REAL EXCHANGE RATES AND BILATERAL TRADE BALANCES: SOME EMPIRICAL EVIDENCE OF MALAYSIA

Transcription

1 REAL EXCHANGE RATES AND BILATERAL TRADE BALANCES: SOME EMPIRICAL EVIDENCE OF MALAYSIA Risalshah Latif Zulkarnain Hatta ABSTRACT This study examines the impact of real exchange rates on the bilateral trade balances of Malaysia with the United State (US), Japan and Singapore. The results of the long-run cointegrating vectors show that a depreciation or devaluation of real exchange rates will improve bilateral trade balances. In the short run, the changes of real foreign money supplies have contributed a lot to the changes of real exchange rates. However, the changes of real exchange rates do not contribute a lot to the changes of bilateral trade balances. On the whole, monetary policies can play a role in the real exchange rate determination and therefore, bilateral trade balance. Keywords: Bilateral trade balance, real exchange rate, income, real money supply, cointegration, generalised impulse response function. INTRODUCTION There are many studies on the impact of exchange rate on the trade balance. However, there is no consensus on the impact of exchange rate on the trade balance (Wilson, 1999, 2001; Baharumshah, 2001; Bahmani-Oskooee & Ratha, 2004; Devereux & Genberg, 2007: 571; Yusoff, 2009). 1,2 According to the elasticity approach of the trade balance determination, an increase in the exchange rate (depreciation or devaluation) will lead to an increase in the trade balance. Nonetheless, this will happen when the Marshall-Lerner condition holds, that is, the sum of the elasticity of import demand and the elasticity of export supply is greater than unity. In the short run, an increase in the exchange rate will make trade balance become worse and then over time, a depreciation of exchange rate will improve trade balance. This

2 Risalshah Latif & Zulkarnain Hatta pattern of the relationship between trade balance and the exchange rate is called the J-curve phenomenon (Daniels & Vanhoose, 2005: 279). 3 Devereux and Genberg (2007) show no evidence of the impact of the real exchange rate on the trade balance. On the other hand, Qiao (2007: 26) shows that depreciation of the exchange rate would lead to an increase in the trade balance of a debtor country. However, an appreciation of the exchange rate may or may not decrease the trade balance of a creditor country. The monetary approach of the exchange rate determination argues that differences between the quantity of money demanded and the quantity of money supplied will lead to changes in the exchange rate, which in turn will lead changes in the trade balance. Under a fixed-exchange rate arrangement, the difference between the quantity of money demanded and the quantity of money supplied causes changes in the trade balance. Conversely, under a flexible-exchange rate arrangement, the exchange rate value of the domestic currency changes (Daniels & Vanhoose, 2005: ). Monetary policies are important in the exchange rate determination. More specifically, exchange rate is endogenous to present and future monetary policies (Qiao, 2007: 4 5). Conversely, Zhang and Wan (2008) show that monetary policies would not be able to improve imbalances in the trade balance of China. Lee and Chinn (2006) show that, generally temporary shocks play a larger role in explaining variations in the current account whilst permanent shocks play a larger role in explaining the variations in the real exchange rate. This study examines the impact of the real exchange rates on bilateral trade balances of Malaysia with the United States (US), Japan and Singapore, using an augmented model of Rose and Yellen (1989), by including the real domestic money supply and the real foreign money supply. 4 In this study, the real exchange rate is treated endogenous in the estimation (Qiao, 2007). Therefore, this study provides some evidence of the impact of real money supply on real exchange rate and indirectly, the impact on bilateral trade balance of a country. Conversely, it is treated to be exogenous in the estimation in most previous studies 24

3 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia in the literature (Wilson, 1999, 2001; Baharumshah, 2001; Yusoff, 2009). If the real exchange rate is endogenous, estimating bilateral trade balance by treating the exchange real rate as exogenous will produce biased result. Furthermore, this study examines the impact of real shocks on bilateral trade balances as nominal shocks are said to have less significant impact (Lee & Chinn, 2006; Zhang & Wan, 2008). 5 This study examines bilateral trade balances of Malaysia with the US, Japan and Singapore mainly as they are important trading partners of Malaysia. In 2002, the trade of Malaysia with the US, Japan and Singapore accounted for 48.5 percent of the overall Malaysian trade. In 2008, the trade of Malaysia with these countries was 38 percent of the overall Malaysian trade (Table 1). Bilateral trade balances are examined to avoid asymmetric responses of trade flows to the changes of the real exchange rates across countries (Wilson, 1999: 2). Moreover, a country could run trade deficits with some trading partners and trade surpluses with other trading partners. This may be because of the differences in the composition of bilateral trade. A country may export primary products or raw materials to one partner and manufactures to another. The Dickey and Fuller (1979) (DF) and Phillips and Perron (1988) (PP) unit root test statistics are used to examine the stationarity of the data. The Johansen (1988) cointegration method is used to examine the long-run relationships among the bilateral trade balance, the real exchange rate, domestic income, foreign income, real domestic money supply and real foreign money supply. The generalised forecast error variance decomposition and generalised impulse response function (Koop, Pesaran & Potter, 1996; Pesaran & Shin, 1998) are used to examine the relationships of variables in the estimation. The relative importance of variables in the estimation can be examined. The structure of this article is as follows. Section 2 presents literature review of the trade balance and the exchange rate. Section 3 discusses the methodology and data used in this study. Section 4 discusses the empirical results. Finally, Section 5 provides some concluding remarks. 25

4 Risalshah Latif & Zulkarnain Hatta LITERATURE REVIEW The impact of the exchange rate on the trade balance is subject to debate. Devereux and Genberg (2007) developed an open macroeconomics model to examine the impact of exchange rate on trade balance. More specifically, the model focuses on current account between an advanced economy and an emerging market economy. The model incorporates three realistic asymmetries between the two economies, namely emerging market economy sets export prices in dollars, holds a large fraction of dollars as gross assets, and uses a substantial fraction of intermediate imports in export production. It is shown that the appreciation of currency of an emerging market economy may not able to reduce its current account. In fact, it may create an increase in its current account. The model is used to examine the impact of an appreciation in renminbi on the current account of China to the US. The results show that it is unlikely that an appreciation in renminbi will lead to a decrease in the current account of China to the US. Qiao (2007: 26) examines the impact of the exchange rate on the trade balance of China. The study shows that an increase in the exchange rate (depreciation) would lead to an increase in trade balance of a debtor country. However, a decrease in the xchange rate (appreciation) may or may not decrease the trade balance of a creditor country. Moreover, the study shows that it is important to consider other effects such as wealth effect, investment effect and indirect investment effect when examining the elasticity approach of the exchange rate determination. The appreciation of renminbi is not able to reduce the trade balance of China, but it motivates the inflow of hot money to speculate on further appreciation, which creates more problem to domestic monetary control. Ogawa and Kudo (2007) examines the impact of a depreciation in the US dollar on East Asian currencies. The results show that a significant depreciation of the US dollar will be necessary in order to reduce the deficit in current account. The responses of the East Asian currencies to a sudden and sharp depreciation of the US dollar will differ with countries because of the different degree of linkages of the East Asian currencies to the US dollar. They propose a regional coordination of the exchange rate policy is necessary to the East Asian countries to respond appropriately to a possible depreciation of the US dollar. 26

5 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia Zhang and Wan (2008) argues that there are two groups of factors that affect trade balance, namely, factors that affect the trade balance permanently and factors that affect the trade balance temporarily. They examine the impact of the real effective exchange rate on the trade balance of China over the period of They use a vector autoregressive (VAR) model, which includes trade balance, trade ratio, real domestic income, real foreign income, relative demand and real effective exchange rate. The results suggest that the changes in the trade balance of China are the result of real shocks. Thus, the trade balance may not be amenable by monetary measures. Moreover, relative income is more important than relative price in the trade balance determination of China. This implies that the trade pattern between China and its major trading partners is more complementary than competitive. Furthermore, it is found that nominal shocks do leave substantial imprint on the real effective exchange rate. However, there is no strong evidence such shocks can elicit strong substitution effects on the trade balance. A nominal devaluation may be followed by real depreciation and temporary improvement of the trade balance. Nevertheless, the improvement is more a result of expenditure reduction caused by output contraction than a result of expenditure switching. Lee and Chinn (2006) examines the relationship between the real exchange rate and the current account in a framework of inter-temporal open macroeconomics model for G7 countries (the US, Canada, the United Kingdom, Japan, Germany, France and Italy) using quarterly data over the period from 1979/1980 to They use a VAR model, which includes the real exchange rate and the current account. The results show that temporary shocks play a larger role in explaining the variations in the current account, whilst permanent shocks play a larger role in explaining the variations in the real exchange rate, except the US. Moreover, temporary shocks depreciate the real exchange rate and improve current account balance, except the United Kingdom. Thus, it is difficult to uncover the relationship between the real exchange rate and current account. The permanent shocks appreciate the real exchange rate and improve the current account balance for some cases, which is contradicted to many other models. 27

6 Risalshah Latif & Zulkarnain Hatta Bergin (2006) develops a macroeconomics model to examine the impact of exchange rate on the current account. The model incorporates monopolistically-competitive firms, sluggish price setting, capital accumulation subject to adjustment costs and monetary policy in the form of interest rate setting rules. The study examines the current account of the US against the remaining G7 countries using quarterly data over the period from 1973 (quarter I) to 2000 (quarter IV). The results show that deviations from interest rate parity do not seem to be closely related to monetary policy, as has been hypothesised in recent theory, but that these deviations do seem to be related to shifts in the marginal utilities of consumption. Moreover, the model indicates that such interest rate parity shocks are not important in explaining variations in exchange rate observed in the data. However, these shocks are helpful in explaining variations in current account. Yusoff (2009) examines the impact of real exchange rates on bilateral trade balances of Malaysia with the US, Singapore and Japan using the Johansen s cointegration method. The results suggest that bilateral trade balances, real exchange rates and incomes are cointegrated. Moreover, the coefficients of real exchange rates are found to be positive. This implies that the depreciation of real exchange rates can improve the trade balance. Both the foreign and domestic incomes are important determinants of bilateral trade balance. The impulse response analyses tend to indicate that a depreciation of ringgit will immediately deteriorate bilateral trade balances, except in the case of Singapore. More specifically, a clear evidence of the J-curve phenomenon is found only in the case of bilateral trade balance of Malaysia with the US. The study suggests that the devaluation of exchange rate is a potent policy option for Malaysia to enhance its competitiveness in the world market and improve its trade balance whilst revaluation of ringgit may hurt exports and trade balances of Malaysia. METHODOLOGY AND DATA The DF and PP unit root test statistics are used to examine the stationarity of the data. The Johansen s cointegration method is used to test the long-run relationship among the variables in the model. 28

7 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia This method proposes two likelihood ratio tests to test the number of cointegrating vectors, namely, the maximum eigenvalue (λ Max ) and trace (λ Trace ) statistics. The λ Max test statistic tests the null hypothesis (H 0 ) of r cointegrating vectors against the alternative hypothesis (H a ) that there are {r + 1} cointegrating vectors in the model. The λ Trace test statistic tests the H 0 that has at most r cointegrating vectors in the model, that is, the number of cointegrating vectors is less than or equal to r. The critical values for the λ Max and λ Trace test statistics are tabulated in Pesaran, Shin and Smith (2000). If there are two cointegrating vectors in the estimation, the first cointegrating vector shall be identified by normalising the bilateral trade balance, and the second cointegrating vector shall be identified by normalising the real exchange rate. Thus, real exchange rate is treated as endogenous in the estimation. The impact of real money supply on bilateral trade balance can be assessed. More specifically, the first and second cointegrating vectors, respectively, is specified as follows: log (X/M) t = β 10 + β 11 log RER t + β 12 logy t + β 13 logy t + u 1,t (1) log RER t = β 20 + β 21 logy t + β 22 logy t + β 23 logms t + β 24 log MS t + u 2,t (2) where, log is the natural logarithm, X t is exports, M t is imports, RER t is real exchange rate, Y t is domestic income, Yt is foreign income, MS t is real domestic money supply, and u i,t (i = 1, 2) is real foreign money supply, and u i,t (i = 1, 2) is a disturbance term. Moreover, a dummy variable, that is, one for the period of 1997 (quarter III) to 1998 (quarter IV), and the rest are zero, is included in the estimations to capture the influence of the Asian financial crisis ( ). The impact of real exchange rate on bilateral trade balance is ambiguous. However, it is generally argued that an increase in the real exchange rate, or it is said to be depreciation of real exchange rate in this study, will lead to an increase in bilateral trade balance. Thus, the coefficient of the real exchange rate is expected to be positive. If the coefficient of real exchange rate is greater than zero, the Marshall- Lerner condition holds. The impact of domestic income or foreign income on bilateral trade balance is also ambiguous. Generally, an increase in domestic income will lead to more imports of a domestic 29

8 Risalshah Latif & Zulkarnain Hatta country, whilst an increase in foreign income will lead to more exports of a domestic country. Thus, the coefficients of domestic income and foreign income are expected to be negative and positive, respectively. Nonetheless, an increase in domestic income may lead to more exports of a domestic country, or an increase in foreign income may lead to less imports of a foreign country and thus, the coefficients of domestic income and foreign income could be positive and negative, respectively (Onafowora, 2003: 2). The impact of domestic income and foreign income on real exchange rate is ambiguous. An increase in domestic income could lead to more demand, including more imports and therefore, real exchange rate is expected to depreciate. On the other hand, an increase in foreign income could lead to more demand, including more imports and therefore, real exchange rate is expected to appreciate. However, an increase in domestic income could lead to the relative quantity of money demanded to decline and therefore, real exchange rate is expected to appreciate. An increase in domestic money supply would cause depreciation in real exchange rate, whilst an increase in foreign money supply would cause appreciation in real exchange rate. Generally, the coefficients of domestic income and domestic money supply are expected to have a positive sign, whilst the coefficients of foreign income and foreign money supply are expected to have a negative sign (Daniels & Vanhoose, 2005: ). 6 The generalised forecast error variance decomposition and generalised impulse response function (Koop, Pesaran & Potter, 1996; Pesaran & Shin, 1998) are used to examine the relationship of the variables in the estimation. The generalised forecast error variance decomposition identifies the proportion of forecast error variance in one variable caused by the innovations in other variables in the estimation. The relative importance of a set of variables that affect a variance of another variable is identified. The generalised impulse response function traces the dynamic responses of a variable to innovations in other variables in the estimation. The generalised forecast error variance decomposition and generalised impulse response function (Koop, Pesaran & Potter, 1996; Pesaran & Shin, 1998) solve the orthogonalised problem of the forecast error variance decomposition 30

9 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia and impulse response function of Sims (1980). The problem is that the latter approaches are sensitive to the order of the variables in which they enter the VAR. Let, x t is an m 1 vector of jointly-determined dependent variables, which are assumed to be I(1) series, respectively. The infinite moving average representation of x t can be written as: x C C w, t = 1, 2,, T (3) t i t i i 0 i 0 i t i where, C i is m m coefficient matrices, = α β (α and β are m m matrices of full rank r, that is, rank ( ) = r), Λ is an m g matrix of unknown coefficients and w t is an q 1 vector of deterministic and or exogenous variables. The generalised impulse response function, which measures the effect of the shock to the j-th equation in (3) on x t+n can be written as: x j ( n) C e, n = 0, 1, 2, (4) 1/ 2 jj n j where, σ is standard error, Σ is m 1 covariance matrix, and e j is an m 1 selection vector with unity as its j-th element and zeros elsewhere. The generalised forecast error variance decomposition is given by: 1 n ' 2 ii l 0( eicl e j ) ij ( n), i, j = 1,, m n ' ' (5) ( e C C e ) l 0 i l l i Generally, the sum of the generalised forecast error variance m decomposition is not one, that is, ij ( n) 1 as the non-zero covariance j 1 between the original (non-orthogonalised) shocks. The generalised impulse response function (Pesaran & Shin, 1998) is invariant to the ordering of the variables in the VAR model, unlike the orthogonalised impulse response function (Sims, 1980) and thus, it is unique. 7 In this study, bilateral trade balance ((X/M) t ) is expressed by log (X/M) t, where, X t is the bilateral exports (RM millions) and M t is the bilateral imports (RM millions). Bilateral trade balance is expressed 31

10 Risalshah Latif & Zulkarnain Hatta in the ratio, that is, (X/M) t and therefore, it can be expressed in the logarithm, regardless of whether exports are greater or less than imports. Moreover, it is not sensitive to the unit measurement, and also, it could be interpreted as nominal or real trade balance (Shirvani & Wilbratte, 1997: 41). Real exchange rate (RER t ) is expressed by log {ER t (CPI j,t / CPI i,t )}, where, ER t is the exchange rate of country i against Country j, CPI i,t is the Consumer Price Index of Country i (2000 = 100) and CPI j,t is the Consumer Price Index of Country j (2000 = 100). Domestic income (Y t ) and foreign income ( Y t ) are respectively expressed by Industrial Production Index (2000 = 100), except Singapore, which is expressed by Manufacture Production Index (2000 = 100). Industrial Production Index is usually used as a proxy for the measure of income in the literature of bilateral trade balance (Wilson, 1999; Yusoff, 2009). Real domestic money supply (RM millions) (MS t ) and real foreign money supply (RM millions) (MS t ) are expressed by money plus quasi money divided by CPI i,t and CPI j,t, respectively. The values of exports and imports were obtained from Direction of Trade Statistics, the International Monetary Fund (IMF). The rest of the data were obtained from International Financial Statistics, the IMF. The data are quarterly. The sample period is over the period from 1976 (quarter I) to 2008 (quarter IV). All data were seasonal- adjusted, or if not, they are seasonal-adjusted using the Census X12 seasonal adjustment procedure. The Census X12 seasonal adjustment procedure is a procedure used by the United States Census Bureau to adjust seasonality in data. All data were transformed into the natural logarithms before estimation. Figure 1 shows the plots of the natural logarithms of the data. Generally, all the data tended to be nonstationary series and also tended to move in the same direction. EMPIRICAL RESULTS AND DISCUSSION The results of the DF and PP unit root test statistics are reported in Table 2. The lag lengths used to compute the DF unit root test statistics are based on the Schwarz Bayesian Criterion. The lag length used to compute the PP unit root test statistics are based on the Newey-West automatic bandwidth selection, with the maximum lag length set to twelve. Generally, the results of the DF and PP unit root test statistics 32

11 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia show that all the variables are non-stationary in level, but becoming stationary after taking the first differences, except the Japanese real income (no trend), where the DF unit root test statistic shows that there is a unit root, whilst the PP unit root test statistic shows that it is a stationary series. The results of the DF and PP unit root test statistics (trend) also show that there is a unit root in the variable. On the whole, the variable is said to have a unit root in this study. The results of the Johansen s cointegration method are reported in Table 3. The λ Max and λ Trace test statistics are computed with restricted intercepts and no trends. The number of lags is chosen such that the disturbance is white noise. Generally, the λ Max and λ Trace test statistics show that there are two cointegrating vectors for bilateral trade balances of Malaysia with the US and Singapore. For bilateral trade balance of Malaysia with Japan, the λ Max and λ Trace test statistics show that there are at least four cointegrating vectors. Thus, the evidence of two cointegrating vectors for bilateral trade balances of Malaysia with the US, Japan and Singapore cannot be rejected. This study estimates the first cointegrating vector by normalising the bilateral trade balance and the second cointegrating vector by normalising the real exchange rate. The results of the normalised cointegrating vectors are reported in Table 4. The results of the likelihood ratio test statistics, which test that the coefficients of real exchange rate, domestic income, foreign income, real domestic money supply and real foreign money supply are zero, respectively, are reported in Table 5. Generally, the results of the likelihood ratio test statistics are rejected. This implies that real exchange rate, domestic income, foreign income, real domestic money supply and real foreign money supply are important to be included in the estimation. Generally, an increase in the real exchange rate will lead to an increase in the bilateral trade balance. For bilateral trade balances of Malaysia with the US and Singapore, an increase in domestic income will lead to a decrease in bilateral trade balance. Conversely, an increase in foreign income will lead to an increase in bilateral trade balance. For bilateral trade balance of Malaysia with Japan, an increase in domestic income will lead to an increase in the bilateral trade balance whilst an increase in foreign income will lead to a decrease in the bilateral trade balance. 33

12 Risalshah Latif & Zulkarnain Hatta For the real exchange rates of Malaysia with the US, Japan and Singapore, an increase in domestic income will lead to an increase in the real exchange rate, whilst an increase in foreign income will lead to a decrease in the real exchange rate. Nonetheless, the impact of real domestic money supply and real foreign money supply on the real exchange rate are different across countries. For the real exchange rate of Malaysia with the US, an increase in the real domestic money supply or real foreign money supply will lead to an increase in real exchange rate. For the real exchange rate of Malaysia with Japan, an increase in the real domestic money supply will lead to a decrease in the real exchange rate, whilst an increase in real foreign money supply will lead to an increase in the real exchange rate. For real exchange rate of Malaysia with Singapore, an increase in the real domestic money supply or real foreign money supply will lead to a decrease in the real exchange rate. The coefficients of the real exchange rate are all greater than one. Thus, the Marshall-Lerner condition holds. The results of the generalised forecast error variance decompositions for bilateral trade balance and the real exchange rate are reported in Table 6 and Table 7, respectively. The results of the generalised forecast error variance decompositions, which are reported, are based on the 1 5, 10, 15 and 20 horizon periods. The choice of the lag used in the estimation is based on the Schwarz Bayesian Criterion. On the whole, the contribution of bilateral trade balance is mainly from its own innovations. For bilateral trade balances of Malaysia with the US and Japan, the results show that domestic income, real foreign money supply and foreign income are important in explaining the variations in the estimation. The contributions of the real exchange rate and real domestic money supply are relatively small. For the bilateral trade balance of Malaysia with Singapore, the results show that domestic income, foreign income and real domestic money supply are important in explaining variations in the estimation. The contributions of the real exchange rate and the real foreign money supply are relatively small. For real exchange rates of Malaysia with the US, Japan and Singapore, the contribution of real exchange rate is mainly from its own innovations. Moreover, real foreign money supply is important in explaining variations in the estimation. The rest are relatively less important. 34

13 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia The results of the generalised impulse response functions to one standard error shock in the real exchange rate, domestic income or foreign income in the equation for bilateral trade balance are shown in Figure 2. The results of the generalised impulse response functions are plotted over the 20 horizon periods. For bilateral trade balances of Malaysia with the US and Japan, one standard deviation shock of real exchange rate to bilateral trade balance, bilateral trade balance increases over the horizon periods from 0 to about 5 before it dies out. For bilateral trade balance of Malaysia with Singapore, one standard deviation shock of real exchange rate to bilateral trade balance, bilateral trade balance increases and decreases over the horizon periods from 0 to about 5 before it dies out. There is no clear pattern of the responses of bilateral trade balances to the changes of real exchange rates. Generally, the responses of bilateral trade balance to one standard error shock in domestic income, foreign income, real domestic money supply or real foreign money supply are positive and/or negative from 0 to about 5 before it dies out. The finding that bilateral trade balance, real exchange rate, domestic income, foreign income, domestic real money supply and foreign real money supply are cointegrated is consistent with the findings of Baharumshah (2001), Onafowora (2003) and Yusoff (2009), amongst others. Conversely, the finding is contradicted to the finding of Wilson (1991, 2001), amongst others that there is no longrun relationship between bilateral trade balance and its determinants. Moreover, real exchange rates of Malaysia against the US, Japan and Singapore respectively is found to be positive and greater than one, which implies that the Marshall-Lerner condition holds. However, the coefficient of real exchange rate of Malaysia against Japan is found to be statistically insignificant. The coefficients of real exchange rates are greater than one, or is said to be elastic, which implies that devaluation of real exchange rates will lead to an increase in bilateral trade balances in the long run. Ogawa and Kudo (2007), amongst others, report that depreciation of exchange rate will improve deficit in current account. The levels of economic activities are also important for the determination of bilateral trade balances of Malaysia with the US, Japan, and Singapore. Generally, an increase in domestic income will lead to a 35

14 Risalshah Latif & Zulkarnain Hatta decrease in bilateral trade balance, whilst an increase in foreign income will lead to an increase in bilateral trade balance. Qiao (2007: 26), amongst others, proposes to examine the impact of other effects such as wealth effect, investment effect, indirect investment effect on bilateral trade balance. Zhang and Wan (2008) find that the changes in trade balance of China are the result of the real shocks. Domestic income, foreign income, real domestic money supply, and real foreign money supply are found to have an impact on real exchange rate. Thus, they are indirectly having an impact on bilateral trade balance. Lee and Chinn (2006) find that the permanent shocks are important in the real exchange rate determination. Generally, an increase in domestic income will lead to an increase in real exchange rate, whilst an increase in foreign income will lead to a decrease in real exchange rate. Nonetheless, the impact of real domestic money supply and real foreign money supply on real exchange rate are different across countries. In the short run, there is no clear pattern of the J-curve phenomenon. The finding is consistent with the findings of Baharumshah (2001) and Wilson (1991, 2001). They also report no evidence of the J-curve phenomenon for bilateral trade balances of Malaysia with the US and Japan. Conversely, Onafowora (2003) reports that there is the J-curve phenomenon for bilateral trade balances of Malaysia with the US and Japan. Yusoff (2009) also reports that the J-curve phenomenon is found for the bilateral trade balance of Malaysia with the US. This could be the results of different models, sample periods and estimators that are used in the estimation (Bahmani-Oskooee & Ratha, 2004). In the short run, the changes of real foreign money supplies have contributed a lot to the changes of real exchange rates. However, the changes of real exchange rates do not contribute a lot to the changes of bilateral trade balances. Generally, the changes of domestic income and foreign income explain a lot to the changes of bilateral trade balances. On the whole, real money supplies are found to have an impact on real exchange rates and therefore, bilateral trade balances. This study suggests that monetary policy is a potent policy option for Malaysia to enhance its competitiveness in the world market and improve its trade balance. The channel is through influence on the 36

15 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia real exchange rate. However, the impact of monetary policies on real exchange rate is different across countries. A careful consideration of the monetary policy shall be carried out before it is used to influence the trade balance. CONCLUSION This study has investigated the impact of real exchange rates on bilateral trade balances of Malaysia with the US, Japan and Singapore. This study finds that bilateral trade balance, real exchange rate, domestic income, foreign income, real domestic money supply and real foreign money supply are cointegrated. More specifically, there are two cointegrating vectors. Thus, this study normalises the first cointegrating vector by bilateral trade balance and the second cointegrating vector by real exchange rate. Thus, real exchange rate is treated to be endogenous in the estimation. The likelihood ratio test statistics show that real exchange rate, domestic income, foreign income, real domestic money supply and real foreign money supply are important to be included in the estimation. Generally, the results of the normalised cointegrating vectors show that an increase in real exchange rate will lead to an increase in bilateral trade balance. Thus, the Marshall-Lerner condition holds. An increase in domestic income will lead to a decrease in bilateral trade balance whilst an increase in foreign income will lead to an increase in bilateral trade balance. The levels of economic activities are important in the determination of bilateral trade balances. The impact of real domestic money supply and real foreign money supply on real exchange rate are different across countries. In the short run, the changes of real foreign money supplies have contributed a lot to the changes of real exchange rates. Nonetheless, the changes of real exchange rates do not contribute a lot to the changes of bilateral trade balances. On the whole, this study provides some evidence that monetary policy could be a potent policy option to influence trade balance. The channel of monetary policy on trade balance is through real exchange rate. The impact could be different across countries. 37

16 Risalshah Latif & Zulkarnain Hatta ACKNOWLEDGEMENTS The author would like to thank the reviewer of the journal for the comments on an earlier version of this article. All remaining errors are the author s. NOTES 1 Bahmani-Oskooee and Ratha (2004) provide a literature review of the impact of exchange rate on trade balance. Generally, the response of trade balance to currency depreciation does not follow any specific pattern. The results are country specific. 2 There are many plausible reasons for the inconclusive of results such as the different models, sample periods, estimation methods, and measuring variables (Bahmani-Oskooee & Ratha, 2004). 3 The J-curve phenomenon results from the demand for imports and supply of exports being inelastic in the short run. Thus, depreciation or devaluation of exchange rate, the quantity of imports demanded and the quantity of exports supplied do not change much. As a result, trade balance deficit, as valued in the domestic currency widens (Daniels & Vanhoose, 2005: 279). 4 The model proposed by Rose and Yellen (1989) is usually used in the literature of the impact of exchange rate on trade balance. The model is specified as a function of real exchange rate, domestic income, and foreign income (Wilson, 1999, 2001; Baharumshah, 2001; Yusoff, 2009). However, the model does not examine the impact of money supply on trade balance. See Wilson (1999) for the derivation and discussion of the model. 5 The impact of exchange rate on trade balance is ambiguous. Trade balance is said to be affected by real exchange rate rather than nominal exchange rate as trade balance depends on prices of foreign goods relative to domestic goods. Thus, nominal exchange rate has an impact on trade balance if it can influence real exchange rate (Yusoff, 2009). 6 An increase in money supply in domestic country would cause domestic residents decrease their cash balanced by buying more imports and reducing trade balance. However, an increase in money supply could also depreciate real exchange rate and therefore, it is expected to improve trade balance. Thus, the impact of money supply on trade balance is uncertain (Buluswar, Thompson & Upadhyaya, 1996: 430). 7 In the analyse of the generalised impulse response function, a shock of typical size to real exchange rate on bilateral trade balance, which will lead to an increase in bilateral trade balance can be interpreted that the Marshall-Lerner condition holds. REFERENCES Baharumshah, A. Z The Effect of Exchange Rate on Bilateral Trade Balance: New Evidence from Malaysia and Thailand. Asian Economic Journal. 15(3): Bahmani-Oskooee, M., & Ratha, A The J-Curve: A Literature Review. Applied Economics. 36: Bergin, P. R How Well can the New Open Economy Macroeconomics Explain the Exchange Rate and Current Account? Journal of International Money and Finance 25: Buluswar, M. D., Thompson, H., & Upadhyaya, K. P Devaluation and the Trade Balance in India: Stationary and Cointegration. Applied Economics. 28:

17 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia Daniels, J. P., & VanHoose, D International Monetary and Financial Economics (3rd ed.). Mason, Ohio, USA: Thomson South-Western. Devereux, M. B., & Genberg, H Currency Appreciation and Current Account Adjustment. Journal of International Money and Finance. 26(4): Dickey, D. A., & Fuller, W. A Distribution of the Estimators for Autoregressive Time Series with a Unit Root. Journal of the American Statistical Association. 74: Johansen, S Statistical Analysis of Cointegration Vectors. Journal of Economic Dynamics and Control. 12: Koop, G., Pesaran, M. H., & Potter, S. M Impulse Response Analysis in Nonlinear Multivariate Models. Journal of Econometrics. 74: Lee, J., & Chinn, M. D Current Account and Real Exchange Rate Dynamics in the G7 Countries. Journal of International Money and Finance. 25: MacKinnon, J. G Numerical Distribution Functions for Unit Root and Cointegration Tests. Journal of Applied Econometrics. 11: Ogawa, E., & Kudo, T Asymmetric Responses of East Asian Currencies to the US Dollar Depreciation for Reducing the US Current Account Deficits. Journal of Asian Economics. 18: Onafowora, O Exchange Rate and Trade Balance in East Asia: Is there a J-Curve? Economics Bulletin. 5(18):1 13. Pesaran, H., & Shin, Y Generalised Impulse Response Analysis in Linear Multivariate Models. Economics Letters. 58: Pesaran, M. H., Shin, Y., & Smith, R. J Structural Analysis of Vector Error Correction Models with Exogenous I(1) Variables. Journal of Econometrics. 97: Phillips, P. C. B., & Perron, P Testing for a Unit Root in Time Series Regression. Biometrika. 75: Qiao, H Exchange Rates and Trade Balances under the Dollar Standard. Journal of Policy Modeling 29. (5): Rose, A. K., & Yellen, J. L Is there a J-Curve? Journal of Monetary Economics. 24: Shirvani, H., & Wilbratte, B The Relationship between the Real Exchange Rate and the Trade Balance: An Empirical Reassessment. International Economic Journal. 11(1): Sims, C Macroeconomics and Reality. Econometrica. 48:1 48. Wilson, P Exchange Rates and the Trade Balance: The Case of Malaysia from 1970 to Malaysian Journal of Economic Studies. 36(2):1 15. Wilson, P Exchange Rates and Trade Balance for Dynamic Asian Economies - Does the J- Curve Exist for Singapore, Malaysia and Korea? Open Economies Review. 12: Yusoff, M Bilateral Trade Balance, Exchange Rates, and Income: Evidence from Malaysia. Global Economy Journal. 9(4), Article 7. Zhang, Y., & Wan, G Correcting China s Trade Imbalance: Monetary Means will not Suffice. Journal of Policy Modeling. 30(3):

18 Risalshah Latif & Zulkarnain Hatta TABLE 1 Percentage of Malaysian overall trade Year US Japan Singapore Total % 11.3% 17.1% 48.5% % 9.3% 15.6% 44.7% % 10.8% 14.7% 38.0% Source: Direction of Trade Statistics, IMF. Ta b l e 2 The results of the Dickey and Fuller (1979) (DF) and Phillips and Perron (1988) (PP) Unit Root Test Statistics DF - No Trend PP - No trend DF - Trend PP - Trend log (X/M) us,t (1) (5) (1) (5) log (X/M) us,t (0) (7) (0) (8) log (X/M) j,t (3) (2) (3) (0) log (X/M) j,t (2) (4) (2) (3) log (X/M) s,t (0) (7) (0) (5) log (X/M) s,t (0) (6) (0) (6) log RER us,t (1) (0) (1) (12) log RER us,t (0) (8) (0) (12) log RER j,t (1) (4) (1) (4) log RER j,t (0) (4) (0) (1) log RER s,t (1) (4) (1) (2) log RER s,t (0) (12) (0) (12) log Y us,t (2) (5) (2) (5) log Y us,t (1) (5) (1) (5) log Y j,t (1) (2) (1) (2) log Y j,t (0) (5) (0) (5) log Y s,t (3) (2) (2) (1) log Y s,t (2) (4) (2) (5) log Y m,t (0) (2) (0) (2) log Y m,t (0) (4) (0) (5) log MS us,t (1) (4) (1) (4) log MS us,t (0) (6) (0) (6) log MS j,t (1) (5) (1) (5) log MS j,t (0) (0) (0) (1) log MS s,t (1) (7) (1) (7) log MS s,t (0) (5) (0) (5) log MS m,t (1) (3) (1) (3) log MS m,t (0) (6) (0) (6) Notes: (X/M) i,t denotes bilateral trade balance of Malaysia with i (i = the US, Japan and Singapore). RER i,t denotes real exchange rate of Malaysia against i. Y j,t denotes income of j (j = the US, Japan, Singapore and Malaysia). MS j,t denotes real money supply of j. The subscripts us, j, s and m denote the US, Japan, Singapore and Malaysia, respectively. No Trend denotes the DF or PP t-statistic is estimated based on the model including an intercept. Trend denotes the DF or PP t-statistic is estimated based on the model including an intercept and a time trend. Values in parentheses are the lag length used in the estimation of the DF or PP unit root test statistic. Critical values can be obtained from MacKinnon (1996). () denotes significance at the 1% (5%) level. 40

19 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia Ta b l e 3 The results of the Johansen (1988) Likelihood Ratio Test Statistics λ Max Test Statistic H 0 : r=0 r<=1 r<=2 r<=3 r<=4 r<=5 H a : r=1 r=2 r=3 r=4 r=5 r=6 US Japan Singapore c.v. (5%) c.v. (10%) λ Trace Test Statistic H 0 : r=0 r<=1 r<=2 r<=3 r<=4 r<=5 H a : r 1 r 2 r 3 r 4 r 5 r 6 US Japan Singapore c.v. (5%) c.v. (10%) Notes: The VAR = 6, VAR = 12 and VAR = 6 are used in the estimation of the likelihood ratio test statistics for the US, Japan and Singapore, respectively. c.v. denotes critical value. () denotes significance at the 5% (10%) level. Ta b l e 4 The results of the normalised cointegrating vectors US log (X/M) t = log RER t log Y t log Y t log RER t = log Y t log Y t log MS t log MS t Japan log (X/M) t = log RER t log Y t log Y t log RER t = log Y t log Y t log MS t log MS t Singapore log (X/M) t = log RER t log Y t log Y t log RER t = log Y t log Y t log MS t log MS t Note: The VAR = 6, VAR = 12 and VAR = 6 are used in the estimation of the normalised cointegrating vectors for the US, Japan and Singapore, respectively. Ta b l e 5 The Likelihood Ratio Test Statistics. LR1 LR2 LR3 LR4 LR5 US Japan Singapore Notes: LR1, LR2, LR3, LR4, and LR5 test that the coefficients of real exchange rate, domestic income, foreign income, real domestic money supply and real foreign money supply are zero, respectively. () denotes significance at the 1% (5%) level. 41

20 Risalshah Latif & Zulkarnain Hatta Ta b l e 6 The generalised forecast error variance decompositions bilateral trade balances Malaysia US Horizon log (X/M) t log RER t log Y t log MS t log MS t log MS t Malaysia Japan Horizon log (X/M) t log RER t log Y t log Y t log MS t log MS t Malaysia Singapore Horizon log (X/M) t log RER t log Y t logy t log MS t log MS t Note: The VAR = 1 is used in all the estimations. 42

21 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia Ta b l e 7 The generalised forecast error variance decompositions real exchange rates Malaysia US Horizon log (X/M) t log RER t log Y t log Y t log MS t log MS t Malaysia Japan Horizon log (X/M) t log RER t log Y t log Y t log MS t log MS t Malaysia Singapore Horizon log (X/M) t log RER t log Y t log Y t log MS t log MS t Note: The VAR = 1 is used in all the estimations. 43

22 Risalshah Latif & Zulkarnain Hatta US Japan BTB RER Y Y MS MS BTB RER Y Y MS MS Singapore BTB RER Y Y MS MS Fi g u r e 1 Plots of the Natural Logarithms of Bilateral Trade Balance (log (X/M) t ), Real Exchange Rate (log RER t ), Domestic Income (log Y t ), Foreign Income (log Y t ), Real Domestic Money Supply (log MS t ) and Real Foreign Money Supply (log MS t ) Note: BTB = log (X/M) t, RER = log RER t, Y = log Y t, Y = log Y t, MS = log MS t and MS = log MS t. 44

23 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia US 45

24 Risalshah Latif & Zulkarnain Hatta Japan 46

25 Real Exchange Rates and Bilateral Trade Balances: Some Empirical Evidence of Malaysia Figure 2 Japan 47