International Competitiveness and Monetary Policy

Transcription

1 Inernaional Compeiiveness and Moneary Policy Paul R. Bergin Deparmen of Economics, Universiy of California a Davis, and NBER Giancarlo Corsei Cambridge Universiy and CEPR Sepember 6, 204 PRELIMINARY. NOT FOR QUOTE. Absrac A classic quesion in open macro concerns he conribuion of moneary and exchange rae policy o a counry s inernaional compeiiveness. We bring a new perspecive on his quesion by developing an open-economy moneary model ha encompasses comparaive advanage. In each counry here are wo secors producing wo ypes of radables. In one secor (manufacuring), firms produce differeniaed goods under monopolisic compeiion subjec o sunk enry coss and nominal rigidiies; in he oher, firms produce non-differeniaed goods under perfec compeiion. We show ha domesic sabilizaion of oupu gap and marginal coss fosers compeiiveness by encouraging invesmen and enry in he differeniaed goods secor. Conversely, he adopion of a peg or moneary regimes implying limied sabilizaion ends o skew he endogenous expor specializaion oward he non-differeniaed goods. Empirical evidence confirms ha moneary regimes such as exchange rae pegs and inflaion argeing influence he composiion of producion and rade consisen wih he heoreical predicion. Keywords: inernaional coordinaion, moneary policy, producion locaion exernaliy, firm enry, opimal ariff JEL classificaion: F4 We hank Yuan Liu, Riccardo Trezzi and Jasmine Xiao for excellen research assisance. We hank our discussan Maeo Cacciaore and Hélène Rey, as well as seminar paricipans a he 203 NBER Summer Insiue, he 203 Inernaional Finance and Macro Finance Workshop a Sciences Po Paris, he 204 ASSA meeings, he Bank of England, he London Business School, Norges Bank Conference The Role of Moneary Policy Revisied, he Naional Universiy of Singapore, and he Universiy of Wisconsin for commens. Paul R. Bergin, Deparmen of Economics, Universiy of California a Davis, One Shields Ave., Davis, CA Phone: (530) prbergin@ucdavis.edu. Giancarlo Corsei, Faculy of Economics, Cambridge Universiy Sidgwick Avenue Cambridge CB3 9DD Unied Kingdom. Phone: +44(0) gc422@cam.ac.uk.

2 . Inroducion This paper brings a new perspecive on how moneary and exchange rae policy may affec economic aciviy and social welfare by supporing a counry s inernaional compeiiveness. The convenional approach in he Keynesian radiion emphasizes he compeiive gains from currency devaluaion, as a way o lower he relaive cos of producion, and herefore he inernaional price of domesic goods, over he span of ime ha prices and wages are sicky in local currency. Conversely, he New Open Economy Macroeconomics (NOEM) or he new-keynesian lieraure emphasize ha moneary policy conribues o welfare by improving a counry s erms of rade. As his ypically means a higher inernaional price of home goods, he main ransmission mechanism appears o be he opposie of improving compeiiveness. In his paper, we show ha hese differen perspecives naurally reconcile once he moneary model is appropriaely developed as o accoun for comparaive advanage, reconnecing open-macro o rade. Our main argumen is ha moneary policy aimed a sabilizing marginal coss and demand condiions a an aggregae level is likely o have asymmeric effecs across secors, more consequenial for firms facing higher enry coss and nominal price disorions---usually firms producing differeniaed manufacuring goods. Via his channel, moneary rules ha sysemaically reac o counry-specific shocks (hus weakening or srenghening he exchange rae in response o cyclical disurbances) can be expeced o influence he composiion of producion and expors beyond he shor run. Exchange rae and moneary regimes can hus have long-lasing effecs on a counry s inernaional compeiveness. On heoreical grounds, we draw on rade heory o model open economies wih incomplee specializaion across wo radable secors: in one secor, convenionally idenified wih manufacuring, firms produce an endogenous se of differeniaed varieies operaing In virually all conribuions o he new-open economy macroeconomics and New-Keynesian lieraure, he rade-off beween oupu gap and exchange rae sabilizaion is mainly modeled emphasizing a ermsof-rade exernaliy (see Obsfeld and Rogoff (2000) and Corsei and Peseni (200,2005), Canzoneri e al. (2005) in he NOEM lieraure, as well as Benigno and Benigno (2003), and Corsei e al. (200) in he New-keynesian lieraure, among ohers). Provided he demand for expors and impors is relaively elasic, an appreciaion of he erms of rade of manufacuring allows consumers o subsiue manufacuring impors for domesic manufacuring goods, wihou appreciable effecs in he marginal uiliy of consumpion, while reducing he disuiliy of labor. The opposie is rue if he rade elasiciy is low.

3 under imperfec compeiion; in he oher secor, firms produce non-differeniaed goods under perfec compeiion. Our argumen builds on he idea ha, because manufacuring indusries supplying differeniaed goods are ypically associaed wih a combinaion of monopoly power, price sickiness and sunk (enry) invesmen in esablishing a differeniaed produc, he need o susain upfron enry coss and o ake forward looking pricing decisions makes hem more sensiive o uncerainy abou fuure macroeconomic shocks han are oher indusries. We argue ha, by conribuing o domesic macroeconomic sabiliy, efficien moneary policy creaes favorable condiions for such indusries, wih long-lasing effecs on heir compeiiveness. Furhermore, o he exen ha he producion of manufacuring differeniaed goods bears social benefis for he counry no inernalized by individual firms, he welfare gains from efficien sabilizaion could poenially be larger han accouned for by sandard models. 2 These gains would however maerialize by virue of wha is bes described as a pro-compeiive effec of sabilizaion, raher han in erms of compeiive devaluaions and/or erms of rade manipulaion emphasized in exising lieraure. Our heoreical sraegy consiss of embedding a wo-secor marke srucure in a sochasic general equilibrium moneary model. In he presence of nominal rigidiies, as shown in previous work carried ou in a closed economy conex (such as Bergin and Corsei (2008) and Bilbiie, Ghironi and Meliz (2008)), uncerainy can imply he analog of a risk premium in a firm s prices, depending on he covariance of demand and marginal coss, i.e. he variabiliy of he ex-pos markups. 3 By affecing his covariance, moneary policy rules can conribue o firms seing low, compeiive prices on average, which in urn boos domesic demand. Low prices and higher domesic demand foser he comparaive advanage of he counry in producing and exporing differeniaed manufacuring goods. Endogenous firm enry amplifies he relocaion of producion beween counries. A key esable implicaion of he heory is ha counries wih a reduced abiliy o sabilize macro shocks, say, because of he adopion of a fixed exchange rae regime, will end 2 For example, produc differeniaion is associaed in he lieraure wih addiional rade and search coss, due o he lack of an organized exchange or reference prices ha are ypical for non-differeniaed goods; producing his good domesically avoids hese addiional rade coss. 3 See also Corsei and Peseni (2005) and Obsfeld and Rogoff (2000). 2

4 (all else equal) o specialize away from differeniaed goods, relaive o he counries wih an independen moneary policy. We es his predicion by conducing panel regressions of he composiion of expors o he U.S. by counry, on he exchange rae and moneary regime of ha counry, including a number of conrols. Through exensive analysis, we find ha a peg reduces he share of differeniaed goods in expors by abou 3-5 percenage poins. This resul is robus o changing he reference sample (e.g. o excluding oil expors and oil exporing counries), as well as o adoping alernaive classificaions of he exchange rae regime, and/or of insrumens designed o conrol for he endogeneiy of a peg. Conversely, we find ha adoping an inflaion-argeing regime raises he differeniaed share a similar amoun. We should sress here ha our heoreical and empirical conribuion is enirely disinc from he macroeconomic lieraure sressing and esing he effec of exchange rae volailiy on he volume of expors, wih ofen inconclusive resuls; we insead provide heoreical argumens and evidence, in favor of he view ha exchange rae and moneary regimes have appreciable effecs on he composiion of expors. Our heoreical resuls sugges an insighful re-inerpreaion of he differen ways in which he radiional (Mundell-Fleming) and he new (NOEM) lieraure characerize he policy rade-off beween oupu gap sabilizaion and he erms of rade. According o our model, sabilizaion policies have pro-compeiive effecs by helping firms in he manufacuring secor o keep average markups low, which in urn may/may no resul is low average inernaional relaive prices of domesically produced manufacuring. Wheher or no he inernaional price of domesic manufacuring falls wih well-designed sabilizaion policies, however, overall erms of rade of he counry become sronger. This is because efficien sabilizaion policies favor he producion and expors of high value-added manufacuring goods, raising heir weigh in he baske of expors. So, while lower manufacuring markups ends o (and under cerain condiions do) worsen a counry s erms of rade, he endogenous shif in he composiion of expors couneracs his effec, and causes an overall improvemen. Our resuls concerning he pro-compeiive effecs of moneary sabilizaion are hus fully consisen wih a key conclusion 3

5 in he NOEM lieraure, ha efficien sabilizaion may srenghen he erms of rade of a counry. 4 Wihou loss of generaliy, we model he social benefis from gaining comparaive advanage in he manufacuring secor assuming a producion relocaion exernaliy, as developed in Ossa (20) in a rade conex. In he presence of such exernaliy, acquiring a larger share of he world producion of differeniaed goods produces welfare gains due o savings on rade coss. In his paper, we show ha a producion relocaion exernaliy has a macroeconomic dimension, since i poenially causes moneary policy o have firs-order effecs on welfare via is influence on he composiion of oupu and expor. We sress, however, ha our main conclusions would also apply o oher ypes of exernaliies, impinging on he welfare implicaions of acquiring comparaive advanage in manufacuring. 5 The ex is srucured as follows. The nex secion describes he model. Secion 3 derives some analyical resuls for a special case, and secion 4 uses sochasic simulaions o demonsrae a broader se of implicaions Secion 5 presens empirical evidence in suppor of he heory. Secion 6 concludes. 2. Model Consider a model of wo counries, home and foreign. While mos of he model is a sandard wo-counry DSGE moneary model, here is a key novel elemen, in he way we 4 Our heory has implicaions for he analysis of cross-border policy cooperaion. Namely, he impac of moneary policy on rade and producion paerns creaes welfare incenives o deviae from moneary rules ha are efficien from a global perspecive, defining a policy game over comparaive advanages. While relaed o he NOEM lieraure sudying sraegic policy and coordinaion (see e.g. Benigno and Benigno (2003), Corsei and Peseni (2005), Corsei e al. (200), Obsfeld and Rogoff (2002) and Suherland (2004)), he mechanism producing gains from cooperaion in our model are differen. Our argumen is also concepually disinc from he conclusions of he lieraure assuming a raded and a non-raded goods secor (see e.g. Canzoneri e. al 2005). Gains from coordinaion in his case may resul from rade-offs in sabilizing marginal coss across wo secors in each counry, poenially creaing sronger cross-border spillovers han in he sandard model where all goods are raded. Our work is insead relaed o Corsei e. al (2007), which considers he role of he home marke effec in a real rade model, as well as Ghironi and Meliz (2005). We differ in modeling economies wih wo radable secors, as well as considering he implicaions of price sickiness and moneary policy. 5 There are a number of conribuions sudying he effecs of moneary policy regimes on enry, see, e.g., Cavallari (200), or he effec of exchange rae policy on rade, see, e.g., Saiger and Sykes (200). However, o our knowledge, no sudy has focused on compeiiveness encompassing producion relocaion. 4

6 specify he goods marke srucure. Namely, wo ypes of radable goods can be produced in eiher counry. The firs ype of good comes in differeniaed varieies produced under monopolisic compeiion. This is he marke wih firm enry and prese prices. The second ype of good is produced by perfecly compeiive firms, and is modeled according o he sandard specificaion in real business cycle models. For his good, here is perfec subsiuabiliy among producers wihin a counry (indeed, he good is produced under perfec compeiion), bu imperfec subsiuabiliy across counries, as summarized by an Armingon elasiciy. 2.. Goods marke srucure Households consume goods from wo secors. The D secor consiss of differeniaed goods, associaed wih manufacuring, which are produced by n and n monopolisically compeiive firms in he home and foreign counry, respecively---from now on, foreign variables will be denoed wih an aserisk. The N secor consiss of non-differeniaed goods, produced by perfecly compeiive firms. The home and foreign versions of he N and D good are imperfec subsiues for each oher, wih elasiciy ϕ and η, respecively. The overall consumpion index is specified: C C D, C N,, where n n CD, ch dh c f df 0 0 is he index over he home and foreign varieies of manufacuring good, c(h) and c(f), and C N, is he index over non-differeniaed goods.. The corresponding price index is C H, C F, 5

7 P P P D, D,, () where P D n p h n p f (2) is he index over he prices of all varieies of home and foreign manufacuring goods, and N H F P P P (3) is he index over he prices of home and foreign non-differeniaed goods. These definiions imply relaive demand funcions for domesic residens: c() h p h / P C (4) D D c( f) p f / P C (5) D D C PC / P (6) D D C PC / P. (7) N N H H N N C P / P C (8) / F F N N C P P C (9) 2.2. Home household problem The represenaive home household derives uiliy from consumpion (C), holding real money balances (M/P), and disuiliy from labor (l). The household derives income by selling labor a he nominal wage rae (W), receiving real profis from home firms h, and ineres income on holding domesic bonds (ib), which are in zero ne supply. Inernaional rade in goods is balanced since here is no inernaional rade in asses. They pay lump-sum axes (T). Household opimizaion for he home counry may be wrien: 6

8 M max E0 UC, l, 0 P subjec o he budge consrain: Uiliy is defined by n P C W l hdh W K M M B i B T. 0 M U C l ln P Defining PC, opimizaion implies an ineremporal Euler equaion:, i E (0) a labor supply condiion: W L () and a money demand condiion: M i i. (2) The problem and firs order condiions above are analogous for he foreign household Home firm problem and expor enry condiion In he differeniaed goods secor, producion is linear in labor: D y h l h, (3) where l(h) is he labor employed by firm h, and D is sochasic echnology common o all producion firms in he counry. Expors involve an iceberg rade cos, D, so ha yh d h D d h, (4) 7

9 where AC, K, d h c ( h) d ( h) d ( h) is oal demand for he produc in he home counry, for use in consumpion, adjusmen coss, and enry coss, respecively; corresponding demand for home goods abroad. Firm profis are compued as: sunk cos, h p hd h e p hd h Wy h/ AC h p, 8 d h is he. (5) There is free enry ino he expor marke wih a one-period lag subjec o a one-ime K. The sunk cos is composed of a fixed proporion of labor unis ( K ) and differeniaed goods unis (- K ). I is assumed ha a fracion of all firms mus exogenously exi each period. Le n represen he number of firms, and define new enrans o he expor marke, ne, by he flow condiion: n n ne. (6) The value funcion of firms ha ener period as an exporer may be represened as he discouned sum of profis of domesic sales and expor sales, s s vh E sh. s0 Firms ener unil he poin ha firm value equals he enry cos: KW K PDK v h. (7) The goods porion of enry cos uses a composie of differeniaed goods, following he consumpion index: d () /, h p h P nek (8) K D K d ( f) p f / P nek. (9) K D K The home firm h ses a price p(h) in domesic currency unis for domesic sales. Under he assumpion of producer currency pricing, his implies a foreign currency price p h p h / e, (20) M where he nominal exchange rae, e, is defined as home currency unis per foreign currency uni. Firms face a nominal cos of adjusing prices AC h h p h P 2 p 2 p hy h. (2)

10 We follow Bilbiie e. al (2008) in making he simplifying assumpion ha new firm enrans inheri from he price hisory of incumbens he same price adjusmen cos, and so make he same price seing decision. The aggregae value of adjusmen coss are: AC h nac h. (22) The adjusmen cos uses final goods, and he composiion follows ha assumed for consumers in equaion (4)-(9): AC, M ACD, d () h p h / P D (23), d ( f) p f / P D (24) AC M AC D D P AC P (25) AC, M / D DAC, N PAC / PN (26) AC, H H / N AC, N D P P D (27) / AC, F F N AC, N D P P D. (28) Maximizing firm value subjec o he consrains above leads o he price seing equaion: 2 2 W p h p h p h ph ph 2 p h p h p h E p h p h p h p h where he opimal pricing is a funcion of he average price of world manufacuring: 2,,, PM, CD, DAC, D K nek D e P M C D D AC D K nek.. In he second secor firms are assumed o be perfecly compeiive in producing a good differeniaed only by counry of origin. The producion funcion for he home nondiffereniaed good is linear in labor: y H, H, H, (29) l, (30) 9

11 where H, is subjec o shocks. I follows ha he price of he homogeneous goods in he home marke is equal o marginal coss: p W /. (3) H, H, An iceberg rade cos specific o he non-differeniaed secor implies prices of he home good abroad are p p H, H, N / e. (32) Analogous condiions apply o he foreign non-differeniaed secor Governmen The model absracs from public consumpion expendiure, so ha he governmen uses seigniorage revenues and axes o finance ransfers, assumed o be lump sum. The home governmen faces he budge consrain: M M T. (33) 0 Since in his ex we do no address issues in he design of opimal sabilizaion, we specify moneary policy primarily in he form of a Taylor rule: p p h Y i i p h Y Y. (34) This Taylor rule defines inflaion in erms of differeniaed goods producer prices, and defines Y a measure of oupu defined as: Y p h y h dh p y P n H H 0 /. In running he model, we will use eiher his definiion of oupu, or a narrower definiion, including only manufacuring. Given our calibraion of he Taylor rule, wih a high coefficien on inflaion, his will be immaerial for our resuls. The foreign counry will in general be assumed in simulaions o follow an exchange rae peg: e e. (35) 0

12 2.5. Marke clearing The marke clearing condiion for he manufacuring goods marke is given in equaion (4) above. Marke clearing for he non-differeniaed goods marke requires: Labor marke clearing requires: Bond marke clearing requires: y C C D D (36) H, H, H, AC, H, AC, H, y C C D D. (37) F, F, F, AC, F, AC, F, n, l h dh l nek l. (38) H 0 Balance of paymens in his case requires balanced goods rade: n B 0. (39) n K AC, p h c h d h d hdh p f c f dk f dac, f df 0 0 (40) PH CH DAC, H, PF CF DAC, F, Equilibrium definiion and shocks: The produciviy shocks follow he join log normal disribuion: logd log D logd log D logd log D logd log D logh logh logh logh log log log log F F F F Wih he covariance marix E ' Relaive price measures: Along wih he real exchange rae ( ep / P), we repor wo alernaive measures of inernaional prices. Firs, as common pracice in he producion of saisics on inernaional

13 relaive prices, we compue he erms of rade weighing goods wih heir respecive expendiure shares: TOTS ph () H H H, Fep ( f) F ep F, p, (4) where he weighs are: p () h n c h d Kh d AC, h K AC, H H AC, H, H p h n c h d h d h P C D () p( f) n c f dk f dac, f F p( f) n c f d f d f P C D K AC, F F AC, F, (4a). (4b) Following he rade lieraure, however, we also compue he erms of rade as he raio of exfacory prices se by home firms relaive o foreign firms in he manufacuring secor: TOTM p h e p f ( )/ ( ). 6 This measure ignores he non-differeniaed good secor. 3. Analyical Resuls Insighs from a Simple Version of he Model In his secion, we will sudy he consequences of alernaive sabilizaion policy regimes on compeiiveness and welfare in he wo economies, working ou a simplified version of he model ha is amenable o analyical resuls. Despie a number of simplifying assumpions needed o make he model racable, we will be able o derive key predicions ha remain valid in our more general model specificaion. The secion is organized as follows. Afer saing he main assumpions, we will characerize he flex-price allocaion, and analyze he consequences of nominal rigidiies. In doing so, we will firs clarify he mechanism by which macro uncerainy abou demand and marginal coss impinge on firms pricing. We will hen esablish ha symmeric moneary policy rules of full markup sabilizaion suppor he flex-price allocaion, showing ha, in our simplified model, hese policies are he same as in he baseline NOEM model wihou enry (see, e.g., Corsei and Peseni 2005). We conclude wih he analysis of a unilaeral peg. 6 This is he same definiion used in Ossa (20), hough in our case i does no imply he erms of rade are consan a uniy, because moneary policy does affec facory prices. See also Helpman and Krugman (989), as well as Campolmi e al. (202). 2

14 3. Model Specificaion In order o illusrae he moneary ransmission mechanism as clearly as possible, we will make he following assumpions. Firs, we posi ha manufacuring firms operae for one period only (implying in he enry condiion), and symmerically prese prices over he same horizon. Enry coss are in labor unis only. Second, we simplify he non-differeniaed good by seing is rade coss o zero ( N 0 ) and leing he elasiciy subsiuion beween home and foreign goods approach infiniy ( ). This implies he secor acs like a homogeneous good, an assumpion frequenly made in he rade lieraure. 7 Correspondingly, we limi produciviy shocks in he Differeniaed good secor o he i.i.d. case, while produciviy in he Non-differeniaed good secor is no subjec o shocks. Finally, we assume ha uiliy is log in consumpion and linear in leisure. Under hese assumpions, i is convenien o represen policy as a rule ha responds direcly o shocks:,, where PC (see Bergin and Corsei 2006). The firms problem becomes max p h E h. The opimal prese price in he domesic marke is: p E h. (42) E The home enry condiion is a funcion of price seing and he exchange rae: K E p h p h (43) where upon appropriae subsiuions (deailed in he appendix) can be wrien as: 7 Differen from he rade lieraure, however, we do rea his secor as an inegral par of he (general) equilibrium allocaion, e.g., expors/impors of he homogeneous good secor eners he erms of rade of he counry. 3

15 n p (h) n p ( f ) e n p (h) n p ( f ) e Provided ha he price seing rules can be expressed as funcions of he exogenous shocks and policy seings (a condiion saisfied in several useful cases), he home and foreign equilibrium enry condiions along wih he exchange rae soluion above comprise a hree equaion sysem in he hree variables: e, n and n. This sysem admis analyical soluions for several configuraions of he policy rules. Before proceeding, we should sress wo key implicaions of our assumpions. Since boh economies produce he same homogeneous good wih idenical echnology under perfec. compeiion, wih no rade coss in his secor, arbirage ensures ha P e P D D. Using he labor supply condiion () for he case of a uniary elasiciy ( 0 ), he exchange rae may be expressed as: e p D W p D W PC P C. (44) As long as boh economies produce he homogeneous goods, he exchange rae is hus deermined hrough arbirage in he perfecly compeiive secor of he goods marke. Given symmeric echnology in labor inpu only, he law of one price implies ha nominal wages are equalized (once expressed in a common currency) across he border. By he equilibrium condiion in he labor marke wih wih an infinie labor supply elasiciy, hen, he exchange rae is a funcion of he raio of nominal consumpion demands (and hence is he raio of he moneary policy sance variables). A furher noable implicaion is ha risk sharing is perfec regardless of he srucure of financial markes and/or he way we specify producion and rade in he oher secor. Rewrie he above equaion as: e P P rer C C. 4

16 This is he risk sharing condiion implied by complee asse markes for he case of log uiliy: home consumpion rises relaive o foreign consumpion only in hose saes of he world in which is relaive price (i.e. he real exchange rae) is weak. In our economy, risk sharing is complee per effec of nominal wage equalizaion (due o rade in a single homogenous good whose producion is no subjec o shocks), even in he absence of rade in financial asses Flexible prices versus nominal rigidiies If firms are able o se prices afer observing (produciviy) shocks, wih consan demand elasiciy, managers will charge a consan markup over marginal coss, which in our case coincide wih uni labor coss: p flex h flex p f. Subsiuing ino he enry condiion (27) above, i is easy o derive he equilibrium number of firms under price (and wage) flexibiliy: n flex n flex 2 q E Since in our simple model we assume i.i.d. shocks and he number of firms is predeermined, n and n are no ime-varying (do no respond o curren produciviy). Wih no moneary policy response o shocks (, as none is needed under flexible prices), he exchange rae will be consan a e /. Consider now nominal rigidiies impinging on prices of manufacuring goods. To bes appreciae he ransmission mechanism underlying our resuls, i is convenien o rewrie (42) as follows: 2( ). p h E Cov E (42 ) 5

17 The covariance erm on he RHS of his expression shows ha opimal prese pricing depends on he comovemens of nominal wages and labor produciviy (deermining firms marginal coss κμ+/α +), and he world price of manufacuring, scaled by he expeced price of world manufacuring. Since boh marginal coss and he world price of manufacuring are funcions of moneary sances, he effec of sabilizaion rules on pricing crucially hinges upon heir effec on his covariance erm. To appreciae he consequences on opimal pricing and enry, wihou loss of generaliy se, as before,, implying a consan exchange rae ( e / ). Wih i.i.d. shocks, here are no dynamics in predeermined variables such as prices and numbers of firms. We can hus solve for boh analyically. As regards prices, firms prese hem opimally charging he consan, equilibrium markup over expeced marginal coss: no sab p h E no sab p f E. Noe ha, wih a moneary policy unresponsive o produciviy shocks, hese opimal pricing decisions do no depend on he world wide price of manufacuring erm Ω (hence do no vary wih rade coss and firms enry), as hey do in he general case. The number of firms can be compued by subsiuing hese prices ino he enry condiion (43), so o obain: no n sab no n sab q. Inuiively, wih prese prices and no change in he exchange rae, here is no change in he erms of rade or he real exchange rae. For a given moneary sance, here is no change in consumpion demands, and so no change in he level of producion in any good. An i.i.d. shock lowering produciviy in he home manufacuring secor necessarily leads o an increase in he level of employmen in he same secor (no compensaed by a change in employmen in he oher secors of he economy). Firms end up producing a high marginal coss and hus subopimally low markups, as nominal rigidiies preven hem from re-pricing and scaling down producion. By he same oken, given nominal prices and demand, an economy experiencing a rise in produciviy will end up producing oo lile a low marginal coss, hence subopimally high markups. These wo conrasing effecs do no wash ou on average. 6

18 From he above pricing funcions, i is easy o see ha, in a regime of non-coningen moneary policy, a high variance of produciviy shocks leads firms o opimally prese relaively high prices --- inducing a risk-premium like erm in pricing. As discussed in Corsei and Peseni (2005) and Bergin and Corsei (2008), given nominal demand, high prese prices allow firms o conain overproducion when low produciviy squeezes markups, rebalancing demand across saes of naure. High average markups, in urn, end o reduce demand, producion and employmen on average, discouraging enry Markup and oupu gap sabilizaion Since he model posis ha he homogenous good secor operaes under perfec compeiion and flexible prices, here is no rade-off in sabilizing oupu across differen secors. I is herefore possible o replicae he flex-price allocaion under he following simple moneary policy rule: he moneary sance in each counry moves in proporion o produciviy in he differeniaed good secor:,. The exchange rae in his case is no consan, bu coningen on produciviy differenials. Namely, he home currency depreciaes in response o an asymmeric rise in home produciviy: e. The acive moneary policy jus described specifically affecs opimal pricing by firms. By ensuring ha he nominal marginal coss κμ/α remain consan, he above policy ensures ha he covariance erm in (see (42 )) is zero, hus insulaing he opimal price prese by home manufacuring firms from he variance of produciviy shocks. 8 The price firms prese is hus lower han in an economy wih no sabilizaion: sab no sab p h p h, 8 As is well undersood, he policy works as follows: in response o a incipien fall in domesic marginal coss domesic demand and a real depreciaion boos foreign demand for domesic produc. As nominal wages rise wih aggregae demand, marginal coss are compleely sabilized a a higher level of producion. Vice versa, by curbing domesic demand and appreciaing he currency when marginal coss are rising, moneary policy can preven overheaing, driving down demand and nominal wages. Again, marginal coss are compleely sabilized as a resul. 7

19 given ha, by Jensen s inequaliy, E. Noe ha, by sabilizing marginal E coss compleely, such policy sabilizes markups a heir flex-price equilibrium level. As shown in he appendix, i follows ha he number of manufacuring firms is he same as under flexible prices: 9 n sab flex n. Despie nominal rigidiies, policy makers are able o sabilize he oupu gap relaive o he naural-rae, flex-price allocaion --- a resul ha is familiar from he classical NOEM lieraure (wihou enry) assuming ha prices are sicky in he currency of he producers (Corsei and Peseni (200, 2005) and Devereux and Engel (2003), among ohers) Inefficien sabilizaion: he case of a unilaeral currency peg The key insigh from he model follows from he analysis of asymmeric policies. Consider he case in which one counry adops a currency peg, while he oher sick o efficien inward-looking policy rules. Namely, we posi ha he home governmen fully sabilizes is oupu gap, while he foreign counry mainains is exchange rae fixed agains he home currency:, e so. 0 Under he asymmeric policy scenario of a currency peg jus described, he opimally prese prices of domesically and foreign produced differeniaed goods are, respecively: p h, p f E. While he home policy makers manage o sabilize he markup of manufacuring firms compleely, he foreign firms producing under he peg regime face sochasic marginal coss 9 As discussed in he appendix, i is no possible o deermine analyically wheher symmeric sabilizaion policies raise he number of firms compared o he no sabilizaion case. Model simulaions sugges ha here is no posiive effec for log uiliy, and a small posiive effec for CES uiliy wih a higher elasiciy of subsiuion. Noneheless, we are able o provide below an analyical demonsraion of asymmeric sabilizaion, which is our main objecive. 0 A relaed exercise consiss of assuming ha he foreign counry keeps is money growh consan ( ) while home carries ou is sabilizaion policy as above. 8

20 driven by shocks o produciviy, boh domesically and abroad. Wih i.i.d. shocks, prese prices will be increasing in he erm E(/α+), as in he no sabilizaion case. The equilibrium number of firms n and n insead solve he following wo-equaion sysem: where A E n An q n Bn A B q n An n Bn, B E 9. While i is no possible o solve for he number of firms in closed form, he sysem above does allow one o prove ha n>n flex > n (see he appendix). Oher hings equal, he consrain on macroeconomic sabilizaion implied by a currency peg ends o reduce he size of he manufacuring secor in he foreign counry: here will be fewer firms charging higher prices. The home counry s manufacuring secor correspondingly expands. In oher words, he counry pegging is currency will end o specialize in he homogeneous good secor. To undersand he mechanism: insofar as inefficien sabilizaion under a peg raises markups and exacerbae monopolisic disorions in he foreign manufacuring secor, foreign consumpion boh falls overall, and shifs owards he non-differeniaed good secor, reducing he incenive for firms o ener. In his process, he erms of rade of he Foreign counry weakens (alhough he welfare-relevan real exchange rae appreciaes driven by he fall in domesic varieies available o he consumers). Correspondingly, in he Home counry, sronger erms of rade boos consumpion, raising world demand for Home-produced manufacuring. As a resul, wih a foreign counry passively pegging is currency, here are exra benefis for he home counry from being able o pursue sabilizaion policies. Lack of sabilizaion abroad undermines he foreign comparaive advanage in indusries ha are sensiive o shocks and uncerainy, providing a boos o he home domesic manufacuring secor. This secor expands driven by higher Home demand overall, and fills par of he gap in manufacuring producion no longer supplied by foreign firms. A he same ime, he shifing

21 paern of specializaion ensures ha he home demand for he homogeneous good is saisfied via addiional impors from foreign. 4. Numerical simulaion We now evaluae he quaniaive implicaions of he core mechanism described in he simplified model, by conducing sochasic simulaions of he full model. Relaive o he simple model, in our economy, firs, facor prices (wages) are no equalized (in erms of he nondiffereniaed good) across borders. Second, depending on he inerplay of he elasiciies governing saving, risk aversion, labor supply and cross-border rade, a lower average markup in Home pricing does no necessarily resul in lower inernaional prices of manufacuring (since he increase in he demand for he Home goods ends o raise producion coss). Third, enry coss are specified in erms of goods raher han labor, and hus may move wih inernaional prices. Finally shocks are no resriced o be i.i.d.. Despie hese differences, we will show ha key resuls from he simple version of our model coninue o hold in he general one. Namely, i will sill be rue ha, if he foreign counry moves from efficien sabilizaion o a peg, while he Home counry sicks o efficien sabilizaion rules (a) he Foreign average markups in manufacuring will end o increase and (b) here will be producion relocaion---firm enry in he foreign counry will fall on average, while enry in he Home counry will rise on average (correspondingly, average consumpion will rise a Home relaive o Foreign). We will see ha, on average, his relocaion will be associaed wih a Home erms of rade improvemens (while he Home welfare-relevan real exchange rae depreciaes). The inernaional price of Home manufacuring goods may or may no fall, depending on Home coss. We firs discuss our calibraion of he model, hen presen our main resuls. In relaed work (Bergin and Corsei 2008), we sress a reason o prefer our specificaion in Secion 2, as he alernaive implies ha, counerfacually, moneary expansions should be sysemaically associaed wih exi. 20

22 4.. Calibraion The model is calibraed for an annual frequency, o mach he frequency of he daa available for secoral produciviy used o calibrae our shocks. The ime preferences is calibraed a We calibrae risk aversion a he usual value of / 2 supply elasiciy is se a /.9 from Hall (2009).. Labor The price sickiness parameer is se a 77, following he esimae in Ireland (200). The deah rae is se a 0., which is four imes he sandard rae of o reflec he annual frequency. As is sandard, he sunk cos of enry is normalized o he value of. To calibrae he differeniaed and non-differeniaed secors we draw on Rauch (999). Differeniaed goods represen around half of U.S. rade in value, so we calibrae = The home share of non-differeniaed goods is se a 0.5, which implies a rade share of abou 30%, given he rade coss and elasiciies below. To se he elasiciies of subsiuion for he Differeniaed and Non-differeniaed goods we draw on he esimaes by Broda and Weinsein (2006), classified by secors based on Rauch (999). The Broda and Weinsein (2006) esimae of he elasiciy of subsiuion beween differeniaed goods varieies is 5.2 (he sample period is ). The corresponding elasiciy of subsiuion for nondiffereniaed commodiies is 5.3. To calibrae rade coss, we need o hink beyond coss associaed wih jus ransporaion. These are ofen hough o be higher for commodiies han for high value differeniaed goods. As Rauch (999) poins ou, differeniaed goods involve search and maching coss, whereas commodiies and goods raded on an organized exchange wih a published reference price avoid such coss. Esimaes are available for he ariff equivalen of language coss, wih a value of % in Hummels (999) or 6% in Anderson and van Wincoop (2004), so we use 8% in beween. Since Obsfeld and Rogoff (2000) recommend a calibraion of oal rade coss a 6%, our calibraion implies ha half of his is due o language and 2 Values vary by year and by wheher a conservaive or liberal aggregaion is used. For example, in 980, he conservaive aggregae indicaes a differeniaed goods share of 5.5%, and he liberal 48.9%. 2

23 maching coss, and he oher half due o ransporaion. This implies a calibraion of D = 0.6 for differeniaed goods, and N =0.08 for non-differeniaed goods. The parameers in he home moneary policy rule are deermined by he values ha maximize home uiliy. As ypically found, he opimal weigh on inflaion is he maximum value considered in he grid search ( P =000), and he opimal value on oupu is Y =0. The foreign counry is assumed o peg is exchange rae a pariy wih he home counry: e=. To our knowledge, no one else who has calibraed a DSGE model wih secoral shocks disinc o differeniaed and nondiffereniaed goods. Time series for secoral produciviies are available on an annual basis from he Groningen Growh and Developmen Cenre (GGDC), on an annual basis Daa for he U.S. is used o calibrae shocks o he Home counry, and an aggregae of he EU 0 for he foreign counry. TFP is calculaed on a value added basis. For each counry, he differeniaed goods secor includes oal manufacuring excluding wood, chemical, minerals, and basic meals; he nondiffereniaed goods secor includes agriculure, mining, and subcaegories of manufacuring excluded from he differeniaed secor. To calculae he weigh of each subsecor wihin he differeniaed (or non-differeniaed) secor, we use he 995 gross value added (a curren prices) of each subsecor divided by he oal value added for he differeniaed (or non-differeniaed) secor. Afer aking logs of he weighed series, we derend each series using he HP filer. Parameers and Ω, repored in Table, are obained from running a VAR() on he four de-rended series Simulaion Resuls We sar by analyzing he dynamics of he benchmark model in response o a one sandard deviaion posiive shock o home differeniaed goods produciviy. Assuming efficien sabilizaion policy in boh counries, for each variable of ineres, Fig. plos he percenage deviaion from he uncondiional mean. As Home policymakers reac o he shock by expanding domesic demand and depreciaing he exchange rae, Home producion emporarily shifs from he non-differeniaed goods secor o differeniaed goods---non- 22

24 differeniae oupu acually falls. Due o posiive correlaion in shocks across counries, foreign producion of differeniaed goods also rises, bu by a smaller amoun han in he home counry. In conras wih he home counry, foreign producion in non-differeniaed goods rises. This shif in producion of differeniaed goods is mirrored in a rise in he number of firms in boh counries, by a larger amoun a home. Table 2 characerizes uncondiional means of key variables coming from sochasic simulaions of a second order approximaion of he benchmark model. Column () repors he values of hese means when boh counries use sabilizaion policy, column (2) when he foreign counry adops an exchange rae peg; in column (3) we repor he percen change beween he previous columns when foreign swiches from inflaion sabilizaion o a peg, accouning for changes across sochasic seady saes. Noe ha counry means are no compleely symmeric even under he symmeric policies in column (), due o he asymmeries in he calibraed shocks. The simulaion resuls fully confirm he main analyical insighs from he previous secion. When he foreign counry pegs, average producion of he differeniaed good shifs away from he foreign counry and oward he home counry; he foreign counry insead has higher producion of he non-differeniaed good. This shif in producion is refleced in a.3 percen fall in he number of foreign differeniaed goods firms, corresponding o a. percen rise a home. The share of differeniaed goods in expors falls by.2% in he Foreign counry, while rises by.% in he Home counry. Consisen wih he ransmission mechanism highlighed by he analyical resuls based on he simplified version of our model, wha drives he Foreign loss in he differeniaed goods marke share is a higher markup charged by differeniaed goods producers on average. Even if falling wages in he foreign counry lead o falling prices of all foreign goods, he price of differeniaed goods rises relaive o boh wages and non-differeniaed goods. Anoher noable implicaion of he simulaion is ha he erms of rade including he homogenous good, TOTS, acually worsen for he foreign counry when i abandons efficien sabilizaion policy. This sands in conras wih he convenionally defined erms of rade including only differeniaed goods, TOTM, which does no worsen, reflecing he rise in 23

25 foreign markups on differeniaed expors. The conrasing behavior of he TOTS is due o a composiion effec: he shif in foreign expor share away from differeniaed goods means hese more cosly goods receive a smaller weigh in he average price of foreign expors and a larger weigh in he average price of foreign impors. The able also indicaes ha relaive consumpion levels are affeced, wih a fall in foreign consumpion and arise a Home. Table 3 summarizes he robusness of our key resuls for alernaive specificaions and calibraions of he model. To save space, we only repor he percenage change in number of firms and percen change in differeniaed expor share when he foreign counry swiches from inflaion sabilizaion o exchange rae peg. We repor hese differences boh cumulaed and by counry, as he laer will be useful in making comparisons o he empirical secion o follow. The firs column reieraes he key resul for he benchmark case from Table 2, adding ha he oal change in differeniaed expor share beween counries is 2.2 percenage poins. Column (2) indicaes ha he percen changes in differeniaed expor share shrink by more han an order of magniude when he number of firms is held fixed exogenously. We conclude ha he endogenous shif in number of firms beween counries is essenial for he change in moneary policy o ranslae ino quaniaively meaningful effecs on expor shares. The assumpion of balanced rade plays no key role in our resuls: columns (3) and (4) indicae ha he model keeps predicing a significan reallocaion of producion and a change in he composiion of rade when agens can rade non-coningen bonds or have access o complee asse markes---in he appendix we deail changes in our model specificaion o inroduce hese alernaive asse marke arrangemens. Columns (5) and (6) show ha, as expeced, he size of reallocaion can be larger when he labor supply elasiciy is infinie, as we assumed in he analyical secion above. A rising labor supply elasiciy is no he only facor in amplificaion, hough. Our sronges resul obains when we se an infinie labor supply elasiciy ogeher wih a relaive risk aversion coefficien of 5, as in, e.g., Fernandez- Villaverde e al. (20). This combinaion of elasiciies grealy magnifies he effecs of policy on he rade shares: he convexiy of he uiliy funcion ends o raise he risk-like covariance erm in pricing, magnifying he differenial effecs of a peg; an elasic labor supply helps smooh ou reallocaion and conain he effecs of a large increase in enry on local coss. When 24

26 he foreign counry adops a peg, he share of differeniaed goods in foreign expor drops 9.7 %, and ha of he home counry rises 5.7%. The las hree columns of Table 3 analyze he role of enry coss. The Table shows ha, when enry coss are specified in labor unis, magniudes of percen changes in firm number and expor shares shrink. Noe ha wages end o rise in he counry wih more firm enry: if coss are in labor unis, his effec ends o counerac he incenive o se up new producion lines. Conversely, when enry coss depend on domesic manufacuring goods prices raher han wages, so ha enry coss are valued in he same unis in which a firm evaluaes is profis and firm value, resuls are very similar o our benchmark---in which he iniial invesmen also includes expendiure on impors. This suggess ha our benchmark resul is no driven by he Home erms of rade appreciaion, driving down he porion of enry coss falling on impored goods. Enry coss are obviously consequenial. The final column shows ha he srong resuls obained in column (7) for he case of a high labor elasiciy and a high risk aversion combined become closer o our benchmark when he enry cos is in labor unis, raher han unis of a firm s good. The difference is ha he iniial invesmen no longer hinges on falling prices in goods. In column (7) indeed, when more firms ener, he home price index of differeniaed goods falls, driving down enry coss, in urn inducing furher enry, and so on. 5. Empirical evidence To carry ou an empirical exploraion of our main argumen, we focus on a key esable implicaion of he model: counries wih moneary policy focused on domesic macro sabilizaion will have greaer specializaion of producion and expor in differeniaed producs, relaive o counries wih moneary policy driven insead by he objecive of mainaining a fixed exchange rae. Our empirical sraegy consiss of aking he U.S. as he base counry, and run panel regressions o es wheher counries wih independen moneary policy end o have a greaer share of heir expors o he U.S. in indusries classified as differeniaed. 25

27 5.. Daa consrucion and descripion To disinguish counries wih or wihou moneary policy independence, we use informaion on boh he exchange rae and he moneary regimes. Namely, we proxy lack of independence by he adopion of a pegged exchange rae regime. A he oher exreme, we consider a counry as pursuing independen moneary policy when i formally embraces an inflaion argeing regime. For he classificaion of exchange rae regimes, we rely on wo sources. The Inernaional Moneary Fund produces a classificaion of exchange rae regimes based upon he observed degree of exchange rae flexibiliy and he exisence of formal or informal commimens o exchange rae pahs. The definiion of peg includes counries wih no separae legal ender, currency board arrangemens, exchange rae bands, or crawling pegs; his excludes counries classified as managed floaing and independen floaing. We will also consider he classificaion sysem of Shambaugh (2004), which idenifies a peg if a counry ses is ineres raes sysemaically following he policy decision in a base counry. One advanage of his classificaion focuses on moneary independence raher han exchange rae regime per se, and our heory says pegs maer mainly for he loss of moneary independence. For example, Germany was classified as having moneary independence despie paricipaing in pegs over he sample, because i aced as a leader wihin a pegging block. Furher counries where capial conrols insulaed domesic moneary policy from global marke pressure are also classified as having moneary independence. Furher, China is classified as having moneary independence in much of he sample. We dae he adopion of an inflaion argeing (IT) regime drawing on Roger (2009). This auhor disinguishes wo phases in he adopion process: an iniial disinflaion period, ha lass for one or more years; and a sable IT regime --- possibly moivaing he use of a dynamic specificaion of he regression model. Around 20 counries adop an IT regime in he years covered by our sample, boh high and middle-income ones. To idenify expors of differeniaed goods, we rely on Rauch (999). This auhor provides a classificaion of 4-digi SITC indusries in erms of he degree of differeniaion among producs. Some producs are raded on organized exchanges, while some ohers have 26

28 reference prices published in rade journals. Those producs for which neiher is rue are classified as differeniaed. Roughly 58% of he indusries fall ino he differeniaed caegory. Trade daa come from he World Trade Flows Daabase (see Feensra, e al., 2005). Expors o he U.S. (in dollars) are available disaggregaed by counry and by four-digi indusry, on an annual basis for he period The se of counries covered boh by he rade daa and exchange rae classificaion number 64. The sample years are deermined by he availabiliy of U.S. disaggregaed impor daa, covering he period Empirical Specificaion Our dependen variable is he share of differeniaed goods in expors. Le xij denoe he dollar value of expors in indusry i from counry j o he U.S. in year. Le DIF akes he value of for a differeniaed indusry and 0 oherwise. For counry j in year, we define a measure of he share of differeniaed goods in he overall expors of a counry o he U.S.: SDIF j i x i ij 27 DIF i x ij. The index akes values on he coninuous inerval beween 0 and. In some of our experimens we will resric aenion o manufacuring expors only. In his case, we will only consider xij if belonging o SITC secor wih code saring wih 5 hrough 8. When using he exchange rae regime as an indicaor of consrains on moneary sabilizaion, our regression specificaion is SDIF j 0 MR j 2 X j j j,. (32) MR j PEG j, IT j where he moneary regime MR dummy is, respecively, PEG or IT, X is vecor of addiional variables ha we may include in he analysis as addiional conrols and χ are counry and year fixed effecs. When we proxy moneary independence wih he exchange rae regime, PEG akes he value of for a fixed exchange rae and 0 oherwise. For he case of inflaion argeing, he dummy variable PEG replaced by IT --- which akes he value of in he years in which a counry adops ha moneary regime, 0 oherwise. In some of our specificaions, we