The Impact of Yuan/Ringgit on Bilateral Trade Balance of China and Malaysia

Transcription

1 MPRA Munich Personal RePEc Archive The Impac of Yuan/Ringgi on Bilaeral Trade Balance of China and Malaysia Chee Wooi Hooy and Tze-Haw Chan School of Managemen, USM, Cenre for Globalizaion and Susainabiliy Research, MMU 1 Ocober 2008 Online a hps://mpra.ub.uni-muenchen.de/11306/ MPRA Paper No , posed 3 November :54 UTC

2 The Impac of Yuan/Ringgi on Bilaeral Trade Balance of China and Malaysia Hooy Chee Wooi Finance Secion, School of Managemen Universii Sains Malaysia Chan Tze Haw Cenre for Globalizaion and Susainabiliy Research Faculy of Business and Law Mulimedia Universiy Absrac The exposure o exchange raes remains an unresolved issue in inernaional rade lieraure. The issue is paricularly relevan o China and Malaysia, whom relaxed heir USD pegging he same day in he mid of Our paper invesigaes he exchange rae exposure of China-Malaysian bilaeral rade balance over he las 20 years using a sandard rade balance equaion which is a funcion of local income, foreign income, and he bilaeral real exchange raes of yuan/ringgi. Our modeling is somewha differen wih he lieraure where we ake ino accoun he srucural breaks of he 1997 Asian currency crisis as well as he fixed-exchange rae regime adoped by he Malaysia. Wih high frequency monhly sample (Jan1990-Jan2008), we documened GAR effec in he rade model. Taking ha ino consideraion, our resul shows ha real exchange raes do play a role in he bilaeral rade of China-Malaysia. The long run exchange rae elasiciy is consisen wih he Marshall-Lerner condiion. However, he shor run J-curve phenomenon is somewha inconclusive. Keywords: rade, exchange raes exposure, J-curve, srucural breaks, GAR. JEL Classificaions: F31 *Corresponding auhor: Dr. Hooy Chee-Wooi Finance Secion, School of Managemen Universii Sains Malaysia USM, Penang, Malaysia Tel: ; Fax: cwhooy@usm.my 1

3 1. Inroducion Since he 1980s, China's economic reforms have been insrumenal in promoing is rade secor. The coninuous high growh of China has had a significan impac on he world economy, paricularly in he Eas Asian region. A subsanial amoun of China s growh over he pas decade has semmed from he coninued surge in her rade surplus due o undervaluaion of renminbi over he years. In a broader perspecive, China has hampers he prospec of her Asian neighbors from leing heir currencies o rise very much agains he USD o avoid losing compeiive posiion agains China. Malaysia, as one of he major rading parners of boh China and he US, is unexcepionally facing he challenge. Along he lines of rade liberalizaion among he Eas Asian economies, Malaysia has acively paricipaed in boh global and regional rading aciviies. The liberalizaion process since 1970s brings o a slash in impor ariff and non-ariff barriers and promoed a surge in her rade wih indusrial counries. The reducion in capial resricion and invesmen friendly policy has also araced inflow of FDI from he indusrial counries and spurred he growh and diversiy of Malaysian economy especially during However, he rising of China in he 1990s was commonly claimed o have divesed away her rade and capial resources because boh Malaysia and China shared quie a comparable facor endowmen raios, range of expor producs (mainly in manufacuring produc), as well as similar direcion of rade o he US and Japan. The recen devaluaion and USD de-pegging of boh China renmimbi and Malaysian rinngi on July 2005 have open a new scenario o he rade secor in boh counries. Since he opening of mainland Chinese economy in 1978, renmimbi was pegged o he USD, and a dual-rack currency sysem was insiued, where renminbi is only usable locally while foreign exchange cerificaes are forced on foreigners. China abolished he dual-rack sysem and inroduced single free floaing currency effecive January 1, 1994 and he renminbi urn freely converible under curren accoun ransacion effecive December In he decade unil 2005, renminbi was ighly pegged a yuan o he USD. On July 21, 2005 People s Bank of China announced he 2.1% revaluaion o 8.11 yuan per USD and move from USD pegging o managed floaing based on a baske of foreign currencies. To dae, he yuan is raded a around 6.95 yuan (June 2008), appreciaed abou 16% since The Malaysian ringgi was rading as a free floa currency a around RM2.50 per USD since early 1970s. During he 1997 Asian financial crisis, afer he sharp depreciaion of ringgi o around RM 4.00 wihin a year, Bank Negara Malaysia (BNM) decided o peg ringgi o he USD in Sepember 1998 a RM3.80. On July 21, 2005, BNM responded o China s de-pegging announcemen wihin an hour by announcing he end of he 7-year pegging. Similar o China, BNM allows he ringgi o operae in a managed floaing sysem based on a baske of several major currencies. The ringgi has appreciaed 1.3% o RM3.75 in a shor period of ime and now has reach RM3.2 (June 2008), abou 15.6% appreciaed from he pegged level a 2005, a value quie near o renminbi appreciaion. The close ied of ringgi o renmimbi implied he Malaysian governmen regards vary seriously on he real exchange rae value of he Malaysian ringgi agains any appreciaion of renminbi as i could hreaen he balance of paymen of Malaysia economy. However, he effecs of currency devaluaion on rade balance sill remain an unsolved issue as he impac of exchange raes mechanism is far from perfecly undersood. Conroversies were abounded, and heoreical as well as applied quesions 2

4 have been raised among academia and policy makers. The issue coninues o be relevan o he undersanding of he exchange rae dynamics and he formulaion of rade policies, paricularly for Malaysia and China. Boh he wo economies were commied o he expor-led growh policy based on he mainenance of heir undervalued currencies bu again boh also have recenly succumbed o he revaluaion pressure by releasing heir pegging agains USD. The issue of exchange rae devaluaion on inernaional rade has long been a major opic of sudy in inernaional economics. The elasiciy approach o balance of paymen was made well known as Marshall-Lerner condiion (MLC henceforh) 1 ha becomes he underlying assumpions of currency devaluaion policy. The foreign exchange insabiliy during he pos-breon Wood era offers an excellen opporuniy o invesigae how exchange rae changes affec rade flows, and wheher he currency devaluaion is expansionary or conracionary. Mos early sudies focused on he US and developed naions (see Krugman and Baldwin, 1987; Rose and Yellen, 1989; Noland, 1989; Rose, 1991; among ohers) bu he findings are a bes mixed when aggregae rade daa were used. Some of hem ried o avoid he aggregaion bias problem by employing daa beween one counry and each of her rading parners a he bilaeral level, and found some suppors for favorable impac of currency depreciaion on rade balances. On he oher hand, lieraure on developing Asian naions show beer suppors for he MLC as long run feaures and J-curve 2 as shor run phenomenon (see iner alia, Himarios, 1989; Hsing and Savvides, 1996; Bahmani-Oskooee and Janardhanan, 1994). Ineresing findings are repored by recen Asian sudies ha consider he crisis experience. For insance, Bahmani-Oskooee and Mieza (2003) find ha devaluaions have been conracionary for Indonesia and Malaysia, bu expansionary for he Philippines and Thailand. Onafowora (2003) employs a coinegraion approach o find ha bilaeral rade, real exchange raes, domesic and foreign incomes are bounded by long run relaionship and confirms he shor run J-curve effec. Bahmani-Oskooee and Wang (2006) employ disaggregae quarerly daa o discover ha he Chinese income insead of Chinese yuan has played he major role in he Chinese rade balance deerminaion. Chinese yuan depreciaion only shows favorable impac on rade balance in 4 ou of 13 major rading parners, including he US. The J-curve, however, is rarely suppored. Apparenly, a presen sage, neiher he heoreical nor he empirical works have esablished definiively wheher currency devaluaion (nominal or real) has caused rade expansion or rade deerioraion, or even if exchange raes play a role in 1 Under MLC, o ge a posiive effec from devaluaion, a necessary condiion is ha he demand elasiciy of boh expors and impors mus exceed one. There is an excess supply of currencies when he exchange rae is above he equilibrium level and excess demand when i is below. Only wih his condiion a nominal devaluaion will affec real exchange raes o enhance compeiiveness and hence improves rade balance. 2 The J-curve sands as a shor-run deparure from Marshall-Lerner condiion. A usual raionale for i is ha exchange rae depreciaion iniially means cheaper expors and more expensive impors, making he curren accoun worse (a bigger defici or smaller surplus). Afer a while he volume of expors will sar o rise because of heir lower price o foreign buyers, and domesic consumers will buy fewer of he coslier impors. Evenually he rade balance will improve. 3

5 deermining rade flows. The issue has become more vial following he recen developmen of regional episodes. Wih he China's recen accession o WTO (November 2001) and he emergence of ASEAN10+3 Free Trade Area due o he Chiang Mai Iniiaive (May 2000) 3 and he Bali Dialogue (Ocober 2003) 4, he need for an amendmen of regional rade policy and currency arrangemens anchoring by China is well undersood, bu less being invesigaed. This sudy invesigaes he dynamic nexus of bilaeral rade balance-exchange raes, wih respec o Malaysia and China. Boh economies are of differen regulaory regimes, differen degrees of developmen and rade openness, bu wihin a comparable exchange rae regime. Our analyses ake concerns of he possible ransmission channels via macro-variables (e.g. domesic oupu, foreign income) as in sandard inernaional rade model. The J-curve effec is invesigaed as well, wihin he unresriced Vecor Auoregressive (VAR) framework. The presen sudy is organized in he following manner. In secion 2, a heoreical rade model is presened which forms he basis of our empirical model for esing he exchange rae impac. This is followed by our empirical esimaion procedures and daa descripion repored in Secion 3. Esimaion resuls are presened and discussed in secion 4. Finally, conclusions are drawn in he closing secion. 2. The Trade Balance Model for China-Malaysia We posi ha he demand for China impor depends upon he relaive price of income of Malaysia and China, expressed as follow: IM IM Y, RP, (1) ( ) ( ) where IM ( ) represen China demand for impors from Malaysia, Y refers o China real income, and RP is he relaive price of goods beween China and Malaysia. Leing E = he nominal exchange rae, defined as he domesic price of foreign currency, he relaive price of impored goods can be expressed as: P RP FX RFX P (2) 3 During he Asian Developmen Bank meeing on 6 h May 2000, Chiang Mai (Thailand), ASEAN-10, China, Japan, and Souh Korea (collecively known as ASEAN10+3) agreed o creae a nework of regional currency swap arrangemens, associaed wih surveillance and monioring mechanisms. The iniiaives began o ake concree when muliple counries signed swap arrangemens in 2001, some wih ceilings as high as $3 billion. These eye-caching iniiaives parallel plans by China and Souheas Asian counries o form a Free Trade Area, ongoing sub-regional economic developmen projecs and he effors o regularize meeings among finance and rade officials, have consiued owards regional inegraion. 4 During he 9 h ASEAN Summi on 7-8 Ocober 2003, Bali (Indonesia), leaders from ASEAN, China, India, Japan and Souh Korea have expressed heir srong suppor for he Bali Concord II as a solid plaform o achieve an ASEAN Communiy based on poliical-securiy, economic and socio-culural cooperaion. Despie he counermand of rans-naional crimes/ errorism and communicable diseases, hese counries have propounded he economic inegraion of ASEAN (a regional and sub-regional level) and he esablishmen of Asian Bond as an alernaive for regional financing. 4

6 where P P is he raio of China and Malaysia price indexes of all goods, nominal exchange raes of China yuan over Malaysian ringgi and FX is he RFX corresponding real exchange raes, defined as he relaive price of domesic o foreign goods. Wih real exchange raes hus defined, a decrease in is value indicaes a real devaluaion of he domesic currency. Subsiuing (2) ino (1), we obain: IM ( ) IM ( ) Y, RFX (3) On he conrary, China expor o Malaysia depends upon Malaysian income, as well as he relaive price of goods beween China and Malaysia: EX ( ) EX ( ) Y, RP (4) Again, based on (2) we can rewrie he funcion as: EX ( ) EX ( ) Y, RFX (5) We hus derive China s rade balance wih Malaysia, TB ( ), as he following funcion: EX ( ) TB ( ) Y, Y, RFX IM (6) ( ) The above model expresses he balance of rade as a funcion of he real exchange rae and he levels of boh China and Malaysia incomes. Taking naural logarihm of boh sides, exemped he counry noaions, and adding a sochasic erm o capure shorerm deparures from long run equilibrium, he empirical model for China-Malaysia rade is obained: ln( TB ) ln( Y ) ln( Y RFX (7) 0 1, 2, ) 3 where ln represens naural logarihm, ln( TB ) is calculaed from ln( EX ) ln( IM ), represens a whie noise process, 0 is he inercep and 1, 2, 3 are he parameer o be esimaed. Noe ha expressing he rade balance as he raio of expors o impors allows all variables o be expressed in log form and obviaes he need for an appropriae price index o performing our basic saisical ess. Given he definiion of he real exchange raes, he sign of 3 is negaive if he Marshall Lerner condiion holds, ha is, if a real devaluaion of he domesic currency improves he rade balance. 3. The Empirical Tesing Procedure The relaionship beween rade balance wih income and exchange raes is considered using ime series regression analyical framework. Our approach is a 2-sep procedure. 5

7 The firs is o idenify and filer any rend and srucural breaks in all he series involved in order o avoid possible spurious regression problem. The rend problem is paricularly concern for he indusrial producion series. The srucural breaks occurred due o he recen Asian currency crisis, as well as he fixed exchange rae regime adoped by he Malaysian governmen during he period Sep 1, 1998 o July 21, The filering process is done hrough running a simple regression on all he involved variables, as shown followings: Z a0 b1trend b2d1997, b3drmfx, e (8) where Z includes TB, he rade balance raio, RFX, he real exchange raes, and Y, and Y,, incomes of China and Malaysia, respecively. The binary variable D 1997, akes uniy value of one for he period July 1997 o Augus 1998 and zero oherwise. For D,, he binary variables akes value of one for he period RMFX Malaysia ringgi was fixed o RM3.80/USD, i.e. from Sepember 1998 o July 2005, and zero oherwise. If hese series are indeed conaminaed wih he rends and srucural breaks, he residuals of he regression will be colleced o replace he original series and regressed in model (7) as follows: * * * * ln( TB ) ln( Y ) ln( Y ) RFX (9) 0 1, 2, The aserisk marks define he de-rended series ha is also free from srucural breaks. In addiion, we also conduced a saionariy es developed by Kwiakowski-Phillips- Schmid-Shin (1992) o verify he uni roo problem. Since all he righ hand side variables are demeaned, we can expec 0 no significanly differen from zero. For diagnosic checking purposes, o ensure ha he specificaion of he mean equaion of model (9) is free from auocorrelaion problem we refer o Durbin-Wason es on AR(1), F-es on he join significance of all of he slope coefficiens in he regression, and he Ljung-Box Q-saisics on he residual for higher AR erms. As we are using high frequency monhly observaions for all series, our leas square regression migh have exposure o auoregressive condiional heeroskedasiciy (AR) effecs. We es he effecs by a Lagrange muliplier (LM) es proposed by Engle (1982). The AR effec is also examined hrough he Ljung-Box Q-saisics on he squared residuals. Finally Jarque-Bera normaliy es is also examined o affirm ha our regression is consisen wih he sandard regression assumpions. To race he possibiliy of J curve, we run a vecor auoregressive model (VAR) o examine he sequenial impac of exchange raes on rade balance, assuming he effecs of he income variable, i.e. Y and Y, o be exogenous. The VAR specificaion is given by he sysem of regressions as following: TB 1TB i 2Y, i 3Y, i 4, i RFX i v1 i1 j RFX 1TB i 2Y, i 3Y, i 4, i RFX i v2 i1 j1 6 3 (10a) (10b) The responses of he rade balance from he real exchange rae shocks are examined using he generalized impulses procedure as described by Pesaran and Shin (1998), which does no depend on he ordering of he variables a he righ hand side of he

8 VAR equaions. The generalized impulse responses from he real exchange rae shocks o he rade balance, as saed in (10a), is basically he orhogonal se of innovaions derived by applying a variable specific Cholesky facor of he residual covariance marix compued wih he rade balance a he op of he Cholesky ordering. Our analyses are all based on high frequency monhly daa. The sample period spanned from January 1990 o January Real exchange raes are compiled by having he nominal exchange raes (local currency/usd) adjused for relaive price changes which is proxy by consumer price index (CPI) series; whereas rade balance raios are compued based on he USD denominaed expor and impor series. The income for China and Malaysia are represened by heir indusrial producion (IP) indices as GDP is no available for high frequency monhly observaion. All rade series are sourced from he Direcion of Trade Saisics compiled by Inernaional Moneary Fund while he CPI, IP and exchange raes are sourced from DaaSream. 4. Empirical Resuls and Discussion Descripive saisics for all he series are repored in panel A of Table 1. All he ime series basically are no univariae normal. To avoid spurious regression problem, he saionariy of all he series are examined using he Augmened Dickey Fuller (ADF) uni roo es for boh inercep and inercep plus rend models. The ADF resuls sugges ha only he real exchange rae series of yuan/ringgi is no saionary. In panel B of Table 1, he correlaion marix is displayed. Generally, here is no mulicolineariy problem, excep ha he indusrial producion of Malaysia is moderaely correlaed wih he real exchange raes. This is no serious as he value is sill below The resuls of applying model (8) on all he ime series are repored in Table 2. All involved series are saionary bu hey are sill highly exposed o he rend and srucural breaks dummy variables. The Asian currency crisis has a significan posiive impac on he bilaeral China-Malaysia rade balance, where China-Malaysia rade balance was shown o improve abou 5% as compared o before crisis. During he fixed ringgi regime, however, he rade balance deerioraed more han 25% as compared o before crisis. The linear ime rend is posiive and significan for all excep he rade balance series, which is negaively significan, suggesing a reducion in he bilaeral China-Malaysia rade balance over ime. As a resul of significan loadings of he ime read and dummy variables, we decided o replace he original series wih hese residual series colleced from model (8). To be cauious, we also repor he ADF uni-roo resuls on hese new series a he las column in Table 2. As hese residuals basically represen he componen of he original ime series afer removing he mean, ime rend, he 1997 srucural break, and he fixed exchange rae regime over , he esed ADF model excludes he drif erm and ime rend. The uni roo ess suppor hese series are all saionary, including he real exchange raes. This implies ha he non-saionary behavior of real exchange raes repored in uni roo ess in Table 1 is due o srucural breaks. 7

9 Table 1 Descripive Saisics and Correlaion Panel A: Summary of Descripive Saisics Saisics ln(ex/im) ln(ip China ) ln(ip Malaysia ) RFX Mean Maximum Minimum Sd. Dev Skewness Kurosis Normaliy * *** *** *** (0.0729) (0.0000) (0.0011) (0.0000) Uni Roo *** *** *** Uni Roo *** ** *** Panel B: Correlaion Saisics Variable ln(ex/im) ln(ip China ) ln(ip Malaysia ) RFX ln(ex/im) 1 ln(ip China ) ln(ip Malaysia ) RFX Noe: Figures in he parenhesis are probabiliy values. Sd. Dev. denoes sandard deviaion. Aserisks *, ** and *** denoe significance a he 10%, 5% and 1% levels, respecively. Normaliy refers o Jarque- Bera normaliy es, where rejecion of hull hypohesis implies non-normal disribuion. Tes for saionariy es refers o Augmened Uni Roo (ADF) es, where Uni Roo 1 is he model wih inercep only and Uni Roo 2 is he model wih inercep and ime rend. Rejecion of null hypohesis reflecs saionariy. Table 2 De-rending and Removal of Srucural Breaks Based on Model (8) Y Inercep Trend D97 Dfix Adj R 2 Uni Roo ln(ex/im) *** *** *** (0.0312) (0.0003) (0.0643) (0.0367) ln(ip China ) *** *** *** *** *** (0.0065) (0.0001) (0.0134) (0.0076) ln(ip Malaysia ) *** *** *** ** *** (0.0099) (0.0001) (0.0204) (0.0116) RFX *** *** * *** (0.0229) (0.0002) (0.0471) (0.0269) Noe: Figures in he parenhesis are sandard errors. Aserisks ** and *** denoe significance a he 5% and 1% levels, respecively. RFX denoes he real exchange raes calculaed using he formula ln([cpi Mal *FX Yuan/Ringgi ]/CPI China ). Adj R 2 denoes adjused R 2. Uni Roo refers o ADF es on he residuals of model (9) wih inercep only, and Uni Roo 2 on he residuals of model (8) wih no inercep and no ime rend. Rejecion of null hypohesis reflecs saionariy. The coefficiens esimaed for model (9) are repored in panel A of Table 3. We repor hree model esimaes. The firs model, OLS, is he simple leas square. This model is subjec o various diagnosic problems. The R 2 is low, he residual and residual square series are serially correlaed, he error process conen AR effecs and he error disribuion is no normal. As a resul, we proceed o he second model, which is 8

10 basically model (9) accouning for a sandard GAR(1,1) specificaion. This model basically akes care of he AR effecs; however, i is sill subjec o serial correlaion in he residual, non-normaliy error and worse, insignifican F-es. Thus we proceed augmen he model o include auoregressive (AR) erms o model he correlaion in he error process. This hird model, which is ermed as AR-GAR, is free from all he above menioned diagnosic problems. The AR-GAR model is also beer in he sense ha i provides beer goodness-of-fi, and lower AIC and SBC values. Thus, our discussion following will focus only on he AR-GAR model. Table 3 Regression Resuls for Long Run Elasiciy Panel A: Coefficien Esimaes OLS GAR AR-GAR C (0.0149) (0.0153) (0.0268) IP (0.3207) (0.3358) (0.3224)** IP (0.2900)** (0.2810)* (0.2807) REX (0.1253)*** (0.1258)** (0.1978)** AR(1) (0.0575)*** AR(2) (0.0658)** W (0.0017)** (0.0001)** AR (0.0751)** (0.0175)** GAR (0.0878)*** (0.0189)*** Panel B Diagnosics Saisics Adj R SSR LogL AIC SBC DW F-saisic (0.0019)*** (0.1561) (0.0000)*** Q(4) (0.0000)*** (0.0000)*** (0.2990) Q(12) (0.0000)*** (0.0000)*** (0.1240) Q 2 (4) (0.0000)*** (0.6340) (0.3560) Q 2 (12) (0.0000)*** (0.9610) (0.7430) AR (0.0000)*** (0.1603) (0.1492) Normaliy (0.0288)** (0.0431)** (0.8075) Noe: Figures in he parenhesis are probabiliy values. Aserisks *, ** and *** denoe significance a he 10%, 5% and 1% levels, respecively. Adj R 2, SSR, LogL, AIC, SBC and DW denoe adjused R 2, sum of square residual, log likelihood value, Akaike informaion crierion and Schwarz informaion crierion, and Durbin-Wason es, respecively. Q(4) and Q 2 (4) refers o Ljung-Box Q-saisics for checking auocorrelaions in he residuals and squared residuals, respecively, for lags up o one quarer (4 monhs) and Q(12) and Q 2 (12) for lags up o one year (12 monhs). Normaliy and AR refer o Jarque-Bera normaliy es, and AR Lagrange muliplier es by Engle (1982), respecively. Tes for saionariy refers o uni roo es by Kwiakowski-Phillips-Schmid-Shin (1992), where Saionariy 1 is he model wih inercep only and Saionariy 1 is he model wih inercep and ime rend. Rejecion of null hypohesis reflecs uni roo. For he AR-GAR model, all he coefficiens esimaed are saisically significan a 95% confiden level or higher. Only wo parameers are insignifican, he inercep and he coefficien for Malaysian income, Y,. As expeced, wih he demeaned righ hand side variables, here is no saisical evidence o infer ha 0 is significanly differen 9

11 from zero. The coefficien for Malaysian income 2 does no have he correc direcion bu since i is insignifican, his is no really a maer. For he res, he direcion are mosly consisen wih heory, so as he magniude. The resul suggess ha for every 1% higher in China income, he rade balance will deeriorae for abou 0.7%. This implies ha when he income of China s people increases, hey migh have demand more impors from Malaysia and hus worsen he bilaeral rade balance wih Malaysia. For he real exchange raes, he esimae shows ha for 1% devaluaion in yuan/ringgi real exchange raes, bilaeral China-Malaysia rade balance will improve for abou 0.4%. Basically a devaluaion of yuan/ringgi means cheaper expors o Malaysia and more expensive impors from Malaysia, making China s curren accoun improved, i.e., eiher a bigger surplus defici or a smaller defici. However, he resul implies ha he process of rade adjusmen is inelasic o heir real exchange raes. This could be he resul of incomplee pass-hrough of he currency effecs on prices, he resisan in consumers o fully adjus o he new prices, or boh. Wih a posiive impac documened here for devaluaion on rade, we hus can conclude ha MLC is me, alhough he currency effec is inelasic. Besides, he rade balance process is also significanly depending on previous performances, wih higher persisency from las monh rade figure. The componen of he variance process, i.e. he consan variance, AR and GAR erms are all saisically significan. This implies ha here is baseline volailiy in he rade flows beween China and Malaysia, and he magniude of he rade balance flucuaion ends o persis from he previous volailiy. The race wheher he currency effecs follows a J-curve phenomenon, we plo he generalized impulse responses of China-Malaysia rade balance o uni shocks in real yuan/ringgi raes using an unresriced VAR model as explained in secion 3.2, assuming he income series are exogenous. As we are analyzing monhly observaions, he shor run dynamics of rade adjusmen o shocks in he real exchange raes is raced as far as 24 monhs o allow us o compare our resul here wih he lieraure ha predominanly based on quarerly daa. By heory, if J-curve is presen, a counry is able o correc exernal imbalances via exchange rae devaluaion afer emporal adjusmens of exernal compeiiveness, or oherwise. As shown in Figure 1, here is no clear indicaion of J-curve effec. The Chinese- Malaysian rade balance series depiced immediae posiive adjusmen o real exchange shocks from an iniially negaive posiion. A 1% real depreciaion of renminbi brings o a maximum of 4.5% improvemen in rade balance. The correcion of rade reduces afer he 3 rd monh and he impac of real depreciaion die ou gradually afer 14 monhs. In oher words, he volume effec occurs faser han price effec bu afer a moderae ime period, he price effecs become large enough o offse he volume effec ha he rade balance improvemens due o real depreciaion die off. 10

12 Figure 1 Response of Trade Balance o Real Exchange Rae Shocks Noe: The responses of China-Malaysia rade balance is raced up o 24 monhs. The impulse is generalized one sandard error of yuan/ringgi real exchange raes derived from he 12-lag VAR modeling as shown in (10a) and (10b). 5. Conclusion This paper deals wih currency exposure of bilaeral rade balance beween China and Malaysia. We are moivaed by he fac ha boh China and Malaysia, emerging and open economies, whom wen hrough similar currency regime over he las decade, has relaxed heir pegging o USD exacly he same day in Our sample covered he las 20 years of monhly frequency daa. We follow a sandard rade balance model relaing bilaeral rade o local and foreign incomes and heir bilaeral real exchange raes. One of our conribuions in empirical modeling is ha our modeling akes ino accoun he srucural impac of he 1997 Asian currency crisis on boh China and Malaysia, as well as he period of pegging regime adoped by Malaysia during Our resul shows ha real exchange raes play a significan role in he bilaeral rade of China-Malaysia. The Marshall-Lerner condiion is parially me and he currency effec is inelasic. However, he J-curve phenomenon is somewha unobserved hrough he generalized impulse response analysis. The real depreciaion of Chinese Yuan poses an immediae correcion of he Chinese-Malaysia rade imbalances bu he effec does no las long. Addiionally, he coefficiens on domesic and foreign income show consisen signs o hose prediced by economic heory where he China-Malaysia bilaeral rading is demand driven bu he income effec is greaer for Malaysia. All in all, Malaysia holds beer gains in he bilaeral rading wih China. 11

13 References Bahmani-Oskooee, M. and Mieza, I. (2003) Are Devaluaions Expansionary or Conracionary? A Survey Aricle. Economic Issues, 8(2), pp Bahmani-Oskooee, M. and Wang, Y. (2006) The J Curve: China Versus Her Trading Parners. Bullein of Economic Research, 58(4), pp Bahmani-Oskooee, M. and Janardhanan, A. (1994) Shor -Run versus Long-Run Effecs of Devaluaion: Error Correcion Modeling and Coinegraion. Easern Economic Journal, 20, pp Engle, R.F. (1982) Auoregressive Condiional Heeroskedasiciy wih Esimaes of he Variance of Unied Kingdom Inflaion. Economerica, 50, pp Himarios, D. (1989) Do Devaluaions Improve he Trade Balance? The Evidence Revisied. Economic Inquiry, 27, pp Hsing, H.M. and Savvides, A. (1996) Does A J -Curve Exis for Korea and Taiwan? Open Economies Review, 7, pp Krugman, P. and Baldwin, R.E. (1987) The Persisence of he US Trade Defici. Brookings Papers on Economic Aciviy, 1-2, pp Kwiakowski, D., Phillips, P.C.B., Schmid, P. and Shin, Y. (1992) Tesing he Null Hypohesis of Saionary Agains he Alernaive of a Uni Roo. Journal of Economerics, 54, pp Noland, M. (1989) Japanese Trade Elasiciies and he J-Curve. The Review of Economics and Saisics, 71, pp Onafowora, O. (2003) Exchange rae and rade balance in Eas Asia: is here a J-curve? Economics Bullein, 5, pp Pesaran, M.H. and Shin, Y. (1998) Generalised Impulse Response Analysis in Linear Mulivariae Models. Economics Leers, 58, pp Rose, A.K. (1991) The Role of Exchange Raes in a Popular Model of Inernaional Trade: Does he Marshall-Lerner Condiion Hold? Journal of Inernaional Economics, 30, pp Rose, A.K. and Yellen, J.L. (1989) Is There a J-Curve? Journal of Moneary Economics, 24, pp