Is the liberalization policy effective on improving bivariate cointegration of current accounts, foreign exchange, stock prices? Further evidence from Asian markets

This paper fist examines three set of bivariate cointegrations between any two of current accounts, stock markets, and currency exchange markets in 10 Asian countries. Furthermore, this work examined the effect of country characters on this bivariate cointegration. Our findings suggest that for three sets of cointegration test, each sample country at least exists one cointegration. India consistently exhibited a bi-directional causal relationship between any two of three indicators. Unlike Pan et al. (Int Rev Econ Financ 16:503–520, 2007 and Phylaktis and Ravazzolo (J Int Money Financ 24:1031–1053, 2005), this study found that such cointegration is influenced by three characteristics: capital control; flexibility in foreign exchange rates; and the ratio of trade to GDP. These characteristics are the result of liberalization in each Asian country. This implies that liberalization policies are effective on improving the cointegration between any two of financial markets and current account for 10 Asian countries.


Introduction
In Asian countries, a series of liberalization policies have been launched which contain relaxation of foreign capital controls and flexible exchange rate regimes Since 1990s. These polices could induce variation of foreign exchange market and stock markets, and bring the risk of international investment through portfolio diversification. These influences have increased the interest of academics and practitioners in studying the interrelation between two financial markets. Recent numerous studies propose empirical evidence of a linkage between exchange rate and stock price for Asian economies (Phylaktis and Ravazzolo 2005;Pan et al. 2007;Economic Indicator Financial Market Fig. 1 The framework of interrelationship among stock market, foreign exchange market, and current account Tai 2007;Lee and Yoon 2007). Phylaktis and Ravazzolo (2005) found that the degree of foreign exchange restrictions is not an important determinant of the relation between stock and foreign exchange market. This makes the effect of liberalization policy on two markets not to be detected easily. In other words, if openness degree of foreign exchange significantly influenced the relation between the two markets, it could be proven that liberalization policies had been effective, and vice versa. For example, when a country with greater flexibility in foreign exchange policy seeks to strengthen the linkage between the two markets, then liberalization policies would be more effective, and vice versa. Phylaktis and Ravazzolo (2005) findings might be the omission of an important variable which serves as a conduit between markets. What a economic variable is appropriate for entering into the two financial market?
In previous studies, current account interacts with stock price as well as exchange rate closely. Recently global current account imbalance which mostly stems from large deficit in U.S. and surplus in Asian economics has attracted substantial attention among academics and policy makers. Following such tendency, the issues about current account have become an important subject of recent literatures. Numerous studies proposed evidences on a linkage between current account and exchange rate (Baharumshah and Masih 2005;Miller 2005;Freund 2005, Bergin andSheffrin 2006;Devereux and Genberg 2007). A number of researches compared stock price with exchange rate shocks to explore which is more important for current account adjustment (Fratzscher et al. 2010). Previous studies analyzed theoretical connection of security prices and current account (Caballero et al. 2006a,b;Kitamura 2009). However, there is yet no systematic evidence on the three sets of relationship between any two of current account, stock price, and exchange rate. Hence, this paper fist examines three sets of bivariate cointegration between any two of three economic variables. Figure 1 diplays the bivariate frameworks including the relationships among exchange rate market, stock market, current account.
Moreover, each Asian economy's liberalization policy would bring about three country characters, i.e., degree of capital restrictions, foreign exchange flexibility, trade ratio to GDP. Based on the goods market theory, for a country with high trade ratio and high exchange rate exposure, it has a strong bi-directional relation between stock price and exchange rate. The portfolio balance approach indicated that, a country with freely floating exchange rate and less capital control, exhibits a more significant bi-directional relation between stock price and exchange rate. However, recent studies could not completely obtain evidences in favor of theoretical prediction (Phylaktis and Ravazzolo 2005;Pan et al. 2007). Previous studies supported that among three characteristics, two (i.e., degree of capital control, and the ratio of trade to GDP) have influence on the relationship of foreign exchange and stock markets. Phylaktis and Ravazzolo (2005) propose openness in exports and imports to explain a linkage between foreign exchange market and stock market, yet fail to examine this characteristic (i.e., trade ratio). Similar to Phylaktis and Ravazzolo (2005), Pan et al. (2007) suggested that capital control could weaken this cointegration, but there is no examination in this characteristic (i.e., degree of capital control). As the early mentioned, if three characteristics significantly influenced the relation between the two financial markets, it could be proven that liberalization policies were effective, and vice versa. Thus, to detect the effect of liberalization policy for Asian economies, secondly, we examined the effect of country characters on the three sets of bivariate cointegration. That is, we attempt to answer a question, "do various country characters arising from liberalization policies have different cointegration relationship between any two of exchange rate, stock price, and current account?". If the answer is positive, this implies that liberalization policies are effective on improving the relationship between financial markets and current account for 10 Asian countries. This paper begins by extending the proposition of Phylaktis and Ravazzolo (2005) that financial integration is closely related to economic integration. Because current accounts signify the volume of external trade with other countries and measures the degree to which the two countries are integrated, this paper introduce current accounts to the relationship between stock markets and foreign exchange markets, and examine three kinds of relationship. We proposed a bivariate framework connecting financial integration with economic integration. Our findings present that first, for three cases concerning the relationship between any two of three indicators, except that Korea and the Philippines have two cointegrations consistently, and other eight countries at least have one cointegration. For the three cases, India consistently exhibited a bi-directional causal relationship between any two of three indicators.
Second, unlike assertion of Phylaktis and Ravazzolo (2005) and Pan et al. (2007), showing that whether exchange rate and stock prices are cointegrated or not was not influenced by foreign exchange regimes. Our finding implies that three country characteristics (i.e., the degree to which capital is controlled, flexibility in foreign exchange mechanisms, the ratio of trade to GDP) indeed influence these cointegrations. With regard to the direction of influence, our findings show that a relaxation of restrictions on exchange rates and capital control enhance the cointegration in favor of the portfolio balance approach. Moreover, we found that a stronger relationship between current account and exchange rate occurs in the countries with a higher ratio of trade, being closely consistent with the proposition of the good market theory. This implies that liberalization policies are effective on improving the relationship between any two of financial markets and current account for 10 Asian countries.

The relationship of exchange rate and stock price
With regard to a relationship between stock price and exchange rate, two theories are as follows. One is the goods market theory (Dornbusch and Fischer 1980) proposes that a change in exchange rate affect international competitiveness, trade balance and output of a country, thereby influencing firms' cash flow and stock prices. Other is the portfolio balance approach suggests that, exchange rates are determined by stock market mechanisms (Branson 1983;Frankel 1983). 1 Numerous works have provided evidences which have been quite mixed for the sign and causal direction. 2 Several empirical studies present evidence on Asian countries in favor of a causal relation between exchange rate and stock price (Wu 2000;Ramasamy and Yeung 2002;Chiang and Yang 2003;Phylaktis and Ravazzolo 2005;Pan et al. 2007;Tai 2007;Lee and Yoon 2007;Yau and Nieh 2009;Zhao 2010). 3 2.2 The relationship of current account and exchange rate Bergin and Sheffrin (2000) proposed a present value model of current account incorporated with exchange rate to improve ability of forecasting in current account. Their model provides evidence that a causal relationship from exchange rate to current account, which is supported by recent studies (Iscan 2003;Gruber 2004). On the contrary, monetary model augmented with current account provides theoretical basis of a causal relationship from current account to exchange rate. Prior empirical studies 4 show supportive evidences on this model. Along this line, numerous evidences on recent studies that exchange rates of Asian economies cointegrate with current account were found continuously (Husted and MacDonald 1999;Chinn 2000;Baharumshah and Masih 2005).

The relationship of current account and stock price
Previous studies continuously verify a cause relationship from current account to stock price. Mercereau (2003a,b) proposed a notion that current account can predict performance of stock market. Recently, numerous studies support a linkage between current account and stock price. (Caballero et al. 2006a,b;Kraay and Ventura 2005;Kitamura 2009;Fratzscher et al. 2010). In contrast, a cause relationship from stock price to current account could be stated from three theoretical aspects: the wealth effect, uncertainty, a leading indicator effect. Hall (1978) illustrated the wealth effect, indicating that stock price adjustment would alter permanent income and consumption of people in real world, based on permanent income hypothesis. Romer (1990) stated that investor uncertainty resulting from deterioration in stock market would reduce consumption on durable goods, and decrease balance of current account. With regard to a leading indicator effect, Poterba and Samwick (1995) indicated that according to permanent income theory, stock price could predict future income, and further lead to variation in current account.
2.4 Influence of country characteristics on relationship between any two of stock prices, exchange rates, and current accounts A number of theories and previous studies indicated influence of three country characteristics for each Asian country (i.e., flexibility in foreign exchange regimes, degree to capital control, and the ratio of trade to GDP) on the relationship between any two of three economic indicators.
A portfolio balance approach indicates that in countries without a free floating exchange rate, fluctuations in exchange rates are not necessarily met with a corresponding movement in stock prices. This suggests that a free exchange rate regime could enhance the relationship between stock prices and exchange rates. However, recent studies tend to not support this theoretical prediction. Phylaktis and Ravazzolo (2005) found that a relaxation in restrictions to foreign exchange was not a determinant of a link between foreign exchange and domestic stock markets. Hence, they stated free trade of import and export in five Asian countries as a possible explanation contributing to this linkage. Their proposition is in accordance with a goods market theory, which indicates a stronger bi-directional causal relationship between exchange rates and stock prices for countries with higher levels of trade (Pan et al. 2007, p. 513). Similar to Phylaktis and Ravazzolo (2005) and Pan et al. (2007) proposed evidence on whether exchange rate and stock prices are cointegrated or not was not influenced by foreign exchange arrangements. They yet suggested that capital control could weaken this cointegration. In sum, empirical researches supported that among three characteristics, two (i.e., degree to capital control, and the ratio of trade to GDP) have influence on the relationship of foreign exchange and stock markets.
Recent study obtained evidence that increasing the ratio of trade would improve the relationship of exchange rate and current account. Miller (2005) found that implementing policy to increase trade alters exchange rates, thereby helping to improve an imbalance in current accounts. We proceed to discuss the influence of three characteristics on the relationship of current account and stock price. According to wealth effect of Hall (1978), in countries with a free exchange rate and less control of capital, a rise in stock prices facilitates lending to international financial markets, which smoothes the consumption in those countries, causing a shift in current accounts. In a country with a high trade to GDP ratio, any change in current accounts increases import and export trade, further stimulating cash flow and bolstering stock prices.

Research design
3.1 Data description Table 1 shows data series which this paper utilized for 10 sample Asian countries, Taiwan, Hong Kong, Japan, Singapore, Malaysia, Thailand, Philippines, Korea, Indian, and Indonesia. Sample period spans from January, 1970 to July, 2010. Except that Hong Kong, Japan, Singapore are developed countries, other countries are developing ones which often are regarded as emerging markets. Data series were extracted from AREMOS Statistical Data Bank of Taiwan's Ministry of Education. This study adopts 4,625 monthly observations of stock price Note: the data that this paper utilized for 10 Asian countries during sample period from January, 1970 to July, 2010.were presented in Table 1. This study divided the data into two subsets: one is developing countries including Taiwan, Malaysia, Thailand, Philippines, Korea, Indian, and Indonesia. Other is developed countries consisting of Hong Kong, Japan, Singapore. The 4,625 monthly observations of individual country indices, real exchange rate as local currency per U.S. dollar, and 1,301 quarterly observations of current account, were selected from AREMOS database. To compare the impact of 1997 Asian financial crisis, this paper divided all data into two subsample periods, pre-1997 and post-1997 indices and real exchange rate as local currency per U.S. dollar, and 1915 quarterly ones of current accounts. Liberalization policy which Asian economies have launched since 1990s would reflect three country characters, i.e., degree of capital control restrictions, foreign exchange flexibility, and trade ratio to GDP. Hence, Table 2 reports three country characters for each Asian economy, which contain exchange rate arrangement, capital mobility controls and international trade size. Developing countries except Korea, exhibit managed floating foreign exchange and moderate (or strong) capital control. Except Malaysia, these developing ones display low ratio of trade to GDP consistently. Developed countries, Hong Kong, Japan and Singapore, show no capital control, and they present high ratio of international trade except Japan. Note: The ratio of trade to GDP is measured by the ratio of exports and imports divided by GDP. The data were collected from AREMOS and Global Financial Database. The exchange rate regimes and capital control are from World Currency Yearbook. "H" and "L" denotes high and low ratio of trade respectively. Moderate and strong capital control is attributed to "H", and none capital control is denoted by "L". The ratio of trade to GDP above two belongs to high degree, otherwise is low degree Figure 2 shows the series trends of current account, stock price indices, and exchange rates for 10 Asian economies. This study's sample periods from 1970 to 2010 include 1997 Asian financial crisis and 2007 subprime mortgage financial storm, and hence can capture a full impact before and after two events on three indicators. Most countries show that after 1997, three variables series exhibit an up-and-down pattern dramatically. The drops in stock price indices accompany depreciation in currency, and current account variation, which are consistent with theoretical expectation of the portfolio balance approach.

Unit root tests of augmented Dickey-Fuller and Phillips-Perron
In order to avoid spurious regression problem of Granger and Newbold (1974) from non-stationary variables, we implemented the Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit root tests to examine current account, exchange rate and stock price index. The optimal lag length in the ADF regressions was determined by Akaike Information Criterion and Schwartz Bayesian Criterion. Table 3 shows the result of unit root tests, revealing that for each country, at least two of three variables exhibit I (0), i.e., non-stationary. This suggests a possibility of cointegration between any two of three economic indicators.

Cointegration methodology
This paper aims to examine bi-directional cointegration relationship between any two of stock market index prices, exchange rates and current account series. In this study, cointegration method was employed to explore, which was proposed by Johansen (1988) and Johansen and Juselius (1990) including maximum likelihood ratio test and trace test.
where CA t denotes current account, P t denotes stock market index prices and E t denotes real exchange rate for 10 Asian economies. Null hypothesis of trace statistics  P −2.621 −6.283 * * * −2.803 −6.331 * * * I(1) * , * * , * * * denote the null hypothesis that unit-root exists is rejected at 10, 5, 1% statistical significance level. CA, E, P denote current account, exchange rate, and stock price indices respectively. The ADF and PP unit root tests are under the null hypothesis of unit root (H0: unit root) that its critical value is decided on the critical value table trace=−T n i=r+1 ln(1 − λ i ) presents that there are at most r cointegration vectors 0 ≤ r ≤ n and (n-r) common stochastic trend. If Y t is cointegrated, a vector error correction model (VECM) can be generated by: where μ is 2 × 1 vector of drift, π are 2 × 2 matrices of parameters, π Y t−1 = α βYt−1 is error correction term, and e t is 2 × 1 white noise vector.

Empirical result
4.1 Cointegration between any two of current account, stock price, and exchange rate Table 4 provides the results of the Johansen-Juselius cointegration test for 10 Asian countries. Each has a long-run cointegration relationship between exchange rate and stock price for seven countries except two cointegrations which each has for Korea, the Philippines, and Thailand. Specifically, conforming to Yau and Nieh (2009), showing that Taiwan has one cointegration between stock price and exchange rate (NTD/USD). For the case of current account and exchange rate, Korea, Malaysia, Indonesia, the Philippines have two conitegration relationships. Other countries have one for each. For Taiwan, Korea, Indonesia, the Philippines, current account and stock price of each country are cointegrated with two vectors, while other six countries have one cointegrating for each. In sum, for three cases, except that Korea and the Philippines have two cointegrations consistently, other sample countries at least have one cointegration. Table 5 reports that the results of exclusion restrictions and weak exogeneity tests. We found that a cointegration exists between stock prices and exchange rates. In Table 5, Singapore, Korea, India, the Philippines, and Thailand show that the exclusion hypotheses (β E = 0 and β P = 0) were rejected at 1, 5, and 10% significance level, respectively. This implies that exchange rates and stock prices were cointegrated over the long-term. In Table 6, these countries' coefficients of stock price in cointegration vector presented significant at 1, 5, 10% level. Singapore, India, the Philippines, and Thailand, their two hypotheses of weak exogeneity (α E = 0 and α P = 0) were rejected, suggesting that exchange rates and stock prices have a bi-directional causal relationship adjusted toward long-term equilibrium. This finding is consistent with the prediction of the good market theory and portfolio balance approach, and supports previous results concerning a connection between exchange rates and stock prices. Table 6 shows that the largest cointegration coefficient of stock price was found in Japan at a 1% significance level.
In Table 5, for Hong Kong, Japan, Malaysia, and the Philippines, two exclusion hypotheses (β ca = 0 and β E = 0) were rejected at 1, 5, and 10% significance level, respectively. This implies that current accounts and exchange rates were cointegrated. In Table 6, these countries' coefficients of exchange rate in cointegration vector presented significant at 1, 5, 10% level. For Taiwan, Korea, Malaysia, India, and Thailand, their two hypotheses of weak exogeneity (α ca = 0 and α E = 0) were rejected, revealing that current accounts and exchange rates had bi-directional causal relationships following equilibrium adjustment over the long-term. Malaysia is only one country which has both cointegration and bi-directional cause relationship. This finding upholds the monetary model and augmented present value model of current accounts, and provides evidence to support the previous results in which exchange rates and current accounts in Asian economies are cointegrated. In Table 6, the largest cointegration coefficient of exchange rates (3,221,556) occurred in Thailand at 1% significance level, implying that a 1% change in exchange rates would result in 32215 changes in the multiplier in current accounts.
Further evidence was found in current accounts and stock prices. In Table 5, Singapore, India and Philippines demonstrated that the exclusion hypotheses of β ca = 0 and β P = 0 were rejected at 1, 5, and 10% significance level, respectively. This implies that current Note: The optimal lengths of lags reported are chosen by Schwartz information, Akaike information, LR statistics criterion. Bivariate Vector Autoregression (BVAR) is estimated through residual auto-correlation test (Q-test), residual normality test (Jarque-Bera test) Trace statistics indicates that either developed countries or developing countries in 10 Asian countries, at least have one cointegrating vectors at 1% significance level accounts and stock prices are cointegrated over the long-term. In Table 6, these countries' coefficients of stock price in cointegration vector presented significant at 1, 5, 10% level. For Korea, Malaysia, India, and Thailand, their two hypotheses of weak exogeneity (α ca = 0 and α P = 0) were rejected, suggesting that current accounts and stock prices had a bi-directional causal relationship adjusted toward long run equilibrium. India is only one country which has both cointegration and bi-directional cause relationship. This result reconciles with the prediction of the background theory (i.e., wealth effect, uncertainty, and leading indicator effect), and supports previous empirical results concerning a connection between current Developed countries Developing countries (emerging market)   Note: If LR statistics is larger than χ 2 (1) statistics with one degree of freedom, then it implies that the hypothesis (H 0 ) is rejected. * * * , * * , * denote statistic significance at 1, 5, 10% level respectively accounts and stock prices. In Table 6, the largest cointegration coefficient of stock prices (−6588) was found in Thailand at a 1% significance level.

4.2
The effect of country characteristics on cointegration between any two of stock prices, exchange rates, and current accounts

The effect of relaxation of foreign exchange and capital control on the cointegration
Since the 1990s, the liberalization policy which Asian economies launched has presented three economic characteristics unique to each country, i.e., the degree to capital control, Note: When a country has the cointegration between exchange rate and current account, the coefficients a 1 of exchange rate in E t = a 0 +a 1 P t +ε t present statistic significance at 1, 5, 10% level. When a country has the cointegration between stock price and exchange rate, the coefficients a 1 of exchange rate in C A t = a 0 +a 1 E t +ε t exhibit statistic significance at 1, 5, 10% level. If there exists the cointegration between stock price and current account, then the coefficients a 1 of stock price in C A t = a 0 + a 1 P t + ε t show statistic significance at 1, 5, 10% level flexibility in foreign exchange mechanisms, and the ratio of trade to GDP. Our findings support the contention that relaxization of foreign exchange rates and capital control influences the cointegration between any two of the three variables (i.e., current account, stock price, exchange rate). Ten Asian countries, with various foreign exchange regimes and varying degrees of capital control had different cointegration relationships. Panel A in Table 4 shows that, for the cointegration between exchange rate and stock price, Taiwan, Malaysia, India, and Indonesia had more tightly constrained exchange rate regimes and more capital control, and exhibited one cointegration while Korea had free exchange rate regime and far less capital control, exhibiting two cointegration relationships. Panel B in Table 4 shows that, for the cointegration between current account and exchange rate, Taiwan, India, and Thailand had more tightly constrained exchange rate regimes and more capital control, and exhibited one cointegration while Singapore,Korea had free exchange rate regime and far less capital control, exhibiting two cointegration relationships. Panel C in Table 4 shows that, for the cointegration between current account and stock price, Malaysia, India, and Thailand had more tightly constrained exchange rate regimes and more capital control, and exhibited one cointegration while Korea had free exchange rate regime and far less capital control, exhibiting two cointegration relationships. These findings uphold the portfolio balance approach showing that a relaxation of restrictions on exchange rates can enhance the relationship between stock prices and exchange rates. Table 4 show that the 10 economies, with various proportions of international trade, differed in the cointegration between any two of three economic indicators, supporting the contention that the ratio of international trade influences the cointegration. In Table 4 panel A, for the cointegration between exchange rate and stock price, Korea, the Philippines and Thailand with a low ratio of trade had two cointegrations while three economies (Hong Kong, Malaysia, and Singapore) with high ratios of trade, had one cointegration relationships. This result is contrary to the goods market theory, showing a stronger causal relationship between exchange rates and stock prices in countries with a higher proportion of trade. In Table 4 panel B, for the cointegration between current account and exchange rate, four countries (Japan, Taiwan, India and Thailand) with a low ratio of trade had one cointegration while Malaysia and Singapore with high ratios of trade, had two cointegration relationships. This result closely advocates the assertion of goods market theory that a high share of share can enhance a linkage between financial markets. In Table 4 panel C, for the cointegration between current account and stock price, four countries (Taiwan, Korea, Indonesia and the Philippines) with a low ratio of trade had two cointegrations while Hong Kong, Malaysia and Singapore with high ratios of trade, had one cointegration relationships. This result is closely opposite to the contention of goods market theory that a high share of share can enhance a linkage between financial markets.

Conclusions
Unlike previous studies exploring relationship between any two of current accounts, stock prices and exchange rates respectively, this article examines the cointegration between any two of the three economic indicators simultaneously. We employ the cointegration methodology proposed by Johansen (1988) and Johansen and Juselius (1990) to explore 10 Asian countries over the period from January 1970 to July 2010. Our findings reveal that first, for three cases concerning the relationship between any two of three indicators, except that Korea and the Philippines have two cointegrations consistently, and other eight countries at least have one cointegration. For the three cases, India consistently exhibited a bi-directional causal relationship between any two of three indicators. Second, sample countries which have different foreign exchange regimes and the degree to capital control, and ratio of trade to GDP, had different cointegration relationships. Our finding presents an implication that three country characteristics (i.e., the degree to which capital is controlled, flexibility in foreign exchange mechanisms, the ratio of trade to GDP) indeed influence these cointegrations. With regard to the direction of influence, our findings show that a relaxation of restrictions on foreign exchange rates and capital control enhance the cointegration between any two of the three economic indicators in favor of the portfolio balance approach. Moreover, we found that a stronger relationship between current account and exchange rate occurs in the countries with a higher ratio of trade, being closely consistent with the proposition of the good market theory. These results differ from assertion of Phylaktis and Ravazzolo (2005) that degree to which foreign exchange is restricted is not necessarily a condition for a linkage between foreign exchange and stock market. Our finding also differs from the contention from Pan et al. (2007) that exchange rate mechanisms do not influence whether stock prices and exchange rates are cointegrated (see Pan et al. 2007, p. 512). To sum up, by introducing current accounts to a linkage between the two financial markets, this paper constructs a bivariate VECM framework to provide evidence of cointegration relationship between any two of three economic indicators (i.e., current account, stock markets, and foreign exchange markets) in Asian countries, in favor of the proposition that liberalization policy have impact on the cointegration of current account and two financial markets in Asian economies.
Further research might accumulate numerous and reliable evidence on the studies about interrelationship between any two of stock prices, exchange rates, and current accounts by using cointegration methods. This study verifies the cointegration of seven Asian developing countries, yet little evidence occurs in developed countries. Hence, next question "what about developed or industrial countries are?" would be leaved for future research. We recommend that analysis of cointegration could be implemented by employing alternative data from multi-countries and a variety of securities in Europe and America.