ASIAN ECONOMIC INTEGRATION AND STOCK MARKET COMOVEMENT. Abstract

Transcription

1 The Journal of Financial Research Vol. XXV, No. 1 Pages Spring 2002 ASIAN ECONOMIC INTEGRATION AND STOCK MARKET COMOVEMENT Robert Johnson University of San Diego Luc Soenen California Polytechnic University, San Luis Obispo Abstract Using daily returns from 1988 to 1998, we investigate to what degree twelve equity markets in Asia are integrated with Japan s equity market and examine the factors that affect the level of economic integration. We find that the equity markets of Australia, China, Hong Kong, Malaysia, New Zealand, and Singapore are highly integrated with the stock market in Japan. There is also evidence that these Asian markets become more integrated over time, especially since A higher import share as well as a greater differential in inflation rates, real interest rates, and gross domestic product growth rates have negative effects on stock market comovements between country pairs. Conversely, increased export share by Asian economies to Japan and greater foreign direct investment from Japan to other Asian economies contribute to greater comovement. JEL Classifications: F36, G15, F02, F15 I. Introduction Capital market integration provides the opportunity for better diversification as investors shift to higher risk and expected return projects because they are able to diversify their overall risk (Obtsfeld 1994). International market integration has been the subject of considerable empirical investigation. Because expected returns and variances are required to construct optimal risk and return portfolios, investors, portfolio managers, and financial market regulators can benefit from new insights into the comovements among international equity markets. We first examine to what degree equity markets in Asia are integrated with Japan s equity market. The relatively large size of the Japanese economy and The authors wish to thank William T. Moore, editor of the JFR, Paul D. Koch, and an anonymous referee for many helpful comments that substantially improved this article. 141

2 142 The Journal of Financial Research the comparatively close proximity of the other Asian equity markets suggest a high degree of integration. On the other hand, the rapid economic growth of the emerging economies in the region combined with a prolonged economic slump in Japan could have resulted in divergent behavior. In addition, the emerging markets of Asia seem to be looking more to the west and less to Japan for trade. With the exception of China, the share of Japan s imports coming from other Asian economies has declined over the years, whereas the share of Japan s exports to these countries continues to rise. From a trade perspective, the trend for the Asian economies may be less, rather than more, economic integration. We next examine the extent to which macroeconomic variables that are usually associated with economic integration explain the changes in the degree of stock market integration. As the degree of economic integration varies over time for a given pair of countries, we expect the extent of equity market integration to vary systematically. The financial literature offers much research on the risk reduction benefits of international diversification (e.g., Grubel 1968; Agmon and Lessard 1977). Harvey (1993) points out that emerging markets have high average returns, low overall volatility, low exposure to world factors, and little integration. Higher returns and lower risk can be obtained by incorporating emerging market stocks in investors portfolios. Numerous studies investigate the transmission mechanism of stock price movements across international equity markets. Hamao, Masulis, and Ng (1990), using an autoregressive heteroskedastic (ARCH) model, report evidence of a price spillover effect from New York to Tokyo. In a subsequent study, Becker, Finnerty, and Tucker (1992), document that consistent with market efficiency, the Japanese market reacts within the first hour to previous returns in the United States. Employing a multivariate generalized autoregressive heteroskedastic (GARCH-M) model, Theodossiou and Lee (1993) find that statistically significant mean spillovers radiate from stock markets of the United States to the United Kingdom, Canada, and Germany, and then from the stock markets of Japan to Germany. Significant volatility spillovers radiate from the U.S. stock market to all four stock markets, from the U.K. stock market to the Canadian stock market, and from the German stock market to the Japanese stock market. Ng (2000) examines the magnitude and changing nature of the return and volatility spillovers from Japan and the United States to six Pacific-Basin equity markets. In addition to the effect of the world factors, he finds significant spillovers from the region to many of the Pacific-Basin countries. Using monthly excess returns for seven major European countries from 1970 to 1990, Longin and Solnik (1995) find that cross-country stock market correlations increase over time but are larger when large shocks occur. In a subsequent study, Karolyi and Stulz (1996), investigating daily return comovements between Japanese and U.S. stocks from 1988 to 1992, find evidence that correlations and covariances are high when markets move a lot. This suggests international diversification does not provide as much diversification against large shocks to national

3 Asian Economic Integration 143 equity markets as expected. Although international capital markets are perceived to have become more highly integrated, Bekaert and Harvey (1995) find time-varying integration for several countries. Using Geweke measures of feedback, we find a statistically high percentage of contemporaneous association between the twelve equity markets in Asia and Japan. The same-day intermarket responses are the most significant for Australia, China, Hong Kong, Malaysia, New Zealand, and Singapore. This result suggests a high degree of both market integration and market efficiency, as these markets interact significantly on the same day. Our empirical results indicate these Asian equity markets become more integrated over time, especially since A pooled time-series regression model across all possible pairs of these Asian markets, using Geweke feedback measures as the dependent variable and various economic factors as independent variables, shows that a higher import share as well as a greater differential in inflation rates, real interest rates, and gross domestic product (GDP) growth rates have negative effects on stock market comovements. Conversely, increased export share by Asian economies to Japan and greater foreign direct investment (FDI) from Japan to other Asian economies contribute to greater comovement. II. Data and Test Methods The data include Morgan Stanley Capital International s daily closing stock index values for thirteen national equity markets (Australia, China, Hong Kong, India, Indonesia, Japan, Korea, Malaysia, New Zealand, Philippines, Singapore, Taiwan, and Thailand) from 1988 through The following variables are used as possible economic determinants of international integration. All of the data are obtained from Datastream. Xij/Xi = exports from country i to country j (Japan), as a percentage of i s total exports; Xji/X j = exports from country j to country i, as a percentage of j s total exports; Mij/Mi = imports of country i from country j, as a percentage of i s total imports; Mji/M j = imports of country j from country i as a percentage of j s total imports; DINFLij = inflation differential between markets i and j; DRINTij = real interest rate differential between markets i and j; DXR ij = percentage change in the bilateral exchange rate of currency i in terms of j (Japanese yen);

4 144 The Journal of Financial Research SDXRij = volatility in the bilateral exchange rate of currency i in terms of j (Japanese yen); DMVij = percentage of world equity market share in country i minus that in j; DGDPij = the difference between the annual growth rate of GDP in country i minus that in j; FDIij = foreign direct investment from j to country i as a percentage of j s total foreign direct investment; and T t = trend variable for time t (i.e., 1, 2,..., 11). Geweke (1982) develops measures of feedback based on log likelihood ratio statistics, which provide a cardinal measure of the degree of comovement. An increase (decrease) in a Geweke measure, from year to year, reflects the magnitude of increase (decrease) of stock market integration for that pair of countries. The Geweke measures of feedback provide a more appropriate framework than the vector autoregression (VAR) model. The VAR approach is deficient in its failure to incorporate potential long-term relations and, therefore, may suffer from specification bias (Mukherjee and Naka 1995). Three Geweke measures of feedback are estimated for each Asian market paired with Japan, and for each year. Each Geweke measure represents a log likelihood ratio statistic for a null hypothesis, devised to test whether the equity market returns move together on the same day (H 1 ) or whether the Japanese market leads (H 2 ) or lags (H 3 ) other Asian markets by one to five days. The three hypotheses are stated as follows: H 1 : There is no contemporaneous relationship between r it and r jt on the same day. H 2 : The variable r jt does not lead r it across days. H 3 : The variable r it does not lead r jt across days. The variable r it represents the daily equity return in another Asian country and r jt represents the daily equity return in Japan. The asymptotic distribution of each Geweke feedback measure is also known under the alternative hypothesis that feedback is present. 1 This approach provides an important advantage over other means for testing the hypotheses, such as the Wald F-test, because Geweke measures are cardinal measures of comovement that allow us to determine the economic causes of greater comovement in stock market returns in two countries. 1 The Appendix presents the development of the Geweke measures for each of the three hypotheses. The reader can also refer to Geweke (1982) and Bracker, Docking, and Koch (1999) for the theoretical development of this measure.

5 Asian Economic Integration 145 Because the fundamental link between financial integration and economic growth is well documented (see Pagano 1993), we expect equity market integration to vary systematically with the degree of economic integration for a given pair of countries. The estimated annual Geweke contemporaneous feedback measures (GCFMij) are used as the dependent variable in a pooled time-series regression model across pairs of twelve Asian equity markets and Japan to estimate the influence of macroeconomic determinants on evolution in stock market integration. We specify a set of macroeconomic variables hypothesized to influence the degree of economic integration for each pair of countries. Our model extends the analysis of Bracker, Docking, and Koch (1999) for the United States versus eight other developed country markets in two important ways. First, we analyze the economic basis for comovement between emerging markets and the developed Japanese economy. Second, we add two important variables to the econometric model. The first variable is the annual growth rate of GDP in each Asian country minus the annual growth rate of GDP in Japan. This variable accounts for the effect of divergent growth patterns in the region. The second variable is FDI between Japan and the region, to account for the increased economic integration usually associated with FDI activity. Stock market performance of a country is an indication of future economic conditions. The prospects for exporting to that country should be positively related to its economic performance. Therefore, if country i sells a greater share of its total exports to country j (if Xij/Xi is larger), the stock market of country i should be more sensitive to country j s stock market movements. Similar reasoning leads to the prediction of a positive coefficient on Xji/X j as well. Two opposing hypotheses apply to the expected relation between the GCFMij and the share of country i s total imports coming from country j (Mij/Mi). One could argue that when the economy (and stock market) of country i does well, its consumers and producers acquire more imports from its trading partners, so the economy of country j and its stock market do well. Thus, we might expect greater import dependence of country i from country j to be associated with greater integration between stock markets i and j (i.e., a positive coefficient on Mij/Mi). Conversely, a reduction in Mij/Mi might signal improved prospects for exporting firms in country i and might therefore evoke a greater response in country i s stock market. Thus, the ability of country i s exporting firms to compete with firms from country j might be inversely related to its import dependence with country j (i.e., a negative coefficient on Mij/Mi). The combined result of these two effects may be either a positive or negative influence on the extent of stock market integration. An analogous argument leads to the prediction of a positive or negative coefficient on country j s import dependence on country i (i.e., Mji/M j). Purchasing power parity (PPP) requires that changes in relative inflation between two countries be offset by exchange rate changes. Frequent and large changes in the inflation differential between country i and country j (i.e., DINFLij), or in

6 146 The Journal of Financial Research the exchange rate between country i and country j (i.e., DXRij), can cause deviations in the terms of trade. Together, these two variables account for potential deviations from PPP (changes in real exchange rates over time). Similarly, real interest differentials DRINTij and exchange rate changes (i.e., DXRij) will combine to encompass deviations from interest rate parity (IRP). Any deviation from PPP or IRP may be expected to affect the profitability of firms in either national stock market (Bodurtha, Cho, and Senbet 1989). Thus, greater divergence in inflation rates, real interest rates, or currency valuation is likely associated with less comovement across equity markets on the same day, and we expect the coefficients on the three variables to be negative in the regression model. In addition, economic costs are imposed on economic entities in both countries by greater volatility in the bilateral exchange rate of currency i in terms of j (i.e., SDXRij). This uncertainty is expected to reduce economic and stock market integration between the two markets. The relative size of two countries stock markets may increase or decrease the degree of comovement between pairs of equity markets. As some of these emerging equity markets are significantly smaller because of bad economic policy, corruption, and lack of transparency, it would be unlikely for these markets to move in tandem. Alternately, the relative size and importance of one country s market might influence another country s market. Thus, we are unable to predict the sign of MVji. We expect that faster growing economies will become increasingly more independent of large economies like Japan and therefore experience less comovement with the Japanese stock market, unless the faster growth rate is the result of increased economic integration between the two markets. Because other factors in the model (e.g., export share, import share, and FDI) account for economic integration trends, we expect the difference between the annual growth rate of GDP in country i minus that in country j (i.e., DGDPij) to have a negative effect on stock market comovement. An increase in FDI from country j to country i, as a percentage of i s total FDI (i.e., FDIji), signals a growing internationalization of country i s economy and closer economic ties to Japan. Therefore, we expect a positive coefficient on FDIji. The relation between GCFMij and FDIji is nonlinear, so log(fdiji) is used in the regression model. III. Results The Geweke measures of contemporaneous feedback are reported in Table 1. They are calculated using daily rates of change in the equity price index (log(p t /P t 1 )) for twelve Asian markets versus Japan s market (data availability for the emerging economies in the sample limited the analysis to eleven years). Each statistic has an approximate χ 2 distribution with 1 degree of freedom under the hypothesis of no contemporaneous relation.

7 TABLE 1. Geweke Contemporaneous Feedback Measures Between Asian Markets and Japan. AU JP CH JP HK JP IN JP ID JP KO JP MY JP NZ JP PH JP SP JP TW JP TH JP NA 33.8 NA NA 5.7 NA NA 78.0 NA NA 72.4 NA NA 5.2 NA Notes: This table presents the Geweke measures of contemporaneous feedback described in the Appendix, representing likelihood ratio test statistics of the null hypothesis that there is no contemporaneous relation between the daily stock market returns in country i and Japan (JP), where i represents Australia (AU), China (CH), Hong Kong (HK), India (IN), Indonesia (ID), Korea (KO), Malaysia (MY), New Zealand (NZ), Philippines (PH), Singapore (SP), Taiwan (TW), and Thailand (TH). Each statistic has an approximate χ 2 distribution with 1 degree of freedom under the null hypothesis of no contemporaneous relation. Significant at the 1% level. Significant at the 5% level. Significant at the 10% level. Asian Economic Integration 147

8 148 The Journal of Financial Research The 122 Geweke contemporaneous feedback measures (GCFMij) 2 in Table 1 (twelve countries times eleven years minus ten because the China and India data were not available for the first five years) have an average value of Of these 122 measures, 72% are significant at the 10% level, 69% are significant at the 5% level, and 62% are significant at the 1% level. The high percentage of significant Geweke measures is deceptive because there is large variation in the degree of contemporaneous comovement of markets between different country pairs. For example, Australia, China, Hong Kong, Malaysia, New Zealand, and Singapore show significant contemporaneous comovement (5% level) for at least 82% of the years from 1988 to 1998 and have an average GCFMij of These countries experience substantive market comovement with Japan. A similar multilateral interaction is reported by Eun and Shim (1989) for the United States and eight other equity markets Australia, Canada, France, Germany, Hong Kong, Japan, Switzerland, and the United Kingdom for 1980 to They find that innovations in the U.S. stock market are rapidly transmitted to the rest of the world, whereas innovations in other national markets do not have much effect on the U.S. market. Other Asian markets paired with Japan show significant contemporaneous comovement less often at the 5% level (e.g., India, 17%; Korea, 18%; Indonesia, 36%; Taiwan, 45%; Philippines, 55%; Thailand, 55%) and have an average GCFMij of 5.9. For these countries we cannot reject the hypothesis of no contemporaneous comovement for most years, and the magnitude of comovement is generally small. Bracker, Docking, and Koch (1999) find a tendency for intensified comovement across the equity markets of Australia, Canada, Germany, Hong Kong, Japan, Singapore, Switzerland, the United Kingdom, and the United States from 1972 to Our results give only partial support for their conclusion. Figure Ia displays a nearest neighbor fit polynomial regression line between the Geweke contemporaneous feedback measures (GCFMij) and time (T). The regression line suggests that the degree of contemporaneous comovement between different Asian markets and Japan increases between 1988 and 1991 (when all pairs are significant), only to fall to a low in 1993 (when only China has a significant contemporaneous comovement with Japan), and then to rise to 1991 levels again in Table 2 contains the Geweke unidirectional feedback measures from Japan to the other twelve Asian markets (GUFMji). These measures record the magnitude of the association between changes in Japan s daily equity returns and the daily equity returns in each of the twelve Asian markets, one to five days later. In general, the measures are small and only 21 of the 122 measures are significant at the 5% level. 2 All variables ending in the combinations ij, ji, i, or j represent annual time-series measures where i represents one of the twelve Asian (non-japan) countries and j represents Japan.

9 Asian Economic Integration 149 Note: Time (T), plotted on the horizontal axis, ranges from 1 (1988) to 11 (1998). Figure Ia. Geweke Contemporaneous Feedback Measures Between Twelve Asian Equity Markets and Japan ( ). Similarly, Table 3 contains the Geweke unidirectional feedback measures from the other twelve Asian markets to Japan (GUFMij). These measures reflect the magnitude of the association between changes in the daily equity returns in each of the twelve Asian markets and Japan s daily equity returns, one to five days later. The measures tend to be small and only 15 of the 122 measures are significant at the 5% level. In general, the hypothesis of no lagged effect from Japan s market to the other Asian markets and vice versa cannot be rejected. Figure Ib displays a nearest neighbor fit polynomial regression line between the Geweke unidirectional feedback measures (GUFMij and GUFMji) and time (T). The regression line suggests that the degree of unidirectional feedback between Japan and other Asian markets is low and changes little between 1988 and The fact that only a few of the Geweke unidirectional feedback measures are significant, combined with the large number of cases where the Geweke contemporaneous feedback measures are significant, indicates that the markets in this region display a high degree of efficiency. As in prior studies for the U.S. stock market (Eun and Shim 1989; Koch and Koch 1991; Bracker, Docking, and Koch 1999), there are few intermarket adjustments continuing beyond the same calendar day that might offer exploitable profit opportunities. The observed response pattern is consistent with the notion of informationally efficient international stock markets. This concludes the first stage of our analysis. Stage two of our analysis scrutinizes the variation in stock market integration in an attempt to investigate why different pairs of equity markets experience changing degrees of stock market integration over time. The financial literature reports empirical evidence supporting

10 TABLE 2. Geweke Unidirectional Feedback Measures from Japan to Asian Markets, One to Five Days Later. JP AU JP CH JP HK JP IN JP ID JP KO JP MY JP NZ JP PH JP SP JP TW JP TH NA 12.9 NA NA 9.2 NA NA 13.0 NA NA 1.7 NA NA 7.9 NA Notes: This table presents the Geweke measures of unidirectional feedback described in the Appendix, representing likelihood ratio test statistics of the null hypothesis that there is no unidirectional feedback between the daily stock market returns in Japan (JP) and country i one to five days later, where i represents Australia (AU), China (CH), Hong Kong (HK), India (IN), Indonesia (ID), Korea (KO), Malaysia (MY), New Zealand (NZ), Philippines (PH), Singapore (SP), Taiwan (TW), and Thailand (TH). Each statistic has an approximate χ 2 distribution with 5 degrees of freedom under the null hypothesis that Japan s market does not lead other Asian markets over a period of one to five days. 150 The Journal of Financial Research Significant at the 1% level. Significant at the 5% level. Significant at the 10% level.

11 TABLE 3. Geweke Unidirectional Feedback Measures from Asian Markets to Japan, One to Five Days Later. AU JP CH JP HK JP IN JP ID JP KO JP MY JP NZ JP PH JP SP JP TW JP TH JP NA 16.9 NA NA 5.3 NA NA 9.6 NA NA 7.7 NA NA 3.9 NA Notes: This table presents the Geweke measures of unidirectional feedback described in the Appendix, representing likelihood ratio test statistics of the null hypothesis that there is no unidirectional feedback between the daily stock market returns in each of the other Asian market i and Japan (JP), one to five days later, where i represents Australia (AU), China (CH), Hong Kong (HK), India (IN), Indonesia (ID), Korea (KO), Malaysia (MY), New Zealand (NZ), Philippines (PH), Singapore (SP), Taiwan (TW), and Thailand (TH). Each statistic has an approximate χ 2 distribution with 5 degrees of freedom under the hypothesis that other Asian markets do not lead Japan s market over a period of one to five days. Significant at the 1% level. Significant at the 5% level. Significant at the 10% level. Asian Economic Integration 151

12 152 The Journal of Financial Research Note: Time (T), plotted on the horizontal axis, ranges from 1 (1988) to 11 (1998). Figure Ib. Geweke Unidirectional Feedback Measures Between Twelve Asian Equity Markets and Japan, One to Five Days Later. the association between changes in stock prices and measures of real economic activity (Chen, Roll, and Ross 1986; Elton and Gruber 1988; Mukherjee and Naka 1995; among others). Table 4 reports the results of the pooled regression analysis over the eleven-year sample period for the twelve country pairs. The model explains more than one-third of the variation in contemporaneous comovement between Asian stock markets and the Japanese stock market. The variables in the model are jointly significant at the 1% level, and seven of the twelve variables in the model are individually significant at the 10% level. All of the significant variables have the expected effect on stock market comovement. In general, a higher import share as well as a greater differential in inflation rates, real interest rates, and GDP growth rates have a negative effect on stock market comovements between country pairs. Conversely, increased export share by Asian economies to Japan and greater FDI from Japan to other Asian economies contribute to greater comovement. Our results provide evidence consistent with Asprem (1989) for ten European countries from 1968 to Asprem shows that employment, imports, inflation, and interest rates are inversely related to stock prices. In a similar study, Bracker, Docking, and Koch (1999) find bilateral import dependence, among other factors, to be significantly associated with the extent of stock market integration over time. The share of Japanese imports falls for most Asian economies, whereas the share of Japanese imports from many of the other Asian markets rises over the

13 Asian Economic Integration 153 TABLE 4. Results of the Pooled Regression Analysis. Dependent Variable GCFMij Intercept (5.63) Xij/Xi 0.75 (1.79) Xji/X j ( 0.72) Mij/Mi 1.02 ( 2.72) Mji/M j 2.32 ( 1.86) DINFLij 3.20 ( 6.10) DRINTij 3.56 ( 5.23) DXRij 0.13 (0.86) SDXRij 0.05 (0.09) MVji ( 0.43) DGDPij 1.86 ( 3.74) LogFDIji 9.86 (4.48) T t 1.17 (1.45) R Adjusted R F-statistic 6.85 Notes: This table presents the results of the pooled regression for the model described in section II. GCFMij = Geweke contemporaneous feedback measure from country i to j (Japan); Xij/Xi = exports from country i to j, as a percentage of i s total exports; Xji/X j = exports from j to country i, as a percentage of j s total exports; Mij/Mi = imports of country i from j, as a percentage of i s total imports; Mji/M j = imports of j from country i as a percentage of j s total imports; DINFLij = inflation differential between markets i and j; DRINTij = real interest rate differential between markets i and j; DXRij = percentage change in the bilateral exchange rate of currency i in terms of j (Japanese yen); SDXRij = volatility measured as the standard deviation in the bilateral exchange rate of currency i in terms of j (Japanese yen); DMVij = percentage of world equity market share in country i minus that in j; DGDPij = the difference between the annual growth rate of gross domestic product in country i minus that in j; LogFDIij = the natural logarithm of FDI from j to country i as a percentage of j s total FDI; T t = trend variable for time t (i.e., 1, 2,..., 11). The t-statistics are in parentheses. Significant at the 1% level. Significant at the 5% level. Significant at the 10% level. period. These two effects may largely offset each other. Negative real interest rate differentials tend to offset large positive inflation rate differentials in Hong Kong (and China for 1993 to 1995). However, both differentials are generally high for India, Indonesia, Philippines, and Thailand, whereas Australia and New Zealand tend to have high real interest rate differentials during the entire period. Thus, these two factors account for a significant reduction in comovement over the period. Neither DXRij nor SDXRij prove to be significant in any of the equations when the other is removed; thus, multicollinearity is not the reason for the insignificant effects of these two variables. Rapid economic growth in many Asian economies relative to Japan leads to less influence from Japan s stock market. Finally, stock market comovement in the region is significantly increased by Japan s large FDI in the region. This is especially true for countries like Australia, China, Hong Kong, Indonesia, and Thailand, which are receiving the bulk of it.

14 154 The Journal of Financial Research IV. Conclusions We examine the degree of cross-country return comovement for twelve Asian equity markets with the Japanese stock market. First, using Geweke measures of feedback, we find a high percentage (62% at the 1% level) of contemporaneous association between the twelve Asian equity markets paired with Japan. The same-day intermarket responses are significant for Australia, China, Hong Kong, Malaysia, New Zealand, and Singapore (at least 82%). This result suggests a high degree of both market integration and market efficiency, as these pairs of markets interact significantly on the same day. This finding is consistent with the observation made by Eun and Shim (1989), Becker, Finnerty, and Tucker (1992), and Koch and Koch (1991) that there are few significant cross-market adjustments taking place beyond one day. Similar to Bracker, Docking, and Koch (1999) for the United States and eight other developed country markets, investors in Asian markets find few intermarket arbitrage opportunities beyond twenty-four hours. Second, the degree of contemporaneous comovement between different Asian markets and Japan increases between 1988 and 1991 only to fall to a low in 1993, and then to rise to 1991 levels again in Third, results for same-day relations indicate that several macroeconomic variables are significantly associated with the extent of equity market integration over time. In general, a higher import share as well as a greater differential in inflation rates, real interest rates, and GDP growth rates have a negative effect on stock market comovements between country pairs. Conversely, increased export share by Asian economies to Japan and greater FDI from Japan to other Asian economies contribute to greater comovement. Appendix We specify a model where the daily stock returns in one country are a function of the lagged returns in another country s market and its own past returns. The following system of two equations is estimated as seemingly unrelated regressions. Under H 1, H 2, and H 3, the system of equations in (A.1) and (A.2) becomes: M 2 M 1 r 1t = α 0 + E k r 2t k + F k r 1t k + ε 1t var(ε 1t ) = σε1 2 (A.1) k=1 k=1 M 2 M 1 r 2t = β 0 + G k r 1t k + H k r 2t k + ε 2t var(ε 2t ) = σε2 2 (A.2) k=1 k=1 Y =determinant of covariance matrix Y = cov(ε 1t, ε 2t ), where r it represents the daily stock market return in country i (i = 1, 2). The

15 Asian Economic Integration 155 disturbance terms ε it are assumed to be distributed N(0, σi 2 ) i = 1, 2, not autocorrelated, and contemporaneously correlated with each other in σ 12 (Judge et al. 1988). Potential market trends and day-of-the-week effects are accounted for with dummy variables. M 1 is set to ten business days and M 2 to five business days. The results are robust for higher values of M 1 and M 2,butt-tests and Wald tests of coefficients on additional days added to M 1 and M 2 are generally insignificant. As a general rule, adding values of M 1 and M 2 beyond 10/5 does not systematically change the significance of the observed Geweke measures. Our objective is to measure the degree to which daily stock returns move together in the two countries on the same day and the degree to which daily stock returns in the two countries lead and lag each other. For example, the coefficients E k in equation (A.1) display how the second stock market leads the first market across days. Similarly, the coefficients G k in (A.2) reflect how the first market leads the second across days. The contemporaneous correlation across error terms in the covariance matrix captures the relation on the same day. These considerations lead us to specify and test the following three null hypotheses: H 1 : There is no contemporaneous relation between r 1t and r 2t on the same day. H 2 : The variable r 2t does not lead r 1t across days (E k = 0, for any k). H 3 : The variable r 1t does not lead r 2t across days (G k = 0, for any k). Under H 1, H 2, and H 3, the system of equations in (A.1) and (A.2) becomes: M 1 r 1t = α 0 + F k r 1t k + µ 1t, Var(µ 1t ) = σµ1 2 (A.3) M 1 r 2t = β 0 + k=1 k=1 H k r 2t k + µ 2t, Var(µ 2t ) = σ 2 µ2, with Cov(µ 1t, µ 2t ) = 0. (A.4) Equations (A.1) and (A.2) are estimated as a system of seemingly unrelated regressions (Judge et al. 1988), and equations (A.3) and (A.4) are estimated with ordinary least squares. The likelihood ratio test formulated with the estimated residual variances and covariances of the restricted equations (i.e., (A.1) and (A.2)) and unrestricted equations (i.e., (A.3) and (A.4)) can be used to test hypotheses H 1, H 2, and H 3. These likelihood ratio test statistics form the Geweke (1982) measures of feedback: GCFM 1 2 = (n) ln[(σ 2 µ1 σ 2 µ2 )/ Y ] is distributed approximately χ 2 with 1 degree of freedom under H 1 ; GUFM 2 1 = (n) ln(σ 2 µ1 /σ 2 ε1 ) is distributed approximately χ 2 with M 2 degrees of freedom under H 2 ;

16 156 The Journal of Financial Research GUFM 1 2 = (n) ln(σ 2 µ2 /σ 2 ε2 ) is distributed approximately χ 2 with M 2 degrees of freedom under H 3. The variable GCFM 1 2 represents the estimated Geweke contemporaneous feedback measure between countries 1 and 2; GUFM i j represents the estimated Geweke unidirectional feedback measure from country i to country j; n is the sample size; Y is the estimated determinant of the covariance matrix Y ; σij 2 represents the estimated variance of the residuals from equations (A.1) to (A.4) where i = ε and µ and j = 1 and 2; and M 2 equals 5. References Agmon, T. and D. R. Lessard, 1977, Investor recognition of corporate international diversification, Journal of Finance 32, Asprem, M., 1989, Stock prices, asset portfolios and macroeconomic variables in ten European countries, Journal of Banking and Finance 13, Becker, K. G., J. E. Finnerty, and A. L. Tucker, 1992, The intraday interdependence structure between U.S. and Japanese equity markets, Journal of Financial Research 25, Bekaert, G. and C. Harvey, 1995, Time-varying world market integration, Journal of Finance 50, Bodurtha, J. Jr., D. Cho, and L. Senbet, 1989, Economic forces and the stock market: An international perspective, Global Finance Journal 1, Bracker, K., D. Docking, and P. Koch, 1999, Economic determinants of evolution in international stock market integration, Journal of Empirical Finance 6, Chen, N. F., R. Roll, and S. A. Ross, 1986, Economic forces and the stock market, Journal of Business 59, Elton, E. J. and M. Gruber, 1988, A multi-index risk model of the Japanese stock market, Japan and the World Economy 1, Eun, C. S. and S. Shim, 1989, International transmission of stock market movements, Journal of Financial and Quantitative Analysis 24, Geweke, J., 1982, Measurement of linear dependence and feedback between multiple time series, Journal of the American Statistical Association 77, Grubel, H. G., 1968, International diversified portfolios: Welfare gains and capital flows, American Economic Review 58, Hamao, Y. R., R. W. Masulis, and V. K. Ng, 1990, Correlation in price changes and volatility across international stock markets, Review of Financial Studies 3, Harvey, C. R., 1993, Portfolio enhancement using emerging markets and conditioning information, in S. Claessens and S. Gooptu, eds.: Portfolio Investment in Developing Countries (The World Bank Discussion Series, The World Bank, Washington, DC). Judge, G., R. Griffiths, W. Hill, H. Lutkepohl, and T. Lee, 1988, Introduction to Theory and Practice of Econometrics (Wiley, New York). Karolyi, G. and R. Stulz, 1996, Why do markets move together? An investigation of U.S.-Japan stock return comovements, Journal of Finance 51, Koch, P. D. and T. W. Koch, 1991, Evolution in dynamic linkages across daily national stock indices, Journal of International Money and Finance 10, Longin, F. and B. Solnik, 1995, Is the correlation in international equity returns constant: , Journal of International Money and Finance 14, Mukherjee, T. K. and A. Naka, 1995, Dynamic relations between macroeconomic variables and the Japanese stock market: An application of a vector error correction model, Journal of Financial Research 18,

17 Asian Economic Integration 157 Ng, A., 2000, Volatility spillover effects from Japan and the US to the Pacific-Basin, Journal of International Money and Finance 19, Obstfeld, M., 1994, Risk taking, global diversification, and growth, American Economic Review 84, Pagano, M., 1993, Financial markets and growth: An overview, European Economic Review 37, Theodossiou, P. and U. Lee, 1993, Mean and volatility spillovers across major national stock markets: Further empirical evidence, Journal of Financial Research 16,