he performance of a sric inflaion argeing rule and a pegged exchange rae. When pass-hrough is insananeous, a policy maker canno really sabilise CPI inflaion wihou sabilising he exchange rae. Bu wih delayed pass-hrough, his is quie possible. Under he sric inflaion argeing rule, here is a big iniial depreciaion in he nominal exchange rae, far larger han he exchange rae response when he inflaion argeing rule is applied under full pass-hrough. The resul is ha here is a subsanial real depreciaion under inflaion argeing. Bu a big real depreciaion, by generaing a large expeced appreciaion, allows he policyraising he real deb burden) is removed. Bu since he overall ne worh effecs do no affec he choice of moneary policy, removing his channel canno aler he resuls. 12 In his figure i is assumed ha he collaeral consrains on invesmen financing are absen. The case wih financing consrains has very similar resuls. 25

26 maker o cushion he impac of he shock on he real ineres rae. As a resul, he fall in oal absorpion and GDP, and he rise in he rade balance is much less han in he case of immediae pass-hrough. The absence of pass-hrough herefore raionalises he use of sric inflaion argeing in an emerging marke, a leas for dealing wih shocks o he foreign ineres rae. CPI inflaion argeing becomes much closer o he policy of non-raded inflaion argeing. Non-raded inflaion argeing, as before, acs so as o sabilise oupu, by generaing subsanial movemens in he real exchange rae. Boh policy rules operae by acively employing he nominal exchange rae in order o sabilise he effecive real ineres rae. I is ineresing o noe here ha while he sric `expendiure-swiching mechanism for he exchange rae is grealy diminished, since nominal exchange rae changes no longer aler relaive prices facing consumers and firms, here is sill a criical role played by he exchange rae in conrolling effecive real ineres raes. An alernaive perspecive is o noe ha while he law of one price relaionship no longer holds insananeously, he ineres rae pariy relaionship is sill an imporan macroeconomic linkage. A corollary of hese resuls is ha he inflaion oupu volailiy rade-off is alered by presence he delayed pass-hrough. Wih full pass-hrough, we noed ha he policy of sabilising non-raded goods inflaion cushions he impac of an ineres rae shock on GDP. Bu his can only be done by allowing a large iniial burs of inflaion, following up he exchange rae depreciaion. An exchange rae peg, on he oher hand, sabilises inflaion, bu de-sabilises GDP. Bu Figures 3 now shows us ha boh GDP and inflaion can be subsanially sabilised simulaneously, using eiher a sric inflaion argeing rule, or a non-raded inflaion argeing rule. Indeed, we see from he Figure ha he response of inflaion under an exchange rae peg is 26

27 now, in absolue erms, as grea as ha under he non-raded inflaion arge rule. In he simulaions carried ou below, we will see his poin more generally; delayed passhrough allows he policy maker o sabilise oupu wihou high inflaion variance. Terms of Trade Shocks Figures 4-6 illusrae he effec of persisen negaive shock o he erms of rade. In his ype of model, a erms of rade shock is essenially equivalen o a negaive produciviy shock in he expor secor: he reurn o invesmen declines. This also generaes a negaive wealh effec, leading o a decline in consumpion. Wih Keynesian price rigidiies, he fall in absorpion leads o a fall in oupu in he non-raded secor, a real exchange rae depreciaion, and a fall in GDP. The nonraded inflaion arge generaes a couneracing moneary expansion, which effecively sabilises GDP, bu he exchange rae peg and he sric inflaion arge involve a fall in oupu. Noe ha in general, he magniude of response o a erms of rade shock is smaller han he response o an ineres rae shock. There is only a very small real exchange rae depreciaion and essenially no real ineres rae response o he shock. As was he case for he ineres rae shock, we see ha he inroducion of credi consrains in Figure 5 does no aler he qualiaive paern of responses o a erms of rade shock bu jus acs as an amplificaion device. As before, incomplee pass-hrough in impor prices significanly alers he relaive performance of he alernaive moneary rules: in paricular, inflaion argeing performs much beer in erms of sabilizing oupu. As before, we also observe much larger real exchange rae movemens for he acivis moneary regimes, bu much smaller inflaion volailiy. 27

28 Domesic Moneary Shocks One well-known case where a fixed exchange rae regime may be beneficial is in he presence of domesic moneary shocks. By keeping he exchange rae fixed, he moneary auhoriy prevens hese shocks from affecing aggregae demand and real domesic magniudes. How would he presence of such shocks impac on he comparison beween fixed exchange raes and inflaion argeing in our analysis? The answer is ha hey would no impac on he comparison a all. A domesic moneary shock, for insance a shock o he ineres rae rule in equaion (31) above, would be eliminaed equally under a sric inflaion argeing policy, a nonraded goods inflaion argeing policy, or a fixed exchange rae regime. By following a rule o sricly mainain eiher CPI inflaion argeing or non-raded goods inflaion, he moneary auhoriy would preven he ineres rae shock from having any influence on he domesic economy. 13 Overall Regime Evaluaion We now urn o an evaluaion of he overall performance of alernaive policy regimes in responding o exernal shocks. To obain empirical variances, covariances and auocorrelaions for he shock processes, we ran a quarerly VAR sysem over o for he US real ineres rae and he erms of rade for he Asia region in he IMF s Inernaional Financial Saisics. The resuls are shown in Table 2 and indicae ha here is a low correlaion beween shocks o he real ineres rae and erms of rade. Boh ypes of disurbance have similar variances bu erms of rade shocks end o be more persisen han ineres rae shocks (auoregressive coefficiens of 0.59 and 0.28 respecively). 28

29 Table 3 shows, for each of our four scenarios, he sandard deviaions of key macroeconomic variables when he model is driven by he shock processes esimaed in he VAR exercise. In addiion, Table 3 repors a welfare comparison across regimes, where he loss under a fixed exchange rae is normalized o 100 in each case. The welfare measure is calculaed as a second order approximaion o expeced uiliy, following he procedure used in Woodford (1999). In case I (no credi consrains and full pass hrough), he policy of non-raded goods inflaion argeing delivers much lower oupu volailiy han he oher rules. I is apparen ha he big difference beween he rules lies in he differences in he variabiliy of invesmen. Since he policy ha sabilises non-raded inflaion also ends o sabilise real ineres raes, he volailiy of invesmen is reduced considerably under his policy. However, a he same ime, his policy generaes a much higher volailiy of inflaion, he nominal exchange rae, and he real exchange rae han eiher he price sabiliy rule or he pegged exchange rae rule. In erms of expeced uiliy, he non-raded goods inflaion argeing clearly dominaes boh oher rules. Moreover, he price sabiliy rule is only slighly differen in erms of volailiy of oupu, consumpion invesmen ec, from he pegged exchange rae rule. This is no surprising given he high rae of exchange rae pass-hrough in his case. Comparing cases I and II, he main impac of he inroducion of credi consrains is ha oupu, employmen, and invesmen are significanly more volaile. Bu as we saw in he figures described above, he volailiy of prices is unchanged. Inflaion, he real exchange rae, and he nominal exchange rae have he same volailiy as in case I. A key implicaion of case II also is ha he welfare ranking of alernaive moneary rules is no alered by he inroducion of financing consrains. 13 However, in responding o nominal shocks, he peg ou-performs a pure floa (i.e. a fixed money 29

30 Non-raded goods inflaion argeing is sill clearly he mos preferred moneary rule. Afer ha comes he CPI inflaion argeing, and finally, a pegged exchange rae. Case III illusraes he impac of incomplee pass-hrough. This has a dramaic effec on he workings of he moneary policy rules. Oupu volailiy is lowered significanly for he wo ypes of inflaion argeing rules. Especially in he case of CPI inflaion argeing, oupu volailiy is lowered by 60 percen. Invesmen volailiy is now lowered for boh ypes of inflaion argeing. Employmen volailiy changes in a slighly differen way however. For he CPI inflaion argeing, employmen volailiy falls by abou 50 percen. This is because here are sicky prices in he raded goods secor now, and he CPI inflaion argeing ends o sabilise a combinaion of he oupu in boh raded and he non-raded secor. The non-raded goods inflaion argeing policy however, leads o higher employmen volailiy now ha in he case wih full pass-hrough. This is inuiive, since focusing exclusively on sabilising he mark-up in non-raded goods ends o creae excessive employmen reallocaion across secors. Finally, real and nominal exchange rae volailiy increases quie subsanially when pass-hrough is delayed, for boh ypes of inflaion argeing. In welfare erms, case III shows ha he presence of delayed pass-hrough causes he rankings of policy rules o be reversed. Now he CPI inflaion argeing rule does beer han he non-raded goods inflaion rule. Inuiively, he exclusive focus on condiions in he non-raded secor is no longer an opimal sraegy for he moneary auhoriy, since his leads o excessive movemens in he real exchange rae and in resource reallocaions beween he raded and non-raded secors. sock rule) and more-resriced moneary policy sraegies such as a Taylor rule ha responds only o inflaion and oupu deviaions. 30

31 5 Conclusions This paper has conduced an invesigaion of exchange rae regimes and alernaive moneary policy rules for an emerging marke economy ha is subjec o a volaile exernal environmen in he form of shocks o world ineres raes and he erms of rade, and when he economy is consrained by exernal financing riskpremia associaed wih domesic ne worh. We saw ha he paricular moneary policy rule being followed is as imporan as wheher he exchange rae is fixed or flexible. One key finding is ha degree of pass-hrough in impor prices is cenral in deermining he sabilizaion properies of an inflaion argeing regime. Accordingly, a high prioriy for (heoreical and empirical) research is o undersand he deerminans of he degree of pass hrough. Here, candidae variables include he level of rend inflaion, policy credibiliy, policy uncerainy and he compeiive srucure of goods markes. A second key finding is ha financial disorions amplify exernal shocks bu have lile impac on he ranking of alernaive policy regimes: liabiliy dollarizaion and he inroducion of a financial acceleraor channel does no make a fixed exchange rae a superior regime in erms of macroeconomic sabilizaion. 31

32 References Aghion, Philippe, Philippe Bacchea, and Abhiji Banerjee (2001) Currency Crises and Moneary Policy in an Economy wih Credi Consrains, European Economic Review 47, 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, Amserdam: Norh Holland, Calvo, Guillermo (1983), Saggered Prices in a Uiliy Maximizing Framework, Journal of Moneary Economics 12, Calvo, Guillermo (1999), On Dollarizaion, mimeo, Universiy of Maryland. Calvo, Guillermo and Carmen Reinhar (2001), Fear of Floaing, mimeo, Universiy of Maryland. Cespedes, Luis Felipe, Robero Chang, and Andres Velasco (2001a), Balance Shees and Exchange Rae Policy, mimeo, Harvard Universiy. Cespedes, Luis Felipe, Robero Chang, and Andres Velasco (2001b), Dollarizaion of Liabiliies, Ne Worh Effecs and Opimal Moneary Policy, mimeo, Harvard Universiy. Chang, Robero and Andres Velasco (1999), Liquidiy Crises in Emerging Markes: Theory and Policy, NBER Working Paper No Chari, V.V. Parick J. Kehoe, and Ellen McGraen (1998), Moneary Shocks and Real Exchange Raes in Sicky Price Models of he Inernaional Business Cycle, Federal Reserve Bank of Minneapolis Research Deparmen Saff Repor No. 223, February. Chrisiano, Larry J, Marin Eichenbaum and Charles L.Evans (1997), Sicky Price and Limied Paricipaion Models of Money: A Comparison, European Economic Review 41, Clarida, Richard, Jordi Gali and Mark Gerler (1999), The Science of Moneary Policy: A New Keynesian Perspecive, Journal of Economic Lieraure 37, Cook, David and Michael B. Devereux (2001), The Macroeconomics of Inernaional Financial Panics, mimeo UBC. De Gregorio, José, Albero Giovannini and Holger Wolf (1994), Inernaional Evidence on Tradables and Nonradables Inflaion, European Economic Review 38,