Global Monetary Policy Shocks in the G5: a SVAR Approach

Transcription

1 Global Moneary Policy Shocks in he G5: a SVAR Approach Joao Miguel Sousa 1 and Andrea Zaghini 2 Absrac The paper consrucs a global moneary aggregae, namely he sum of he key moneary aggregaes of he G5 economies (US, Euro area, Japan, UK, and Canada), and analyses is indicaor properies for global oupu and inflaion. Using a srucural VAR approach we find ha afer a moneary policy shock oupu declines emporarily, wih he downward effec reaching a peak wihin he second year, and he global moneary aggregae drops significanly. In addiion, he price level rises permanenly in response o a posiive shock o he global liquidiy aggregae. The similariy of our resuls wih hose found in counry sudies migh suppors he use of a global moneary aggregae as a summary measure of worldwide moneary rends. JEL classificaion: E52, F1. Keywords: Moneary policy, Srucural VAR, Global economy. * We are graeful o Alessandro Calza, Michael Ehrmann, Leonardo Gambacora, Livio Sracca and o paricipans he XII Inernaional Tor Vergaa Conference on Banking and Finance and he XIX European Economic Associaion Congress. This paper does no necessarily reflec he views of he Banco de Porugal and of he Banca d'ialia. 1 Banco de Porugal (Joao.Miguel.Sousa@bporugal.p) 2 Banca d Ialia (andrea.zaghini@bancadialia.i) 1

2 1. Inroducion Over recen years, moneary rends in he major indusrialized counries have exhibied many similariies. Moreover, several commenaries sugges ha global liquidiy significanly affecs financial condiions in inernaional markes and ha liquidiy developmens in one financial cener can influence financial condiions elsewhere. However, so far only a limied number of sudies have examined he role of shocks o moneary aggregaes in driving business flucuaions or, more generally, in influencing he behavior of global macroeconomic and financial variables. Afer he early aemp by McKinnon (1982), in which he auhor sudied he effecs of changes in he world money supply on US price inflaion, only recenly sound economeric works emerged on he issue of cross-counry moneary spillovers (Baks and Kramer, 1999; Kim, 21; Holman and Neumann, 22; Canova, 25). This paper follows his recen srand of research and goes a sep furher by sudying he effecs of shocks o global money on global inflaion and oupu. The analysis is based on an aggregaed model of he mos indusrialized counries. The basic idea is o use global (G5) moneary aggregaes in order o pool informaion from several counries o assess o wha exen sylized facs in closed economies can be exended o a broader global model. In fac, such global aggregaes are likely o inernalize crosscounry movemens in moneary aggregaes (due o capial flows beween he differen regions) ha may make he link beween money and inflaion and oupu more difficul o disenangle in he single counry case. A furher moivaion of his paper is o invesigae o wha exen a global moneary aggregae can be useful for analyzing inernaional liquidiy condiions. Several insiuions (ECB, IMF, OECD) have used global aggregaes as summary measures of he worldwide moneary siuaion. Since we hink ha such use deserves a rigorous es of he informaion conen of global moneary indicaors, we propose in his paper an economeric analysis 2

3 based on he srucural VAR approach. In fac, a mere correlaion beween global money and global inflaion and/or oupu is no sufficien o deermine he direcion of he relaionship beween he variables given ha hey are all endogenously deermined. In his regard, he srucural VAR approach is a more powerful mehodology o invesigae his link, as i conrols for he ineracions beween he variables allowing us o provide a more appropriae assessmen of he conribuion of moneary shocks o global oupu and inflaion. The paper is organized as follows: Secion 2 provides some informaion and sylized facs abou global liquidiy and is relaionship wih oher aggregaed variables, Secion 3 presens he empirical framework of he SVAR analysis. In Secion 4 we inroduce our global model wih aggregae variables for he group of he G5 counries. In Secion 5 we perform a robusness check while Secion 6 concludes. 2. Preliminary evidence on global liquidiy The global liquidiy aggregae analyzed in his paper is consruced as a sum of he reference moneary aggregaes for he US, he Euro area, Japan, he UK and Canada using exchange raes vis-à-vis he euro based on purchasing power pariies o conver hem ino a common currency (see daa Annex for furher deails). 1 In order o ge some insighs ino he series underlying he global liquidiy aggregae, Figure 1 plos he annual growh rae of broad moneary aggregaes in differen counries in domesic currency. As can be seen in he char, alhough money growh raes in he differen areas seem o be only weakly correlaed in he shor-run, in longer periods here appears o be clear co-movemens. In paricular, he 198s are characerized by higher moneary growh almos everywhere, while in he 199s average money growh was lower in all he economies considered. In addiion, in he early 199s here is a clear slowdown in money growh ha 1 The moneary aggregaes have been chosen by looking a he mos common indicaors used in he economeric lieraure, for each counry, when dealing wih money demand and supply. See Sims and Zha (1998), Hendry and Mizon (1998), Coenen and Vega (21), Yamada (2) and Holman and Neumann (22), among ohers. 3

4 has occurred a abou he same ime in mos of he counries. Only in he Euro area, such slowdown ook place somewha laer, in he years More recenly, here seems o be a similar behavior in money growh inernaionally, wih he rise in moneary growh from he end of 21 onwards occurring roughly a he same ime in mos counries. Alhough counry specific facors have played a role, developmens in he Euro area, he US and he UK have shared common feaures. In he US, he srong growh in M2, while parly relaed o he increased refinancing aciviy of morgage loans, also refleced he low level of shor-erm ineres raes and precauionary demand for asses included in M2. In he UK, he srong growh in M4 was relaed o he srenghening of economic aciviy bu also o precauionary savings and a higher aversion by he public o inves in equiy. US Japan UK Canada Euro Area Q3 85Q3 87Q3 89Q3 91Q3 93Q3 95Q3 97Q3 99Q3 1Q3 3Q3 Fig. 1. Four quarer moving average of annual money growh in differen economies. GL4 Euro Area M Q1 86Q1 88Q1 9Q1 92Q1 94Q1 96Q1 98Q1 Q1 2Q1 Fig. 2. Non-euro area moneary aggregae a PPP exchange raes (GL4) and euro area M3 (annual growh raes). 4

5 For insance, he exisence of co-movemens beween Euro area M3 and broad money in he oher four economies becomes clearer if M3 growh is compared o an aggregaion of he moneary aggregaes of he non-euro area counries (convered ino euro wih PPP exchange raes: GL4). This is shown in Figure 2 wih he deails of he aggregaion are repored in he Appendix. From he char i can be seen ha wih he excepion of a few years in he early 199s (perhaps relaed o he ERM crisis and he slowdown in M2 in he US which led o insabiliy in money demand in ha period), here is a posiive correlaion beween he wo series, suggesing he exisence of a mechanism able o correc inernaional differenials in moneary growh hrough changes in he exchange rae and/or he moneary aggregaes of he differen counries. The co-movemen of he wo series has been remarkably close in recen years. GDP deflaor (LHS) Broad money grow (RHS) Q1 86Q1 88Q1 9Q1 92Q1 94Q1 96Q1 98Q1 Q1 2Q1 Fig. 3. Global liquidiy (GL5) and global inflaion (four quarer moving average of annual growh raes). Figure 3 shows he developmens of he annual growh rae of a global liquidiy aggregae including all he five economies (GL5) and he global inflaion rae (measured by he annual growh rae of he GDP deflaor of he counries considered). Even hough here are several periods when he developmens in he wo variables appear no o be srongly relaed, overall here is a posiive correlaion beween global inflaion and global liquidiy. The char also suggess ha he decline in he growh of global liquidiy preceded he 5

6 disinflaionary period in he firs half of he 199s. However, he relaion beween he wo variables from mid-1995 onwards is no so clear as while he growh of global liquidiy increased, global inflaion coninued o decline and sared o rise only in 21. Real GDP (LHS) Real GL (RHS) Q1 85Q3 87Q1 88Q3 9Q1 91Q3 93Q1 94Q3 96Q1 97Q3 99Q1 Q3 2Q1 3Q3 Fig. 4. Four quarer moving average of he annual growh of real global liquidiy (GL5R) and real GDP. The developmens in he real global liquidiy aggregae and in global real GDP growh are shown in Figure 4. As in he case of inflaion, a posiive correlaion beween real global liquidiy and economic aciviy emerges: periods when real GDP growh is relaively high (low) appear o coincide wih periods where he growh of real broad global liquidiy is also relaively high (low). The main excepion is he period from mid-21 onwards, during which he annual growh of real global liquidiy increased subsanially while global real GDP growh declined significanly. Thus, he srong urbulence in financial markes noably following he 11 Sepember erroris aacks and, more recenly, relaed o he worldwide heighened economic, financial and geopoliical uncerainy seems o have led o an increased preference of economic agens worldwide for holding liquid and safe asses, such as hose included in he broad moneary aggregaes. However, as already menioned in he inroducion, he evidence of possible sable relaionships among aggregae variables as summarised above deserves a more sound economeric analysis. In paricular, i is imporan o ake ino accoun he developmen of 6

7 ineres raes worldwide and he evoluion of commodiies prices o disenangle he possible linkages among global money, oupu and prices. In he res of he paper we propose a more accurae empirical framework of analysis in which evaluae hese relaionships. 3. Empirical framework There are several advanages in relying on he srucural VAR mehodology for he analysis of he effecs of moneary policy changes. In paricular, i allows modelling nonrecursive srucures of he economy wih a parsimonious se of variables and i faciliaes he inerpreaion of he conemporaneous correlaions among disurbances. 2 Consider he following model as in he radiional reduced-form VAR analysis: L Y u (1) where Y is a vecor of macroeconomic variables and LY is a marix polynomial in he lag operaor L for which L. I The sandard hypoheses hold for he residuals: Eu ' E( uus ) when s when s (2) (3) Condiion (3) implies ha here is no serial correlaion among disurbances bu, a he same ime, conemporaneous correlaion is allowed. In a sandard VAR framework, simulaneous relaionships are hen condensed in he variance-covariance marix, making he economic inerpreaion of hese relaionships quie difficul. In order o ransform he original VAR ino a model in which disurbances are orhogonal, he SVAR approach proposes o sar from he rue srucural form model. 3 For he same vecor Y of variables in (1) consider he following dynamic model: 2 For a comprehensive ex-book reference see Amisano and Giannini (1997). 3 Sims (198) suggesed o rely on he Cholesky decomposiion of he variance-covariance marix, hrough a lower-riangular marix P such ha =PP. However, he Cholesky decomposiion is no an a-heoreical approach. The lower riangulariy of P implies a recursive scheme among he variables (he Wold causal chain) ha has clear economic implicaions and has o be empirically esed as any oher relaionship. 7

8 K L Y e (4) k K L K i1 A L i i where K is an n n non singular marix. The conemporaneous relaions are hus direcly explained in K. The srucural model is linked o he reduced form (1) by: K i K i (5) Ku e E ' ' ' ' Ku u K KK Ee e I (6) (7) Given ha is a symmeric marix, he maximum likelihood esimaes of he reduced form model give rise o an insufficien number of parameer for he exac recovering of he srucural form. 4 The SVAR mehodology suggess o impose resricions on he conemporaneous srucural parameers only, so ha reasonable economic srucures migh be derived. 5 The fac ha only conemporaneous resricions are imposed however does no imply ha here is no feedback among variables. In he (S)VAR srucure he lagged values ener each equaion and hus all variables are linked ogeher. 4. A global approach In his secion we propose a unified G5 model including aggregaed variables from he mos indusrialized areas (US, Euro area, Japan, UK and Canada). In paricular, we ry o idenify a common moneary policy shock in his enlarged framework. Of course we are aware ha here is neiher a common moneary policy nor any broad policy coordinaion a such aggregae level. However, his approach migh help in solving he problem of 4 In he srucural form of he generic model of lag-lengh k here are kn²+n² free parameers belonging o K and K i, while from he esimaes of i and one ges only kn²+n(n+1)/2 values. 5 This specificaion scheme is he mos used in he moneary policy analysis: see among ohers Gordon and Leeper (1994), Sims and Zha (1998), Leeper and Roush (23) for he US and Kim and Roubini (2), Mojon and Peersman (23), Dedola and Lippi (25) for oher counries. 8

9 endogeneiy in open-economy single counry models. In fac he possible endogeneiy from moneary policy shocks derives from a following he leader behavior by which a given counry (he follower) always adjuss is moneary policy sance accordingly o he decisions of a leader. Thus, a shock o moneary policy in he follower economy migh no be exogenous bu only a reacion o he leader counry. Grilli and Roubini (1995) find some evidence of he US being he leader inernaionally a he G7 level, while many applicaions for EU counries sugges ha Germany has been, a leas for he ERM-period, he European leader. 6 As a basis of our analysis of global liquidiy shocks in he G5 counries, we need a benchmark model able o properly explain he macroeconomic shor- and long-erm dynamics of each counry under consideraion. Our firs choice is based on a specificaion proposed by Kim (1999) for each of he G7 counries and laer esed for he Euro area as a whole by Peersman and Smes (23) and Sousa and Zaghini (24). However, since he specificaion esimaed in he laer wo papers slighly differs from he one in Kim (1999), we propose a given se of resricions ha is, also for he Euro area, as close as possible o he one by Kim (1999). In paricular, we will keep he quarerly frequency of he daa and four ou of five variables as in Peersman and Smes (23) and Sousa and Zaghini (24), bu, following Kim (1999), we will use a oal commodiies index convered ino domesic currency insead of he effecive exchange rae of he euro. Thus, he vecor of endogenous variables is as follows: Y YR5, PI5, GL5, SR5, TC where YR5 is he real GDP, PI5 is he consumer price index, GL5 is he moneary aggregae and SR5 is he average shor-erm ineres rae of he G5 area, whereas TC is he commodiies price index in euro (see daa Annex for deails). 6 See for insance Kim and Roubini (2) and he discussion in Dornbusch e al. (1998). 9

10 The idenificaion scheme used is characerized by he following non-recursive srucure of he kind Ku = e : 1 k k k k k k k k k u u u k45u 1 u YR5 PI 5 GL5 SR5 TC e e e e e YR5 PI 5 GL5 SR5 TC (8) The firs wo equaions deal wih he non-policy variables wih GDP and prices responding o financial innovaions (money, ineres rae and commodiies price) only wih a lag. The hird and fourh equaions include he global liquidiy aggregae and he conemporaneous cenral bank reacion funcions. The global liquidiy aggregae is assumed o reac in he same quarer o shocks on real income, he price index and he shor-erm ineres rae, while he moneary auhoriy feedback rule relaes he ineres rae o conemporaneous changes in he moneary aggregae and he commodiies price. As in Sims and Zha (1998), he choice of his moneary policy reacion funcion is based on he assumpion of informaion delays ha do no allow he moneary policy o respond wihin he same period o price level and oupu developmens. More in deails: published daa on money and he exchange rae are available wihin he period bu reliable daa on oupu and prices are no. In he fifh equaion he oal commodiies index reacs o conemporaneous changes in all he oher variables. However, concerning he money supply equaion in paricular, he basic idea underlying he SVAR approach is ha no all changes in he moneary policy sance reflec a sysemaic response o variaions in he sae of he economy: he unaccouned aleraion is referred o as a moneary policy shock. This shock can be inerpreed as a deviaion of he ineres rae from he moneary policy rule ha he cenral bank is assumed o be following. 7 Esimaions are based on quarerly daa over he period from 198 Q1 o 21 Q4. Daa are expressed in logarihmic form and are seasonally adjused, excep he ineres raes which 7 See Chrisiano, Eichenbaum and Evans (1998) for possible explanaions on his poin. 1

11 are in levels. A consan and a linear rend are added o he model. Sandard informaion ess hined o a 4-lag lengh for he VAR. 8 Impulse Responses.1 of Oupu o Shor rae of Liquidiy o Shor rae of Prices o Shor rae.24 of Shor rae o Shor rae Fig.5. Moneary Policy Shock. Figure 5 depics he impulse response funcions o an unexpeced moneary policy shock. We can see ha, as in a single counry model, real GDP decreases a impac bu hen ends o recover o is iniial level and ha global liquidiy quickly drops afer an increase in shorerm raes and he effec is long-lasing. Only he price index does no respond in he expeced way. The figure highlighs a clear price puzzle : for 2 years prices increase and only hereafer sar o decline. One possible explanaion of he price puzzle ha is usually found in VAR sudies is he omission of a variable useful in forecasing inflaion (such as he commodiy price index) which implies ha endogenous responses o expeced inflaion increases will misakenly be aken as moneary policy shocks (see Giordani, 24). In he case of our model, he absence of an exchange rae erm in he specificaion may resul problemaic, since we ake he G5 economies as our world economy bu he currencies of omied counries may sill play a relevan role for G5 inflaion rends. In addiion, G5 8 In paricular, he LR es, he Akaike informaion crierion and he FPE es poined o a 4-lag lengh, while he Schwarz and HQ crieria suggesed a smaller number of lags. 11

12 counries oher han US have been implicily and explicily concerned abou he effecs of a depreciaion of heir currency on heir inflaion raes for a leas par of he ime period we are considering. Thus he model is no able o conrol for he par of ineres raes movemens ha are sysemaic responses o a depreciaion of he domesic currency and which may differ from counry o counry. Finally, i could be he case ha, as suggesed by Giordani (24), one should use he oupu gap insead of he level of real oupu in he VAR, as ha migh solve he price puzzle. Impulse Responses.1 of Oupu o To Comm of Prices o To Comm Fig.6. Toal Commodiy Index Shock. Figure 6 shows he impulse responses of oupu and prices o a shock o he oal commodiy index. As seen in he char, an exogenous increase in commodiy prices emporarily decreases global oupu. This effec is only significan afer a year and a half afer he shock. In he long-run, oupu reurns o he iniial level. As for prices, he reacion of he price level appears o be immediae, wih an increase one quarer afer he shock. As wih real GDP, he effec on global prices peaks afer five quarers afer which he price level gradually reurns o he iniial level. 12

13 Impulse Responses.2 of Oupu o Liquidiy of Prices o Liquidiy Fig. 7. Global Liquidiy Shock. The reacions of real GDP and prices o a shock in he global liquidiy are repored in Figure 7. Given ha we rely on broad measures of he moneary aggregae, we inerpre our global liquidiy shocks as a money demand shock, namely an exogenous increase in agens' preferences for liquidiy. As seen in he figure, an increase in he global moneary aggregae has a posiive impac on real GDP in he shor-run, ha however disappears in he mediumo long-run. As for prices, he effec is negligible in he firs 6 quarers, bu soon afer becomes significanly posiive and permanen. Again he resuls are akin o hose obainable from a single counry model. We now analyze he resuls concerning he sources of oupu and price flucuaions. As for global GDP (Figure 8), he forecas error variance decomposiion shows ha he conribuion of unexpeced shocks o shor-erm raes is raher limied in he shor-run bu i quickly increases over ime. A he end of he second year he conribuion o oupu volailiy is already above 2% and i remains slighly below ha hreshold ill he end of he sample period. Even hough oupu iself and prices explain always he vas majoriy of he flucuaions a any horizon, he role of global liquidiy and ha of commodiies price conribue significanly from he second year onwards. 13

14 1% 9% 8% 7% 6% 5% 4% 3% 2% 1% % YR5 PI5 G5T SR5 TC Fig.8. GDP Variance Decomposiion. 1% 9% 8% 7% 6% 5% 4% 3% 2% 1% % YR5 PI5 G5T SR5 TC Fig.9. CPI Variance Decomposiion. Figure 9 depics he conribuions o price flucuaions. As one would expeced, in he shor-run prices are quie sicky, heir own conribuion sill accouns for more han 2/3 of oal volailiy a he end of he second year. However, shor-erm raes, global liquidiy and 14

15 especially commodiies prices gain relaive weigh srongly from he beginning of he hird year, reaching a cumulaive share of 5% in he las wo quarers. The conribuion of GDP is relaively limied. 5. Robusness In order o have an idea on how he dynamics of he global model are influenced by he imes series aggregaion procedure here implemened, i migh be worh o look a he impulse responses funcions counry by counry. In fac, if he ransmission mechanism is similar across he G5 counries, he global approach provides a measure of he effecs of a given shock which is as good as ha obainable by oher esimaion mehods. Euro area USA Japan UK Canada Global Average Fig. 1. Impulse responses of real GDP o a moneary policy shock. We adoped he same idenificaion paern of he global model for he domesic variables in naional currency of each counry. Once obained he impulse responses we could hen compue he average response o a shock o he moneary policy auhoriy insrumen. Following he spiri of he mean group esimaor proposed by Pesaran and Smih (1995), we averaged he impulse responses by a simple mean. As an example, Figure 1 shows he real oupu response o a posiive moneary policy shock (an increase of 1-ime he sandard 15

16 deviaion of he shor-erm ineres rae) for he US, he Euro area, Japan, he UK and Canada ogeher wih he average cross-counry effec and he impulse response semming from he global framework. The shape of he curve is indeed similar for he whole group of counries and he wo average measures ; only Japan seems o suffer a longer conracionary effec from a ighening of he moneary sance. However, he dynamics seem o be similar concerning boh he shape and he magniude of he oscillaions when we consider he resul of he wo aggregaions, hus somehow supporing he validiy of he global framework here implemened..6 USA Euro Area Japan UK Canada Fig. 11. Impulse responses of real GDP o a liquidiy shock. Since he focus of his sudy is on he influence of moneary liquidiy on oher macroeconomic variables, we replicae Figure 7 for each counry. In paricular, Figure 11 and Figure 12 show he impulse responses of oupu and prices o a shock in he domesic moneary aggregae, respecively. As for he reacion of real GDP, we can observe ha Canada responds in relaively quick fashion o he shock in he domesic liquidiy, while he impac for he US seems o be indeed marginal. For he oher counries he reacions are in line wih wha expeced and wih he global esimae: here is a posiive emporary increase, which ends o be reabsorbed wihin he 4-year horizon. 16

17 USA Euro Area Japan UK Canada Fig.12. Impulse responses of prices o a liquidiy shock. Also he resuls semming from Figure 12 are in line wih he economic wisdom. The impac of a liquidiy expansion is ransmied o prices only wih a lag also a naional level. In he US he impac seems o be faser han in he oher economies, wih a peak reached already in he second year. However, he differen speed a which each counry adjuss o he liquidiy shock and he ime-lag necessary o impac on prices dynamics confirm he resuls of a recen srand of he economic research. Several conribuions sugges ha he one-o-one relaion beween money and prices derived from he quaniy heory, while being generally valid in he long-run, may no hold over shorer periods of 2 o 3 years (Bachmeier and Swanson, 25; Bruggeman e al., 25; Roffia and Zaghini, 25). This in urn implies ha he leading properies of money for inflaion may no be srong over a relaively shor horizon. 9 Thus, signals emerging from moneary developmens, while mainaining heir long-run srucural leading properies for prices, have o be suppored by oher kind of analysis in he shor-run, for insance indicaors of he business-cycle posiion of an economy. 9 De Grauwe and Polan (25) sugges insead ha he quaniy equaion relaionship may no hold in a framework of very low inflaion. 17

18 One ineresing aspec of he above resuls is ha here is a grea deal of similariy in he empirical responses of he macroeconomic variables o liquidiy shocks across counries. This seems o sugges ha he role of money in he economy ends no o differ markedly across he counries considered. In urn, according o hese resuls, here are no srong reasons for he differences observed in he imporance given o moneary aggregaes in he moneary policy sraegy of major cenral banks Conclusion This paper explicily modelled a global G5 framework relying on a common srucural idenificaion scheme ha worked well for each single area. The global framework poins o a srong similariy in he behavior of aggregae variables compared o single economy models. In paricular, afer a moneary policy shock oupu declines only emporarily, wih he downward effec reaching a peak wihin he second year, and he global moneary aggregae drops significanly. In addiion, when analyzing he impac of an unexpeced change in he G5 moneary aggregae on GDP and inflaion, he resuls are as one would expec from he corresponding responses in a single economy. In fac, he global price level rises permanenly and real global oupu only emporarily in response o a posiive shock o global liquidiy. Also as far as volailiy is concerned, he resuls of he decomposiion of he forecas error variance show ha global real oupu and price flucuaions can be explained in he aggregaed framework in he same vein as in single counry model. The overall similariy in he resuls in he aggregaed G5 framework relaive o single counry models suggess ha a global liquidiy measure migh be used, ogeher wih radiional cross-counry variables, as an indicaor of global moneary condiions in abou he same way he domesic moneary aggregaes are used in he single economy case. 1 For insance, moneary aggregaes are given a prominen role in he moneary policy sraegy of he ECB while hey rarely play a role in he moneary policy of he US Federal Reserve. 18

19 As furher exensions of his work, i would be useful o perform a horough analysis of he leading indicaor properies of global liquidiy for prices and economic aciviy aking ino accoun also he forecas horizon. A he same ime anoher possible exension would be o sudy he join developmens in global asses price and global liquidiy possibly in a larger group of counries including also emerging marke economies. 19

20 Appendix The moneary aggregaes used in he consrucion of he broad measure of global liquidiy were M3 for he Euro area, M2 for he US, M2 plus cerificaes of deposis for Japan, M4 for he UK and M2+ for Canada. These aggregaes have been chosen as hey are he key broad moneary aggregaes in he differen counries from a moneary policy poin of view. The global aggregaes were consruced by convering each naional aggregae ino euros using PPP exchange raes. The formula used is he following: GL5 5 i1 M i E ppp i, eur where M i represens each naional moneary aggregae and i eur E, ppp is he corresponding counry s PPP exchange rae vis-à-vis he euro. The PPP exchange rae is based on relaive PPP aking he nominal exchange rae of January 1999 of each counry agains he euro as he basis. Thus, his procedure does no guaranee ha absolue PPP holds. However, for he purpose of his sudy, he level of he exchange rae used o consruc he global liquidiy is relaively no imporan as only he changes over ime of he global liquidiy aggregae will maer in he esimaion of he model. One limiaion in he consrucion of he global liquidiy aggregae as done above is ha he resuling aggregae will be raher sensiive o he definiion of he moneary aggregaes used o consruc i. As here may be problems of comparabiliy beween he aggregaes used for he differen counries, he weighs may no reflec appropriaely he differences in he imporance of each counry. This is paricularly he case for Japan and he US, wih he former counry having over same periods a larger share in he global liquidiy aggregae han he laer. Therefore, we have also consruced a differen measure of global liquidiy using GDP weighs. The formula is he following: 2

21 GL5 5 i1 GDP MIndexi GDP i eur all E i, eur ppp where GDP i represens nominal GDP of counry i expressed in naional currency and is he aggregae GDP of he whole se of counries obained as he sum of each counry s GDP convered ino euros wih PPP exchange raes. MIndex i is he index of he moneary aggregae in counry i. For each counry his index equals 1 in January 1999 and grows a he same rae as he moneary aggregaes denominaed in naional currency used for each counry. Figure A.1 shows he difference beween he wo series. As can be seen in he char, mos of he ime differences are limied. eur GDP all Q1 99Q2 97Q3 95Q4 94Q1 92Q2 9Q3 88Q4 87Q1 85Q2 83Q3 Broad money growh, simple sum Broad money growh, GDP weighs Fig. A.1. Global liquidiy compued wih differen weighs. In he case of he oher variables, namely he shor-erm ineres rae, real GDP and he GDP deflaor, he compuaion of global aggregaes was done by making recourse o GDP weighs obained using PPP exchange raes o conver each naional nominal GDP ino euro. Table A1 provides an overview of he series used in his sudy. All series are seasonally adjused excep ineres raes. 21

22 Table A1 Variable Definiion Source Broad moneary aggregaes Euro area: M3 US: M2 Japan: M2 plus cerificaes of deposi UK: M4 Canada: M2+ Real GDP, GDP deflaor and CPI Euro area (HICP) US Japan Canada UK Shor-erm ineres raes Euro area: hree-monh inerbank rae (unil 29 December 1998); hree-monh EURIBOR (hereafer) US Japan Canada UK ECB US Federal Reserve Board (press release H6) Bank of Japan Bank of England Bank of Canada Eurosa OECD Main Econ. Indicaors OECD Main Econ. Indicaors OECD Main Econ. Indicaors OECD Main Econ. indicaors ECB Toal commodiy prices Commodiy Price Index HWWA Real (CPI-based) effecive ECB exchange rae of he euro Aggregaion of he bilaeral exchange rae of he euro agains 12 parner counries OECD Main Econ. Indicaors OECD Main Econ. Indicaors OECD Main Econ. Indicaors OECD Main Econ. indicaors 22

23 References Amisano, G., Giannini, C., Topics in srucural VAR economerics. Springer, Heidelberg and New York. Bachmeier, L.J., Swanson, N.R., 25. Predicing inflaion: does he quaniy heory help? Economic Inquiry 11, Baks, K., Kramer, C., Global liquidiy and asse prices: measuremen, implicaions and spillovers. Inernaional Moneary Fund. Working Paper 1999/168. Bruggeman, A., Camba-Méndez, G., Fischer, B., Sousa, J., 25. Srucural filers for moneary analysis. European Cenral Bank. Working Paper 47. Canova, F., 25. The ransmission of US shocks o Lain America. Journal of Applied Economerics 2, Chrisiano, L., Eichenbaum, M., Evans, C., Moneary policy shocks: wha have we learned and o wha end? In: Taylor, J., Woodford, M. (Eds.), Handbook of Macroeconomics, Norh Holland, Amserdam. Coenen, G., Vega, J.L., 21. The demand for M3 in he Euro Area. Journal of Applied Economerics 16, De Grauwe, P., Polan, M., 25. Is inflaion always and everywhere a moneary phenomenon? Scandinavian Journal of Economics 17, Dedola, L., Lippi, F., 25. The moneary ransmission mechanism: evidence from he indusries of five OECD counries. European Economic Review 49, Dornbusch, R., Favero, C., Giavazzi, F., The immediae challenge for he European Cenral Bank. Economic Policy 13, Giordani, P., 24. An alernaive explanaion of he price puzzle. Journal of Moneary Economics 51, Gordon, D.B., Leeper, E.M., The dynamic impacs of moneary policy: an exercise in enaive idenificaion. Journal of Poliical Economy 12, Grilli, V., Roubini, N., Liquidiy and exchange raes: puzzling evidence from he G7 counries. Salomon Cener, New York Universiy. Working Paper 31. Hendry, D.F., Mizon, G.E., Exogeneiy, causaliy, and co-breaking in economic policy analysis of a small economeric model of money in he UK. Empirical Economics 23, Holman, J.A., Neumann, R.M., 22. Evidence on he cross-counry ransmission of moneary shocks. Applied Economics 34, Kim, S., Do moneary policy shocks maer in he G-7 counries? Using common idenifying assumpions abou moneary policy across counries. Journal of Inernaional Economics 48, Kim, S., 21. Inernaional ransmission of US moneary policy shocks: evidence from VARs. Journal of Moneary Economics 48, Kim, S., Roubini, N., 2. Exchange rae anomalies in he indusrial counries: a soluion wih a srucural VAR approach. Journal of Moneary Economics 45,

24 Leeper, E.M., Roush, J.E., 23. Puing M back in moneary policy. Journal of Money, Credi, and Banking 35, McKinnon, R.I., Currency subsiuion and insabiliy in he world dollar sandard. American Economic Review 72, Mojon, B., Peersman, G., 23. A VAR descripion of he effecs of moneary policy in he individual counries of he Euro Area. In: Angeloni, I., Kashya, A., Mojon, B. (Eds.), Moneary policy ransmission in he Euro Area, Cambridge Universiy Press, Cambridge. Peersman, G., Smes, F., 23. The moneary ransmission mechanism in he Euro Area: more evidence from VAR analysis. In: Angeloni, I., Kashyap, A., Mojon, B. (Eds.), Moneary policy ransmission in he Euro Area, Cambridge Universiy Press, Cambridge. Pesaran, M.H., Smih, R., Esimaing long-run relaionship from dynamic heerogeneous panels. Journal of Economerics 68, Roffia, B., Zaghini, A., 25. Excess money growh and inflaion dynamics. European Cenral Bank and Bank of Ialy (manuscrip). Sims, C.A., 198. Macroeconomics and realiy. Economerica 48, Sims, C.A., Zha, T.A., Does moneary policy generae recessions? Working Paper 12, Federal Reserve Bank of Alana. Sousa, J., Zaghini, A., 24. Moneary policy shocks in he Euro Area and global liquidiy spillovers. European Cenral Bank. Working Paper 39. Yamada, H., 2. M2 demand relaion and effecive exchange rae in Japan. Applied Economic Leers 7,