Vol. 3, No. 8 Inernaional Journal of Business and Managemen Real Exchange Rae and Trade Balance Relaionship: An Empirical Sudy on Malaysia Ng Yuen-Ling Faculy of Accounancy and Managemen, Universii Tunku Abdul Rahman, Bander Sungai Long, 43000 Selangor, MALAYSIA. Har Wai-Mun (Corresponding auhor) Faculy of Accounancy and Managemen, Universii Tunku Abdul Rahman, Bander Sungai Long, 43000 Selangor, MALAYSIA. E-mail: harwm@mail.uar.edu.my Tan Geoi-Mei Faculy of Accounancy and Managemen, Universii Tunku Abdul Rahman, Bander Sungai Long, 43000 Selangor, MALAYSIA. Absrac This paper aemps o idenify he relaionship beween he real exchange rae and rade balance in Malaysia from year 1955 o 2006. This sudy uses Uni Roo Tess, Coinegraion echniques, Engle-Granger es, Vecor Error Correcion Model (VECM), and impulse response analyses. The main findings of his paper are: (i) long run relaionship exiss beween rade balance and exchange rae. Oher imporan variables ha deermine rade balance such as domesic income shows a long run posiive relaionship beween rade balances, and foreign income shows a long run negaive relaionship (ii) he real exchange rae is an imporan variable o he rade balance, and devaluaion will improve rade balance in he long run, hus consisen wih Marshall-Lerner condiion (iii) he resuls indicae no J-curve effec in Malaysia case. Keywords: Exchange rae, Trade balance, Devaluaion, Coinegraion, Malaysian economy 1. Inroducion Depreciaion of he currency has grea impacs o rade balance bu he impac may vary, probably due o differen level of economic developmen. One of he prominen impacs is he Marshall-Lerner condiion, which represens ha real depreciaion leads o increases he rade balance in he long run if sum up value of impor and expor demand elasiciy exceed one. Real depreciaion improves he rade balance hrough wo differen channels. Firsly, increase quaniy of expor. Depreciaion of he currency reveals he domesic goods cheaper as compared o he foreign goods, hus making expor more compeiive. Secondly, quaniy of impors decreases, as impor is relaively more expensive. Alernaively, amoun of expor and impor may no responsive a iniial period of depreciaion. Thus, rade balance may be worsening firs due o decrease in value of expor and increase in value of impor bu improves afer some ime. This make scenario knows as J-curve. 2. Objecive of Sudy The main objecives of his paper, herefore aims (i) o sudy he relaionship beween exchange rae and rade balance in Malaysia, and (ii) o invesigae wheher Marshall-Lerner condiion and J-curve exis, boh for he period 1955-2006. The res of he paper is srucured as follow: Secion 3, here will have review on lieraures. Secion 4 will be he heoreical framework and mehodology. Secion 5 will be he resul and inerpreaion and finally is Secion 6 will be he conclusion from his sudy. 3. Lieraure Review Hernan Rincon (1999) examined he relaionship beween rade balance and exchange rae es for Marshall-Lerner condiion in Colombia using Johansen-Juselius mehod. His empirical resuls provided significan eviden for he Marshall-Lerner condiion. Shirvani and Wilbrae (1997), Akbosanci s (2002) and Liu, Fan and Shek (2006) also found Marshall-Lerner condiion hold in heir respecive sudies. Shirvani and Wilbrae (1997) examined he relaionship beween rade balance and real exchange rae in Unied Saes and he G7 counries of Canada, France, Germany, Ialy, Japan, Unied Kingdom and Unied Saes, Akbosanci s (2002) in Turkey while Liu, Fan and Shek 130
Inernaional Journal of Business and Managemen Augus, 2008 (2006) in Hong Kong. Besides, Onafowora (2003) repored significan relaionship exis for hree ASEAN counries of Thailand, Malaysia, and Indonesia in heir bilaeral rade o Unied Saes and Japan. In conras, Rose (1991) repored he Marshall-Lerner condiion does no exis in five major OECD counries (Unied Kingdom, Canada, Germany, Japan, and he Unied Saes). Her resuls also showed insignifican relaionship beween rade balance and exchange rae, hus implying ha devaluaion could no improve rade balance in he long run. Rose (1991) prediced i would reveal significan hrough rade balance reaed as exogenous wih respec o he exchange rae. Using coinegraion es, Haemi and Irandous s (2005) sudy showed Sweden did no saisfy Marshall-Lerner condiion. This migh be due o he rade balance in Sweden is no sensiive in real exchange rae bu only sensiive in changes in income. Wilson and Kua (2001) examined he relaionship case beween Singapore and Unied Saes. Their resul indicaed exchange rae does no have significan impac on he bilaeral rade balance. Liew, Lim, and Hussain (2003) sudied he relaionship beween real exchange rae and balance of rade based in ASEAN counries. They suggesed ha rade balance is affeced by a real money raher han he exchange rae. Thorbecke (2006) indicaed ha he change in he exchange rae could affec rade wihin Asia. His empirical sudies demonsraed an appreciaion in Indonesia, Malaysia, and Thailand would decline expor. As for lieraures on he J-curve effec, Ahmad and Yang (2004) examined he hypohesis of J-curve on China s bilaeral rade wih he G-7 counries by using and found no evidence characerize of J-curve effec. Moffe (1989) examined empirical evidence for he rade price (price of he expor and impor) and he quaniies (quaniies of expor and impor) of he Unied Saes o deermine wheher J-curve exiss or no from he period of 1967 o 1987. Repored resul indicaed ha, dollar depreciaion leads o impor quaniies decrease, bu i simulaneous decrease in he quaniies of expor. Based on he J-curve heory, depreciaion leads impors o decrease and expors increase. Expors decrease in his case, so, i resembles of sine wave raher han a J-shape. Rose and Yellen (1989), using ordinary leas square (OLS) and coinegraion es repored no response of he rade balance o he real exchange rae in he Unied Saes. Meanwhile, Bahmani-Oskooee and Raha (2007) examined he bilaeral rade beween Sweden and her 17 rading parners and analyzed he real depreciaion of shor run effec and he long run effec. Their long run resul concluded ha real depreciaion of he currency only sufficien in five cases, which is in he rade balance beween Sweden and Ausria, Denmark, Ialy, Neherlands, and he Unied Kingdom. Shor run resul has effecs on he rade balance in 14 ou of he 17 cases. Neverheless, Sugema (2005), examining he deerminae of he rade balance and crisis adjusmen in Indonesia hrough exchange rae make a cauion poin on he issue of effeciveness of exchange rae depreciaion in improving rade balance in he long run. Sugema (2005) claimed ha exchange rae migh overshoo if he rade balance is no sensiive wih he depreciaion. 4. Theoreical Framework and Mehodology The modeling he rade balance in his paper follows similar equaion chosen from Shirvani and Wilbrae (1997), Baharumshah (2001), Gomez and Alvarez-Ude (2006), which emphasized in exchange rae on bilaeral rade balance evidence. Equilibrium goods marke in an open economy can be described by he following equaions: Y = C( Y T ) I ( Y, r) G IM ( Y, ε ) X ( Y *, ε ) which Y represens oal domesic income, C represens consumer spending, and T represens income ax, I, represens invesmen, r known as ineres rae, G represens governmen spending, represens real exchange rae, IM represens impor, X represens expor, and Y* represens foreign income. Signs in bracke (below he equaion) indicae relaionships for respecive facors. Consumers spending (C) which funcion as oal income subrac income ax, which i knows as disposal income (Y-T). Higher disposal income lead o higher consumer spending besides o increase oal domesic income, herefore, posiive relaionship incurs beween oal domesic income and consumer spending. Invesmen (I) is a funcion of oal income and ineres rae. Naions would invesmen more if increase in he oal personal income. Thus, i shows posiive relaionship beween invesmen and oal income. Besides ha, ineres rae migh effec invesmen decision. Lower ineres rae reduces cos for capial, hus aracs more invesor come and invess. For ha reason i shows negaive relaionship beween invesmen and ineres rae. In oher word, higher ineres rae would decrease oal domesic invesmen. For he real exchange rae equal he nominal exchange rae (E) muliple he foreign price level (P*) and divided by he domesic price level. Nominal exchange rae (E) is defined as he number of uni domesic currency exchange for one uni of foreign currency, giving: = (EP*)/ P (1) 131
Vol. 3, No. 8 Inernaional Journal of Business and Managemen Impor (IM) is influenced by domesic income or oupu (Y). Higher domesic income leads o high impors. So, i shows posiive relaionship. In addiional, impor has negaive relaionship wih oal domesic income; quaniy of impor also depends on he real exchange rae ( ). Higher ( ) leads o lower quaniy of impors because of he foreign goods relaively more expensive. Expor (X) depends on he foreign income (Y*) and real exchange rae ( ). High he foreign income leads o increase in foreign demand for all goods and services as a resul increase expors. On he oher hands, increase in real exchange rae, he relaive price of foreign goods in erms of domesic goods also leads o increase in expor. I is showing posiive relaionship beween rade balance and foreign income, real exchange rae. As he objecive is o examine rade balance (ne expor, NX) and exchange rae, oher variables are assumed consan. The ne expor is: NX X IM By subsiuing he funcion of expor and impor ino equaion (2), i shows NX X ( Y*, ε ) IM ( Y, ε ) Afer ha, subsiue equaion (1) ino equaion (3) EP * EP * NX = X ( Y*, ) IM ( Y, ) (4) P P Assume EP*/P is saionary, we can rewrie he equaion (4) as NX = NX ( Y, Y*, ε ) (5) Therefore, equaion (6) expresses he balance of rade as a funcion of he levels of domesic and foreign income and he real exchange rae. * ln 0 β1 β2 β3 TB = β ln RER u (6) where ln represens naural logarihm, u is assumed o be a whie-noise process, and rade balance, TB, represens as he raio of expors o impors allows all variables o be explained in logarihm form and removes he need for appropriae price index o explain he rade balance in real erm. In his research, real exchange rae, RER, expresses by Ringgi Malaysia (RM) agains Unied Saes Dollar (US$) and Y* expresses as gross domesic produc of Unied Saes. Following classical heory, he sign of 1 could be eiher posiive or negaive. If he esimae of 1 would be expeced o be negaive which means ha an increase in Malaysian real income, Y, increases impors volume. However, if he esimae of 1 would be expeced o be posiive which means ha an increase in Y he is due o an increase in he producion of impor-subsiued goods. Similarly he esimae of 2 could be eiher posiive or negaive. I he sign of 2 would be depended on wheher he supply side facors dominae demand side facors. Marshall-Lerner heory holds when 3 is posiive indicaing ha depreciaion leads o improve he rade balance for Malaysia. The annual daa used o model his equaion from year 1955 o 2006 obained from Inernaional Moneary Fund (IMF). During ha period, Malaysia has some dramaic change in real exchange rae and rade imbalance. Hence his provides an excellen research condiion o examine wheher he changes in real exchange rae affec he volume of rade. The rade balance, domesic and foreign incomes are in real erms; he consumer price index (CPI) acs as he price deflaor. Uni roo es is used o es of saionary. Following he work of Baharumshah (2001) and Sugema (2005), Augmened Dickey-Fuller (ADF) es and Philips-Perron (PP) es is applied for esing saionariy in economic daa. If ADF es and PP es show differen resuls, he Kwiakowski-Philips-Schmid-Shin (KPSS) es is used as decisive resuls. In order o solve he spurious regression problem and violaion assumpions of he Classical Regression Model, coinegraion analysis used o examine he long-run relaionship beween TB, RER, Y and Y *. To es for coinegraion, hree mehods are used. There are Engle-Granger Tes, Error Correcion Model, and Johansen-Juselius Tes. In order o know he disequilibrium error, we rewrie equaion (6) as: u = ln RER (7) * lntb β0 β1 β2 β3 In order o perform Engle-Granger Tes, he order of inegraion of he esimaed residual, u, should be esed. If here is a coinegraing regression, hen disequilibrium errors in equaion (7) should form a saionary ime series, and have a zero mean, he u should be saionary, I(0) wih E(u ) = 0. The long run equilibrium may be rarely observed, bu here is a endency o move owards equilibrium. Thus, Error Correcion Model is used o represen he long-run (saic) and shor-run (dynamic) relaionships beween rade balance, real exchange rae, domesic and foreign income. According o Baharumshah (2001), Onafowora (2003), (2) (3) 132
Inernaional Journal of Business and Managemen Augus, 2008 Ahmad and Yang (2004) and Sugema (2005), Vecor Error Correcion Model is suiable o esimae he effec of exchange rae on rade balance. The equaion (8) represens Error Correcion Model as: lntb = lagged ( TB, RER, Y, Y * ) λ u v (8) where u -1 represens he residual erm a -1 in long erm. Boh, Engle-Granger Tes and Vecor Error Correcion Model (VECM), are es for wheher he long-run relaionship exiss in equaion only. Following he work of Shirvani and Wilbrae (1997), Baharumshah (2001), Onafowora (2003), Gomez and Alvarez-Ude (2006), Johansen-Juselius es is used o perform hypohesis ess abou he number of he long-run relaionship exiss in equaion. To use Johansen-Juselius s mehod, he Vecor Auoregressive (VAR) of he form needed o urn firs, Z = β 1Z β2z 2 Κ βkz k v, = 1, Κ, T (9) ino a Vecor Error Correcion Model (VECM), which can be wrien as Z = Z -k 1 Z -1 2 Z -2 k-1 Z -(k-1) v (10) The es for coinegraion beween he Z is calculaed by looking a he rank of he marix via is eigenvalues. The rank of a marix is equal o he number of is characerisic roos (eigenvalues) ha are differen from zero. represens how many linear combinaions of Z are saionary. The vecor Z included of rade balance (TB), real exchange rae (RER), domesic income (Y) and foreign income (Y*). Thus, Z = [TB RER Y Y*]. We have chosen he number of lags based on Akaike Informaion Crierion (AIC) and Schwarz Crierion (SIC). Nex, following Johansen-Juselius s approach, he number of coinegraing equilibrium relaionship beween he logarihms of rade balance, domesic and foreign naional income and real exchange rae should be esed. Two saisics for coinegraion used: he race saisic, race, and he maximal-eigenvalue saisic, max. Boh es saisics are he esimaed value for he ih ordered eigenvalue from he marix. The r se from zero o k 1, where k = 4 (k represens he number of endogenous variables in his research). For race saisic, he es saisic for coinegraion is formulaed as λ race g ( r) = T ln(1 ˆ λ ) i= r 1 i where T represens he sample size, r represens number of long run relaionship exis, and λˆ represens he eigunvalue. For race saisic, he null hypohesis is he number of coinegraing vecors is less han equal o r agains an unspecified alernaive. If race equal o zero, all he i equal o zero, so i is a join es. For maximal-eigenvalue saisic, he es saisic for coinegraion is formulaed as λ (, 1) ln(1 ˆ max r r = T λr 1) The null hypohesis for maximal-eigenvalue saisic is he number of coinegraing vecors is r agains an alernaive of r 1. Before forecasing wih he final model, i is necessary o perform various diagnosic ess o verify he adequacy of represenaion of he model. To es he parameer, he -es is used. In order o es he direcion of causaliy beween wo variables, he Pairwise Granger Causaliy Tes is used. For analyzing he residual, Pormaneau Auocorrelaions (Q) es, Auocorrelaion LM (LM) es, Whie heeroskedasiciy (Whie), and Jarque-Bera residual normaliy es via Cholesky (JB CHOL ) and Urzua (JB URZ ) facorizaions are applied. Impulse response analysis provides he informaion abou ineracion among he variables in he sysem, herefore used for forecas purpose. Impulse response funcions seek he effecs of a shock o endogenous variable on he oher variables in he sysem. According o Gomez and Alvarez-Ude (2006), he impulse response funcion map ou he dynamic response of rade balance o Cholesky one sandard deviaion real exchange rae innovaion. Following works of Baharumshah (2001), Akbosanci (2002), Onafowora (2003), Sugema (2005) and Gomez and Alvarez-Ude (2006), Impulse Response Funcion used o deermine wheher J-curve heory exiss in Malaysia. 5. Research resul Table 1 repors he resuls of he ADF ess and PP ess for uni roo on boh he level and he firs difference of he variables (for all ables, refer Appendix). The null hypohesis in ADF ess and PP ess are ha he variables follow a difference saionary process is esed. Boh ADF ess and PP ess show ha ln RER and ln Y are inegraed of order one in levels, I (1), and ln TB is saionary in level form, I (0). The resul of ADF es shows ha ln Y* is inegraed of order wo in levels, I (2), in inercep wihou rend model; however, he resul of PP es shows ha ln Y* is inegraed of order one in levels, I (1), in inercep wihou rend model. In order o confirm he number of order inegraion for ln Y*, he Kwiakowski-Philips-Schmid-Shin (KPSS) es was used. The null hypohesis in 133
Vol. 3, No. 8 Inernaional Journal of Business and Managemen KPSS ess is ha he variables follow a saionary process is esed. The KPSS resul shows ha ln Y* is inegraed of order one in levels, I (1) in inercep wihou rend model (see Table 2). In conclusion, he es saisics indicae ha ln RER, ln Y, and ln Y* are inegraed of order one in levels, I (1), while ln TB is saionary in level form, I (0). The Engle-Granger long-run coinegraion es he mulivariae sysem o see wheher here exis any linear combinaions of he four variables ha have common rend. In resul, he error erm in long run, u, is saionary in level form, I (0) (See Table 3). This means ha has linear combinaions beween ln TB, ln RER, ln Y and ln Y*. Thus, we conclude ha long-run relaionship exiss beween he variables in he model. The long erm effec of rade balance on he normalizing variable is explained by he coefficien of variable in he coinegraing vecor afer normalizing for rade balance (see Table 4). The posiive sign on he real exchange rae (RER) variable represens a devaluaion of currency causes an improvemen in rade balance in long run. Based on classical heory, he sign on domesic income (Y) o domesic rade balance is uncerain, i s depending upon wheher i represens he level of economic aciviy or i may also be seen as a supply variable measuring he supply of exporable. The posiive sign on he domesic income represens an increase in domesic income leads o an improvemen in rade balance in he long run. Usually, he sign on he foreign income (Y*) o domesic rade balance should be posiive. Theory suggess ha he volume of expors (impors) o a foreign counry (domesic counry) ough o increase as he real income and purchasing power of he rading parner (domesic economy) rises, and vice versa. However, he resul shows ha he negaive sign on he foreign income (Unied Saes) implies ha a rise in foreign income leads o a decrease domesic rade balance. This maybe because he rise in foreign real income is due o an increase in he foreign producion of impor-subsiue goods, hus, heir impors may decline as income increases. Afer esimaed he long run relaionship beween rade balance, real exchange rae, domesic income, and foreign income, he error correcion model (ECM) used for esimaion. Based on resul, lag one is chosen based on Akaike Informaion Crierion (AIC) and Schwarz crierion (SIC). All variables (excluded consan erm) is saisically significan a 95% confidence level. The resul of error correcion model (ECM) shown as: * lntb = 0.0295 0.2411 lntb 0.2572 lnrer 0.6481 1.2616 0. 5237u = (-0.8631) (2.0837) * (1.7934) * (-3.5955) ** (2.3071) * (-4.9323) ** where *, ** denoe significance a he 5% and 1% level of significance respecively. The resul of diagnosic checking shown ha well-behaved residuals in all period (see Table 5).The resul of Pairwise Granger Causaliy shown ha here is evidence saisically Granger causal effec running from he real exchange rae o he rade balance a 10 percen level. There has unidirecional causaliy from he real exchange rae o he rade balance exiss. These resul also sugges ha he direcion of causaliy is from he domesic income o he rade balance is significan a 1 percen level. There also shown ha unidirecional causaliy from he domesic income o he rade balance exiss. These resuls also shown ha here is evidence saisically Granger causal effec running from he foreign income (Unied Saes) o he rade balance a 5 percen level and unidirecional causaliy from foreign income (Unied Saes) o rade balance. Since he Johansen-Juselius es is quie sensiive o he lag lengh seleced, he mos commonly used crierions such as AIC and SIC are uilized o deermine he proper lag lengh, all of which sugges ha one lag be included. The resuls of he Johansen-Juselius es are repored in Table 6. In Trace es, i indicaes one coinegraing equaion a he 5% and 1% level. In Max-eigenvalue es, i indicaes one coinegraing equaion a he 5% and no coinegraion a 1% level. Impulse response funcion used o provide informaion abou he shor-erm responses for rade balances. To es wheher J-curve effecs exis in Malaysia, we examine he response of rade balance o innovaion in real exchange rae. If he response of rade balance o depreciaion has shown a J-shape indicaing ha J-curve effecs exis in Malaysia. This means ha depreciaion would worsen he rade balance firs and hen having improvemens in rade balance afer several periods. The impulse response funcion of rade balance o shock in he real exchange rae is shown in Figure 1. From Figure 1, we know ha rade balance increases quickly o respond he innovaion due o depreciaion in nex wo year. Afer ha, rade balance has improved slowly down from year 2 o year 7. And hen, he shock has coninuing effec permanenly. From Figure 1, he impac does no follow he classical J-curve paern. Thus, J-curve hypohesis is invalid for Malaysia case. 6. Conclusion In order o es wheher Marshall-Lerner condiion and J-curve effecs exis, his research sudied he shor run and long run effec of he real exchange rae on he Malaysian rade balance in a dynamic model. In his research, he resuls suppor he empirical validiy of he Marshall-Lerner condiion hrough VECM, indicaing ha depreciaion has improved he rade balance. This resul has furher confirms hrough he empirical work repored by Baharumshah (2001). The empirical work for differen se of counries ha repored by Shirvani and Wilbrae 134
Inernaional Journal of Business and Managemen Augus, 2008 (1997), Sugema (2005), Akbosanci (2002) and Thorbecke (2006) are also suggesed Marshall-Lerner condiion exiss. However, VECM analysis does no find he evidence of he shor erm worsening of rade balance suggesed by he J-curve effecs. Thus, by using impulse response funcions, he resul show ha Malaysian rade balance has no followed he J-curve paern of adjusmen or in anoher words, he resul show no evidence for he J-curve hypohesis. This resul is consisen wih Baharumshah (2001). The empirical work for differen se of counries ha repored by Rose and Yellen (1989), Akbosanci (2002), Ahmad and Yang (2004), Gomez and Alvarez-Ude (2006), also suggesed ha no evidence of J-curve effecs. As implicaion, in order o achieve he desired effecs on rade balance, he counries should depend on policy ha focusing on he variable of real exchange rae, which is he nominal exchange rae o aggregae price level. A he same ime, he devaluaion-based policies (affeced hrough changes in nominal exchange rae) mus cooperae wih sabilizaion policies (o ensure domesic price level sabiliy) o achieve he desired level of rade balance. However, devaluaion-based policies had caused some problem. Devaluaion-based policies would cause increases in he cos of impor. This migh lead o impor inflaion ha would damage he domesic firms ha use impored inpus. Besides ha, he devaluaion-based policies may no effecive in improving rade balance if oher counries also apply he devaluaion-based policies a he same ime. On he oher hand, he counries should implemen he policy ha focuses on he producion of impored-subsiued goods. Impor-subsiuion policy may work well in improving domesic income and rade balance. References Ahmad, J. & Yang, J. (2004). Esimaion of he J-curve in China. Economics Series Eas Wes cener working papers. Akbosanci, E. (2002). Dynamics of he rade balance: he Turkish J-curve. Economic research cener working papers in economics, 0105. Baharumshah, A.Z. (2001). The effec of exchange rae on bilaeral rade balance: New evidence from Malaysia and Thailand. Asian Economic Journal, Vol.15 (3), 291-312. Bahmani-Oskooee, M. & Raha, A. (2007). The bilaeral J-curve: Sweden Versus her 17 Major rading parners. Inernaional Journal of Applied Economics, 4(1), 1-13. Gomez, D.M., & Alvarez-Ude, G.F. (2006). Exchange rae policy and rade balance: A coinegraion analysis of he Argenine experience since 1962. MPRA Paper 151, Universiy Library of Munich, Germany. Haemi-J, A., & Irandous, M (2005). Bilaeral Trade Elasicies: Sweden Versus Her Major Trading Parners. Available: www.arpejournal.com/aprevolume3number2/haemi-j-irandous.pdf. Hernan Rinco, C. (1999). Tesing he shor-and-long run exchange rae effecs on rade balance: The case of Colombia. Borradores De Economia 003561, Banco De La Republica. Liew, K.S., Lim, K.P., & Hussain, H. (2003). Exchange rae and rade balance relaionship: The experience of ASEAN counries. EconWPA, Inernaional Trade wih number 0307003. Liu, L.G., Fan, K., & Shek, J. (2006). Hong Kong s Trade Panerns and Trade Elasiciies. Available :www.info.gov.hk/hkma/eng/research/rm18-2006.pdf. Moffe, M.H. (1989). The J-curve revisied: an empirical examinaion for he Unied Saes. Journal of Inernaional Money and Finance, 8, 425-444. Onafowora, O. (2003). Exchange rae and rade balance in Eas Asia: Is here a J-curve. Economic Bullein, Vol. 5, No.18, pp. 1-13. Rose, A.K. (1991). The role of exchange raes in a popular model of inernaional rade, Does he Marshall Lerner condiion hold. Journal of Inernaional Economics, 30, 301-316. Rose, A.K., & Yellen, J.L. (1989). Is here a J-curve. Journal of Moneary Economics, 24, 53-68. Shirvani, H., & Wilbrae, B. (1997). The relaionship beween he real exchange rae and he rade balance: An empirical reassessmen. Inernaional Economic Journal, Vol. 11(1), 39-51. Sugema, I. (2005). The deerminans of rade balance and adjusmen o he crisis in Indonesia. Cenre for inenaional economics sudies, No.0508. Thorbecke, W. (2006). The effec of exchange rae changes on rade in Eas Asia. RIETI Discussion Paper Series 05-E-009. 135
Vol. 3, No. 8 Inernaional Journal of Business and Managemen Wilson, P., & Kua, C.T. (2001). Exchange raes and he rade balance: he case of Singapore 1970 o 1996. Journal of Asian Economics, 12, 47-63. Figure 1. Response of LTB o Cholesky One S.D. LRER Innovaion Appendix Table 1. Tesing for Uni Roo (ADF & PP es) ADF sa. PP sa. Inercep Inercep Inercep Inercep Variables & no rend & rend & no rend & rend ln TB -4.0302** -3.9965* -4.0030** -3.9081* (0.0027) (0.0151) (0.0029) (0.0187) ln RER -0.8049-2.2823-0.6402-2.2851 (0.8091) (0.4355) (0.8520) (0.4341) ln Y 0.6961-2.7771 0.5595-2.8662 (0.9910) (0.2122) (0.9872) (0.1818) ln Y* -1.0674-0.4965-0.8322-0.7387 (0.7215) (0.9806) (0.8012) (0.9644) ln TB -4.8441** -4.8861** -8.9251** -8.8943** (0.0003) (0.0014) (0.0000) (0.0000) ln RER -7.3252** -7.2673** -7.5435** -7.6354** (0.0000) (0.0000) (0.0000) (0.0000) ln Y -5.7823** -5.7587** -5.7317** -5.7092** (0.0000) (0.0001) (0.0000) (0.0001) ln Y* -2.0718-4.2251** -4.2324** -4.3117** (0.2566) (0.0082) (0.0015) (0.0065) Noe: *, ** denoe significance a he 5% and 1% level of significance respecively. ( ) denoes he p-value. Table 2. Kwiakowski-Philips-Schmid-Shin (KPSS) es (Inercep wihou rend) Variable KPSS Sa. 1%CV 5%CV 10%CV ln Y* 0.2514 0.7390 0.4630 0.3470 Noe: 1%CV, 5%CV, and 10%CV 136 sand for 1% criical values, 5% criical values, and 10% criical values.
Inernaional Journal of Business and Managemen Augus, 2008 Table 3. Resul of Engle-Granger Tes Variable ADF sa. -saisic Prob. u -3.9711 0.0002 Table 4. Esimaed Coinegraed Vecors in Johansen ln TB ln RER ln Y lny* Consan Msia/US -1.0000 0.0851 0.0992-0.1324 0.0401 Noe: The esimaed coefficiens were obained by normalizing he rade balance variable. Table 5.Diagnosic Checking A. Residuals-Diagnosic Views H 0 : non auocorrelaion H 0 : normaliy H 0 : homoscedasiciy Q LM JB Chol JB Urz Whie 18.95 13.70 184.68 323.46 230.56 B. Pairwise Granger Causaliy Resul Based on VECM [Msia/US(lag 1)] x 2 -saisics (p-value) ln TB ln RER ln Y ln Y* Dependen Variable ln TB - 3.21(0.07) * 12.92(0.00) *** 5.32(0.02) ** ln RER 0.11(0.73) - 0.00(0.99) 0.07(0.78) ln Y 0.20(0.64) 0.75(0.38) - 1.95(0.16) ln Y* 0.30(0.58) 0.10(0.91) 0.04(0.82) - Noe: The x 2 (Wald) saisics for he join significance of each of he oher lagged endogenous variables in ha equaion. ( ) denoes he p-value. *, **, *** denoe significance a he 10%, 5% and 1% level of significance respecively. Table 6. Tesing for Coinegraion (Full Sample) lag 1 H 0 H 1 5%CV 1%CV Trace sa. r 0 r>0 47.21 54.46 57.4729** r 1 r>2 29.68 35.65 26.4101 Max r=0 r=1 27.07 32.24 31.0627* r=1 r=2 20.97 25.52 16.4820 Noe: * (**) denoes rejecion of he hypohesis a 5 %( 1%) level. 5%CV and 1%CV sand for 5% criical values and 1% criical values. Chosen r o denoe number of coinegraing equaion under boh ess. 137