ARE MALAYSIAN EXPORTS AND IMPORTS COINTEGRATED?

Transcription

1 Sunway College Journal 1, 29 38(2004) ARE MALAYSIAN EXPORTS AND IMPORTS COINTEGRATED? CHOONG CHEE KEONG a Universii Tunku Abdul Rahman SOO SIEW CHOO Monash Universiy Malaysia ZULKORNAIN YUSOP Universii Pura Malaysia ABSTRACT This paper invesigaes he long-run relaionship beween Malaysian expors and impors. Towards his end, mulivariae coinegraion echniques have been applied. For all measures (or cases), boh he unresriced and resriced coinegraion resuls demonsrae ha here exiss a long-run relaionship beween Malaysian expors and impors. Simply pu, boh he variables will converge owards equilibrium in he long run which indicaes he effeciveness of Malaysia s long-erm macroeconomic planning in sabilising rade balance. Thus, he Malaysian economy does no violae is iner-emporal budge consrain. Key words: Trade balance, Iner-emporal budge consrain, coinegraion. INTRODUCTION Malaysia, as a small counry wih an open economy, is highly dependen on foreign rade. Any changes in he inernaional marke, eiher hrough price of commodiies or inernaional demand, and domesic macroeconomic policies boh fiscal and moneary policies will grealy affec expors, impors and economic growh. Malaysia s dependency on rade is illusraed in Table 1. In he fifh column, we find ha he dependency raio (expors plus impors divided by nominal GDP) has been increasing from 1980 o 2000, and is expeced o grow furher in he fuure. This implies ha he Malaysian economy is highly dependen on foreign rade. During he Asian financial crisis of , Malaysia s foreign rade volume (boh expors and impors) was double ha of is Gross Domesic Produc (GDP). The relaionship beween expors and impors is presened in Figures 1 o 4 for 1959 hrough In all four figures, expors and impors move closely ogeher from 1959 o From 1998, however, boh series begin o drif away from each oher and flucuae unil I is srongly believed ha he shor-run disequilibrium is due o some forms of exernal and inernal shocks in he Malaysian economy. The Asian financial crisis ha firs eruped in July 1997 in Thailand before affecing Malaysia in he subsequen monhs, for example, is viewed as a source of disequilibrium for Malaysia s rade balance. Early in he crisis, Malaysia s macroeconomic policy underwen a a choongck@mail.uar.edu.my.

2 30 Choong C. K., Soo S. C. and Zulkornian Yusop U-urn from iniially conracionary fiscal and moneary policies o expansionary policies. Laer, he conroversial capial conrols were implemened, followed by he dramaic swich o a pegged exchange rae regime on 1 Sepember 1998 deemed an economic heresy by he world. These moves have had a grea impac on he counry s rade balance. These policy changes raise wo quesions: (1) are he flucuaions beween expors and impors susainable in he long erm, and (2) do hese macroeconomic policies effecively influence he rade balance? This sudy aims o invesigae and o answer he quesions posed. In he following secion, we discuss a simple heoreical model beween expors and impors. Nex, some economeric echniques are described. Finally, he empirical resuls are repored. Year Table 1. Malaysia s Dependency Raio, GDP Expors (RM Millions) (RM Millions) (a) (b) (c) Impors (RM Millions) Dependency Raio [(b) + (c)]/ (a) Source: Inernaional Moneary Fund ( ), Inernaional Financial Saisics, various issues. SIMPLE THEORETICAL MODEL Following Hused (1992), we consider consumers who live in a small, open economy wih no governmen inervenion. The consumers are assumed o maximise heir uiliy funcion subec o a budge consrain, and hey borrow and lend in inernaional markes a a predeermined world ineres rae o achieve maximum uiliy. The consumers revenues consis of an endowmen of oupus and profis disribued from firms. These revenues are used for consumpion and saving. Hence, he individual s curren period budge consrain is as follows: C = Y + B I ( 1+ r ) B 1 (1) where C is curren consumpion; Y is oupu level; I is invesmen; r is he curren world ineres rae; B is he inernaional borrowings; and ( 1+ r ) B 1 is he deb of he previous period, which corresponds o he counry s exernal deb.

3 Sunway College Journal 1, 29 38(2004) 31 Because condiion (1) mus hold for every ime period, he iner-emporal budge consrain is obained by he summaion of all individuals budge consrain in he economy: B = = 1 λ TB + lim λ B (2) n n n where TB = X M ( = Y C I ) indicaes he rade balance in period, X is expors, M equals impors, λ = 1 (1 + r ) and is he discoun facor defined as he 0 produc of he firs values of λ. When he las erm of equaion (2) equals o zero, hen a counry s borrowing (lending) is exacly he same as he presen value of he fuure rade surpluses (deficis). We can derive he esable model by rearranging equaion (1) as Z ( 1+ r) B = X + B (3) + 1 = M + ( r + r) B where Z 1 and i is assumed ha he world ineres rae is saionary wih uncondiional mean r. According o Hakkio and Rush (1991), equaion (3) can be expressed in more deail: + [ X + Z+ ] lim B 1 M + r B = X + λ + λ + = 0 1 (4) where λ = 1 (1 + r) and is he firs-difference operaor. The lef-hand side represens expendiure on impors as well as ineres paymens (receips) on ne foreign deb (asses). If X is subraced from boh sides of (4) and each side is muliplied by minus one, hen he lef-hand side becomes he counry s curren accoun. Assuming boh X and Z are I(1) variables (ha is, hey are non-saionary a level form), hen equaion (4) can be re-expressed as: + X α + MM lim λ B + ε (5) = + MM ; α = [(1 + r) 2 1 /r](α 2 α 1 ) and ε = λ ( ε ε ) where = M + r B α is he drif parameers (possibly equal o zero) and ε are saionary process. From equaion (5), since, hen he limi erm will become zero. Hence, equaion (5) can be resaed as a sandardised regression: X = a + b * MM + e (6) Equaion (6) saes ha a counry saisfies is iner-emporal budge consrain if he esimaed coefficien of MM equals o uniy ( b = 1) and e is whie noise disurbance erm and saionary. If boh he condiions are valid, hen expors and impors are coinegraed.

4 32 Choong C. K., Soo S. C. and Zulkornian Yusop RM Millions EXP IMP Year Figure 1. Nominal Value of Expors and Impors (RM). Source: Inernaional Moneary Fund ( ), Inernaional Financial Saisics, various issues US$ Millions EXP IMP Year Figure 2. Nominal Value of Expors and Impors (US$). Source: Inernaional Moneary Fund ( ), Inernaional Financial Saisics, various issues.

5 Sunway College Journal 1, 29 38(2004) 33 Real Value (RM Millions) REXP RIMP Year Figure 3. Real Value of Expors and Impors (RM). Source: Inernaional Moneary Fund ( ), Inernaional Financial Saisics, various issues Real Value (US$ Millions) REXP RIMP Year Figure 4. Real Value of Expors and Impors (US$). Source: Inernaional Moneary Fund ( ), Inernaional Financial Saisics, various issues. METHODOLOGY Many economic ime series clearly display rend and seasonal componens and some also exhibi common feaures over ime such as saionariy. However, i is only recenly ha ime series economericians have formalized in economeric models he concep of common co-movemens a paricular frequencies, as well as he idea ha common facors may influence he rend componen of some macroeconomic variables. Neverheless, he saisical underpinnings of ime series analysis require daa o be saionary. This requires a firs-difference for mos macroeconomic series before esimaing he economic models.

6 34 Choong C. K., Soo S. C. and Zulkornian Yusop Hence, he significance of deecing and recifying he rend componen in macroeconomic daa is sufficienly indicaed. If wo or more variables have a common rend, hen causaliy mus exis in a leas one direcion (Engle and Granger, 1987). Moreover, if a rend componen exiss beween wo variables, boh variables will move ogeher owards a long-erm equilibrium. Many series, which separaely are non-saionary, when oined linearly have a long-erm equilibrium relaionship (Engle and Granger, 1987). Hence, boh series are said o be coinegraed. Coinegraion ess are concerned wih he long-erm behaviour among he componens of parially non-saionary ime series which is an indicaion of a common rend componen. In oher words, coinegraion is he saisical approach ha ess for he exisence of a long-run equilibrium relaionship among non-saionary variables inegraed of he same order. Two non-saionary series are said o be inegraed if here exiss a linear combinaion of he wo series. For example, by using sandard OLS regression echniques, we can esimae he parameers of he coinegraing regression and calculae he residual erms, U, where: Y = α + βx + U (7) Suppose ha he variables are firs-order inegraed, I(1), bu ha here exis values of α and β such ha U = Y α βx ~ I(0), wih zero mean, ha is, E ( Z ) = 0. Then we conclude ha he variables are coinegraed. According o he mulivariae model of Johansen and Juselius (1990), he following vecor auoregressive models are esimaed: X = Π1 X 1 + Π 2X Π k X k + ν (8) where X is an N 1 vecor of I(1) variables. The long-run equilibrium associaed wih (2) is ΠX = 0, where he long-run coefficien marix Π is defined as Π = I Π1 Π 2... Π k (9) The long-run coinegraing marix Π is an N N marix whose rank deermines he number of coinegraing vecors, say p. If we define wo marices α and β (boh N r ) ' such ha Π = αβ, where he prime denoes he marix ranspose, he row of β will be from he p coinegraing vecors. Johansen and Juselius (1990) have inroduced wo likelihood raio ess in deermining he number of coinegraing vecors, namely, he maximum-eigenvalue and race ess. In esing he long-run relaionship beween expors and impors by applying coinegraion procedures, four differen measures will be employed for each variable. Following Bahmani-Oskooee and Rhee (1997), he firs measure is he nominal value of expors and impors in erms of he Malaysian ringgi, denoed by C1EXP and C1IMP. The second measure is also a nominal erm, bu denominaed in U.S. dollars, denoed by C2EXP and C2IMP. The hird measure for each variable is he real expors as well as impors in erms of a consan ringgi. These measures can be obained by dividing C1EXP and C1IMP by he consumer price index (CPI), denoed by C3EXP and C3IMP. The fourh measure is also in real erms, where we de flae C2EXP and C2IMP by he CPI. The real expors and impors measured in U.S. dollars are denoed by C4EXP and C4IMP, respecively. These four measures are crucial in reflecing he effeciveness of macroeconomic policies. Dividing he measures wih he Malaysian ringgi and he U.S. dollar is significan

7 Sunway College Journal 1, 29 38(2004) 35 in reflecing he effeciveness of he exchange rae policy or regime ha has been implemened. Moreover, boh expors and impors expressed in nominal and real erms aim o show he success of he Mala ysian governmen in sabilising and conrolling aggregae price level (inflaion), economic growh and rade policies. Therefore, if he resuls of all measures provide similar conclusions, hen we can conclude ha he Malaysian governmen has been implemening effecive macroeconomic policies. Conversely, if he resuls of he measures give differen findings, hen we can deduce which ype of policy is more effecive. Annual daa were used for he period and he daa were colleced from Inernaional Financial Saisics published by Inernaional Moneary Fund (IMF). EMPIRICAL RESULTS AND INTERPRETATION The resuls of he Augmened Dickey-Fuller (ADF) uni roo es are demonsraed in Table 2. We found ha all series are non-saionary a heir level form since he null hypohesis of non-saionariy canno be reeced a he 5% significance level. However, he null hypohesis is easily reeced a he 5% significance level afer he firs difference. Simply, he variables are now I(1) variables. Variable Table 2. Uni Roo Resuls (lag-lengh, k = 1 ) ADF Tes Level Consan, No Consan Trend wih Trend Firs Difference Consan, No Trend Consan wih Trend Sample Period: C1EXP ** ** C1IMP ** ** C2EXP ** ** C2IMP ** ** C3EXP ** ** C3IMP ** ** C4EXP ** ** C4IMP ** ** Noe: The null hypohesis is ha he series is non-saionary, or conains a uni roo. The reecion of he null hypohesis for he ADF es is based on he MacKinnon criical values; ** indicaes he reecion of he null hypohesis of non-saionariy a 5% significance level. Wih he same order of inegraion beween he variables, we have saisfied he condiion for he Johansen and Juselius mulivariae coinegraion es in invesigaing he long-run relaionship beween expors and impors for all cases. In Table 3 (Panel A), we can reec he null hypohesis of r = 0 agains is alernaive r = 1, for all measuremens, or cases a he 95% confidence level. However, we failed o reec he null hypohesis of r

8 36 Choong C. K., Soo S. C. and Zulkornian Yusop 1 agains is alernaive r = 2 a he same level of significance. Therefore, we conclude ha here exiss a single coinegraing vecor in all cases. Table 3. The Resuls of Coinegraion Tes (lag-lengh, k=5) Variable Null Alernaive λ-max 95% Trace 95% Hypohesis Hypohesis Tes Panel A: Case 1: r = 0 r = ** ** 15.4 C1EXP & C1IMP r 1 r = Case 2: r = 0 r = ** ** 15.4 C2EXP & C2IMP r 1 r = Case 3: r = 0 r = ** ** 15.4 C3EXP & C3IMP r 1 r = Case 4: r = 0 r = ** ** 15.4 C4EXP & C4IMP r 1 r = Panel B: Case C1EXP C1IMP C2EXP C2IMP C3EXP C3IMP C4EXP C4IMP ** indicaes he reecion of he null hypohesis of non-saionariy a 5% significance level. In addiion, here appears o be a posiive relaionship beween expors and impors in he long erm, as indicaed by he esimaed coinegraing vecors for all cases, which range from o (Table 3, Panel B). From he esimaed coinegraing vecors, which are posiive and close o uniy, we inerpre hese findings as Malaysia s adherence o he inernaional budge consrain. To confirm his conclusion, we conduced he resriced coinegraion es o examine he one-o-one relaionship beween expors and impors. In Table 4, for all cases, we were unable o reec he null hypohesis of a uniy relaionship beween hese wo variables. In oher words, he Malaysian governmen has been playing a crucial role in sabilising he rade balance (expors and impors), and all of Malaysia s macroeconomic policies have been effecive in leading expors and impors ino a long-run seady-sae equilibrium relaionship.

9 Sunway College Journal 1, 29 38(2004) 37 Table 4. The Resuls of Resriced Coinegraion Tes (lag-lengh, k=5) Case Likelihood Raio Decision (F saisics) [0.2908] Do no reec H [0.2890] Do no reec H [0.2074] Do no reec H [0.2775] Do no reec H 0 Noe: Figures in [] are marginal level. Null hypohesis: The esimaed coefficien of impor is equal o uniy. The resuls discussed, however, do no reflec he overall picure of he expor-impor disequilibrium in he shor run. The shor-run disequilibrium, as demonsraed in Figures 1 and 4, brings forh he quesion of wheher i is driven by expors or impors. Therefore, he Granger causaliy es wihin he vecor error correcion model (VECM) was applied o examine he dynamic shor-run causaliy beween hese wo variables. The causaliy resuls are repored in Table 5. Several findings are o be noed. Firs, unidirecional causaliy running from expors o impors in Case 1 and Case 3 and saisically significan a 5 percen significance level. In conras, Case 2 and Case 4 are insignifican even a 10% significance level. The error correcion erms in all cases, however, are highly significan a 5% level of significance. Second, shor-run causaliy relaionship beween wo variables does no exis in Cases 2 and 4, even a 10% significance level, which implies ha hese variables drif apar from each oher and ha here is no error correcion mechanism in correcing hese flucuaions. Finally, we can conclude ha he disequilibrium is paricularly creaed by he insabiliy in he foreign exchange marke. Neverheless, he Malaysian governmen has succeeded in sabilising domesic macroeconomic environmen wih a sable level of inflaion. Table 5. Granger Causaliy Resuls based on Vecor Error Correcion Model (Uniform lag lengh, k=5) Independen Variables Dependen Variables D(EXP) D(IMP) ECT F-saisic (Significance level) -saisic Case 1: D(EXP) D(IMP) ** ** Case 2: D(EXP) D(IMP) ** Case 3: D(EXP) D(IMP) ** ** **

10 38 Choong C. K., Soo S. C. and Zulkornian Yusop Table 5 (coninued) Dependen Variables Independen Variables D(EXP) D(IMP) ECT F-saisic (Significance level) -saisic Case 4: D(EXP) D(IMP) ** Noes: The F-saisics ess he oin significance of he lagged values of he independen variables, and - saisics ess he significance of he error correcion erm (ECT). The aserisks indicae he following levels of significance: *10%, **5% and ***1%. CONCLUSION The main aim of his sudy is o invesigae he long-run relaionship beween Malaysia s expors and impors by applying well-developed economeric echniques. Using coinegraion procedures, we found ha Malaysia s expors and impors will converge in he long run. This means ha he shor-run flucuaions beween expors and impors are no susainable since in he long run; expors and impors will evenually converge owards an equilibrium sae. The ulimae convergence owards equilibrium signifies he overall effeciveness of Malaysia s pas macroeconomic policies in sabilising rade condiions, which does no exceed he iner-emporal budge. REFERENCES Bahmani-Oskooee, M., & Rhee, H. (1997). Are impors and expors of Korea coinegraed? Inernaional Economic Journal, 11, Engle, R. F., & Granger, C. W. J. (1987). Coinegraion and error correcion: Represenaion, esimaion and esing. Economerica, 55, Hakkio, C., & Rush, M. (1991). Is he budge defici oo large? Economic Inquiry, 29(3), Hused, S. (1992). The emerging U.S. curren accoun defici in he 1980s: A coinegraion analysis. The Review of Economics and Saisics, 74(1), Inernaional Moneary Fund ( ). Inernaional Financial Saisics, various issues. Washingon D.C.: IMF. Johansen, S., & Juselius, K. (1990). Maximum likelihood esimaion and inference on coinegraion wih applicaions o he demand for money. Oxford Bullein of Economics and Saisics, 52,