DRAFT PLEASE DO NOT QUOTE OR DISTRIBUTE

Transcription

1 Does Asia s choice of exchange rae regime affec urope s exposure o US shocks? Boan Markovic* and Laura ovoledo** Working aper no. * Bank of ngland hreadneedle Sree London C2R 8AH. mail: boan.markovic@bankofengland.co.uk. ** he Universiy of Reading mail: l.povoledo@reading.ac.uk. DRAF LAS DO NO QUO OR DISRIBU he views expressed in his paper are hose of he auhors and no necessarily hose of he Bank of ngland. We wish o hank Marin Brooke Karen Dury and Francesco Giavazzi for useful commens and Mike Smih for valuable research assisance. < > his paper was finalised on dd monh 200x. he Bank of ngland s working paper series is exernally refereed. Informaion on he Bank s working paper series can be found a ublicaions Group Bank of ngland hreadneedle Sree London C2R 8AH; elephone 44 (0) fax 44 (0) mapublicaions@bankofengland.co.uk. Bank of ngland 2006 ISSN

2 Conens Absrac 3 Summary 4 1 Inroducion 6 2 he Model Households Firms Governmens 12 3 Calibraion 13 4 he ransmission of US shocks o urope osiive produciviy shock in he US raded goods secor osiive produciviy shock in he US non-raded goods secor Negaive demand shock in he US 22 5 he impac on volailiies and exposure o foreign shocks 24 6 Sensiiviy analysis 27 7 Conclusions 30 Appendix A: Impulse responses o a posiive produciviy shock in he US raded secor 32 Appendix B: Impulse responses o a posiive produciviy shock in he US N- secor 35 Appendix C: Impulse responses o a negaive demand shock in he US 38 References 41 2

3 Absrac In his paper we use a sylised hree-counry model o analyse how he ransmission of US shocks o urope migh be affeced by Asia s choice of exchange rae regime. We find ha if Asia decides o peg her exchange rae o he dollar he impac of US shocks on uropean oupu and inflaion is likely o be bigger han i oherwise would have been. his happens because wihou nominal exchange rae flexibiliy Asian firms reac o he shocks originaing in he US by implemening significan price adusmens which in urn affec urope s relaive compeiive posiion. On he heoreical side our resuls conribue o he lieraure by suggesing ha he shock insulaion propery of floaing exchange raes exends beyond he wo counries ha have currencies ha are free o move. he ransmission of shocks beween wo counries can also be dampened by he choice of floaing exchange raes in a hird counry. On he pracical side our resuls sugges ha if China did evenually decide o floa her currency urope s exposure o US shocks would decrease. Neverheless our resuls also sugges ha he overall fall in volailiy of urope s inflaion and oupu following China s floaing migh be modes in size. Key words: Inernaional ransmission xchange rae regimes Asia urope. JL classificaion: F41. 3

4 Summary he exen o which Asia s choice of exchange rae regime affecs urope s exposure o US shocks is a perinen issue o examine a presen for wo reasons. Firs a number of commenaors have suggesed ha he Unied Saes large curren accoun defici is unsusainable and will likely unwind (perhaps caused by shocks in he US). Second he Chinese auhoriies have made a number of suggesions over he pas year ha hey migh allow greaer flexibiliy in he exchange rae movemens of he renminbi. We use a hree-counry model symmerically calibraed for he US urope and Asia o analyse he effecs on urope of US shocks and compare wo cases: (1) when he currency of he Asian bloc is pegged o he US dollar (2) when he Asian currency freely floas agains boh he US and uropean currencies Our model capures several channels by which shocks in he US affec aggregae demand for urope s oupu. Firs any shock ha raises US consumpion increases US demand for worldwide and hus urope s oupu he aggregae world demand effec. Aggregae demand for urope s oupu is furher affeced by he consequen reacion of consumpion in urope and consumpion in Asia. Second aggregae demand for urope s oupu depends no only on aggregae world consumpion bu also on he allocaion of consumpion across counries because households are biased owards consuming domesic producs. Hence a uni increase in consumpion in Asia has a differen effec on demand for urope s oupu han a uni increase in consumpion in urope. hird he allocaion of consumpion over ime (consumpion smoohing) also affecs he demand for urope s oupu. he srengh of hese channels will depend in par on he choice of he exchange rae regime. Our resuls show ha Asia s choice of he exchange rae regime does no significanly affec urope s exposure o US shocks in he case of some shocks (namely he demand shock and he produciviy shock in he US raded goods secor) bu has a significan effec in case of oher shocks (namely he produciviy shock in he US non-raded goods secor). When nominal exchange raes canno be used as a buffer for shocks Asian firms reac o shocks originaing in he US by implemening large price adusmens (since prices now have o do all he work) and his in urn srongly affecs urope s relaive compeiive posiion. his adusmen is especially pronounced afer a shock occurring in he US non-raded goods secor. Our model shows ha he adusmen of Asian prices iniially dominaes he oher effecs induced by he pegging of he Asian 4

5 currency. Overall he fixed exchange rae in Asia increases he exposure of oupu and inflaion in urope o shocks originaing in he US. his resul suggess ha if China evenually decided o floa her currency urope s exposure o US shocks would decrease. Bu our resuls also indicae ha for a reasonable calibraion he overall volailiy of urope s oupu and inflaion depends mainly on domesic and no foreign shocks. herefore even hese significan changes in urope s exposure o US shocks following China s floaing migh no have large effecs on he overall volailiy of urope s inflaion and oupu. 5

6 1 Inroducion he inernaional economic environmen has been dramaically ransformed in recen years by increasing global financial inegraion and by he rapid growh of several Asian economies. Boh evens profoundly affec macroeconomic inerdependence. For example as China evaluaes wheher o coninue having a fixed exchange rae agains he dollar or o allow he renminbi o floa many commenaors have asked wha he consequences of his decision will be for he oher counries in he world economy. In oher words boh economic commenaors and policymakers recognise ha he configuraion of he inernaional moneary sysem (which is deermined by exchange rae arrangemens) affecs he inernaional ransmission of shocks. A he broad level of simplificaion one could characerise he inernaional moneary sysem as consising of hree large blocs: he euro he US dollar and a bloc of Asian currencies ha have been fixed (o varying degrees) agains he dollar. 1 Given he size of he Asian bloc he choice of exchange rae regime in Asia may affec he ransmission of shocks beween he US and he euro bloc. his is because any US shock riggers an immediae reacion from he Asian cenral banks if he laer are pegging heir currencies 2 whereas if heir exchange raes are floaing hen Asian cenral banks do no necessarily need o ake acion. he purpose of his paper is o analyse how urope s oupu and inflaion respond o shocks originaing in he US and compare wo cases: (1) when Asia pegs her currency o he dollar and (2) when he Asian currency freely floas. We choose o invesigae he ransmission of US shocks o urope because he empirical lieraure has found ha urope is more affeced by US shocks han vice versa (Aris Osborn and erez 2004). o conduc our analysis we use a hree-counry model ha parallels he Cenre-eriphery model of Corsei eseni Roubini and ille (2000) which we exend by adding a non-raded goods secor in each counry by inroducing dynamics and by adding exogenous supply and demand shocks. A leas since Friedman s (1953) pioneering conribuion 3 here has been a vigorous debae on he impac of exchange rae regimes on he inernaional ransmission of shocks. We aim o conribue o his debae from an enirely differen perspecive. Raher han analysing how he choice of exchange rae regime in one counry affecs is exposure o foreign shocks we analyse how he ransmission of shocks beween wo counries is affeced by he choice of he exchange rae regime in a hird counry. 1 Mervyn King he Inernaional Moneary Sysem remarks a Advancing nerprise 2005 conference London. Available a hp:// 2 In order o defend a peg o he dollar Asian cenral banks have o move heir ineres raes in line wih prevailing ineres raes in he US. 3 Friedman was he firs o argue ha flexible exchange raes insulae a counry from foreign shocks. 6

7 o our knowledge here are no oher papers ha analyse his issue direcly alhough some sudies have in a more indirec manner looked ino i. For example using a hree-counry model Obsfeld and Rogoff (2005) show ha in he even of an unspecified shock ha balances he US curren accoun he euro area effecive (rade-weighed) real exchange rae would appreciae more if Asia kep her exchange rae pegged o he dollar han if Asia decided o floa. Our aim is o expand heir analysis in a number of ways: (1) by using a dynamic seing (2) by specifying he naure of he shocks and (3) by looking a he effecs of he regime choice on oher variables. he hree-counry model in his paper has imperfecly subsiuable consumpion goods a raded and a non-raded goods secor in each counry nominal rigidiies and a moneary policy auhoriy. We consider only supply (produciviy) and demand (consumpion preference) shocks originaing in he US. Using boh impulse responses and variance decomposiions we examine he ransmission of hese shocks o urope under wo scenarios: when Asia pegs her currency o he dollar and when he Asian currency freely floas. Our main resul is as follows. If Asia keeps her exchange rae pegged o he dollar boh oupu and inflaion in urope respond more o US shocks. his happens because in he absence of nominal exchange rae adusmens Asian firms reac o he US shocks by implemening large price changes ha srongly affec urope s relaive compeiive posiion. Imporanly he impulse responses show ha he adusmen of Asian prices is especially pronounced afer a shock occurring in he US non-raded goods secor. his paricular finding is very robus o changes in various parameer values. Asia s adopion of a fixed peg vis-à-vis he dollar has wo imporan consequences for he ransmission of US shocks o urope. Firsly if he Asian currency is pegged o he dollar he nominal ineres rae in Asia mus move in line wih he US nominal ineres rae. Secondly US shocks do no only affec he euro-dollar bilaeral exchange rae bu hey also affec he exchange rae of he euro vis-à-vis he Asian currency. However our general equilibrium model shows ha he large adusmen in Asian prices iniially prevails over hese wo effecs. he res of he paper is organised in he following way. Secion 2 presens he model while Secion 3 discusses he calibraion of he parameer values. Secion 4 analyses he ransmission of each considered US shock o urope and Secion 5 examines he impac of Asia s choice of exchange rae regime on volailiies of urope s oupu and inflaion and is exposure o US shocks. Secion 6 conducs some sensiiviy or robusness checks. Finally Secion 7 concludes. 7

8 2 he Model o conduc our analysis we use a dynamic sochasic general equilibrium model wih nominal rigidiies. I consiss of hree counries or regions namely urope () he US (U) and Asia (A). We assume ha urope has size γ γ he US has size γ - γ γ and Asia has size 1 - γ. 4 he agens in he economy are households firms and governmens. Households in each counry consume a variey of differeniaed goods each produced by a single monopolisic firm. ach firm produces only one good which can be eiher raded or non-raded. In each counry a coninuum of raded and a coninuum of non-raded varieies are produced. he size 5 of he raded and he non-raded goods secors in each counry is consan. Goods are imperfec subsiues in consumpion. Households in each counry firs allocae heir expendiure beween raded and non-raded goods and hen hey allocae heir expendiure beween raded goods of differen counries. Finally households allocae heir expendiure among all he raded and he non-raded individual goods varieies. Nominal rigidiies are modelled by assuming ha firms pay a nonlinear adusmen cos whenever hey change heir prices as in Ireland (2001); moreover prices are sicky in he producer s currency. 6 We consider only supply (produciviy) and demand shocks Households Households derive uiliy from an index of consumpion goods (C) real money balances (M/) and leisure (1 - H). he represenaive household in counry ( = U or A) has he following lifeime uiliy funcion: U 0 = = 0 β ω 1 ξ 1 ξ C M ( C ) χ M ( 1 H ) 1 ξ C 1 ξ M ϕ 1 ξ where β is he discoun facor and ω is a counry-specific exogenous sochasic process ha affecs consumpion preferences. We inerpre a shock o his process as a demand shock. 8 he parameers 1 ξ H H 4 he counries are named only for exposiional purposes. Furher deails are available in he calibraion secion. 5 he size of a secor is he measure of goods produced or firms operaing in ha secor. 6 hus we do no consider he case of local currency pricing. he price of any counry s raded good is idenical in all counries (he law of one price holds) bu he non-raded goods secor and he home bias in preferences imply ha purchasing power pariy does no hold. 7 Resricing he number of shocks cerainly faciliaes he analysis bu on he oher hand i may go a he expense of realism. For his reason we decide o keep a limied bu sufficienly represenaive menu of shocks by assuming ha uncerainy can eiher originae from he supply or he demand side of he economy. 8

9 9 C ξ M ξ H ξ χ and ϕ are posiive and assumed o be equal in all counries. We also assume ha labour can freely move across secors and households can conemporaneously supply heir labour o he raded and non-raded goods secors. oal hours worked H are given by he sum of hours worked in boh he raded () and non-raded (N) goods secors: N H H H = where ( ) = S dz z H H and ( ) = S N N N N N dz z H H are aggregaes across firms in each secor. 9 Households can smooh heir consumpion by buying or selling he domesic and he foreign non-saeconingen nominal bonds. hey can also inves in money holdings (M) bu no in physical capial. We also inroduce a small quadraic cos of holding bonds 10 which ensures saionariy of he model and a well-defined seady sae. 11 he budge consrain of he represenaive household in counry a dae is: ( ) ( ) ( ) N A A U U U A A U U U U R W H M B i B i B i C M B B B B B B Π Π = = ε ε ε ε ε ε η ε ε ε η ε ε ε ε η ε ε where B U B and A B denoe he holdings of counry U and A s nominal bonds (all denominaed in he issuer s currency) which pay a rae of ineres i U i and A i respecively. Bonds are in zero ne supply worldwide. he parameer η governs he size of he cos of holding bonds. he variable is he aggregae price level in counry and ε is he price of counry s currency in erms of counry A s currency (herefore 1 = A ε ). Lump-sum governmen ransfers are denoed wih W is he hourly 8 Neiss and Nelson (2003) have demonsraed ha his disurbance o consumpion preferences is he equivalen of a real demand or IS shock. 9 he se of firms operaing in he raded (non-raded) goods secor in counry is denoed wih S ( N S ). he variables z and N z are indices for firms. 10 Our model allows for a non-zero exogenous seady-sae holding of bonds. herefore his is a cos of holding a volume of bonds differen o he seady sae volume. Bu as he bond porfolio choice is exogenous in order o keep he model parsimonious we assume ha bond holdings are zero in he seady sae. 11 his poin has been demonsraed by Schmi-Grohe and Uribe (2003).

10 wage rae (equal in boh secors) secors and Π and R are lump-sum rebaes. 12 Π denoe real profis from he ownership of firms in he wo N Consumers in each counry have preferences over domesic and foreign raded goods and also nonraded goods. references over goods are described by CS aggregaors. he consumpion index C in each counry is given by: C = α 1 µ µ 1 1 ( ) µ C ( 1 α ) µ ( C ) N µ 1 µ µ µ 1 where C and C are consumpion sub-indices ha represen respecively he oal consumpion of N raded and non-raded goods and α is he share of raded in oal goods. Households in each counry firs allocae heir expendiure on raded goods beween counry A s goods and he bundle formed by and U goods which ogeher make up he periphery () 13 hen beween and U raded goods: 1 ρ 1 1 ( ) ( ) ρ A γ ρ [( γ ) ] ρ C h ( C ) ρ 1 = 1 ρ A C A for = U : h = 1 C = 1 ψ 1 [( ) ] ψ γ h ( C ) ψ ( γ ) 1 ρ ρ 1 1 U [ ] ψ h ( C ) U ψ 1 ψ ψ ψ 1 U for = U A : h = 1. For = A : h = 1 he elasiciies of subsiuion µ (beween and N goods) ρ (beween A and raded goods) and ψ (beween and U raded goods) and he weighs α γ and γ are he same for all counries. he parameers h h U and h A are free and allow us o inroduce home bias in consumpion. 14 he consumpion sub-indices for he individual varieies are given by similar CS aggregaors wih he same elasiciy of subsiuion θ for he raded and non-raded goods secors in all counries. A higher θ 12 Households pay he cos of holding bonds o financial inermediaries locaed in heir own counry and firms pay he cos of adusing heir prices o he issuers of new menus. We assume ha financial inermediaries and he issuers of menus are locaed in each one of he hree counries and are owned by households herefore we add o he budge consrain an allinclusive lump-sum rebae or ren R. 13 herefore he degree of subsiuabiliy beween uropean and US goods ψ can be differen from he degree of subsiuabiliy beween Asia s goods and he bundle formed by uropean and US goods ρ. 14 he assumpion ha he raded goods produced by urope and he US form a disinc consumpion bundle separae from Asian goods is made in order o check wheher our resuls can be alered by a higher elasiciy of subsiuion in 10

11 implies higher subsiuabiliy among individual varieies. herefore θ is inversely relaed o he degree of monopolisic compeiion in he goods marke. Le N and denoe he price indices ha mach up he consumpion U A aggregaors illusraed above. he price indices are defined as he minimal expendiures needed o buy one uni of corresponding consumpion bundles hey are obained in he sandard way from he household s preferences. 2.2 Firms ach firm is a monopolisic supplier of a single differeniaed produc which is produced using labour as he only inpu. he firms producion funcion is affeced by an exogenous produciviy shock common o all firms in he same secor. Since we absrac from physical capial and regard i as fixed in he shor run we assume ha he marginal produciviy of labour is decreasing. 15 he producion funcions of he individual firms goods secor in counry a dae are given by: z and z operaing in he raded and non-raded N ~ ( z ) = A [ H ( z )] ζ Y ~ ( z ) = A [ H ( z )] ζ Y N N N N ~ ~ where ( z ) and ( z ) Y N Y N N denoe he oupu produced by he individual firms and A and A are secor-specific exogenous processes ha affec he produciviy of labour. We inerpre he shocks o hese processes as supply shocks. he parameer ζ measures he elasiciy of oupu wih respec o labour and we assume i o be equal across secors and across counries. Since labour is compleely flexible beween secors raded and non-raded secor firms pay he same hourly real wage rae W. N rices adus sluggishly o changes in demand. In paricular we assume ha firms have o pay a nonlinear adusmen cos whenever hey wan o change heir prices. For example a firm z operaing in urope s raded goods secor will choose he price of her own differeniaed good by maximising profi: consumpion beween uropean and US goods which could be usified by he similariy of he indusrial srucure ( new goods in boh urope and he US maure goods in Asia). 15 As observed by resco (1986): he capial sock varies lile cyclically and is nearly uncorrelaed wih oupu. A common assumpion in he new open economy lieraure is consan marginal labour produciviy and consan reurns o scale bu he assumpion of imperfec compeiion also makes i possible o allow for decreasing marginal labour produciviy. 11

12 12 ( ) ( ) ( ) ( ) [ ] i i i i i i i i i W z H z SR C C C C = 0 / / ξ ξ ω ω β for every [ ) 0. he variable ( ) z SR denoes he sales revenues ne of he adusmen cos and ( ) z Y ~ he demand for he good produced by firm z : ( ) ( ) [ ] ( ) ( ) z Y z z z SR ~ 1 Γ = ( ) ( ) ( ) ( ) [ ] A U C C C z z Y 1 1 ~ γ γ γ γ γ γ γ αγ θ = rice-adusmen coss incurred by firm z are a funcion of he change in her own price level: 16 ( ) ( ) ( ) = Γ z z z π φ where π >1 denoes he gross seady-sae rae of inflaion in counry. Firms maximise he expeced discouned profis subec o heir producion funcion and he demand for heir own differeniaed good. Non-raded goods secor firms and firms locaed in counry U and A solve a similar maximisaion problem. 2.3 Governmens In all counries he governmen represens boh he fiscal and moneary auhoriy. Seigniorage revenues are redisribued o he households in he form of ne ransfers and in each counry he governmen runs a balanced budge in every period: M M = 1 We assume ha moneary policy follows an ineres rae feedback rule specified in erms of curren economic condiions and ha he cenral bank arges only he inflaion rae. hus in each counry he nominal ineres rae is se according o he following equaion: Our specificaion of he adusmen cos funcion is close o Ireland (2001) excep for he absence of a componen ha depends on he change in he firm s inflaion rae. We have chosen o neglec ha componen because Ireland s esimaes indicae ha he daa prefer he specificaion in which adusmen coss apply only o changes in he price level bu no o changes in he inflaion rae.

13 ρi ( 1 i ) = ( 1 i ) 1 ( 1 i ) π π ρπ 1 ρi where ρ and i ρ are posiive parameers π i is he seady-sae nominal ineres rae and π is counry s gross inflaion rae a dae. We absrac from he specificaion of an opimal rule and se he inflaion arge equal o is seady-sae gross rae π. When Asia pegs her currency o he dollar her moneary policy rule is simply replaced by he requiremen o mainain he peg. 3 Calibraion We conduc our analysis using a calibraed version of he model choosing he srucural parameers from he lieraure wih he aim of approximaing he salien feaures of he US uropean and Asian economies. Our main reference for he calibraed parameer values is Obsfeld and Rogoff s hreecounry model (2005) bu we also conduc a sensiiviy analysis in Secion 6. All he calibraed parameer values are symmeric across counries and repored in able 1. able 1: arameer values references β = 0.99 ξ C = 5 ξ H = 2.33 φ = Relaive sizes γ = 1/3 γ = 2/3 Consumpion indices α = 0.36 µ = 1 h = 4.67 h U = 4.67 h A = 4.67 ρ = 2 ψ = 2 θ = 7.88 roducion ζ = 0.8 Adusmen coss φ = 77 η = Ineres rae rule ρ i = 0.81 ρ π = 1.52 xogenous processes: AR coefficiens Sd deviaions: urope Sd deviaions: US ρ = 0.81 σ = U σ = 0.01 ρ N = 0.81 ρ ω = 0.33 σ N = U σ N = 0.01 σ ω = 0.03 U σ ω = 0.02 Because of he uncerainy abou he properies of he shocks and he values for he parameers ha appropriaely characerise urope US and Asia we prefer o make only minimal resricions. In Secion 6 we check wheher our main resuls are robus o changes in parameer values. In his way alhough 17 We have chosen his funcional form so as o reproduce a sandard linear feedback rule (wih smooh ineres rae adusmen) in boh a linear and a log-linear approximaed soluion of he model. 13

14 our research is largely moivaed by China s presen dilemma over fixed or flexible exchange raes we are also able o assess wheher a hird counry s choice of exchange rae regime can affec he ransmission of shocks beween any wo oher counries in general erms. Since he model canno be solved in closed-form we obain our resuls by means of a numerical approximaed soluion. In paricular we find he seady sae of he nonlinear model using Mahemaica 18 and hen we use DYNAR o compue he firs-order approximae soluion of he model he impulse response funcions he second-order momens and he variance decomposiions. 19 he discoun facor β is se so as o yield an annual real ineres rae of four percen in he seady sae and he risk aversion parameer ξ is aken from Chari Kehoe and McGraan (2002). he parameer φ C is calibraed so ha in he seady sae he fracion of ime ha households in all counries or regions spend working is hen given his choice of average hours worked in he seady sae he parameer ξ H is calibraed so ha in he seady sae he Frisch elasiciy of labour supply is equal o 1 he value suggesed by Galí Gerler and Lopez-Salido (2005). he ineres rae feedback rule makes he money demand equaion redundan (money demand aduss auomaically o money supply) herefore he parameers χ and ξ M need no be calibraed. he parameers γ and γ govern he size of he hree regions in he model wih γ denoing he size of urope and γ denoing he combined size of urope and he US. In paricular boh γ and γ simulaneously accomplish several roles wihin he model since hey give he measure of households firms and produc varieies in each region in addiion o being he weighs of he consumpion indices for raded goods. herefore γ and γ could be calibraed using populaion daa GD daa or rade daa leading o very differen calibraions. 20 Because of his ambiguiy we decide o ake a neural or inermediae sance and we assume ha all regions have he same size and ha all oher maor parameers are symmeric across all counries bu we conduc sensiiviy analysis wih respec o γ in Secion 6. his approach has an addiional benefi keeping he maor parameers same across all counries allows clearer undersanding of he ransmission mechanism because following he US shock he only difference beween urope s and Asia s reacion is due o heir differen choice of he exchange rae regime. he criical parameers in he model are µ he elasiciy of subsiuion in consumpion beween raded and non-raded goods ψ he elasiciy of subsiuion beween uropean and US raded goods and ρ Wolfram Research Inc. 19 Deails on solving non-linear models using DYNAR can be found for example in Juillard (2001). 20 For example in erms of populaion China is approximaely 4 imes as large as eiher urope or he US bu only 0.14 imes as large in erms of GD. 14

15 he elasiciy of subsiuion beween Asian goods and he bundle formed by uropean and US raded goods. he values for hese elasiciies are all aken from Obsfeld and Rogoff (2005). 21 he home bias parameers h h U and h A are chosen so as ge close o he values suggesed by Obsfeld and Rogoff 22 given our choice of counry weighs γ and γ. he share of raded goods α is based upon he esimaes of Spange and Zabczyk (2006) and he parameer θ is aken from Roemberg and Woodford (1998). However since he elasiciies of subsiuion he degree of home bias and he share of raded goods in oal consumpion are crucial for he ransmission of shocks across counries we conduc exensive sensiiviy analysis wih respec o hese parameers in Secion 6. he elasiciy of oupu wih respec o hours ζ is calibraed so ha given he mark-up 23 in he seady sae he share of wages in he value of oupu is equal o 0.7 a value ha is aken from Roemberg and Woodford (1999). Our choice of he price-adusmen cos parameer φ is based upon he esimaes of Ireland (2001) while η he parameer ha governs he size of he bond-adusmen coss is aken from Ghironi and Meliz (2005). he coefficiens of he ineres rae feedback rule ρ and i ρ are se in accordance wih he IMF π calculaions. 24 We oo assume ha hese coefficiens are he same across counries or regions and in Secion 6 we conduc some sensiiviy checks. We decide o adop an agnosic approach and assume ha he exogenous processes ha affec demand and produciviy are he same across counries. hey are all assumed o follow saionary AR(1) processes wih zero mean and he shocks o hese processes are uncorrelaed. Our research quesion is resriced o be he analysis of how he ransmission of US shocks o urope is affeced by he choice of he exchange rae regime in Asia. herefore for he sake of clariy we decide o absrac from all shocks originaing in Asia. he auoregressive parameers are aken from Benigno and hoenissen (2003) and Neiss and Nelson (2003). We normalise he sandard deviaions of he wo produciviy shocks as being half he sandard deviaion of he demand shock so as o avoid giving more imporance o eiher demand or supply shocks. Finally we choose he sandard deviaions of he wo demand shocks so as o approximae he sandard deviaion of oupu and inflaion in urope and he US. 21 Since he possibiliy of modelling he subsiuabiliy beween uropean and US goods separaely from he subsiuabiliy of Asian goods is missing in Obsfeld and Rogoff s hree-counry model we se ψ equal o ρ in our benchmark. 22 Obsfeld and Rogoff (2005) assume ha US uropean and Asian households aach a weigh of 0.7 o heir own domesic producs. Moreover urope and he US place a weigh of 0.1 on each oher s raded goods and a weigh of 0.2 on Asian goods. On he oher hand Asia places a weigh of 0.15 on boh urope and US raded goods. 15

16 4 he ransmission of US shocks o urope In his secion we explain how secor-specific produciviy shocks and demand shocks originaing in he US are ransmied o urope s oupu and inflaion and how he ransmission is affeced by Asia s choice of exchange rae regime. For each shock we firs describe he inernaional ransmission mechanism when all hree currencies freely floa agains each oher a siuaion ha we call he floaing scenario. hen we examine how he inernaional ransmission mechanism changes when we allow Asia o peg her currency o he dollar a siuaion ha we call he hybrid scenario because differen exchange rae regimes coexis. 25 he impulse responses for he main macroeconomic variables described in he previous secions are presened in he appendices. We also presen he responses of he hree bilaeral real exchange raes and he hree bilaeral erms of rade which are defined as follows: ' RR ε ' ' ε o ' ε ' ' ' ε For example a rise in RR is an increase in he price of urope s consumpion baske relaive o he U US consumpion baske herefore we say ha urope s real exchange rae appreciaes and he US real U exchange rae depreciaes. A rise in o is an increase in he price of urope s raded goods relaive o US raded goods herefore we say ha urope s erms of rade improves and he US erms of rade deerioraes. Since ours is a general equilibrium model all he variables are inerrelaed and he channels ha allow he shocks o have an effec on he world economy can be numerous. However as far as he ransmission across counries is concerned our analysis is grealy faciliaed by he recogniion ha in our model here are only wo sources of inernaional linkages namely rade and he asse marke. he inernaional ransmission hrough rade can be summarized as follows. In addiion o having oher direc effecs in he counry of origin a given shock will change he demand or supply of ha counry s raded oupu. Since in our model prices are se in he producers currency and prices are cosly o adus hese changes will affec he demand for ha counry s currency. hus he nominal exchange rae and he compeiiveness of he oher counries raded goods will be affeced by he shock. As a resul oupu in he oher wo counries changes as well. 23 Our choice of θ implies a seady sae mark-up of prices over marginal coss of approximaely 14%. 24 See Globalizaion and xernal Imbalances Chaper 3 World conomic Oulook April One could hink of a hird scenario where Asia pegs o a baske of currencies ha include boh dollar and euro wih equal weighs he Baske scenario. his scenario is simply an average of he wo examined scenarios. 16

17 Furhermore a shock originaing in one counry will affec her desired ineremporal allocaion of consumpion and herefore her demand or supply of he inernaionally raded bonds. For example a counry ha wans o increase fuure consumpion will sar lending and hereby have a higher demand for bonds. Since bonds are in zero ne supply worldwide he adusmen owards equilibrium in he asse marke will require changes in he cos of borrowing and bond holdings in he oher wo counries which in urn will affec heir ineremporal allocaion of consumpion. We recover his mechanism by examining he changes in he real ineres raes which we define as follows: 26 r 1 i B 1 η 1 1 he above definiion of he real ineres rae is useful because consumpion growh in counry is posiive (negaive) whenever r is posiive (negaive). 27 hus changes in r capure he inernaional ransmission of he shock hrough he asse marke. For he purpose of clariy we can summarise rade and asse marke channels by an example ha explains how shocks in he US affec aggregae demand for urope s oupu. Firs any shock ha raises US consumpion increases US demand for worldwide and hus urope s oupu aggregae world demand effec. he aggregae demand for urope s oupu is furher affeced by he consequen reacion of consumpion in urope and consumpion in Asia. Second aggregae demand for urope s oupu depends no only on aggregae world consumpion bu also on allocaion of consumpion across counries because households are biased owards consuming domesic producs. Hence a uni increase in consumpion in Asia has a differen effec on demand for urope s oupu han a uni increase in consumpion in urope. hird he allocaion of consumpion over ime (consumpion smoohing) also affecs he demand for urope s oupu by changing is paern over ime and hus by changing he volailiy of demand. he srengh of many of hese channels depends on he choice of he exchange rae regime. hroughou he analysis we describe he ransmission of each shock originaing in he US by means of a simple sep-by-sep approach. Firs we describe is direc effecs on he US economy. hen we examine is consequences for rade and for he asse marke and from hose consequences we explain why we ge he observed responses of oupu and inflaion in urope. he main ineres of his analysis 26 hus for each counry we consruc a unique measure of he real ineres rae obained from he nominal rae on he domesically-issued bond. Since all bonds are perfec subsiues in equilibrium here are no arbirage opporuniies. 27 he only siuaion when his rule does no hold is when counry is hi by a demand shock. 17

18 lies in he inerpreaion of he resuls presened in Secion 5: wihou an undersanding of he ransmission mechanism we would no be able o explain and deermine he limis of our findings. 4.1 osiive produciviy shock in he US raded goods secor scenario 28 he posiive produciviy shock in he US raded goods secor direcly affecs he US economy. o begin wih all he firms operaing in he raded secor enoy higher produciviy of labour and herefore lower real marginal coss. Under monopolisic compeiion and sicky prices prices are a funcion of expeced fuure marginal coss. herefore he posiive produciviy shock which reduces marginal coss causes a fall in he US radable price index which is also accompanied by an increase in raded oupu. Because of he increase in raded oupu he US aggregae income or GD increases. Lower prices and higher income simulae households demand herefore consumpion increases in he US. Secondly US firms in he raded secor reac o he increase in produciviy by shrinking heir workforce. he resuling fall in labour demand causes a fall in wages herefore marginal coss fall also for he firms operaing in he non-raded secor because he model assumes here is a common wage rae across he raded and non-raded secors. As a resul prices fall also in he non-raded secor. Since prices of boh non-raded and raded goods decrease US CI inflaion falls. he US moneary auhoriy reacs by cuing he nominal ineres rae hus US consumpion demand receives a furher boos from moneary policy. Moreover he shock is ransmied o he oher counries hrough rade and he asse marke. he higher US consumpion will push up demand for all goods and hus increase inernaional rade. Bu since US firms now sell a lower prices inernaional rade is furher affeced; in paricular world demand now swiches owards he cheaper US raded goods. As a resul he demand for dollars increases causing a nominal depreciaion of he oher wo currencies. he appreciaion of he dollar 29 hen causes a decline in he US price of impors and adds o he fall of CI inflaion in he US. he asse marke is also affeced by he raded produciviy shock. In order o spread he benefi (in erms of higher consumpion) of he shock ino he fuure US households lend money abroad and he US curren accoun becomes posiive. Real ineres raes fall in he oher wo counries. 30 Low real 28 Impulse responses are presened in Appendix A. 29 However he acual behaviour of he dollar exchange rae also depends on he magniude of he response of he US moneary policy. In our benchmark moneary policy acually causes an iniial and shor-lived depreciaion of he dollar bu US domesic prices sill fall. 30 his is caused by he higher supply of money coming from he US. here are however very shor-lived iniial increases in urope and Asia s real ineres raes because of falling inflaion raes. 18

19 ineres raes induce he foreign households o fund heir curren consumpion by borrowing from he US. As a resul consumpion in boh urope and Asia increases. he response of uropean oupu depends on he relaive srengh of several facors. 31 On he one hand he worldwide increase in consumpion and he depreciaion of he euro simulaes producion. On he oher hand world demand swiches owards he cheaper US raded goods. We can see from Appendix A ha he laer effec prevails since oupu in urope falls. 32 I is also imporan o recall ha under he floaing scenario urope and Asia are perfecly symmeric. herefore he posiive produciviy shock in he US raded secor affecs hem in he same way. Finally under monopolisic compeiion and sicky prices domesically-se prices depend only on curren and fuure expeced marginal coss which in our se-up depend on boh wages and he level of oupu. 33 Afer a posiive produciviy shock in he US raded secor marginal coss decrease in boh urope and Asia. his plus he fall in he US raded prices explains he decrease in domesic (and CI) inflaion raes in boh counries which in urn induces heir cenral banks o cu he nominal ineres raes. However since urope and Asia do no benefi from he direc effec of he produciviy shock on domesic prices heir fall in inflaion raes and he subsequen expansionary responses from he moneary auhoriies are milder han in he US. scenario Under he hybrid scenario he direc effecs on he US economy of he posiive produciviy shock in he US raded goods secor are he same as before bu he inernaional ransmission changes. Now in order o preserve he fixed exchange rae agains he dollar he nominal ineres rae in Asia mus fall o he same level as he US nominal ineres rae. Alhough in principle his could have implicaions for he ransmission of he shock hrough he asse marke we can see from he impulse responses ha bond holdings in he hree counries are affeced in a very similar way under he wo scenarios; moreover he responses of consumpion in urope and he US are also very similar. herefore we can infer ha in he case of his shock Asia s choice of exchange rae regime does no have a maor influence on he ineremporal allocaion of consumpion in he oher wo counries and ha in pracice he ransmission of he shock hrough he asse marke is no significanly affeced. 31 As he maximisaion problem of he firm illusraes for he special case of raded goods he facors ha affec oupu demand are he relaive prices and he domesic and foreign consumpion of uropean goods. 32 A laer sages oupu increases and consumpion decreases in boh urope and Asia so as o generae enough resources o repay heir borrowing. 33 In our se-up he level of oupu affecs marginal coss in wo ways. Firsly he marginal produciviy of labour is decreasing in oupu. Secondly higher oupu iself implies higher labour demand and herefore higher wages. 19

20 he source of he differences in he responses of uropean oupu and inflaion under he hybrid scenario are he inernaional rade flows. Because he Asian currency canno depreciae vis-à-vis he dollar i mus appreciae vis-à-vis he uropean currency. Naurally hese nominal exchange rae effecs influence he ransmission of he shock hrough inernaional rade. A firs we would expec an increase in he compeiiveness of uropean raded goods in comparison wih Asian raded goods and an increase in urope s oupu a he expense of Asian oupu. However compeiiveness and inernaional rade flows are no only affeced by nominal exchange raes bu also by relaive prices. As we can see from he impulse responses Asian raded goods inflaion is more negaive under he hybrid han under he floaing scenario. his occurs because in he absence of any nominal exchange rae adusmen Asian prices mus fall in order for Asia o remain compeiive. In his case prices ake all he brun of adusmen. he marked fall of Asian prices helps o sabilise he US/Asia erms of rade and allows Asian firms o say compeiive vis-à-vis he US in he face of falling US prices and he fixed exchange rae. On he oher hand he marked fall of he Asian raded goods prices also works in he direcion of increasing he price of uropean goods relaive o Asian goods hus eroding he compeiive gain induced by he depreciaion of he euro owards he Asian currency. Indeed Appendix A shows ha iniially he urope/asia erms of rade increases raher han decreases. As a resul here is an expendiure swiching effec away from uropean goods o Asian goods which causes uropean oupu o fall iniially more under he hybrid han under he floaing scenario. Since oupu falls more fuure expeced marginal coss are lower a leas iniially hus explaining why domesic (and CI) inflaion in urope falls more under he hybrid han under he floaing scenario. Neverheless afer several periods he larger effecive euro depreciaion under he hybrid scenario increases he compeiive posiion of urope s raded goods and his effec dominaes in erms of he overall impac on urope s oupu. Overall he large adusmen of Asian prices under he hybrid scenario iniially increases oupu in Asia and reduces oupu in urope. Bu his effec is shor-lived. Afer several periods a higher effecive euro depreciaion increases he compeiive posiion of he uropean raded goods secor hus increasing is expors and rendering uropean oupu under he hybrid scenario higher han ha under he fixed scenario. Neverheless Appendix A shows ha in our calibraion he quaniaive difference beween he oupu and inflaion reacions in he wo scenarios is iny. 4.2 osiive produciviy shock in he US non-raded goods secor For he sake of concision we prefer o analyse he ransmission of he non-raded produciviy shock mainly by explaining how i differs from he ransmission of he raded produciviy shock. 20

21 scenario 34 he direc effec on he US of he posiive produciviy shock in he US non-raded secor is very similar o ha of he raded produciviy shock. roducion increases and real marginal coss fall riggering a reducion in he price of non-raded goods. Since he US economy now produces more oupu a cheaper prices domesic consumpion increases. US demand gradually swiches o he relaively cheaper nonraded goods hus causing a fall in expeced fuure raded oupu and real marginal coss in he raded goods secor. As a resul US raded goods prices fall oo. he fall in domesic (and as a resul CI) inflaion promps he US cenral bank o lower he nominal ineres rae. As in he case of he raded secor produciviy shock he non-raded produciviy shock is ransmied abroad hrough he asse marke and rade. In order o spread he benefi of increased produciviy and enoy higher consumpion in he fuure US households sar holding a posiive amoun of all bonds while he oher wo counries sar borrowing from he US. As a resul consumpion and real ineres raes in boh urope and Asia increase. Moreover since prices in he US raded secor are falling here is more demand for US goods and he dollar appreciaes (excep for a small and shor-lived iniial depreciaion) as i happened afer he raded produciviy shock. However he non-raded produciviy shock does no have a direc effec on raded prices herefore he fall in US raded secor inflaion is significanly lower han under he raded produciviy shock bu i is more persisen. As a resul he appreciaion of he dollar is sronger under he non-raded produciviy shock han under he raded produciviy shock. Nowihsanding he increase in domesic consumpion and he depreciaion of he euro uropean oupu falls since demand swiches owards he cheaper US raded goods. 35 However uropean oupu recovers in subsequen periods. Because of he decrease in wages and oupu curren and fuure expeced marginal coss fall herefore domesic (and CI) inflaion in urope falls iniially. he fall in inflaion is shor-lived because he depreciaion of he euro increases he cos of impors. Moreover as urope s oupu levels sar o grow so do her marginal coss and herefore prices. 34 Impulse responses are presened in Appendix B. 35 uropean raded oupu falls because worldwide demand gradually swiches o he relaively cheaper US raded goods. Moreover he fall in US raded prices causes a fall in he aggregae raded price index. As a resul uropean households consume less non-raded goods and uropean non-raded oupu falls. 21

22 scenario In he previous secion we saw ha Asia s choice of he exchange rae regime did no significanly change Asia s bond holdings. his is no he case here. Following he non-raded produciviy shock Asia s borrowing is lower under he hybrid han under he floaing scenario and real ineres raes in boh urope and Asia fall more. As a resul urope s consumpion iniially increases more under he hybrid han under he floaing scenario. As i happened in he case of he raded produciviy shock he main driver of he differences in he responses of uropean oupu and inflaion under he hybrid scenario is he lack of a depreciaion of he Asian currency vis-à-vis he dollar and he marked fall in Asian prices. 36 From Appendix B we can noice ha in he hybrid scenario he fall in Asian prices and inflaion raes (boh raded and non-raded) is quie dramaic excep for a small and very shor-lived iniial increase. As before his allows Asian raded secor firms o say compeiive vis-à-vis US firms. Iniially he marked fall in Asian prices induces an expendiure swiching effec from uropean and US goods o Asian goods. As a resul urope s oupu mus fall below he level ha occurs under he floaing scenario. Moreover uropean wages also fall below he level ha occurs under he floaing scenario. As a resul expeced fuure marginal coss are comparaively lower and prices in urope fall more hus explaining why domesic (and CI) inflaion raes in urope iniially fall below he floaing scenario. Neverheless afer several periods urope s oupu benefis from he appreciaion of he Asian currency and hus uropean oupu and inflaion gradually rise above he floaing scenario. By and large he mechanism of he adusmen of inflaion and oupu in urope following a nonraded secor shock in he US is similar o ha caused by he raded secor shock. he large fall in Asian prices iniially reduces he compeiive posiion of urope s raded goods secor and hus her oupu and inflaion bu he laer is recovered afer several periods of greaer effecive euro depreciaion. In conras o he raded secor shock however he quaniaive difference beween he hybrid and floaing scenario is now much larger. 4.3 Negaive demand shock in he US In his secion we prefer o analyse he ransmission of a negaive raher han a posiive US demand shock since a negaive demand shock has he same consequences on he uropean economy (fall in 36 he fall in Asian prices is much more marked under he hybrid scenario because in absence of he nominal exchange rae reacion he whole adusmen mus come hrough prices. 22

23 oupu and inflaion) as he wo posiive produciviy shocks. Alhough less deailed han Secion 4.1 he explanaion offered here follows he same sep-by-sep approach. scenario 37 A negaive demand shock induces a decline of consumpion demand in he US and since oupu is demand-deermined in he model curren and expeced fuure oupu falls. xpeced fuure marginal coss which depend on boh wages and oupu are lower so prices fall oo. US moneary policy reacs by cuing he nominal ineres rae. he demand shock is ransmied o he oher counries hrough he asse marke and rade. Since he negaive demand shock as i is a preference shock reduces he araciveness of consumpion US households lend resources abroad. Real ineres raes fall in he oher wo counries and borrowing becomes more aracive o uropean and Asian households hus consumpion rises in he oher counries. Bu a he same ime he fall in US oupu reduces demand for uropean and Asian raded oupu. Inernaional rade is addiionally affeced since US firms now sell a lower prices. hus demand swiches owards he US raded goods and as a resul he dollar appreciaes alhough only mildly. As wih he oher shocks he response of uropean oupu depends on he sign and he relaive srengh of he changes in all he variables ha affec he demand for uropean oupu: relaive price and domesic and foreign consumpion demands. On he one hand boh he depreciaion of he euro and he increase in domesic consumpion simulae he demand for uropean goods. Bu on he oher hand he fall in boh US prices and US consumpion reduces he demand for uropean goods. I is possible o infer from Appendix C ha he laer wo facors dominae since urope s oupu falls alhough in he long-run i mildly increases. In urn falling oupu and lower wages imply lower curren and expeced fuure marginal coss herefore uropean prices fall and inflaion is negaive. However in subsequen periods inflaion becomes mildly posiive in all hree counries. scenario Asia s choice of exchange rae regime does no aler significanly he ransmission of he demand shock hrough he asse marke. his resul can be inferred similar o he case of he raded produciviy shock by noing ha he responses of bond holdings are lile differen under he wo scenarios. As 37 Impulse responses are presened in Appendix C. 23

24 before he main driver of he differences in he responses of uropean oupu and inflaion under he hybrid scenario is he lack of a depreciaion of he Asian currency vis-à-vis he dollar and he marked fall in Asian prices. As wih he wo produciviy shocks under he hybrid scenario he urope/asia erms of rade iniially increases so urope s compeiive posiion owards Asia deerioraes and urope s oupu iniially falls below he level ha occurs under he floaing scenario. Moreover because oupu and wages fall more curren and expeced fuure marginal coss are comparaively lower herefore domesic (and CI) inflaion raes in urope iniially fall below he level ha occurs under he floaing scenario. However euro depreciaion agains he Asian currency becomes a dominan effec afer several periods and urope s oupu and inflaion under he hybrid scenario increase above he floaing scenario. I is imporan o noe ha he differences in he responses of he uropean variables beween he wo scenarios are again iny under he demand shock. 5 he impac on volailiies and exposure o foreign shocks able 2 repors he sandard deviaions of he main macroeconomic variables in he model under he fixed and floaing scenarios. he sandard deviaions produced by he model are roughly consisen wih he daa excep for he real exchange rae. However he abiliy of sicky price models o reproduce he high volailiy of he real exchange raes found in he daa is a well-known problem in he lieraure. 38 Noably Asia s choice of exchange rae regime does no affec significanly he uropean business cycle since he volailiy of he main macroeconomic variables remains he same under boh scenarios. 39 In order o assess he overall impac of Asia s choice of exchange rae regime we look a he variance decomposiions of oupu and inflaion which are repored in able 3. he resuls show ha if Asia keeps he exchange rae pegged o he dollar urope s exposure o US shocks increases since he percenages of he variance of urope s oupu and inflaion ha are due o US shocks are higher under he hybrid scenario han under he floaing scenario. However in our calibraion US shocks accoun only for a small percenage of he variances of urope s oupu and inflaion and hence urope s overall volailiy is lile affeced Chari Kehoe and McGraan s (2002) model generaes real exchange raes ha are as volaile as in he daa however heir resul is influenced by heir assumpion ha prices are se in he local raher han he producer s currency. 39 We do no aim o measure precisely he quaniaive impac of Asia s choice of exchange rae regime on he oal volailiy of oupu and inflaion in urope. he maor focus of our analysis is on qualiaive raher han quaniaive issues. 40 Very low exposure of euro area oupu and inflaion o US shocks is also found in a recen empirical sudy by De Walque Smes and Wouers (2005). hey esimae ha US shocks accoun for only 0.2% of variance of euro area oupu and 1.2% of variance of euro area inflaion. 24