Staff Paper No. 72 ECONOMIC SHOCKS AND EXCHANGE RATE AS A SHOCK ABSORBER IN INDONESIA AND THAILAND. Siwei Goo and Reza Siregar

Transcription

1 Saff Paper No. 72 ECONOMIC SHOCKS AND EXCHANGE RATE AS A SHOCK ABSORBER IN INDONESIA AND THAILAND Siwei Goo and Reza Siregar

2 Saff Paper No. 72 ECONOMIC SHOCKS AND EXCHANGE RATE AS A SHOCK ABSORBER IN INDONESIA AND THAILAND Siwei Goo and Reza Siregar Augus / Ausralian Chamber of Commerce and Indusry, (Corresponding Auhor) siwei.goo@gmail.com 2/ The Souh Eas Asian Cenral Banks (SEACEN) Research and Training Cenre, Kuala Lumpur, Malaysia. E- mail: reza@seacen.org and rezasiregar@yahoo.com. Acknowledgmen: We would like o hank Michael Aris and Michael Ehrmann for sharing heir compuer code. Excellen research assisance by Bernard Ting is also acknowledged. The iniial draf of he paper was compleed during he corresponding auhor s say a he School of Economics, Universiy of Adelaide, Ausralia, and he second auhor s say a he Inernaional Moneary Fund-Singapore Training Insiue (IMF- STI). The views expressed in his paper are he auhors and do no necessarily represen hose of The SEACEN Cenre, he Ausralian Chamber of Commerce and Indusry, he Universiy of Adelaide and he IMF-STI. ii

3 Absrac This sudy invesigaes he requiremen for he exchange rae o be a shock absorber in Indonesia and Thailand from 1986 o In general, we find ha he economic shocks have predominanly been asymmeric relaive o he US and he Japanese economies. Ye, he weighs aached o he US dollar remain respecably high in he exchange rae managemen of he rupiah and he bah, in paricular for he laer currency, during he pos-1997 crisis. Hence, relinquishing he role of he exchange rae as a shock absorber has been cosly during boh he pre-and he pos-1997 crisis periods for hese Souheas Asian counries. Furhermore, i is arguably more cosly for Thailand during he pos-1997, and for Indonesia during he pre-1997 crisis. JEL Classificaion: C32, E42 and F31 Key Words: Economic Shocks; Shock Absorber; Exchange Rae; Srucural Vecor Auoregression; Indonesia; Thailand iii

4 Table of Conens Page Absrac Table of Conens Lis of Tables Lis of Figures ii iii iv iv 1. Inroducion 1 2. Lieraure Review 2 3. Empirical Mehodologies The Srucure of he VAR Model of Aris and Ehrmann (2006) Impulse Response Funcions Daa Descripion and Uni Roo Properies Domesic Oupu Nominal Exchange Rae Foreign Shor-erm Ineres Rae Domesic Shor-erm Ineres Rae Domesic Price Tes Resuls Esimaes for ω Have he Rupiah and Bah Exchange Raes Been Shock Absorbers? Concluding Remarks 14 References 15 iv

5 Lis of Tables Table 1: Srucural VAR Idenifying Resricions 17 Page Table 2: Table 3: The Esimaes of he Weigh Aached o Exchange Rae Managemen (ω ) 18 Summary of Key Findings on he Role of Exchange Rae as a Shock Absorber 19 Lis of Figures Figure 1: Figure 2: Figure 3: Figure 4: Figure 5: Figure 6: Figure 7: Figure 8: Impulse Response Funcions for he Indonesian Rupiah agains he US Dollar during he Pre-crisis Period 20 Impulse Response Funcions for he Indonesian Rupiah agains he Japanese Yen during he Pre-crisis Period 21 Impulse Response Funcions for he Indonesian Rupiah agains he US Dollar during he Pos-crisis Period 22 Impulse Response Funcions for he Indonesian Rupiah agains he Japanese Yen during he Pos-crisis Period 23 Impulse Response Funcions for he Thai Bah agains he US Dollar during he Pre-crisis Period 24 Impulse Response Funcions for he Thai Bah agains he Japanese Yen during he Pre-crisis Period 25 Impulse Response Funcions for he Thai Bah agains he US Dollar during he Pos Crisis Period 26 Impulse Response Funcion for he Thai Bah agains he Japanese Yen during he Pos-crisis period 27 v

6 1. Inroducion Among a number of perennial macroeconomic policy debaes, he conenion on finding he appropriae exchange rae regime coninues o arac a large number of sudies. In general, hese early works prescribe a more flexible exchange rae for any counry moving o liberalise is domesic economy. Eichengreen (1994), Diaz-Alejandro (1985), Chang and Valesco (2000) and Wyplosz (2001) for insance, conend ha i is crucial o realise ex ane ha liberalisaion rocks foreign exchange markes, and building some form of exchange rae flexibiliy (eiher by floaing or by being prepared o realign pegs) ino he liberalisaion programme is, herefore, essenial. The argumen linking a more flexible exchange rae o economic liberalisaion does no, however, advance wihou rebukes and caveas. The source of financing and he degree of financial openness also maer. Goldfajn and Olivares (2001) conclude ha flexible regimes are viable in financially open economies, provided ha exernal financing is no based on very volaile capial. In he absence of full inernaional financial markes, Devereux (2004) demonsraes heoreically ha, alhough a flexible exchange rae acs perfecly as a shock absorber of global demand shocks, in welfare erms i may in fac be beer off o preven he exchange rae from moving a all (pg.360). Moreover, he characerisics and feaures of real economic shocks, i.e., demand and supply shocks, have also been idenified as oher influenial deerminans of a counry s exchange rae regime. When hese shocks are largely symmeric relaive o he major rading parners, he common view in he lieraure is ha he flexibiliy of he exchange rae policy should, herefore, no be an issue. However, if hey are primarily asymmerical, hen he role of exchange rae as a shock absorber is desirable. Despie he imporance of he issue, paricularly for designing effecive moneary and exchange rae policies, and he numerous works on he developed economies, very lile work has been carried ou on he emerging markes, paricularly on he major Souheas Asian economies ha had experienced severe financial crisis abou a decade ago. 1 Ye, a similar se of moneary and exchange rae policy challenges resurfaced wih he global oubreak of he sub-prime crisis saring In he mids of he global financial marke uncerainies, major Souheas Asian economies, such as Indonesia, Malaysia, Philippines, Thailand and Singapore, have been forced o reassess and adjus heir moneary and exchange rae prioriies. To help fill in he gap in he lieraure, he primary objecive of our paper is o examine he experiences of wo mos-severely affeced economies by he 1997 Eas Asian crisis, namely Indonesia and Thailand, during he pas wo decades. In paricular, we wish o address he following wo quesions: a) Have he economic shocks ha Indonesia and Thailand have had o face relaive o heir wo key rading parners, namely Japan and he US, during he previous wo decades, been predominanly symmerical or asymmerical ones? 1 We will review work being done on hese issues in Secion 2 of he paper.

7 b) Has here been any necessiy for he exchange rae o be a shock absorber in hese wo Souheas Asian economies since he mid-1980s? Has his requiremen become more perinen during he pos-1997 financial crisis? To achieve our objecives, we would employ a srucural VAR framework wih longrun and shor-run resricions of Aris and Ehrmann (2006). This approach incorporaes a mehodology inroduced by Smes (1997) o idenify domesic moneary policy shocks, especially on he weighs ha he moneary auhoriies aached o he exchange rae. These approaches enable us o disenangle he role and conribuion of he exchange rae and ineres rae in moneary policy shocks. A he final sage, he impulse response analyses will be employed o explore he characerisics of he economic shocks and he requiremen for he exchange rae o be a shock absorber in Indonesia and Thailand during he pas wo decades. This paper proceeds as follows. Secion 2 briefly reviews he lieraure while Secion 3 inroduces he empirical approaches of Aris and Ehrmann (2006) and Smes (1997). The following Secion (4) discusses he daa ses and he uni roo properies. Secion 5 on es resuls exensively analyses he empirical findings and addresses he se of quesions posed earlier. The conclusion Secion (6) ends he paper. 2. Lieraure Review Numerous effors have been exended o idenify he ypes of economic shocks and heir impacs on he local economies in he pas wo decades. Mos of hese sudies adoped a srucural VAR framework wih long-run idenifying resricions, inroduced by he seminal work of Blanchard and Quah (1989), as he basic framework o heir empirical works. Lasrapes (1992) for insance, finds ha he real and nominal exchange rae flucuaions during he floaing exchange rae period are due primarily o real shocks in six developed economies, namely he Unied Saes, he Unied Kingdom, Germany, Ialy, Japan and Canada. The sudy, however, fails o disinguish wheher he real shocks originaed from he demand or he supply shocks. Exending he bivariae VAR model of Lasrapes (1992), Clarida and Gali (1994) esimae a hree-equaion open macroeconomic model o idenify hree srucural shocks, i.e., wo real shocks (demand and supply) and moneary (nominal) shocks, o examine he sources of real exchange rae flucuaions in Japan, Germany, Briain and Canada during he pos Breon Woods period. The rivariae VAR model consiss of he relaive oupu growh, he firs difference in real exchange rae and he relaive rae of inflaion, in which all of he hree variables under consideraion are specified as relaive o he corresponding variables of he large neighbours. Based on heir findings, Clarida and Gali (1994) argue ha for Germany and Japan, nominal shocks as well as demand shocks explain a subsanial amoun of he variance in he dollar-deusch mark and he dollar-yen real exchange raes. However, nominal shocks explain very lile of he variance in he real exchange rae for Canada and Briain. Similarly, supply shocks in all cases accoun for very lile variance of he real exchange rae relaive o he dollar. Applying he long-run idenificaion scheme of Clarida and Gali (1994), Chadha and Prasad (1996) and Thomas (1997), examine he relaionship beween he real exchange rae 2

8 and he business cycle in Japan and Sweden respecively, afer he collapse of Breon Woods sysem. In addiion, boh sudies specify he variables in heir VAR sysems in relaive erms o Japan s and Sweden s foreign counerpars. From heir srucural decomposiion, Chadha and Prasad (1996) find ha he variaion in Japan s oupu growh was largely due o supply shocks while demand shocks played an imporan bu only secondary role. The real exchange rae change, on he oher hand, was driven equally by demand and nominal shocks wih supply shocks performing a minor role. Moreover, he variaion in he rae of inflaion was mainly influenced by supply and nominal shocks wih he laer shocks becoming increasingly significan a longer horizons. On he oher hand, Thomas (1997) probes he relaive imporance of real and nominal shocks in explaining he flucuaions of he real exchange rae in Sweden. The policy objecive of he sudy is o assess he poenial cos of giving up exchange rae as he insrumen of he macroeconomic policy adjusmen in Sweden, i.e., as a shock absorber, required for paricipaing in he Economic and Moneary Union (EMU). The sudy finds ha real shocks (demand and supply shocks) accoun for over 60 percen of he forecas error variance of he real exchange rae. Moreover, demand shocks accoun for a significanly higher fracion of he real shocks in Sweden as compared o oher core EMU counries. The sudy conends ha he cos of abandoning he exchange rae as an insrumen of moneary policy is no higher and may be lower for Sweden han for mos of he main EMU counries. An ineresing work by Smes (1997) appraises he role of he European Currency Uni (ECU) exchange rae in he moneary policy sraegy of France, Germany and Ialy. The principal conribuion of his sudy is in is aemp o disinguish beween ineres rae and exchange rae innovaions ha are due o domesic moneary policy shocks and hose ha are due o moneary auhoriies response o exchange rae movemens. The sudy proposes an approach whereby he firs empirical sep is o esimae he weigh of exchange rae in he shor-run reacion funcion of he domesic moneary auhoriy using he insrumen variable echnique, and hen employs he esimaed weighs of exchange rae o idenify he domesic moneary policy shock and is effecs on oupu, prices, ineres rae and exchange rae. The sudy claims ha he responses of oupu, prices, ineres rae and exchange rae o moneary policy shocks in hese hree major European economies are consisen wih he sandard open economy model. In paricular, aking ino accoun he role of exchange rae in he moneary policy formulaion explicily, he sudy is able o solve he exchange rae puzzle, i.e., exchange rae depreciaes insead of appreciaing, following a moneary policy ighening. In heir recen works, Aris and Ehrmann (2006) explore he following imporan policy quesion: Was he exchange rae a shock absorber or a source of is own desabilising shocks in he Unied Kingdom, Canada, Sweden and Denmark? Their sudy applies a srucural VAR framework wih he mix of long run and shor run idenificaion resricions pioneered by Gali (1992). They claim ha exchange rae can only ac as a shock absorber if he domesic economy is hi by an asymmeric shock as compared o is foreign counerpars. Mos of early works, however, failed o deermine if real shocks ha hi domesic economy are symmeric or idiosyncraic in naure, largely due o he fac ha he key variables in heir VAR sysems are consruced in he relaive erm o he corresponding variables of he neighbouring counries (Smes, 1997; Thomas, 1997). Under his ype of consrucion, i becomes difficul o separae wheher shocks originaed from he domesic economy or rading parners economies. 3

9 To overcome his shorcoming, Aris and Ehrmann (2006) include he foreign ineres rae variables ino heir VAR sysems ha are no formulaed in relaive erm. Moreover, Aris and Ehrmann (2006) apply he Smes (1997) approach o disenangle domesic moneary policy shocks from exchange rae shocks using he calculaed weighs which cenral banks aached o exchange rae developmen when seing heir domesic moneary policies. Their sudy concludes ha real shocks are predominanly symmeric relaive o he neighbours in all focused counries excep for he Unied Kingdom. Hence, here is lile need for he exchange rae o ac as a shock absorber in Canada, Sweden and Denmark. In addiion, shocks ha are generaed from he exchange rae marke appear o play a more imporan role in Denmark han oher counries. They, herefore, argue ha moneary union is easier o recommend for Denmark and Sweden han i is for Canada and he Unied Kingdom. Despie hese numerous effors, o our knowledge, very lile work has looked ino hese imporan moneary policy issues for he Eas Asian economies. Haaiseree (1998) examines he role of exchange rae in Thailand s Moneary Condiion Index (MCI) 2, and claims ha during he period from 1990 o 1998, he weigh of he exchange rae in Thailand s MCI is 0.33 for he bah nominal effecive exchange rae, and ha exchange rae plays an increasing role in ransmiing he effecs of moneary policy o he real economy. Using a simple VAR framework, Fung (2002) examines he effecs of moneary policy shocks in seven Eas Asian economies, including Indonesia and Thailand during 1985 o 2001, he pre-1997 and he pos-1997 Asian financial crisis. The sudy employs he Bernanke and Mihov (1998) mehodology o calculae he weigh of he exchange rae in he MCI for he nominal effecive exchange rae o capure explicily he imporance of he exchange rae in he domesic moneary policy measures. The auhor furher assumes ha domesic ineres rae is he policy insrumen and domesic moneary policy shocks do no affec he exchange rae conemporaneously. The resuls, however, were no conclusive, especially for he esimaed weighs of he exchange rae which evenually had o be replaced by weighs of he exchange rae based on he openness of he economy. As saed earlier, our sudy aims a exending and enriching he curren lieraure on he role of he exchange rae in he developing counries by specifically examining he cases of Indonesia and Thailand during he period from 1986 o Applying he srucural VAR idenificaion scheme wih shor-run and long-run idenifying resricions of Aris and Ehrmann (2006), we hope o invesigae he requiremen for he exchange rae o be a shock absorber in hese economies during he las wo decades. Furhermore, by harnessing he approach inroduced by Smes (1997) o esimae he role of exchange rae when idenifying he domesic moneary policy shocks, we can resolve he endogeneiy problem beween he exchange rae and domesic ineres rae in he Haaiseree (1998) and Fung (2002) sudies. 2 The moneary condiion index (MCI) is a weighed average of shor erm ineres rae and exchange rae, and is ofen used o measure he sance of moneary policy in open economies. 4

10 3. Empirical Mehodologies Turning ino he empirical approaches, his Secion would firs summarise he srucural VAR framework wih he long-run and shor-run resricions of Aris and Ehrmann (2006) o idenify five srucural shocks, i.e., he supply and demand shocks, he foreign and domesic moneary policy shocks and he exchange rae shocks. We would hen briefly inroduce he basic frameworks of he Smes (1997) mehodology for esimaing he weighs of exchange rae when he moneary auhoriies in Indonesia and Thailand se heir domesic moneary policy measures. Once he srucural VAR model has been idenified, he impulse response funcions and he forecas error variance decomposiions can hen be esimaed. 3.1 The Srucure of he VAR Model of Aris and Ehrmann (2006) Following Aris and Ehrmann (2006), he srucural VAR model in his sudy can be * represened by x [ y, i, i, p, e ], where all variables excep ineres raes are in naural log. y denoes he domesic oupu growh, i is he domesic shor-erm nominal ineres rae, * i is he foreign shor-erm nominal ineres rae, p denoes he domesic rae of inflaion, and e denoes he rae of appreciaion of he nominal exchange rae of home currency agains is foreign currencies. The domesic economy is subjec o 5 srucural S d m* m e shocks ε [ ε, ε, ε, ε, ε ]. Here, we have wo real shocks, namely he supply shocks ( ε ) and he demand shocks ( ε ), and hree nominal shocks, including he foreign moneary s m* d policy shocks ( ε ), he domesic moneary policy shocks ( ε ), and he exchange rae shocks ( ε ). e As in Gali (1992) and Aris and Ehrmann (2006), our srucural VAR model uilises a combinaion of long-run and shor-run idenifying resricions. To sar wih, he following five ses of resricions are o be imposed in our VAR sysem o recover he srucural model. 3 a) Firs, we impose an orhogonaliy condiion among he five srucural shocks, which implies ha he srucural shocks are muually uncorrelaed. This resricion implies ha he channels hrough which each srucural shock will affec he economy are lef unconsrained. b) The second se of resricions will differeniae he supply shocks from he four remaining shocks by assuming ha only supply shocks have a long-run effec on oupu following Blanchard and Quah (1989). c) The hird se of resricions will disinguish he demand shocks from he remaining hree nominal shocks by assuming ha demand shock is he only shock ha can influence oupu conemporaneously. d) The fourh se of resricions will sor ou he foreign moneary policy shocks from he domesic moneary policy shocks and he exchange rae shocks. Our m 3 The deail implemenaion of he idenifying resricions imposed o recover he srucural model will be fully elaboraed in he nex sub-secion. 5

11 sudy assumes ha foreign ineres rae does no respond conemporaneously o domesic moneary policy shocks or o exchange rae shocks. This resricion is reasonable for he purpose of his sudy since Indonesia and Thailand are arguably small open economies relaive o heir major rading parners, such as he US and Japan. e) Finally, o disenangle he wo ypes of domesic nominal shocks (i.e. domesic moneary policy shocks and exchange rae shocks), we would firs calculae he weigh, ω, which cenral banks aach o exchange rae developmen when seing moneary policy as inroduced by Smes (1997). Once all he effecs of supply, demand and foreign moneary policy shocks on he domesic ineres rae and he exchange rae have been removed, he unexplained componens of he ineres rae and he exchange rae can, herefore, be aribued enirely o domesic moneary policy and exchange rae shocks. The reduced-form empirical model of moneary policy behaviour and he foreign exchange marke can hen be presened as follows: i u e u = ε (1) m α1 ε + α 2 m β1 ε + β 2 e = ε (2) e i e Where: he lef-hand side variables, u and u, are he reduced-form residuals for domesic ineres rae and exchange rae respecively. Equaion 1 denoes ha he cenral bank conrols domesic shor-erm ineres rae and adjuss his insrumen eiher o changes in he sance of domesic moneary policy ( ε ) or in e response o foreign exchange marke disurbances ( ε ). Similarly, he curren exchange rae also depends on domesic moneary policy shocks and exchange rae shocks (Equaion 2). m Solving Equaions (1 and 2) for domesic moneary policy shocks, ε, in erms of he reduced-form ineres rae and exchange rae residuals yields: m β 2 i α 2 e ε = u u (3) α β α β α β α β = 1, β 1 = Normalising ( α 1) and he sum of he weighs on he domesic ineres rae and exchange rae residuals o one, resul in he following expression: ε = ( 1 ω) u + ωu (4) m i e m Equaion 4 can be inerpreed as he shor-run moneary condiion index (MCI). The relaive weigh of he exchange rae in he MCI is given by ω = α 2 /( β 2 α 2 ). In a successful idenificaion scheme, one would expecα 2, which capures he effec of exchange rae shocks on domesic ineres rae, o be non-posiive since an appreciaion of exchange rae should lead o a fall in ineres rae; and β 2, which capures he effec of exchange rae shocks on exchange rae iself, o be posiive. Therefore, he relaive weigh of exchange rae in MCI, ω, should lie beween zero and one. 6

12 If he value of ω is known, he remaining idenificaion problem is solved since one can define he domesic moneary policy shocks according o Equaion 4. One of he main advanages of focusing on his weigh o idenify domesic moneary policy shocks is ha i encompasses no only wo exreme cases of ineres rae and exchange rae argeing bu also he inermediae cases, which is very much relaed o he experiences of Indonesia and Thailand. In a pure ineres rae argeing regime, he relaive weigh of he exchange rae on he MCI will be equal o zero ( ω = 0 ) since he moneary auhoriies ignore he effec of he exchange rae when seing he domesic ineres rae ( α 2 = 0). On he oher hand, in a pure exchange rae argeing regime ( ω = 1), here will be a one-o-one relaionship beween domesic moneary policy shocks and exchange rae innovaions since he cenral bank will no allow he foreign exchange marke disurbances o affec he exchange rae ( β 2 = 0 ). For he case of Indonesia and Thailand, one would expec ha he relaive weigh of he exchange rae in domesic moneary policy measure o be some posiive value, since he moneary auhoriies in hese counries acively manage heir exchange rae. Smes (1997) and Aris and Ehrmann (2006) sugges ha ω can be esimaed by ransforming Equaion 4 ino he following regression model: u i ω = u 1 ω e 1 + ε 1 ω m (5) Equaion 5 implies ha he observed reduced-form residuals for ineres rae ( u ) is explained by he observable reduced-form residuals for exchange rae ( u ) and a random 1 m shock, ε. We esimae Equaion 5 by adoping he Shapiro and Wason (1988) 1 ω sequenial insrumen variable echnique, using shocks o he Ausralia RBA s cash rae and he exchange rae of he deuschemark agains he Ausralian dollar as insrumens. 4 The shocks o he insrumens are obained by regressing each of he insrumen variables on is own lags, he lags of he endogenous variables in he VAR sysems, and he esimaed supply, demand and foreign moneary policy shocks. 5 e i 4 5 For insrumen variables, we need o adop he policy rae and he exchange rae of he major rading parners of Indonesia and Thailand. Since relevan key economic indicaors from he Japanese and he US economies are already considered explicily in our srucural VAR sysem, we op o use he Ausralia RBA s cash rae and he Ausralian dollar agains he deuschemark o be he insrumenal variables for Equaion 5. The Ausralian dollar is chosen based on wo crieria. Firs, he counry is one of he major economic parners of Indonesia and Thailand (refer o Bowman (2005)). Second, he Ausralian exchange rae policy is considered among he mos flexible regime in he world. The lag lengh chosen o esimae ω is repored in Table 2 and hese lags are chosen o minimise he AIC crieria and o ensure ha he residuals are well specified. 7

13 3.1.1 The Specificaion and Implemenaion of Idenifying Resricions As inroduced earlier, [ y, i, i, p, e ] 5 1 vecor. I is assumed ha x is a covariance saionary vecor process, and i has he following vecor moving average represenaion form: * x is a ( ) x = C(L)ε (6) S d m m e where ε [ ε, ε, ε, ε, ε ] is he 5 by 1 vecor of serially uncorrelaed srucural disurbances (i.e. he supply shocks, demand shocks, foreign moneary policy shocks, domesic moneary policy shocks and exchange rae shocks respecively). The 5 5 marix of polynomial lags C( L) [ C ( L)], for i, j = 1,..., 5 is he objec o be esimaed. Once he * ij marix C (L) has been esimaed, he expressions for he levels of differen variables in erms of he curren and lagged values of he srucural disurbances can be recovered. The reduced-form Wold moving average represenaion of x is: x = R(L)u (7) where u is a 5 by 1 vecor of reduced-form disurbance, and R( L) [ R ( L)], for i, j = 1,...,5, R ( 0) = I and R (L) is inverible. Moreover, le Σ denoes he symmeric variance-covariance marix of he reduced-form innovaions, u, ha is, Σ = Euu'. The reduced-form auoregressive represenaion of Equaion 7 is given by: B (L)x = u (8) where B( L) [ B ( L)] for i, j = 1,..., 5, B ( 0) = I and ij 1 B ( L) = R( L). We assume ha here exiss a non-singular marix S, such ha he reduced-form innovaions in u are o be he linear combinaions of he srucural disurbances in ε : u = Sε (9) Therefore, from Equaions (6, 7 and 9): C ( L) = R( L) S (10) 1 Pre-muliplying boh sides of Equaion 8 by S, he vecor auoregressive represenaion of x in erms of he srucural disurbances ε can be derived: A (L)x =ε (11) 1 where A( L) [ A ( L)], for i, j = 1,..., 5 and A (0) S. ij ij 8

14 The srucural model, i.e. he coefficiens of A (L) and C (L), can be idenified once enough resricions are imposed o deermine all componens of he marix S. 6 Turning ino he deail operaion of he idenificaion scheme, he assumpion of muually uncorrelaed shocks, ogeher wih a convenien normalisaion ha each srucural shock has a uni variance, imply ha E εε ' = I. Thus from Equaion 9, SS ' = Σ (12) Equaion 12 represens fifeen resricions on marix S, given Σ. Therefore, en addiional resricions need o be imposed o jus-idenify he srucural model (Table 1). The long run idenifying resricions of R1 o R4, imply ha C 12 ( 1) = C13 (1) = C14 (1) = C15 (1) = 0. Thus given Equaion 10, he following linear resricions on marix S can be derived, respecively: C 1) R (1) S + R (1) S + R (1) S + R (1) S + R (1) S 0 (13) 12 ( = C 1) R (1) S + R (1) S + R (1) S + R (1) S + R (1) S 0 (14) 13 ( = C 1) R (1) S + R (1) S + R (1) S + R (1) S + R (1) S 0 (15) 14 ( = C 1) R (1) S + R (1) S + R (1) S + R (1) S + R (1) S 0 (16) 15 ( = Resricions R5 o R9 imply he following five direc consrains on marix S: S 13 = 0 (17) S 14 = 0 (18) S 15 = 0 (19) S 24 = 0 (20) S 25 = 0 (21) 6 Consisen esimaes of he coefficiens in marix B (L) can be obained from Equaion 8 using he ordinary leas square (OLS) echnique. The esimae of marix R (L) can hen be obained by invering marix B (L). A consisen esimae of he symmeric variance-covariance marix of he reduced form disurbances, Σ, can be compued using he residuals of he OLS regression. From Equaion 10, given he marix S, he marix C (L) can be recovered by pos-muliplying marix R (L) by marix S ; while from Equaions (8, 9 and 11), marix A (L) can be recovered by pre-muliplying marix B (L) by he inversion of marix S. 9

15 By consrucion, elemen S ij measures he conribuion of jh srucural shock o he conemporaneous innovaion in he ih elemen of vecor x, i.e. he reduced-form residuals of ih variables in vecor x. Finally, resricion R10 can be implemened as a linear resricion involving some of he elemens of marix A (0). From Equaions 9 and 11, A (0)u = ε. Therefore, he fourh row in marix A (0) is associaed wih he domesic moneary policy shocks. Given he specificaion of vecor x and Equaion 4, R10 implies he following consrain on marix A (0) : A 43 (0) = 1 ω and A 45 (0) = ω (22) Given 1 S A(0), Equaion 22 maps ino one resricion on he elemen of marix S. Wih he idenifying resricions, he srucural model (Equaions 6 and 11) can be recovered from he reduced-form model (Equaion 8). We esimae he srucural VAR model in urn for Indonesia and Thailand for he pre-crisis period (January 1986 o December 1996) and he pos-crisis period (January 2000 o December 2007). For Indonesia, he model is esimaed using 6 lags for boh he pre- and pos-crisis period; while for Thailand, he model is esimaed using 5 lags and 4 lags respecively for he pre- and pos-crisis period 7. In addiion, since i has been a common pracice in moneary VAR models, he world oil price inflaion and linear ime rend are also included in our VAR model as exogenous variables Impulse Response Funcions Once he srucural VAR model has been idenified, we hen move o examine he wo main quesions posed in he inroducion of his sudy. For ha, we generae he impulse response funcions of each key variable under he presence of he five srucural disurbances in he VAR sysems, for boh he pre- and pos-1997 crisis period. Suiably ransformed, he esimaes of C (L) in Equaion 6 can express all he variables of ineres, i.e., oupu, foreign ineres rae, domesic ineres rae, domesic prices and domesic exchange rae in erms of he sum of he disribued lags of he five srucural disurbances. Formally, z = F(L)ε (23) where z [ y, i, i, p, e ] and F( L) [ F ( L)] for i, j = 1,..., 5. The coefficiens of he * ij polynomial lag F ij (L) give he esimaed dynamic response of vecor z s ih variable o a one sandard deviaion realisaion of he jh srucural disurbance. Noe ha he variables in he 7 8 Since he srucural VAR involved long-run idenifying resricions, his sudy adops he above menioned lag lengh o ensure ha he impulse response funcions are well idenified, alhough in some cases he VAR sysems wih 3 lags have a smaller AIC saisic. The RATS programme code used o esimae his Secion s srucural VAR models is adaped from Aris & Ehrmann (2006). 10

16 impulse response funcion are now presened in levels for oupu, prices and exchange rae raher han in heir growh raes. As discussed, he need for he exchange rae o be a shock absorber arises only when he real economic shocks are asymmerical relaive o he counry s rading parners. Shocks ha are predominanly asymmeric require opposed responses of foreign and domesic moneary policy. Accordingly, he rends of he shor-run impulse response funcions of oupu, foreign ineres rae, domesic ineres rae, prices and exchange rae o he five srucural shocks should reveal informaion needed o address our research queries. 4. Daa Descripion and Uni Roo Properies The monhly daa ses from January 1986 o December 2007 are obained from he Inernaional Financial Saisics of he Inernaional Moneary Fund and he DaaSream. Since his sudy is ineresed in examining he changes in he role of he exchange rae during he pos-crisis period compared o he pre-crisis period, he observaion se will be grouped ino he pre-crisis period from January 1986 o December , and he pos-crisis period from January 2000 o December In addiion, he uni roo properies for each variable will also be discussed. In his sudy, we employ hree differen uni roo ess, namely he Augmened Dickey-Fuller (ADF) es, he Philip Perron (PP) es, and he Kwiakowski, Philips, Schmid and Shin (KPSS) es for robusness Domesic Oupu Since he gross domesic produc (GDP) daa is only available quarerly, he indusrial producion or manufacuring producion index are used as a proxy for domesic oupu (y). The oupu level is expressed in naural log. From he uni roo ess, i can be concluded ha in mos cases, he oupu variable is firs-differenced saionary for Indonesia and Thailand for boh sample periods. 4.2 Nominal Exchange Rae To examine he imporance of he shocks relaive o he wo major rading parners (he US and Japan), our sudy employs wo definiions of nominal exchange rae (e), namely he bilaeral nominal exchange rae agains he US dollar and he bilaeral nominal exchange rae agains he Japanese yen. The nominal exchange raes are expressed as unis of foreign currency for one uni of domesic currency. Thus, an increase in he exchange rae implies a domesic currency appreciaion. Boh nominal exchange rae variables are expressed in naural log. The uni roo ess indicae ha he nominal exchange raes are firs-differenced saionary for boh Souheas Asian economies during boh sample periods. 9 Due o daa availabiliy, he pre-crisis period for Thailand only sars in January For he sake of breviy, he uni roo es resuls will no be repored. However, hey can be made available upon reques. 11

17 4.3 Foreign Shor-erm Ineres Rae The shor-erm foreign ineres rae is he US federal funds rae when he esimaion involves nominal exchange rae agains he US dollar and Japan s overnigh call rae if i involves he Japanese yen. The foreign ineres rae variables are expressed in percenage. Based on he uni-roo esing, we conclude ha foreign ineres raes are saionary. The findings are also suppored by he esimaed impulse responses, showing ha foreign ineres rae reurns o he baseline quickly wih converging error bounds, which is generally no he case if he variables are inegraed (Aris and Ehrmann, 2000). 4.4 Domesic Shor-erm Ineres Rae As in recen sudies on moneary policy for Indonesia and Thailand, his sudy uses he 90-day Cerificaes of Bank Indonesia (SBI) rae and he Bank of Thailand 14-day repurchase (Repo) rae as he proxies of shor-erm ineres rae conrolled by he cenral banks of hese wo economies (Berg e al., 2003; Fane, 2005; Fung, 2002). 11 The domesic ineres rae variables are expressed in percenage. In general, he uni roo ess reveal ha he domesic ineres rae is saionary for boh counries for boh sub-periods. This finding is once again suppored by he impulse response es resuls. 4.5 Domesic Price The domesic price (p) is measured by he consumer price index (CPI) wih year 2000 as he base year for Indonesia and Thailand. The domesic price is expressed in naural log. Mos of he uni roo ess indicae ha domesic price variables of boh economies are firsdifferenced saionary for boh sub-periods. 5. Tes Resuls 5.1 Esimaes for ω The resuls for he weighs he cenral banks aach o he exchange rae (ω ) for he various periods are repored in Table 2. All weighs for boh he rupiah and he bah are found o be significan. Turning firs o he Indonesian rupiah, our resuls provide evidence of less rigid exchange rae policy agains he US dollar during he pos-1997 crisis period. The size of he (ω ) for he period of is repored o be around (0.39), and drops o (0.28) for he pos-1997 crisis period. In conras, he Japanese yen coninued o receive much less, bu relaively unchanged weigh, siing wihin a narrow range of ( ). These resuls seem o suppor he more flexible official (de-jure) regime of exchange rae adoped by Bank Indonesia since Neverheless, he moneary auhoriy in Indonesia coninues o assign a relaively large weigh o he US dollar, suggesing a managed floa regime of he exchange rae agains he US dollar. 11 Since he daa for he Bank of Thailand 14-day repurchase rae is only available from January 1997, he call money rae is used for he period before he firs daa is available. We did no see any significan jump or drop for he dae when he wo raes are combined. 12

18 In addiion, our resuls seem o sugges ha he Bank of Thailand has basically pursued he same exchange rae regime in he pas wo decades. The weigh agains he US dollar during he pos-crisis has very much been equivalen o is pre-crisis level a around Very much he same analyses can be said as well for he weigh of he bah agains he yen, albei wih a sligh drop in (ω ) during he pos-crisis period. These findings are consisen wih hose of Baig (2001) and McKinnon and Schnabl (2004). 5.2 Have he Rupiah and Bah Exchange Raes Been Shock Absorbers? The Case of he Indonesian Rupiah Turning firs o he case of he Indonesian rupiah, he role of he exchange rae as a shock absorber appears o be much more vial during he pre-1997 financial crisis han during he pos-crisis period. As shown in Figures 1 and 2, demand shocks during he period of were predominanly asymmeric relaive o Japan and he US, and herefore requiring opposed moneary policy responses. In conras, demand shocks during he pos period riggered similar moneary policy responses by he domesic moneary auhoriy (via is cerificae of Bank Indonesia (SBI) rae) and is counerpars in he US (he US federal fund rae) and Japan (he overnigh call rae) (Figures 3 and 4). As for he supply shocks, hey were largely symmerical relaive o he US economy, bu asymmerical relaive o he Japanese economy during he pre-crisis period. The reverse is, however, rue for he pos-crisis period. In summary, while hree ou of four real shocks which ook place before 1997 were asymmeric, he majoriy of hese shocks since 2000 require parallel responses of he domesic and foreign moneary policies (Table 3). Given he naure of he shocks, he cos of relinquishing he role of he exchange rae as a shock absorber should, herefore, be more expensive during he pre-1997 crisis relaive o he pos-crisis period. More imporanly, he commimen of Bank Indonesia o inroduce more flexibiliy in is exchange rae policy, paricularly agains he US dollar as indicaed by he lesser weigh in Table 2, is appropriae o address he asymmerical supply shocks relaive o he US economy in recen years The Case of he Thailand Bah Based on he naures of he shocks, he necessiy o inroduce more flexibiliy in he managemen of he Thailand bah during he pos-1997 crisis is arguably more profound han during he pre-1997 crisis (Figures 5-8 and Table 3). During he pre-crisis period, real shocks were mainly symmerical relaive o he Japanese economy. Hence, he role of he exchange rae as a shock absorber here was no required. However, relaive o he US economy, he shocks were predominanly asymmerical, suggesing a respecably high cos of mainaining he sof-us dollar peg policy during he pre-1997 crisis. Since 2000, however, he wo ypes of real shocks have largely been asymmeric, boh relaive o he US and he Japanese economies. This implies ha here is a rising urgency o have a more credible and independen domesic moneary policy o generae opposie responses o he real shocks, and hus allowing he exchange rae o be a more efficien shock absorber. Ye, as repored earlier, he Bank of Thailand coninued o mainain 13

19 relaively high weighs for he US dollar and only moderaely inroduced more flexibiliy of he bah agains he Japanese yen in is overall managemen of exchange rae policy. 6. Concluding Remarks Wih he liberalisaion of he curren and capial accouns, he requiremen for he exchange rae o be an insrumen of macroeconomic policy adjusmen, i.e., a shock absorber, in Indonesia and Thailand, has been well documened. Mos of hese early works examined he rigidiies of he bah and he rupiah, paricularly agains he US dollar, and heir roles in explaining he curren accoun deficis facing hese economies prior o he collapse of he sof-us dollar peg exchange rae regimes. 12 However, hardly any examined he naure of he real shocks in he domesic economies of hese Souheas Asian naions relaive o he counries major rading parners and he necessiy for he exchange rae o be a shock absorber domesically. Our sudy finds ha real shocks in Indonesia and Thailand relaive o is rading parners, namely he US and Japan, have been predominanly asymmerical during boh he pre- and pos-1997 financial crisis. More imporanly, we come o a conclusion ha he cos for relinquishing he role of he exchange rae as a shock absorber is relaively greaer during he pre-crisis period for Indonesia, and during he pos-crisis period for Thailand. 12 See for insance Siregar and Rajan (2005) and Rahmasyah, e.al. (2002). 14

20 References Aris, M., Ehrmann, M., 2000, The Exchange Rae A Shock Absorber of Source of Shocks? A Sudy of Four Open Economies, European Universiy Insiue Working Papers RSC No. 2000/38. Aris, M., Ehrmann, M., 2006, The Exchange Rae - A Shock-Absorber or Source of Shocks? A Sudy of Four Open Economies, Journal of Inernaional Money and Finance 25(6), Baig, T., 2001, Characerizing Exchange Rae Regimes in Pos-Crisis Eas Asia, IMF Working Paper WP/01/152. Berg, A. G., Jarvis, C. J., Sone, M. R., Zanello, A., 2003, Re-esablishing Credible Nominal Anchors Afer a Financial Crisis: A Review of Recen Experience, IMF Working Paper WP/03/76. Bernanke, B. S., Mihov, I., 1998, Measuring Moneary Policy, Quarerly Journal of Economics 113(3), Blanchard, O., Quah, D., 1989, The Dynamic Effecs of Aggregae Demand and Supply Disurbances, American Economic Review 79(4), Bowman, C., 2005, Yen Bloc or Koala Bloc? Currency Relaionships Afer he Eas Asian Crisis, Japan and he World Economy 17, pp Chang, R., Valesco, A., 2000, Exchange Rae Policy for Developing Counries, American Economic Review 90(2), Papers and Proceedings of he 112 h Annual Meeing of he American Economic Associaion, May, Chadha, B., Prasad, E., 1996, Real Exchange Rae Flucuaions and he Business Cycle: Evidence from Japan, IMF Working Paper WP/96/132. Clarida, R., Gali, J., 1994, Sources of Real Exchange Rae Flucuaions: How Imporan Are Nominal Shocks?, Carnegie-Rocheser Conference Series on Public Policy 41, Devereux, M.B., 2004, Should he Exchange Rae Be a Shock Absorber?, Journal of Inernaional Economics 62, Diaz-Alejandro, C., 1985, Good-Bye Financial Repression, Hello Financial Crash, Journal of Developmen Economics 19, Eichengreen, B., 1994, Inernaional Moneary Arrangemens for he 21 s Brookings Insiuion, Washingon DC. Cenury, The Fane, G., 2005, Pos-Crisis Moneary and Exchange Rae Policies in Indonesia, Malaysia and Thailand, Bullein of Indonesian Economic Sudies 41(2),

21 Fung, B. S. C., 2002, A VAR Analysis of he Effecs of Moneary Policy in Eas Asia, BIS Working Papers No.119. Gali, J., 1992, How Well Does he IS-LM Model Fi Poswar US Daa?, Quarerly Journal of Economics 107(2), Goldfajn, I., Olivares, G., 2001, Can Flexible Exchange Raes Sill Work in Financially Open Economies?, G-24 Discussion Paper Series No. 8, Unied Naions Conference on Trade and Developmen, January. Haaiseree, R., 1998, Moneary Condiions and Moneary Policy in Small Open Economies: Empirical Resuls for Thailand Under he Floaing Rae Regime, Paper presened a he 22nd Inernaional Conference on a Macroeconomic Core of an Open Economy for Progressive Indusrialisaion and Developmen in Asia in he New Millennium, Bangkok, Thailand. Lasrapes, W. D., 1992, Sources of Flucuaions in Real and Nominal Exchange Raes, Review of Economics and Saisics 74(3), McKinnon, R. I., Schnabl, G., 2004, The Eas Asian Dollar Sandard, Fear of Floaing, and Original Sin, Review of Developmen Economics 8(3), Rahmasyah, T., Rajaguru, G., Siregar, R., 2002, Exchange Rae Volailiy, Trade & Fixing For Life In Thailand, Japan and he World Economy 14 (4), Shapiro, M. D., Wason, M. W., 1988, Source of Business Cycle Flucuaions, NBER Macroeconomics Annual 3, Siregar, R., Rajaguru, G., 2005, Sources of Variaion beween he Inflaion Raes of Korea, Thailand and Indonesia during he Pos-1997 Crisis, Journal of Policy Modelling 27(7), Smes, F., 1997, Measuring Moneary Policy Shocks in France, Germany and Ialy: The Role of he Exchange Rae, BIS Working Papers No. 42. Thomas, A., 1997, Is he Exchange Rae a Shock Absorber? The Case of Sweden, IMF Working Paper WP/97/176. Wyplosz, C., 2001, How Risky is Financial Liberalizaion in he Developing Counries?, G-24 Discussion Paper Series No. 14, Unied Naions Conference on Trade and Developmen, Sepember. 16

22 Table 1: Srucural VAR Idenifying Resricions Long run resricions: R1: no long run effec of aggregae demand shocks on oupu R2: no long run effec of foreign moneary policy shocks on oupu R3: no long run effec of domesic moneary policy shocks on oupu R4: no long run effec of exchange rae shocks on oupu Shor run resricions: R5: no conemporaneous effec of foreign moneary policy shocks on oupu R6: no conemporaneous effec of domesic moneary policy shocks on oupu R7: no conemporaneous effec of exchange rae shocks on oupu R8: no conemporaneous effec of domesic moneary policy shocks on foreign ineres rae R9: no conemporaneous effec of exchange rae shocks on foreign ineres rae R10: he conemporaneous weigh,ω, which cenral banks aached o exchange rae developmen when seing domesic moneary policy, ha is, m i e ε = ( 1 ω) u + ωu 17

23 Table 2: The Esimaes of he Weigh Aached o Exchange Rae Managemen (ω ) Pre-Crisis: (1986:1-1996:12) Pos-Crisis: (2000:1-2007:12) Indonesia Thailand RP-USD 1 RP-YEN 2 BAHT-USD 3 BAHT-YEN (3.23)*** (1.88)* (3.72)*** (6.24)*** (4.43)*** (2.82)*** (2.22)** (6.24)*** The -saisics are in brackes. ***, **, * indicae significance a 1%, 5% and 10% Noes: 1). The (ω ) for RP-USD is esimaed using 11 lags, excluding he crisis period observaion (1997:1-1999:12); 2). The (ω ) for RP-YEN is esimaed using 10 lags, excluding he crisis period observaion (1997:1-1999:12); 3). The (ω ) for BAHT-USD is esimaed using 7 lags, excluding he crisis period observaion (1997:1-1999:12); 4). The (ω ) for BAHT-YEN is esimaed using 10 lags, excluding he crisis period observaion (1997:1-1999:12); 18

24 Table 3: Summary of Key Findings on he Role of Exchange Rae as a Shock Absorber a) Agains he US$ Indonesia Thailand Pre-crisis Pos-Crisis Pre-crisis Pos-Crisis Supply shocks Symmeric Asymmeric Asymmeric Asymmeric Demand Shocks Asymmeric Symmeric Asymmeric Asymmeric b) Agains he Japanese Yen Supply shocks Asymmeric Symmeric Symmeric Asymmeric Demand Shocks Asymmeric Symmeric Symmeric Asymmeric Main Conclusion The cos of relinquishing he role of exchange rae as a shock absorber has arguably been more cosly in Indonesia during he pre crisis relaive o poscrisis period. The cos of relinquishing he role of exchange rae as a shock absorber should definiely be more cosly in Thailand during he pos crisis period relaive o pre-crisis period. 19

25 Figure 1: Impulse Response Funcions for he Indonesian Rupiah agains he US Dollar during he Pre-crisis Period Impulse Response Funcions wih w=0.39 Indonesia (1986:1 o 1996:12) Exch. Rae Shock O u p u - O u p u - O u p u - O u p u - O u p u - F o re i g n In e re s Do m. In e re s F o re i g n In e re s Do m. In e re s F o re i g n In e re s Do m. In e re s F o re i g n In e re s Do m. In e re s F o re i g n In e re s Do m. In e re s Exch. Rae Shock Exch. Rae Shock Exch. Rae Shock Ex c h a n g e Ra e - - Ex c h a n g e Ra e - - Ex c h a n g e Ra e - - Ex c h a n g e Ra e - - Ex c h a n g e Ra e - - Exch. Rae Shock 20

26 Figure 2: Impulse Response Funcions for he Indonesian Rupiah agains he Japanese Yen during he Pre-crisis Period Impulse Response Funcions wih w=0.04 Indonesia (1986:1 o 1996:12) Exch. Rae Shock O u p u - O u p u - O u p u - O u p u - O u p u - F o re i g n In e re s Do m. In e re s F o re i g n In e re s Do m. In e re s Do m. In e re s F o re i g n In e re s F o re i g n In e re s Do m. In e re s Do m. In e re s F o re i g n In e re s Exch. Rae Shock Exch. Rae Shock Exch. Rae Shock E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e Exch. Rae Shock 21

27 Figure 3: Impulse Response Funcions for he Indonesian Rupiah agains he US Dollar during he Pos-crisis Period Impulse Response Funcions wih w=0.28 Indonesia (2000:1 o 2007:12) Exch. Rae Shock O u p u O u p u O u p u O u p u O u p u F o re i g n I n e re s Do m. In e re s F o re i g n I n e re s Do m. In e re s Do m. In e re s F o re i g n I n e re s F o re i g n I n e re s Do m. In e re s Do m. In e re s F o re i g n I n e re s Exch. Rae Shock Exch. Rae Shock Exch. Rae Shock E x c h a n g e Ra e - - E x c h a n g e Ra e - - E x c h a n g e Ra e - - E x c h a n g e Ra e - - E x c h a n g e Ra e - - Exch. Rae Shock 22

28 Figure 4: Impulse Response Funcions for he Indonesian Rupiah agains he Japanese Yen during he Pos-crisis Period Impulse Response Funcions wih w=0.06 Indonesia (2000:1 o 2007:12) O u p u O u p u O u p u O u p u O u p u Exch. Rae Shock F o re i g n I n e re s Do m. I n e re s F o re i g n I n e re s Do m. I n e re s F o re i g n I n e re s Do m. I n e re s F o re i g n I n e re s Do m. I n e re s Do m. I n e re s F o re i g n I n e re s Exch. Rae Shock Exch. Rae Shock Exch. Rae Shock E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e Exch. Rae Shock 23

29 Figure 5: Impulse Response Funcions for he Thai Bah agains he US Dollar during he Pre-crisis Period Impulse Response Funcion wih w=0.33 THAILAND (1988:1 o 1996:12) O u p u - O u p u - O u p u - O u p u - O u p u - Exch. Rae Shock F o re i g n I n e re s Do m. I n e re s F o re i g n I n e re s Do m. I n e re s F o re i g n I n e re s Do m. I n e re s F o re i g n I n e re s Do m. I n e re s Do m. I n e re s F o re i g n I n e re s Exch. Rae Shock Exch. Rae Shock Exch. Rae Shock E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e E x c h a n g e Ra e Exch. Rae Shock 24