Instantaneous Correlation and Cyclical Convergence in the Euro Zone* Scott Davis a

Transcription

1 Insananeous Correlaion and Cyclical Convergence in he Euro Zone* Sco Davis a Absrac: This paper inroduces he measure of insananeous correlaion o sudy he effecs of a common currency on cyclical correlaion and convergence in he euro zone. Convenional, backwardlooking measures of correlaion are no appropriae for a dynamic currency union. An accurae measure of cyclical correlaion mus accoun for he change in economic srucures due o a common currency and he convergen effecs of hese changing srucures. In addiion o accuraely reporing cyclical correlaion, he measure of insananeous correlaion allows he convergen effecs of increased rade, demand spillovers and indusrial specializaion o be isolaed and measured hrough counerfacual siuaions. * - I would like o hank Andrew Hughes Halle, Mario Crucini, and William Huchinson for many helpful commens and suggesions, and Claire Larson for helping in he ediing process. a - Deparmen of Economics, Vanderbil Universiy, VU Saion B #351819, Nashville, TN , US Tel.: ; fax: ; address: jonahan.s.davis@vanderbil.edu. When he euro became he common European currency, he naional cenral banks of he euro zone ceded moneary independence o he European Cenral Bank. Each counry agreed ha a moneary policy should be adoped ha is bes for he enire euro zone, even if i is no bes for heir own individual economies. Since governmens use moneary policy o smooh business cycle flucuaions and sabilize economies, he loss of an independen moneary policy pus he counry a risk for real insabiliy in he shor and medium erm. This is he cos of a common currency. When he various naional governmens approved he euro hey believed he benefis of a common currency ouweighed he coss. These benefis and coss are difficul o quanify and impossible o compare, bu convergence owards a common European business cycle makes his comparison moo as he coss of a common currency end owards zero. Bu will hese coss end owards zero? Will he euro in fac induce cyclical convergence beween he euro zone 1

2 economies? A common currency will lead o increased rade and a common disribuion of moneary shocks, bu does his rade and disribuion of shocks lead o cyclical convergence? In his original work on he heory of opimum currency areas, Mundell (1961) emphasized he imporance of labor mobiliy in correcing for asymmeric shocks. Various reaies of he las 50 years have removed sauory barriers o he free movemen of labor across Europe, bu significan lingual and culural barriers sill remain. Because of hese barriers, labor mobiliy wihin Europe is limied. In 1999, a he adopion of he euro, only 1.5% of Europeans were aking advanage of he free movemen of labor enabled by he Treay of Rome (OECD Observer 1999), and ha number will no rise as he EU s lingual and culural diversiy increases in he coming years. Eichengreen, in 1991 and again wih Bayoumi, in 1993 found ha here is much less labor mobiliy in he euro zone han in comparable currency areas like he Unied Saes. Therefore if Europe is o funcion as an opimum currency area given he lack of labor mobiliy o correc for low business cycle correlaion, he correlaion beween euro zone business cycles needs o be high. The sudy of insananeous correlaion in he euro zone is he sudy of various shocks ha drive business cycle flucuaions, and he role of economic srucures in conrolling he manifesaion of hose shocks. Many scholars have heorized abou he pah of cyclical convergence in he euro zone. Many of hese heoreical works are conroversial and conradic one anoher, and he various heories and argumens are presened hroughou his paper. The exising empirical lieraure on he opic is jus as conradicory. Kalemli-Ozcan, Sørensen, and Yosha (2001) find evidence ha he divergence arising from indusrial specializaion is greaer han he cyclical convergence ha comes from he disribuion of common shocks hrough increased inernaional rade. Using differen economeric echniques, Alavilla (2004) finds he opposie. He saes ha here is 2

3 currenly some difference in he business cycle flucuaions beween he euro zone economies, bu wih ime he members of he EMU should see heir business cycles converge o a common European business cycle. Ohers have acceped ha in a 12 naion currency block like he euro zone, i is possible for cyclical convergence o exis beween some economies and cyclical divergence o exis beween ohers. This has lead o he core-periphery hypohesis ha is common in he lieraure. Bayoumi and Eichengreen (1993), von Hagen and Neumann (1994), and De Grauwe (1996) have all speculaed or esed empirically o find he exisence of a core of euro zone counries (including Germany, France, Belgium, The Neherlands, and Luxemburg) ha should cyclically converge wih he euro zone average in he fuure, and a periphery of counries (Ialy, Porugal, Spain, Greece, Ausria, Finland, and Ireland) which should see heir business cycles diverge from a euro zone average. This paper does wha he empirical papers on his subjec have no. Oher sudies have used various correlaion models ha are inherenly backward looking. A backward looking measure is no appropriae in a dynamic economic and moneary union like he euro zone. By assigning a sochasic funcional form o business cycle flucuaions, his paper will find he insananeous correlaion beween business cycle flucuaions. This insananeous correlaion is no a backward looking correlaion coefficien based on pas observaions, bu a measure based on he curren economic srucures of wo economies. Wih his model of insananeous correlaion, I am able o isolae he demand spillover and indusrial specializaion effecs of increased rade in various counerfacual siuaions. I can hen compare he converging properies of he wo. I find he resuls o be mixed and raher inuiive. Cyclical correlaion beween he euro zone average and he small, open, core economies of he euro zone is very suscepible o he effecs of demand spillovers. And cyclical 3

4 correlaion beween he euro zone average and he economies of he periphery is decreasing because of indusrial specializaion everywhere excep Ialy. These mixed resuls show ha rade in and of iself will no cause cyclical convergence, for some counries see he convergen impac of demand spillovers rump he poenially divergen impac of indusrial specializaion, bu some see he opposie. The model of insananeous correlaion and how i differs from convenional, backward looking correlaion measures is presened in he firs secion of his paper. In he second secion a sochasic funcional form is given o business cycle flucuaions, and wih his funcional form I esimae he variance of and covariance beween he various shocks o an economy. In he hird secion I examine he effec of increased rade on cyclical correlaion in Europe. I use counerfacual siuaions o isolae he effecs of demand spillovers and indusrial specializaion. Through hese 3 secions i becomes clear ha a common currency alone will no lead o higher cyclical correlaion or cyclical convergence in he euro zone, and may lead o lower cyclical correlaion and cyclical divergence. The euro alone will no ease and eliminae he coss of a common currency, and i may increase he coss as divergence pushes euro zone economies away from a common European business cycle. Secion 1: The Model of Insananeous Correlaion The correlaion coefficien is commonly gauged by measuring he covariance beween and sandard deviaions of lagged observaions. The formula for he correlaion coefficien beween wo variables, X and Y, is as follows: 4

5 T ( X µ )( Y µ ) i X i Y i= 1 ρ T, XY = (1) T T 2 2 ( X i µ X ) ( Y i µ Y ) i= 1 i= 1 Equaion 1 shows ha he measuremen of he correlaion coefficien will depend on pas observaions, in his case T observaions. For many applicaions his measure of he correlaion coefficien is appropriae, bu for some applicaions, especially measuring cyclical convergence in he euro zone, i is no. To analyze cyclical convergence in he euro zone i is necessary o measure he correlaion beween business cycle flucuaions ha is no based on lagged informaion, he insananeous correlaion. This is because, on average, he correlaion beween wo counries business cycles will depend on he inheren srucures of hose economies. Cyclical correlaion should be high beween economies wih similar indusrial srucures, similar degrees of rade openness, and similar moneary policies. Some scholars sudying he common currency mainain ha increased rade and demand spillovers will induce cyclical convergence, ye some mainain ha increased rade and indusrial specializaion could induce cyclical divergence. While hey may disagree on he exac pah of cyclical correlaion in he euro zone, hey agree ha he euro will fundamenally aler he economic srucures of he euro zone economies. Therefore any measure of he correlaion coefficien ha relies on lagged observaions will incorporae informaion from boh before and afer ha aleraion. An insananeous correlaion based only on curren economic srucures can more accuraely measure he cyclical correlaion and hus cyclical convergence/divergence in a currency union where rade openness and indusrial srucures are subjec o rapid change. 5

6 To measure he insananeous correlaion beween wo economies business cycle flucuaions i is necessary o characerize hose business cycle flucuaions as a funcion of various economic shocks. Equaion (2) wries counry A s oupu gap in period as a funcion of various shocks: Y Y * = S + v (2) where he scalar Y A is he acual oupu of counry A in period, and he scalar * Y is he poenial oupu of counry A a ime. S is he Nx1 vecor of shocks common o all counries, and he scalar v A, is a shock specific o counry A a ime. The acual manifesaion of a common shock in he business cycle flucuaions of counry A depends on he Nx1 vecor This vecor is boh counry and period specific.. is a vecor of parameers ha represen he srucure of counry A s economy a ime. The individual componens i, for some i = 1... N could be he size of indusrial secor i in counry A s economy a ime, or i could be a measure of how open counry A is o inernaional rade a ime. For cerain common shocks, could be a binary variable. If he shock i, S i, is he resul of a common union wide moneary policy, hen will ake a value of one if counry A is a member of ha moneary union a ime and zero i, if no. If one believes ha he counry specific shocks, v, accoun for only a small porion of he overall variance of business cycle flucuaions once he many common shocks, S, and heir counry-specific manifesaions, S, are accouned for, hen he model in equaion (2) can be used o characerize muliple counries business cycle flucuaions as a funcion of he same common shocks. Therefore i is possible algebraically o represen he insananeous correlaion 6

7 beween he business cycle flucuaions in counries A and B in ime as a funcion of he economic srucure vecors for j B and he variance-covariance marix relaed o he j, = vecor of common shocks S. Various saisical ideniies yield he following algebraic represenaion of he insananeous correlaion beween he business cycle flucuaions in counries A and B a ime T: T B, T ρ AB, T = (3) ( M )( M ) T where M is he NxN variance-covariance marix associaed wih he vecor of common shocks T M B, T B, T S. ( S S ) M = E (4) This equaion gives an accurae measure of cyclical correlaion based on curren economic srucures. In equaion (3) he measure of cyclical correlaion is based on he variances and covariances of he various shocks and j, for j = B and for somet. The commonly used, backward looking measure of cyclical correlaion in equaion (1) is based on he variances and covariances of he various shocks and for j = B and for all = j, 1...T. The insananeous correlaion is based on for some T while he backward looking correlaion is based on for all = 1... T. Therefore if is changing in, he insananeous correlaion measure in equaion (3) yields an accurae measure a ime T, bu he backward looking measure in equaion (1) does no. This disincion is imporan in he sudy of European inegraion, for he common currency s effecs on he economic srucures of he euro zone economies is well modeled and documened. Over ime, wih he common currency, he economies of he euro zone will rade more, and his increased rade will lead o indusrial specializaion. Thus he euro will change 7

8 in ways ha make i impossible o measure cyclical correlaion by some backward looking measure. Secion 2: Measuring Business Cycle Shocks 2.1 The Business Cycle in a Sochasic Funcional Form The model uilized in his paper is a variaion of a model suggesed, bu no explored by Frankel and Rose (1998). Their model divides GDP growh raes ino hree pars: a long run growh rend, common shocks ha are disribued across all open economies, and indusry specific shocks ha are localized and no evenly disribued. The even disribuion of common shocks across he enire currency union leads o greaer cyclical convergence. The uneven disribuion of indusry specific shocks can lead o cyclical divergence. This secion idenifies he differen common and indusry-specific shocks in each period and hen esimaes he variancecovariance marix associaed wih hese shocks. This variance-covariance marix is hen combined wih ime varying vecors of economic srucures, for some T, o find he insananeous correlaion beween business cycle flucuaions in he euro zone. In his model, he common shocks ake four forms forms, he European common shock, he euro zone common shock, he worldwide shock, and shocks linked o specific indusries. The European common shock is par of a general European business cycle. Is effecs are disribued evenly across all counries regardless of he currency used. The euro zone common shock is par of a euro zone business cycle. Is effecs are disribued evenly across all EMU member economies. The worldwide shock affecs all counries, bu i is linked o inernaional rade. The exen o which he worldwide shock will affec he economy of a paricular counry depends on he openness of ha counry. The indusrial shocks are linked o specific indusries and relae o 8

9 anyhing from a change in preferences for ha indusry s produc o a echnological change in ha indusry. The manifesaion of an indusry-specific shock in a paricular economy depends on he size of ha indusry in a paricular economy. Thus indusrial specializaion will lead o an uneven disribuion of hese shocks. The counry-specific shock explains any par of he business cycle flucuaion no explained by he common and indusry-specific shocks. Thus he model of business cycle flucuaions can be wrien as: Y Y * n i= 1 ( i, U i, ) + Z + E + θ X + V = α (5) The indusry specific shocks, U i, have a coefficien α i, A, which is he share of counry A s oal oupu ha comes from indusry i. Z is a common European shock, and E is a common euro zone shock. These shocks are disribued equally, so a coefficien is no necessary. X is he world-wide economic shock linked o inernaional rade. In he model he coefficien of X is θ A, which is a measure of he relaive openness of a counry s economy. θ A is he sum of expors and impors of counry A divided by he oal GDP of counry A. Finally any random disurbance no idenified as he counry-specific manifesaion of a common shock is considered a counry specific shock, V A. 2.2 Compuing he Variance of and he Covariance beween Economic Shocks This linear model of business cycle flucuaions allows for he idenificaion and analysis of he differen shocks ha can affec a naion s economy. Before hese shocks are found, i is helpful o idenify exacly wha ype of daa is used in he empirical analysis. The daa is obained from Eurosa. The empirical resuls are found using he daa from 29 counries and 2 averages. The counries included are 11 euro zone counies, 9 of he new member saes 1, and 9 1 A lack of necessary daa forced he exclusion of Ireland and Mala from he sudy. 9

10 oher noable European counries 2. This provides nearly a complee sample of European economies. The wo averages are for he EU-25 and for he EU-12. These averages provide an economic picure for all of Europe and for he euro zone. This daa is analyzed over 43 quarers, from he second quarer of 1995 unil he fourh quarer of As menioned earlier, he saisic for relaive openness is consruced by combining a counry s oal impors for he quarer wih is oal expors for he quarer and hen dividing by he GDP for he quarer. The saisics for indusrial shares are consruced by dividing he quarer s oal oupu in a specific indusry by he GDP from ha quarer. To consruc hese saisics on indusrial specializaion, he GDP of each counry in each quarer was separaed ino 6 indusrial caegories. These six indusrial caegories are: Agriculure, Manufacuring, Consrucion, Wholesale and Reail Trade, Financial Services, and Healhcare and Public Services. For noaional breviy and convenience, he equaion for he oupu gap as a funcion of hese 10 paricular shocks (9 common and 1 counry-specific) can be wrien in a vecor form like equaion (2) Y Y * = S + v (6) where: where S = = [ θ D 1 α for i = ] and [ X E Z U for i = ] A D A is a dummy variable equal o one if counry A is a member of he euro zone and zero if no. Taking he firs difference of equaion (6) gives he change in he oupu gap as a funcion of changes in counry-specific shocks and counry-specific manifesaions of common shocks. i, i 2 Swizerland, Denmark, Norway, Croaia, Romania, Iceland, Turkey, Sweden, and Briain 10

11 Y Y * Y Y 1 * 1 = ( S 1S 1 ) + ( v v 1 ) (7) We will approximae and say ha he counry and ime specific parameer vecor,, says relaively consan beween wo sequenial quarers, so he expression for he change in he oupu gap can be wrien as: y S + v (8) where S represen he change in he oupu gap, he change in he vecor of y,, and v common shocks, and he change in he counry specific shocks, respecively. Given observable daa on quarerly changes in each counry s oupu gap, y for all j , and counry and j, = period specific observaions of he parameer vecor for all j , an OLS regression is j, = run o find he incidence of he common and counry specific shocks, S and v A,, in each of he 43 quarers from 1995:2 o 2005:4. Once he quarerly incidences of he vecor of common shocks, S, are esimaed hen hey can be used o find a consisen esimae for he common shock s variance-covariance marix. 1 M ˆ = (9) T ( S S ) T = 1 where Mˆ is an esimae of he variance-covariance marix of he common shocks M. The consisency of he OLS esimae ensures he consisency of his variance-covariance marix: Mˆ T M (10) 11

12 The following 9x9 marix is he esimae for he variance-covariance marix M obained afer running he regression in (8) for 31 counries over 43 quarers. The headings M, C, T, F, and P represen he indusry-specific shocks and correspond o Agriculure, Manufacuring, Consrucion, wholesale and reail Trade, Financial services, and Public services and healhcare. Figure 1, Variance-Covariance Marix for he 1995:2 o 2005:4 period: X E Z A M C T F P X 4.5E E E E E E E E E-04 E 4.7E E E E E E E E E-04 Z -2.6E E E E E E E E E-03 A 7.4E E E E E E E E E-02 M 1.8E E E E E E E E E-03 C 1.3E E E E E E E E E-02 T 5.5E E E E E E E E E-03 F 2.7E E E E E E E E E-03 P 4.2E E E E E E E E E-02 In he regressions ha generaed his marix, he counry specific shocks aced as he residual. The assumpion ha he counry-specific shocks do no conribue much owards he variance of he business cycle flucuaions once he counry-specific manifesaions of common shocks are accouned for requires a high 2 R. This linear model has an 2 R of approximaely 0.7. Thus he linear model wihou counry specific shocks is able o explain 70% of he variance of observed business cycle flucuaions. The daa in figure 1 is he variance-covariance marix ha will be combined wih he vecors of economic srucure o find he insananeous correlaion beween he business cycle flucuaions of he euro zone economies. The formula for insananeous correlaion in (3) assumes ha he variances and covariances do no vary over ime. When he variances of and he covariances beween he shocks are compued for wo differen ime periods and compared, i 12

13 becomes clear ha he assumpion of consan variance-covariance is no unreasonable. Figure 1 shows he variances and covariances for he enire 1995:2 o 2005:4 ime period, bu figures 2 and 3 show he variances and covariances for he 1995:2 o 1998:4 and he 1999:1 o 2005:4 ime periods, respecively. Figure 2, Variance-Covariance Marix for he 1995:2 o 1998:4 period: X E Z A M C T F P X 9.1E E E E E E E E E-04 E 1.1E E E E E E E E E-04 Z -5.4E E E E E E E E E-03 A 1.6E E E E E E E E E-02 M 4.6E E E E E E E E E-03 C -2.2E E E E E E E E E-03 T 7.1E E E E E E E E E-03 F 4.6E E E E E E E E E-03 P 8.8E E E E E E E E E-02 Figure 3, Variance-Covariance Marix for he 1999:1 o 2004:1 period: X E Z A M C T F P X 1.9E E E E E E E E E-04 E 3.1E E E E E E E E E-04 Z -1.1E E E E E E E E E-03 A 3.0E E E E E E E E E-02 M 4.6E E E E E E E E E-03 C 2.2E E E E E E E E E-02 T -3.6E E E E E E E E E-03 F 1.8E E E E E E E E E-02 P 1.6E E E E E E E E E-02 The resuls from he es over he enire period remain inac when he period is divided in wo. Only wo of he 45 differen variances and covariances increase or decrease by a facor of 10 beween he wo periods. This implies ha while here may be marginal saisical changes, over 95% of he variances and covariances are of he same magniude in boh periods. This suppors he claim ha he variances and covariances are reasonably consan over ime, and hus he 13

14 formula in (3) is appropriae for finding he insananeous correlaion beween cyclical flucuaions in he euro zone. Secion 3: Cyclical Correlaion in he Euro Zone 3.1 Compuing Insananeous Correlaion in he Euro Zone The consan and consisen esimaes of he variances of and covariances beween he various shocks and he ime and counry specific measures of economic srucure can be used as inpus in he insananeous correlaion model. The resul is a correlaion beween European business cycles ha is no based on backward looking observaions of he counry specific parameer vecor. If he insananeous correlaion formula is applied o he linear model of European business cycle flucuaions in (5) and (6), and using he marix Mˆ, he 9x9 variancecovariance marix in figure 1 in he las secion, he insananeous correlaion beween he business cycles of counries A and B is: ρ AB, = where : ( ˆ )( ˆ M B, M B, ) = [ θ D 1 α for i = ] for j = B j, j, M ˆ B, j i,j, (11) From he formula in (11) i is easy o calculae he insananeous correlaion beween he business cycles of any wo euro zone counries given heir respecive economic srucure vecors, j,. The beauy of he insananeous correlaion is ha i is specific o a given ime period and pair of counries, so housands of correlaions exis. For reporing purposes, I will only repor he correlaion beween each euro zone counry and he euro zone average for he firs quarer of 1999 and he las quarer of The correlaion beween a counry and he euro zone average is imporan because he ECB ailors moneary policy for he euro zone average, so if a counry s 14

15 business cycles are highly correlaed wih he euro zone average here is a high probabiliy ha hey will receive he correc moneary policy for heir economic condiions. The correlaions beween he business cycles of any wo euro zone counries are available and ineresing, bu i is he lack of a perfec correlaion beween a counry s business cycles and hose of he enire euro zone measures he cos of a common currency. Table 1 repors hese insananeous correlaions for a number of euro zone and noable Wesern European counries for he firs quarer of 1999 and he las quarer of Table 1, Euro zone insananeous correlaions for 2005:4 Correlaion Correlaion in 1999:1 Counry in 2005: Belgium France Germany The Neherlands Ausria Finland Greece Ialy Porugal Spain Briain Denmark Sweden Norway Swizerland These 15 counries are separaed ino 4 groups for reporing purposes. The firs group in he able is he group of counries considered he core of he euro zone. The second group is considered he periphery of he euro zone. The hird group is simply a group of EU member counries ha are no members of he euro zone. And he fourh group is a selecion of a few Wesern European counries ha are no EU members. Clearly, hese insananeous correlaions seem o comply wih anecdoal evidence in favor of a wo-iered euro zone wih a core and a periphery. The 15

16 resuls imply ha he counries in he indusrial hearland of Europe have high correlaion beween heir business cycles and hose of he euro zone average, bu counries away from ha indusrial hearland have business cycles ha are less correlaed. Similarly counries ouside he euro zone have correlaions a or below hose of he periphery. 3.2 Measuring Cyclical Correlaion hrough Counerfacual Siuaions Business cycle correlaion measures he cos of a common currency. Insananeous correlaions are paricularly useful because hey allow one o sudy he effec of changing paricular pars of a counry s economic srucure on correlaion. One can isolae he effec on cyclical correlaion ha comes from a changing economic srucure by calculaing he insananeous correlaion under various, imposed vecors, j,. One can isolae he effec of increased rade openness on cyclical correlaion by calculaing he insananeous correlaion under differen levels of rade openness, or one can esimae he effec of indusrial specializaion by calculaing he insananeous correlaion under differen, hypoheical, indusrial srucures. A common currency will induce more rade beween he member economies. Since currency exchange is a barrier o rade, removing ha barrier faciliaes increased rade beween he euro zone counries. Frankel and Rose (2002) found ha a counry ha eners ino a currency union will see a hreefold increase in is volume of rade wih oher members of ha union. These numbers have since been downgraded, bu i is commonly acceped ha a common currency will lead o increased rade beween he member counries. Given ha a common currency will lead o more rade, he debae focuses on he effec of rade on business cycle correlaion. This debae revolves around wo disinc and muually exclusive heories. Krugman (1993) conends ha increased rade beween wo economies will no in and of iself cause business cycle convergence beween he wo. He speculaes ha increased rade will cause he economies o specialize in he 16

17 indusries for which hey have a comparaive advanage. Thus when he economies of he currency union are sruck by cerain indusry specific shocks, he effecs of hose shocks will be localized and no disribued across all counries in he currency union. These asymmeric reacions o symmeric shocks will creae more cyclical divergence beween he members of he currency union. Frankel and Rose (1998) conend ha common and demand side shocks will be beer disribued hrough he enire currency union hrough he increased rade of goods and services. They acknowledge he fac ha increased rade can lead o indusrial specializaion and more asymmeric shocks, bu hey believe ha symmeric responses o common and demand side shocks will dominae he asymmeric responses o indusry-specific shocks, so he economies of he currency union should see heir business cycles converge o a common, union-wide, business cycle The Effec of Increased Trade on Cyclical Correlaion The effec of increased rade openness on cyclical correlaion can be found by calculaing insananeous correlaions wice. Firs he correlaions are calculaed in each quarer using he observed levels of rade openness, and hen i is calculaed in each quarer using he levels of rade openness from he firs quarer of wih increased rade : j, wihou increased rade : = j, [ θ j, D j 1 αi,j, for i = ] = [ θ D 1 α for i = ] j, 1999:1 j i,j, (12) Thus he wo vecors for j B will be used o compue he insananeous correlaions j, = once given ha he euro induced greaer rade openness and once given ha he euro did no affec rade openness. The lef hand column of able 2 repors he insananeous correlaions in he fourh quarer of 2005 if he euro zone counries do no become more open o rade afer he 3 I should be noed however ha while Frankel and Rose advocae he heory ha membership in a currency union will lead o cyclical converge, hey accep ha here can be oher views and consider i an open quesion. 17

18 firs quarer of 1999, and he righ hand column of able 2 hen repors he absolue change in he correlaions due o increased rade Table 2, Insananeous correlaions in 2005:4 given θ from 1999:1, and he absolue change in correlaion due o increased rade. Trade Conroled Change in Correlaion Correlaion in 2005:4 Counry due o Increased Openness Belgium France Germany The Neherlands Ausria Finland Greece Ialy Porugal Spain Briain Denmark Sweden Norway Swizerland In mos cases, increased rade has lead o increased correlaion, bu some counries have noice a decrease in heir correlaion wih he euro zone average as a resul of increased inernaional rade over he pas 7 years. Also he magniude of he absolue change in correlaion is varied across he sample of counries. The increased rade has grealy improved correlaion in small, open, core economies like Belgium and The Neherlands. A he same ime he impac is no ha grea in larger, less open economies like France and Germany The Effec of Indusrial Specializaion on Cyclical Correlaion No only will a common currency cause increased rade in he euro zone, bu his increased rade will lead o indusrial specializaion as counries specialize in he indusries for 18

19 which hey have a comparaive advanage. This indusrial specializaion leads o greaer cyclical correlaion if a counry specializes like he res of he euro zone, and i leads o less cyclical correlaion if a counry specializes differenly from he res of he euro zone. If he insananeous correlaions are calculaed for 2005:4 given he indusrial srucures from 1999:1 i is like finding he correlaion a he end of 2005 condiional on here having been no indusrial specializaion since he adopion of he euro. In erms of he model, he insananeous correlaions are calculaed for wo values of : wih indusrial specializaion : j, wihou indusrial specializaion : = j, [ θ j, D j 1 αi,j, for i = ] = [ θ D 1 α for i = ] j, j i,j, 1999:1 (13) The insananeous correlaion resuls given indusrial specializaion are conained in able 1, bu correlaions wih no indusrial specializaion are given in he lef hand column of able 3. The absolue change in he correlaion as a resul of indusrial specializaion is found in he righ hand column of able 3. Table 3, Insananeous correlaions in 2005:4 given he indusrial srucures from 1999:1, and he absolue change in correlaion due o indusrial specializaion. Specializaion Conrolled Change in Correlaion Correlaion in 2005:4 Counry due o Increased Specializaion Belgium France Germany The Neherlands Ausria Finland Greece Ialy Porugal Spain Briain Denmark Sweden Norway Swizerland

20 The numbers in he righ hand column of able 3 represen he absolue increase or decrease in correlaion due o indusrial specializaion. If his number is posiive hen he counry s business cycles became more correlaed wih he euro zone average as a resul of indusrial specializaion, and if his number is negaive hen he counry s business cycle became less correlaed wih he euro zone average. As is o be expeced, indusrial specializaion has lead o increased correlaion in some counries and decreased correlaion in ohers. This is because some counries are specializing more like he euro zone average and some are specializing away from he euro zone average. Of he counries in he core, he change in correlaion is for he mos par posiive or negaive and small. This is in direc conras o he counries in he periphery where he change in correlaion is quie significan and negaive in every case excep for Ialy. Ialy is a paricularly ineresing case. While many would place Ialy in he periphery, many would argue ha Ialy is becoming more like he core. These calculaions clearly show ha indusrial specializaion is making Ialy more like he core of he euro zone. These resuls do no give a definiive answer as o which side effec of rade, he demand spillovers or he indusrial specializaion, has a greaer impac on cyclical correlaion. For he majoriy of he counries, he indusrial specializaion is found o have a greaer absolue impac on cyclical correlaion han demand spillovers, bu a few noable excepions challenge his rule. A few core counries like Belgium and France see he effec of demand spillovers rump he effec of indusrial specializaion; his is also rue for Ialy and Ausria in he periphery, as well as Sweden ouside he euro zone. Furher heory is needed o explain why he converging effec of demand spillovers and indusrial specializaion differs so much across counries. 20

21 Summary and Conclusions In order for he euro o funcion as a common European currency, he cos of moneary union in erms of real insabiliy mus come down. This real insabiliy arises because a unionwide moneary policy can never be beer han a counry-specific moneary policy, bu i has he possibiliy of being much worse. How much worse depends on he correlaion beween a counry s business cycle flucuaions and hose of he enire euro zone. This paper has shown, hrough he mehod of insananeous correlaion and various counerfacual calculaions ha a common currency will no sysemaically increase cyclical correlaion. Increased rade beween members has boh cyclically converging and diverging effecs, and he magniude of hose effecs differs across counries. Therefore, for he euro o work as a common currency, somehing mus be done o lessen he severiy of he ineviable real insabiliy. The severiy of his real insabiliy can be lessened by eiher greaer facor mobiliy or fiscal federalism. Greaer facor mobiliy lessens he severiy of real insabiliy because facors of producion can move beween regions a differen poins in he cycle. The high unemploymen and low capial uilizaion raes associaed wih a recession and he high inpu price inflaion associaed wih a boom can be alleviaed as facors move o he laer region from he former. The Treay of Rome mandaes ha here are no sauory barriers o facor mobiliy wihin he EU, bu naional governmens need o ensure ha his mandae is acually saisfied, and provide he necessary ools for labor and capial o easily move beween EU counries. Fiscal federalism lessens he severiy of real insabiliy because money is direcly ransferred beween counries a differen poins in he cycle. This smoohes he cycle as a boom (and resuling high inflaion) is lessened by fiscal ighening, and a recession (and resuling high unemploymen) is lessened by fiscal simulus. This fiscal smoohing is possible in an EU wih a 21

22 federalis sysem wih he power o ransfer funds, bu he curren EU budge is so small ha any aemps a federalis ransfers would have only a marginal effec. A funcioning moneary union in Europe is no possible wihou a funcioning poliical union. Wihou a srong federal sysem wih he power o ax, spend, and ransfer funds, he cos of a common currency in erms of real insabiliy will coninue o far ouweigh he benefis. 22

23 References Alavilla C. Do EMU Members Share he Same Business Cycle?. Journal of Common Marke Sudies, 2004; 42(5); Bayoumi T, Eichengreen B Shocking Aspecs of European Moneary Inegraion. In: Torres F, Giavazzi F (Eds), Adjusmen and Growh in he European Moneary Union. Cambridge Universiy Press, Oxford, pp De Grauwe P. Economics of Moneary Union, 5 h Ed. Oxford Universiy Press, Oxford, De Greuwe, P. Moneary Union and Convergence Economics. European Economic Review 1996; 40; Eichengreen B. Is Europe an Opimum Currency Area? NBER Working Paper, No. 3579; Frankel J, Rose A. The Endogeneiy of he Opimum Currency Area Crieria. Economic Journal 1998; 108(449); Frankel J, Rose A. An Esimae of he Effec of Common Currencies on Trade and Income. Quarerly Journal of Economics 2002; 117; Geing European Workers Moving. OECD Observer Augus 1999; 90 Hughes Halle Piscielli L. Does rade inegraion cause convergence? Economics Leers 2002; 75; Hughes Halle, Piscielli L. EMU in Realiy: The Effec of a Common Moneary Policy on Economies wih Differen Transmission Mechanisms. Empirica 1999; 26; Kalemi-Ozcan S, Sorensen B, Yosha O. Economic Inegraion, Indusrial Specializaion, and he Asymmery of Macroeconomic Flucuaions, Journal of Inernaional Economics 2001; 55; Krugman P. Geography and Trade. MIT Press: Cambridge, MS; Krugman P Lessons of Massachuses for he EMU. In: Torres F, Giavazzi F (Eds), adjusmen and Growh in he European Moneary Union. Cambridge Universiy Press, Oxford, pp Mundell R. A Theory of Opimum Currency Areas. The American Economic Review 161; 51; von Hagen J, Neumann M. Real Exchange Raes Wihin and Beween Currency Areas: How Far Away is EMU? The Review of Economics and Saisics 1994; 76(2);