Printer Friendly

US and domestic market gains and Asian investors' overconfident trading behavior.

We find that subsequent to both US and domestic market gains, both Asian individual and institutional investors increase their trading and that this effect is more pronounced in bull markets, in periods of relatively favorable investor sentiment, in periods of extremely high market returns, and in markets with short-sale constraints. We also find that individual investors trade more in response to market gains than institutional investors. Moreover, we find that further integration of Asian stock markets with US stock markets after the Asian financial crisis in 1998 is an important reason for Asian investors' response to US market gains.

**********

There are a growing number of theoretical models rooted in investor overconfidence to account for the observed excess trading volume in securities markets. Gervais and Odean (2001) propose a self-learning model predicting that investors trade more aggressively following market gains as they overestimate the degree to which they contribute to returns from general market increases and exaggerate their ability to pick winning stocks. Since investors, in aggregate, hold long positions in equity markets, the Gervais and Odean (2001) model provides a testable implication of the overconfidence hypothesis: high market gains are followed by high market trading volume.

Odean (1998b) and Gervais and Odean (2001) argue that people who are more overconfident in their investment abilities are more likely to seek jobs as traders or to actively trade on their own accounts. If so, we can expect to find that financial markets are populated by overconfident investors. Many researchers also argue that overconfident investors can survive and dominate markets in the long run (Kyle and Wang, 1997; Daniel, Hirshleifer, and Subrahmanyam, 1998; Gervais and Odean, 2001). These arguments imply that it is possible to detect investors' aggregate overconfident trading behavior if overconfidence is a systematic cognitive bias. Focusing on aggregate overconfident trading behavior, Statman, Thorley, and Vorkink (2006) and Chuang and Lee (2006) find that high US market-wide returns are followed by high US market-wide trading volume. They interpret their finding as consistent with the theoretical prediction of the Gervais and Odean (2001) model.

Overconfidence is not something unique to US investors. Psychologists have demonstrated that Asians also exhibit overconfidence in general knowledge (e.g., Yates, Lee, and Bush, 1997). This suggests that Asian investors may be a good platform for testing the overconfidence hypothesis. Indeed, Chen et al. (2007), Kim and Nofsinger (2007), and Chuang and Susmel (2011) find evidence that Chinese, Japanese, and Taiwanese investors display significantly overconfident trading behavior, respectively.

Prior studies regarding investors' overconfident trading tend to focus on whether domestic market gains make investors trade with overconfidence in domestic markets. In this article, we conjecture that Asian investors' overconfidence and overconfident trading can be evoked by US market gains for three reasons. First, for the purpose of international diversification, Asian investors have a desire to invest in various US-based investments (French and Poterba, 1991; Bekaert and Urias, 1996). We find that the share of US stocks held by investors in our sample Asian countries steadily increases during our sample period. These investors benefit from US market gains. Then, as argued by Gervais and Odean (2001), the increase in wealth makes them overconfident and they are encouraged to trade more actively in subsequent periods in domestic markets.

Second, with the increasing integration of international equity markets, any news concerning US stock markets may have important implications for other stock markets (Hamao, Masulis, and Ng, 1990; Lin, Engle, and Ito, 1994). This is why the financial media report the performance of US stock markets on a regular basis in Asian countries as this may help Asian investors infer future price movement of their domestic stock markets from the current US stock market price movement. Asian investors, for example, may expect their domestic markets to rise in the near future after observing US market gains. As a consequence, their expectations of gains from local equity investments make Asian investors overconfident. As such, they trade more aggressively in domestic markets subsequent to US market gains.

Third, Baker, Wurgler, and Yuan (2012) show that investor sentiment can be decomposed into global and local components. They find that US capital flows are a key mechanism by which sentiment spreads across markets. They also argue that word-of-mouth and the media may also have a similar effect. By the same token, investor overconfidence may also have global and local components. Given the extensive evidence of the return spillover effect from US stock markets to Asian stock markets, US market gains may be a key factor contributing to the global component of Asian investors' overconfidence encouraging them to trade more frequently in their domestic markets. In addition, psychologists have documented that optimistic people tend to be overconfident (Miller and Ross, 1975). As such, global bullish sentiment may also nourish Asian investors' overconfidence causing them to trade more aggressively in their domestic markets.

Based on the Gervais and Odean (2001) overconfident trading hypothesis, we empirically examine whether investors in the 10 Asian stock markets trade with overconfidence after domestic and US market gains. Consistent with our conjecture, we find that not only domestic market gains, but also US market gains encourage Asian investor overconfidence causing them to trade more aggressively in subsequent periods. Our Granger (1969) causality tests show that US market returns (positively) Granger-cause Asian market returns, even for those markets in which US market returns do not (directly) Granger-cause domestic trading volume. This implies that US market returns may also indirectly affect Asian investors' overconfident trading behavior via their impact on domestic market returns.

Regarding the impact of US market gains on Asian investors' overconfident trading, we find that Asian investors trade more actively subsequent to US market gains when US investor sentiment is high. This finding implies that high US investor sentiment leads Asian investors to expect to gain from their domestic and US-based equity investment and, therefore, trade more actively in domestic markets. Our subperiod analysis also shows that most Asian investors trade more actively after US market gains in the second subperiod in which the integration of Asian stock markets with the US stock markets has become stronger and Asian investors trade more in US stocks than in the first subperiod. As discussed earlier, this implies that Asian investors' overconfident trading increases following US market gains due to the recent trend toward globalization of the markets.

The theoretical literature shows that investors' overconfidence can be affected by short-sale constraints. In Baker and Stein's (2004) model, in the presence of short-sale constraints, less informed investors with positive sentiment dominate the market and smart investors sit on the sidelines, making the less informed investors become more overconfident. Without short-sale constraints, both of these investors actively participate in the market. In this circumstance, the interaction of their sentiment encourages less informed investors to become overconfident only to a lesser degree (see also Hong, Scheinkman, and Xiong, 2006). In a similar vein, Scheinkman and Xiong (2003) show that when investors face short-sale constraints, the asset price bubble increases with their overconfidence and more trading is generated during the bubble period.

To test the impact of short-sale constraints on Asian investors' overconfident trading, we divide our sample countries into two groups based on whether short sales are allowed and practiced in their stock markets (Bris, Goetzmann, and Zhu, 2007). The theory predicts that investors will trade more actively after domestic market gains in the presence of the short-sale constraint than in the absence of it. Consistent with this prediction, we find significant evidence that Asian investors in markets with short-sale constraints, on average, tend to trade more frequently after domestic market gains than those in markets without short-sale constraints. To the best of our knowledge, our paper is the first to empirically address this issue in the context of overconfident trading.

To gain more insight into the overconfident trading behavior of Asian investors, we further analyze how Asian investors behave conditional on investor sentiment, market conditions, and attention-grabbing information proxied by extremely high market returns. These conditional events are suggested by the behavioral theory. Using this conditional framework, first we find that Asian investors tend to trade more actively after domestic market gains when they are more optimistic. Second, we find that Asian investors tend to trade more frequently following domestic (US) market gains when their domestic (US) markets are in bull markets than when their domestic (US) markets are in nonbull markets. And third, we find that Asian investors tend to trade more aggressively when the domestic (US) market experiences extremely high returns than when it does not.

It is well known that institutional trading in emerging markets, such as Asian stock markets, tends to be lower than the US stock market or other developed markets (see, for example, Bae, Kang, and Kim, 2002). Our use of the market-level data may conceal the role institutional investors play in Asian stock markets and, consequently, their overconfident trading behavior. To see whether institutional investors also engage in overconfident trading in Asian stock markets, we collect the data of individual and institutional trading for six of our sample stock markets. We find that both individual and institutional investors trade more in bull markets, in periods of high investor sentiment, and in periods of extremely high market returns and that individual investors tend to trade more actively in these events than institutional investors.

Overall, our study provides evidence that Asian investors trade as overconfidently as US investors, as documented in the prior literature, and that not only domestic market gains but also US market gains make Asian investors trade with overconfidence. Baker et al. (2012) find that investor sentiment can be affected by cross-market factors. We contribute to the literature by showing that cross-market factors, such as US market gains, can affect Asian investors' overconfidence, which further affects their trading behavior in domestic markets. We also provide evidence that Asian investors' overconfident trading behavior increases with short-sale constraints. This finding has an important policy implication. If market policy makers want to protect investors from succumbing to the overconfidence bias to trade too much and too speculatively, removing short-sale constraints will help achieve this goal.

However, there are alternative theories that predict the return-volume relation we document here. For example, to interpret their finding of the positive relation between past returns and current volume from 46 countries, Griffin, Nardari, and Stulz (2007) review and discuss theories that predict such a relation. They note that in addition to overconfidence, there are five alternative theories that provide potential explanations. (1) We explore the possibility of alternative theories to explain our findings. The results from these tests show that the return-volume relation in three of ten sample countries can be explained by the participation effect and informed trading. Taking this into account, we control for the impact of these two alternative theories on the return-volume relation when we investigate the trading behavior of investors in these three countries.

The remainder of the paper is organized as follows. Section I introduces the data, describes our method of detrending the trading volume series to achieve its stationarity, and reports some descriptive statistics of the data. Section II discusses our empirical framework that is devised to investigate whether Asian investors trade more after domestic and US market gains, and presents and discusses the empirical results. Section III reviews the alternative theoretical arguments that imply a positive return-volume relation, discusses the test results of these arguments, and conducts various robustness checks using the individual and institutional trading data and the US capital flow data. Finally, Section IV offers our conclusions.

I. Data and Detrending Trading Volume Series

A. Data

Our data set includes market price indices and trading volumes for 10 Asian stock markets: Hong Kong, Japan, Malaysia, Singapore, Thailand, China, Indonesia, Korea, the Philippines, and Taiwan. The stock indices for the 10 Asian stock markets are the Hang Seng Index (HSI) for Hong Kong, the Tokyo Stock Exchange Price Index (TOPIX) for Japan, the Kuala Lumpur Composite Index (KLSE) for Malaysia, the Strait Times Index (STI) for Singapore, the Stock Exchange of Thailand Index (SET) for Thailand, the Shanghai SE Composite Index (SSEC) for China, the Jakarta Composite Index (JKSE) for Indonesia, the Korea Stock Exchange Composite Index (KOSPI) for Korea, the Philippines Stock Exchange Composite Index (PSECI) for the Philippines, and the Taiwan Weighted Index (TWI) for Taiwan. The market index for the US stock market is the S&P 500.

To investigate whether Asian investors trade more heavily in the presence of short-sale constraints, we divide the sample stock markets into two groups based on whether short sales are allowed and practiced in their stock markets. Following the Bris et al. (2007) classification, Hong Kong, Japan, Malaysia, Singapore, and Thailand are classified as stock markets without short-sale constraints as short sales are allowed and practiced, while China, Indonesia, Korea, the Philippines, and Taiwan are classified as those with short-sale constraints as short sales are prohibited or are allowed but not practiced. Hereafter, we call the markets and investors in the former countries as the ss (short sale) markets and ss investors, respectively, and the markets and investors in the latter countries as the sc (short-sale constraints) markets and sc investors, respectively.

For our analysis, we use weekly observations that are constructed from the daily data extracted from the Datastream International database, and the sample period is from January 1995 to December 2010. The weekly return of each index is computed as the return from Wednesday's closing price to the following Wednesday's closing price (see also Griffin et al., 2007). The Wednesday-to-Wednesday close-to-close returns of each market index are measured as the log difference of the index prices and are expressed in percent. We use the total number of shares traded in a trading day as a measure of raw (or undetrended) trading volume. Following Lo and Wang (2000), the weekly raw trading volume is computed as the log of a summation from the total number of shares traded on Thursday to that traded on the following Wednesday. We match all series of stock returns and raw trading volume as our empirical analyses employ Zellner's (1962) Seemingly Unrelated Regression (SUR) model that requires all variables to have the same number of observations.

B. Detrending Trading Volume Series

Previous work finds significant evidence of both linear and nonlinear time trends in the trading volume series (Gallant, Rossi, and Tauchen, 1992). As a result, many empirical studies on trading volume use some form of detrending to achieve stationarity. In the spirit of Gallant et al. (1992), we detrend the raw trading volume series, [RTV.sub.it], for each Asian stock market i by taking into account the calendar effect on trading volume as follows (see also Lo and Wang, 2000):

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

where the regressors DEC[1.sub.t], ..., DEC[4.sub.t] and JAN[1.sub.t], ..., JAN[4.sub.t], denote weekly indicator variables for the weeks in December and January, respectively; and [MAR.sub.t], ..., [NOV.sub.t], denote monthly indicator variables for the months of March through November, respectively. February is omitted to avoid perfect collinearity. The number of lags in the autoregressive terms, [RTV.sub.it-l], is determined by the Ljung-Box (9-statistic; that is, we add lags until the Ljung-Box statistic shows no autocorrelation of the residual terms for each detrended trading volume series. Throughout our analysis, the resulting detrended trading volume, [AV.sub.it], is used as a measure of trading volume for the Asian stock market i. Hereafter, we refer to the detrended trading volume as trading volume or volume. To facilitate reporting of the estimated coefficients in our tests, the estimated detrended trading volume, [AV.sub.it], is multiplied by 100.

C. Summary Statistics

Table I presents descriptive statistics on the weekly return and trading volume series for 10 Asian market indices and the weekly return series for the S&P 500 for the sample period from January 4, 1995 to December 29, 2010. Specifically, the table reports the mean, standard deviation (SD), minimum and maximum values, the test statistic (/-statistic) of the augmented Dickey-Fuller test (ADF, Dickey and Fuller, 1979), two sums of the correlations, and the results of three Granger-causality tests.

Over the entire sample period, Indonesia has the highest mean return of 0.302%, while Japan has the lowest at -0.079%. Korea is the most volatile market with a standard deviation of 4.654%, while the S&P 500 is the most stable index with a standard deviation of 2.848%. It is worth noting here that Hong Kong, Japan, Malaysia, and Singapore have a smaller standard deviation than China, Indonesia, Korea, and the Philippines. This implies that the stock markets without short-sale constraints are more stable than those with short-sale constraints. The trading volume of all Asian stock markets exhibits a mean equal to zero after detrending. This is because detrended trading volume is the residual term in Equation (1). Trading volume appears to be more stable relative to stock returns for all Asian stock markets; the Philippines and Japan have the highest and lowest volume volatilities of 0.477 and 0.169, respectively. The results of the augmented Dickey-Fuller (ADF) test show that the null hypothesis of a unit root can be rejected for all Asian stock market returns and trading volume, indicating that they are stationary time series.

[[rho].sub.1] and [[rho].sub.2] are used to measure the sum of the correlations between current volume and lagged domestic returns up to three lags, and between current volume and lagged US returns up to three lags, respectively. The table indicates that [[rho].sub.1] are positive for all Asian countries and [[rho].sub.2] are positive for all Asian countries but China, suggesting a positive relation between lagged domestic returns and current volume, and between lagged US returns and current volume for these countries. The formal test of these associations can be found in Causality Tests 1, 2, and 3, which are the test results of the trivariate Granger-causality tests of domestic trading volume, domestic market returns, and US market returns. (2) Specifically, Causality Test 1 examines whether domestic market returns positively Granger-cause domestic trading volume, Causality Test 2 whether US market returns positively Granger-cause domestic trading volume, and Causality Test 3 whether US market returns positively Granger-cause domestic market returns. The results of Causality Test 1 show that the domestic market returns positively Granger-cause domestic trading volume for all Asian countries, except for Singapore. This implies that Asian investors, except for Singapore investors, tend to trade more actively after domestic market gains. The results of Causality Test 2 show that the US market returns positively Granger-cause domestic trading volume for Hong Kong, Singapore, Korea, and Taiwan. This implies that investors in these four countries tend to trade more aggressively after US market gains. The results of Causality Test 3 show that, except for China, the US market returns positively Granger-cause domestic market returns for all the other Asian countries. This indicates that the stock price movements in the US stock markets help predict stock price movements in the domestic stock markets for all countries, except for China. Moreover, this finding provides a basis for our conjecture as to why Asian investors trade more actively subsequent to US market gains.

II. Empirical Frameworks and Results

A. Overconfident Trading across Asian Stock Markets

As noted by Griffin et al. (2007), in addition to investor overconfidence, several alternative theories have been proposed for the positive relation between lagged returns and current volume. We also conduct empirical tests to examine whether these theories provide important determinants of the positive return-volume relation in Section III. Our results suggest that the return-volume relation needs to be controlled for the participation effect and the informed trading effect for Hong Kong and Singapore and for Taiwan, respectively. Since we find some evidence for these alternative theories, we control for these factors when we investigate Asian investors' trading behavior.

Before introducing our empirical models, we discuss two control variables used in all of our tests, including the Granger-causality tests in Table I. Ross (1989) shows that in a frictionless market characterized by an absence of arbitrage opportunities, the rate of information flow is revealed by the degree of the volatility of asset returns. Based on this finding, previous studies employ the absolute value of stock returns as a proxy for information flow to the stock markets (Chuang and Lee, 2006; Chuang and Susmel, 2011). Following previous studies, we use the absolute value of the returns of the Asian stock market i, [absolute value of [AR.sub.it]], and that of the returns of the US stock market, [absolute value of [UR.sub.t]], as proxies for information flows emanating from the Asian stock market i and from the US stock market, respectively, to account for informational trades.

Following Chuang and Susmel (2011), we work with the detrended [absolute value of [AR.sub.it]] and [absolute value of [UR.sub.t]] series since it is well known that [absolute value of [AR.sub.it]] and [absolute value of [UR.sub.t]] are highly serially correlated (Ding, Granger, and Engle, 1993). We follow Pagan and Schwert's (1990) method to filter [absolute value of [AR.sub.it]] and [absolute value of [UR.sub.t]] by regressing these variables on their own lagged values and then take two estimated residuals as the detrended value of [absolute value of [AR.sub.it]] (i.e., [DAR.sub.it]) and that of [absolute value of [UR.sub.t]] (i.e., [DUR.sub.t]), and use them as control variables in our empirical models.

We utilize the SUR model in most of our tests including the Granger-causality tests in Table I. This is because the multivariate SUR model allows us to test the cross-equation restrictions when we examine the impact of short-sale constraints on the trading behavior of ss versus sc investors. To compare the relative degree of the trading intensity of ss versus sc investors, we also standardized the variables in our empirical models.

We estimate the following SUR model to examine Asian investors' trading behavior across countries by regressing [AV.sub.it] on two control variables, [DAR.sub.it] and [DUR.sub.t], and the lagged values of [AR.sub.it] and [UR.sub.t] across the 10 sample stock markets:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (2)

where the subscript i represents the cross-sectional unit of the 10 Asian stock markets; [AV.sub.it] is the trading volume of market i on week t; [AR.sub.it] ([UR.sub.t]) is the stock return of market i (US stock market) on week t; and [DAR.sub.it] ([DUR.sub.t]) is the detrended absolute value of [AR.sub.it] ([UR.sub.t]) on week t. The number of lags in Equation (2) and the following equations is chosen by considering the Akaike (1974) information criterion (AIC) for each market i. In Equation (2), we consider only the current values of the detrended absolute value of domestic stock returns and that of US stock returns as we find that they have only contemporaneous impacts on domestic trading volume.

In Equation (2), the [[beta].sub.ij]([[gamma].sub.ik]) coefficients measure the causal relation from lagged domestic (US) returns to the current volume. If the [[beta].sub.ij]([[gamma].sub.ik]) coefficients are positive and significant as a group, domestic (US) market gains help predict investors in market i to trade with overconfidence in subsequent periods.

To control for the participation effect on the domestic return-volume relation for Hong Kong and Singapore, we modify Equation (2) in such a way that domestic returns are decomposed into two components for these two countries: (1) domestic returns conditional on the state of positive market returns with high volatility and (2) those conditional on the other state. Under this specification, the positive relation between current volume and past returns conditional on the state of positive returns with high volatility is explained by the participation effect, while the relation conditional on the other state is explained by investor overconfidence. Similarly, to control for the informed trading effect on trading volume for Taiwan, we modify Equation (2) in such a way that US returns are decomposed into two components for Taiwan: (1) US returns associated with the high volatility of US returns and (2) those not associated. A positive relation between current volume and past US returns related to high US market volatility is interpreted by the informed trading effect, while the relation unrelated to high US market volatility is interpreted by investor overconfidence. We use similar specifications to control for the participation and informed trading effects for Hong Kong and Singapore and for Taiwan, respectively, to see whether we can find some evidence of investors' overconfident trading behavior in these three stock markets.

We use the sum of the coefficients on lagged domestic (US) returns as the measure of cumulative overconfident trading caused by domestic (US) market gains. The greater the sum is, the higher the degree of overconfident trading. However, some of the coefficients may not be significant at conventional levels. To avoid the potential problem of over-parameterization and to obtain a parsimonious model, financial economists tend to adopt a "general-to-simple" strategy, in which one begins with a general specification and then simplifies it after testing parameter estimates (Ericsson and Marquez, 1993).

Based on this empirical strategy, we use a two-step procedure to estimate the SUR model of Equation (2) and others. For instance, in the first step, the SUR model of Equation (2) is estimated as usual. In the second step, the SUR model estimated from the first-step procedure is re-estimated by dropping all insignificant variables at the 10% significance level.3 The resulting SUR model contains the variables, except for the constant terms, that are significant at least at the 10% level. This assures that our measure of (cumulative) overconfident trading is free from the concern that the influence of the insignificant coefficients on the measure may distort and blur the results. Then, we compare the relative degrees of overconfident trading of ss versus sc investors based on the relative magnitude of the measure of their overconfident trading. If, for example, we find that the sum of the significant [[beta].sub.ij] (or [[beta].sub.ssj]) coefficients divided by the number of the ss markets with the significant [[beta].sub.ij] (or [[beta].sub.ssj]) coefficients (i.e., the average degree of ss investors' overconfident trading due to domestic market gains) is significantly smaller than that of the significant [[beta].sub.ij] (or [[beta].sub.scj]) coefficients divided by the number of the sc markets with the significant [[beta].sub.ij] (or [[beta].sub.scj]) coefficients (i.e., the average degree of sc investors' overconfident trading due to domestic market gains), it provides evidence that the degree of ss investors' overconfident trading due to domestic market gains is less than that of sc investors'.

Table II reports the estimation results of Equation (2). (4) The results show that the sum of the beta coefficients that are significant at least at the 10% level is positive for the three ss countries of Japan (JA), Malaysia (MA), and Thailand (TH) and for all sc countries. This provides evidence that both ss and sc investors trade with overconfidence after domestic market gains. Then we further compare the relative degree of overconfident trading of ss versus sc investors. This is tested based on the chi-squared statistic [[chi square].sub.[beta]] under the null hypothesis that the average of the three sums of the beta coefficients across JA, MA, and TH (i.e., [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ssj]/3) is equal to that of the five sums of the beta coefficients across all sc markets (i.e., [[summation].sub.sc] [[summation].sub.j] [[beta].sub.scj]/5). The observation that the average associated with the three ss markets is smaller than that associated with all sc markets and the rejection of the [[chi square].sub.[beta]] statistic at the 1% level jointly indicate that sc investors, on average, trade with more overconfidence after domestic market gains than ss investors.

Looking at the impact of US market gains on Asian investors' trading behavior, Table II shows that the sum of the gamma coefficients that are significant at least at the 10% level is positive for the three ss countries of Hong Kong (HK), Malaysia (MA), and Singapore (SI) and for one sc country, Korea (KO). This offers evidence that ss investors in Hong Kong, Malaysia, and Singapore and sc investors in Korea trade with overconfidence after U.S. market gains. The [[chi square].sub.[gamma]] statistic is used to test the null hypothesis that the average of the three sums of the gamma coefficients across HK, MA, and SI (i.e., [[summation].sub.ss] [[summation].sub.k] [[gamma].sub.ssk]/3) is equal to the sum of the gamma coefficients for KO (i.e., [[summation].sub.k] [[gamma].sub.sck]). The observation that the [[chi square].sub.[gamma]] statistic is not significant indicates that the degree of overconfident trading due to US market gains exhibits no significant difference across ss investors in Hong Kong, Malaysia, and Singapore and sc investors in Korea. In addition to Table II, we also find little difference in active trading subsequent to US market gains between Asian investors in markets with and without short-sale constraints in the tables that follow with the exception of Table VI. This is probably because short sales are allowed and practiced in the US stock market where stocks are less overpriced incorporating pessimistic investors' valuation.

It is noted from Table II (and the following tables) that, on average, the response of volume to US returns is smaller than that to domestic returns. This makes sense as local market gains should encourage Asian investors to make more overconfident trading in their domestic markets than US market gains.

B. The Effect of Market Integration

Asian financial markets have become increasingly integrated over time with global markets and each other. For example, Jeon, Oh, and Yang (2006) find that the integration of Asian stock markets with US stock markets has become increasingly stronger in recent years, particularly after the Asian financial crisis in 1998. Such integration would encourage Asian investors to trade more heavily in US-based equity investments and pay more attention to US stock markets. We collect the monthly data of the dollar amount of US stocks traded by investors in each of our sample countries from the US Treasury International Capital (TIC) reporting system. In unreported results, we find that the time series of the dollar amount of US stocks traded by Asian investors exhibits a significant linear time trend and that the mean of the dollar amount is significantly higher after 1998 for all sample countries. More importantly, the increasing integration also makes the price movements in the US stock markets more useful in predicting the price movements in Asian stock markets. As a consequence, US market gains may contribute more to the global component of Asian investors' overconfidence. Therefore, we conjecture that the degree of Asian investors' overconfident trading due to US market gains would be more significant after the Asian financial crisis. It is not clear, however, whether such integration would exert its impact on Asian investors' overconfident trading prompted by domestic market gains.

Some researchers also found that the integration of Asian stock markets with US stock markets increased during the global financial market turmoil caused by the subprime mortgage crisis in July 2007 (Diebold and Yilmaz, 2009), which suggests that Asian investors' overconfident trading due to US market gains might also be affected by this episode. However, the local or regional factors such as the Asian financial crisis are expected to affect Asian investors' overconfident trading more than the global factors such as the recent global financial market turmoil.

To examine our conjecture, we divide our sample into three subperiods: (1) the precrisis period (January 1995-December 1998), (2) the postcrisis period (January 1999-June 2007), and (3) periods during and after the global financial market turmoil (July 2007-December 2010). We estimate the following multivariate SUR model over the full sample period by regressing [AV.sub.it] on two control variables, [DAR.sub.it] and [DUR.sub.t], and the lagged values of [AR.sub.it] and [UR.sub.t] with two dummy variables, S[2.sub.t] and S[3.sub.t]:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (3)

where the variables are defined as before and the dummy variable S[2.sub.t](S[3.sub.t]) takes a value of one for the subsample period from January 1999 to June 2007 (from July 2007 to December 2010), and zero otherwise.

In Equation (3), the [[beta].sub.i1j], [[beta].sub.i2j], and [[beta].sub.i3j]([[gamma].sub.i1k], [[gamma].sub.i2k] and [[gamma].sub.i3k]) coefficients measure the relation between lagged domestic (US) returns and current domestic volume in the precrisis period, the postcrisis period, and periods during and after the global financial market turmoil, respectively, for market i. A sharp decline in US and Asian stock markets during the global financial tsunami period is not a good environment for Asian investors to nourish their overconfidence. Consequently, it is reasonable to expect that Asian investors' overconfident trading might not be as evident during this period, although the market linkages between Asian and US stock markets increase. As such, we focus on finding evidence to determine whether Asian investors' overconfident trading due to US market gains is stronger after the Asian financial crisis than before. If our conjecture is correct, we expect to see that the sum of the [[gamma].sub.i1k] coefficients is significantly smaller than that of the [[gamma].sub.i2k] and [[gamma].sub.i3k] coefficients for market i. To facilitate our discussion, the period between January 1999 and December 2010 is simply referred to as the postcrisis period.

Table III reports the estimation results of Equation (3). Since our subperiod test is designed to determine whether the degree of Asian investors' overconfident trading due to US market gains is higher after the Asian financial crisis than before, our discussion of the results in Table III focuses on whether Asian investors' trading pertaining to US market gains and their trading pertaining to domestic market gains can be interpreted analogously.

Consistent with our expectations, Table III indicates that the sum of the gamma coefficients reveals that investors in Hong Kong, Malaysia, Singapore, Thailand, and Korea trade with overconfidence after US market gains only in the postcrisis period. However, the results show that investors in Japan trade with overconfidence after US market gains only in the precrisis period. It is worth noting that we find evidence in Table III that US market gains encourage investors in Japan and Thailand to trade with overconfidence in the pre- and postcrisis period, respectively, that is not found in the results of Table II for the whole sample period.

In addition, in the precrisis period, only 55 investors in Japan trade more actively after US market gains implying that ss investors trade more overconfidently after US market gains than sc investors in this period. We find that Japanese investors trade in US stocks in the precrisis period much more often than investors in the other Asian countries. This may be one reason for the finding that only Japanese investors' overconfident trading is affected by US market gains in this period. In the postcrisis period, ss investors in all 55 markets, except for Japan, and sc investors in Korea trade more actively after US market gains. The average of the four sums of the [[gamma].sub.ss2k] and [[gamma].sub.ss3k] coefficients across the four ss markets (i.e., HK, MA, SI, and TH) is slightly larger than the sum of the [[gamma].sub.scs2k] and [[gamma].sub.sc3k] coefficients for Korea. However, a formal test based on the [[chi square].sub.[gamma](2k+3k)] statistics, which is used to test the null hypothesis that the average across the four ss markets is equal to the sum for Korea, finds no evidence of a significant difference between the degrees of ss versus sc investors' overconfident trading prompted by US market gains in the postcrisis period.

C. The Impact of US Investor Sentiment

Optimistic investors tend to be overconfident (Odean, 1998b; Hirshleifer, 2001). Baker and Stein (2004) theoretically show that when shorting is relatively costly, sentimental investors are inclined to become overconfident and trade more actively when they are optimistic. Baker et al. (2012) present evidence that sentiment is contagious across markets and, as such, has local and global components. If high sentiment spreads from US investors to Asian investors, then Asian investors will feel more optimistic about their domestic and US stock markets and, consequently, become overconfident in their ability to profit from trading in domestic markets. This suggests that Asian investors will trade more aggressively after domestic and US market gains when global sentiment is high. Unfortunately, we cannot explore this issue as we are unable to construct a global sentiment index as Baker et al. (2012) due to data limitations.

However, we can take an alternative approach to see how US sentiment affects Asian investors' trading after US market gains. We argue that even though sentiment does not necessarily spillover from US investors to Asian investors, high US sentiment may also encourage Asian investors to trade more actively after US market gains. As argued by Baker et al. (2012), since it is this optimism that leads stocks to be overvalued, at least in the short run, high US sentiment implies that US investors feel optimistic about future price movement in the US stock markets. Given the predictive power of US stock returns for Asian domestic stock returns, Asian investors may also expect that stock prices will rise in their domestic markets when they are aware of US investors' optimism. This may lead them to further expect that they can gain from their domestic and US-based equity investments, thereby trading more aggressively in domestic markets.

To evaluate the above argument, we use the SUR model to examine the relation between past US returns and current Asian domestic volume conditional on US investor sentiment, proxied by Baker and Wurgler's (2006) monthly US orthogonalized sentiment index that filters out business cycle effects. (5) In this model specification, we define high US investor sentiment as the sentiment index included in the top 30% of its distribution.

To save space, we do not report the estimation results of this test, but they are available from the authors upon request. The most important observation from this test is that consistent with our expectations, we find evidence that investors in Japan, Malaysia, Thailand, and Taiwan trade more aggressively after US market gains only when US investor sentiment is high. We also find that investors in Hong Kong and Indonesia trade more frequently subsequent to US market gains whether US investor sentiment is high or not. Still, we find no evidence that there is a significant difference between the degrees of trading by ss versus sc investors induced by US market gains when US investor sentiment is high.

D. The Impact of Domestic Investor Sentiment

The consumer confidence index is commonly used as a measure of investor sentiment in the literature (Brown and Cliff, 2005; Lemmon and Portniaguina, 2006). As such, we would expect to find that Asian investors will trade more aggressively after domestic market gains when domestic sentiment proxied by the consumer confidence index is high. To explore this issue, we collect the data on the consumer confidence index for China, Hong Kong, Indonesia, Japan, Korea, and Thailand from the Datastream International database and for Taiwan from the Taiwan Economic Journal (TEJ) database. (6)

Since only seven sample countries have the data on the consumer confidence index and their sample periods are different from each other, we use a regression model instead of the SUR model for our analysis. That is, we regress domestic volume on past domestic returns conditional on domestic investor sentiment for each of these countries. In this regression, the conditional variable is a dummy variable that takes a value of one if the consumer confidence index is included in the top 30% of its distribution, and zero otherwise. The regression results, as expected, show that investors in Hong Kong, Indonesia, Japan, and Thailand trade more aggressively after domestic market gains when domestic investor sentiment is high than when it is not. These results provide corroborating evidence for our earlier argument that optimistic investors tend to trade with overconfidence. In addition, the results provide no evidence of a significant difference in the overconfident trading of investors in China, Korea, and Taiwan across investor sentiment states. We do not report these results to conserve space, but they are available upon request from the authors.

E. Overconfident Trading Conditional on Market Conditions

An old Wall Street adage, "Don't confuse brains with a bull market," provides investors with the best warning against becoming overconfident during a bull market. Gervais and Odean (2001) argue that overconfident investors are more likely to trade aggressively and speculatively right after a bull market (see also Odean, 1998b; Daniel, Hirshleifer, and Subrahmanyam, 2001). This implies that if investors trade with overconfidence, they trade more so during bull markets than during nonbull markets, which implies that the positive relation between lagged stock returns and current trading volume should be stronger during bull markets than at other times.

To test this empirical implication, we conduct our analysis of Asian investors' trading conditional on the state of the market. To do so, we need a definition of a bull market for each Asian stock market. The definition of a bull market is, however, somewhat subjective. Following Hardouvelis and Theodossiou (2002), we define a bull market as a period during which there are at least m consecutive positive monthly market returns. The monthly market returns are calculated by averaging weekly market returns. The horizon m of our analysis takes three possible values, m = 3, 4, and 5. We estimate the following SUR conditional on the market states:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (4)

where the variables are defined as before and the dummy variable [AB.sub.it] ([UB.sub.t]) takes a value of one if week t is included in the period of a bull market that is defined as m consecutive positive monthly market returns for market i (US stock market), and zero otherwise.

In Equation (4), the [[beta].sub.i1j]([[gamma].sub.i1k]) and [[beta].sub.i2j]([[gamma].sub.i2k]) coefficients measure the relation between lagged domestic (US) returns and current domestic volume when market i (US stock market) is in bull and in nonbull markets, respectively. If investors trade with more overconfidence in bull markets than in nonbull markets for market;, we expect to observe that the sum of the [[beta].sub.i1j]([[gamma].sub.i1k]) coefficients is significantly greater than that of the [[beta].sub.i2j]([[gamma].sub.i2k]) coefficients.

Table IV reports the estimation results of Equation (4). They are obtained using the bull dummies of [AB.sub.it] and [UB.sub.t] using m = 4. We do not report the results using m = 3 or 5 in Table IV as the results using these two alternative measures of the bull markets are virtually the same. Specifically, Table IV reports four test statistics. First, the [[chi square].sub.[beta]] (1j vs. 2j) statistic is used to test the null hypothesis that the sum of the [[beta].sub.i1j] coefficients is equal to that of the [[beta].sub.i2j] coefficients (i.e., [[summation].sub.j] [[beta].sub.i1j]= [[summation].sub.j] [[beta].sub.i2j]) for each market i. Second, the [[chi square].sub.[gamma]](1k vs. 2k) statistic is used to test the null hypothesis that the sum of the [[gamma].sub.i1k] coefficients is equal to that of the [[gamma].sub.i2k] coefficients (i.e., [[summation].sub.k] [[gamma].sub.i1k] = [[summation].sub.k] [[gamma].sub.i2k]) for HK, MA, SI, and KO. Third, the [[chi square].sub.[beta](1j)] statistic is used to test the null hypothesis that the average of the five sums of the [[beta].sub.ss1j] coefficients across all ss markets is equal to that of the five sums of the [[beta].sub.ss1j] coefficients across all sc markets (i.e., [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ss1j]/5 = [[summation].sub.sc] [[summation].sub.j] [[beta].sub.ss1j]/5). Fourth, the [[chi square].sub.[gamma](1k)] statistic is used to test the null hypothesis that the average of the three sums of the [[gamma].sub.ss1k] coefficients across HK, MA, and, SI is equal to the sum of the [[gamma].sub.sc1k] coefficients for KO (i.e., [[summation].sub.ss] [[summation].sub.k] [[gamma].sub.ss1k]/3 = [[summation].sub.k] [[gamma].sub.sc1k]).

Several noteworthy observations emerge from Table IV. First, we find that [[summation].sub.j] [[beta].sub.i1j] > [[summation].sub.j] [[beta].sub.i2j] and that the [[chi square].sub.[beta]] (1j vs. 2j) statistic rejects the null hypothesis at conventional levels for all of the sample countries, with the exception of Malaysia. These results underscore the idea that Asian investors trade more actively after their domestic market gains when their domestic markets are in bull markets than when they are in nonbull markets. Second, [[summation].sub.k] [[gamma].sub.i1k] is positive and significant at conventional levels and no [[summation].sub.k] [[gamma].sub.i2k] is positive for Hong Kong, Malaysia, Singapore, and Korea. This implies that investors in these countries trade more frequently after US market gains only when the US market is in a bull market.

Third, the finding that [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ss1j]/5 < [[summation].sub.sc] [[summation].sub.j] [[beta].sub.sc1j]/5 with the significant [[chi square].sub.[beta](1j)] statistic at the 5% level indicates that the degree to which sc investors trade more actively after their domestic market gains when their domestic markets are in bull markets is stronger than that of ss investors. Fourth, the finding that [[summation].sub.ss] [[summation].sub.k] [[gamma].sub.ss1k]/3 < [[summation].sub.k] [[gamma].sub.sc1k] with an insignificant [[chi square].sub.[gamma](1k)] statistic indicates that there is little difference in trading between investors in three ss markets and investors in Korea in their response to US market gains when the US stock market is in a bull market.

F. Overconfident Trading Conditional on Extremely High Market Returns

Psychologists find that people tend to be overconfident about their judgments and decisions based on attention-grabbing information. In the finance literature, Barber and Odean (2008) find that individual investors are inclined to buy attention-grabbing stocks like stocks in the news, stocks with extreme returns, and stocks experiencing high abnormal trading volume. Seasholes and Wu (2007) also find that individual investors on the Shanghai Stock Exchange are net buyers the day after a stock hits an upper price limit. These findings imply that in the framework of the Gervais and Odean (2001) overconfidence hypothesis, extremely high returns may encourage investors to become overconfident about their ability to increase their wealth easily and quickly by active trading. This further implies that investors will trade more actively after market gains when the stock market experiences extremely high returns than when it does not. It is important to note that if we find supportive evidence for this implication, it is in conflict with the implication of the disposition effect as disposition investors are eager to sell their holding stocks when they can realize small positive returns.

We define the extremely high market returns as those in the top 10% of its distribution. To see how attention-grabbing information like extremely high market returns affects Asian investors' trading behavior, we estimate the following SUR model conditional on extremely high market returns:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)

where the variables are defined as before and the dummy variable [EAR.sub.it] ([EUR.sub.t]) takes a value of one for week t when the return for market i (US stock market) belongs to the top 10% of its distribution, and zero otherwise. In Equation (5), the [[beta].sub.i1j] [[gamma].sub.i1j] and [[beta].sub.i2j] [[gamma].sub.i2k) coefficients measure the relation between lagged domestic (US) returns and current domestic volume when market i (US stock market) experiences extremely high returns and when it does not, respectively.

Table V reports the estimation results of Equation (5). (7) As in Table IV, Table V reports four test statistics. First, we use the [[chi square].sub.[beta]] (1j vs. 2j) statistic to test the null hypothesis that [[summation].sub.j] [[beta].sub.i1j] = [[summation].sub.j] [[beta].sub.i2j] for all sample markets, except for Hong Kong. Second, we use the [[chi square].sub.[gamma]] (1k vs. 2k) statistic to test the null hypothesis that [[summation].sub.k] [[gamma].sub.i1k] = [[summation].sub.k] [[gamma].sub.i2k] for all sample markets, except for Thailand, China, and Indonesia. Third, we use the [[chi square].sub.[beta](1j)] statistic to test the null hypothesis that [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ss1j]/4 = [[summation].sub.sc] [[summation].sub.j] [[beta].sub.sc1j]/5 across all ss markets, except for Hong Kong, and across all sc markets. Fourth, we use the [[chi square].sub.[gamma](1k)] statistic to test the null hypothesis that [[summation].sub.ss] [[summation].sub.k] [[gamma].sub.ss1k]/3 = [[summation].sub.sc] [[summation].sub.k] [[gamma].sub.sc1k]/3 across three ss markets (i.e., HK, MA, and SI) and across three sc markets (i.e., KO, PH, and TA).

Some important observations are made from Table V. First, we find that the sum of the [[beta].sub.i1j] coefficients is greater than that of the [[beta].sub.i2j] coefficients and the [[chi square].sub.[beta]](1j vs. 2j) statistic rejects the null hypothesis at conventional levels for Singapore, Thailand, China, Indonesia, and Korea. These findings imply that investors in these countries trade more aggressively after their domestic markets experience extremely high returns than when they do not.

Second, the observation that Hong Kong, Malaysia, Singapore, Korea, the Philippines, and Taiwan have the positive, significant sum of the [[gamma].sub.i1k] coefficients indicates that investors in these markets respond only to extremely high returns in the US stock market to trade more aggressively in subsequent periods. However, Japan has a positive, significant sum of the [[gamma].sub.i2k] coefficients, which is inconsistent with our hypothesis.

Third, the result that the [[chi square].sub.[beta](1j)] statistic rejects the null hypothesis at the 1% level, together with the result that the average of the four sums of the [[beta].sub.ss1j] coefficients across the four ss markets is smaller than that of the five sums of the [[beta].sub.sc1j] coefficients across all sc markets, indicates that extremely high returns in domestic markets encourage sc investors to trade more heavily than ss investors. Fourth, the insignificance of the [[chi square].sub.[beta](1k)] statistic indicates that there is little difference between ss and sc investors' trading in their response to extremely high returns in the US stock market.

III. Robustness Checks

A. Alternative Theories

As noted by Griffin et al. (2007), five alternative theories have been proposed for a positive relation between lagged returns and current volume. We address these theories in turn below in order to generate testable implications from them. Then, we conduct empirical tests to examine whether these theories provide important determinants of the positive return-volume relation. To conserve space, we do not report the test results of these alternative theories, but briefly discuss what we find from these tests. These results are also available upon request.

B. Participation Effect

Changes in market participation can be an important determinant of the return-volume relation. For example, Orosel (1998) develops a theory of rational trend chasing in which high market returns lead investors who do not participate in the stock market to increase their expectations of the profitability from market participation. In equilibrium, participation in the stock market rises following an up market and falls following a down market, and fluctuations in market participation increase the volatility of stock prices. As argued by Griffin et al. (2007), the key point of this result is that good news for the stock market induces investors to participate more actively in the market. This is especially true when the news is more difficult to interpret and there is more uncertainty about future fundamentals. Consistent with this prediction, Hoffmann, Post, and Pennings (2012) offer evidence that investors with high levels and upward revisions of return expectations tend to be less risk averse and, consequently, trade more actively even when the market is more volatile. As such, we would expect participation effects to be stronger in up markets with high volatility.

To test the participation effect, we run the multivariate SUR model conditional on up markets with high volatility across the ten sample countries. (8) We find that investors in Hong Kong and Singapore trade more following local market gains in up markets with high volatility, which is consistent with the implication of the participation effect. As such, we have controlled for this effect only for Hong Kong and Singapore when examining the trading behavior of investors in these two countries in Section II.

C. Informed Trading and Short-Sale Costs

Wang (1994) shows that informed traders tend to trade gradually in order to prevent their private information from being revealed too quickly. Consequently, private information will be incorporated into prices slowly, so that increased buying activity by informed traders will follow price increases when they acquire favorable private information. Admati and Pfleiderer (1988) show that both trading volume and asset volatility increase with informed trading and that informed traders prefer to trade when trading costs are low in the sense of high volume and price variability. Diamond and Verrecchia (1987) show that the speed at which adverse information is incorporated into prices will be slower than favorable information when short sales are prohibited. Based on this concept, Griffin et al. (2007) argue that fewer trades occur when informed traders receive adverse information, even though short sale is not prohibited but is simply more costly than buying trades. Consequently, it is expected that the return-volume relation is stronger when the market is more volatile as long as short sales are costly.

We estimate the multivariate SUR model conditional on high market volatility across the ten sample countries to test the prediction based on the informed trading theory, together with short sale costs. (9) Our test here focuses on whether investors trade more actively after market gains in high-volatility states than in low- and medium-volatility states. We find that Taiwanese investors trade more actively following US market gains when the US stock market is in high-volatility market states than when it is in the other two states, which is consistent with the implication of the informed trading theory when taking into account short sale costs. Therefore, this effect has been controlled only for Taiwan when we investigate Taiwanese investors' trading behavior in Section II.

D. Momentum Trading by Uninformed Traders and Short-Sale Restrictions

If prices adjust gradually to new information, past price movements may contain signals about the private information of informed traders (Grossman and Stiglitz, 1980). Under this circumstance, uninformed traders will find it optimal to adopt trend-following (momentum) trading strategies. For example, Brennan and Cao (1997) present a theoretical model and empirical evidence in support of the view that foreign investors should pursue momentum strategies due to their informational disadvantage relative to domestic investors. Since it costs more to take a short position than a long position, it is expected that uninformed trend followers will trade more in up markets than in down markets.

In a widely cited study, DeLong, Shleifer, Summers, and Waldmann (1990) develop a positive feedback trading model in which rational speculators jump on the bandwagon rather than buck the trend. In their model, trading by uninformed positive feedback traders, as well as rational speculators, moves prices away from fundamental values, which further triggers more positive feedback trading for which the decision to trade is conditional on past price movements. Consequently, their model implies a positive causal relation between returns and volume in either direction. Jennings, Starks, and Fellingham (1981) extend Copeland's (1976) sequential information arrival model by incorporating real world margin constraints and short sales. Due to the sequential information flow, investors become informed in a sequential pattern and trade in the same direction. Their model implies that trading volume that is generated when a previously uninformed trader responds pessimistically to the news is less than that when the trader responds optimistically. Moreover, the sequential information arrival process allows trading volume to have predictive power for returns and vice versa. In sum, in the presence of a short-sale restriction, the trend-following trading strategies adopted by uninformed traders imply a bi-directional causal relation between returns and volume (Chuang and Lee, 2006; Chuang and Susmel, 2011).

We perform trivariate Granger-causality tests to test the momentum trading theory. The results provide no evidence of the causality running from volume to returns for any sample countries, even though we do find evidence of the causality running from returns to volume for all sample countries, except for Singapore. Moreover, in the presence of short-sale restrictions, the momentum trading theory implies that there is less trading following negative returns (Griffin et al., 2007). To test this implication, we examine the return-volume relation conditional on the sign of past returns and compare whether negative past returns result in a larger magnitude of volume decreases than volume increases associated with positive past returns. Inconsistent with this prediction, we do not find that this is the case for our sample countries. Instead, we find that the response of volume to positive past returns is larger in magnitude than that to negative past returns for a few countries. These additional results, together with no bi-directional causal relation, suggest that the momentum trading theory, taking into account short-sale restrictions, may not provide an explanation for the return-volume relation for our sample countries.

E. Liquidity Effect

The notion that market liquidity declines after large negative market returns has received much attention in the recent literature (Hameed, Kang, and Viswanathan, 2010). A positive relation between lagged returns and current volume may arise from this phenomenon. For example, in the coordination failure model of Bernardo and Welch (2004), a significant decrease in stock prices makes market makers, as liquidity providers, more risk averse and increases the cost and risk of providing liquidity as they become more capital constrained. In this circumstance, the fear of future adverse liquidity shocks will lead to a decline in their willingness to provide liquidity (Jones and Lipson, 2001). Suppose that market makers naturally have net long positions. As stock prices drop, market makers increase their inventory positions and therefore their inventory-holding cost and risk also increase. As such, they become less willing to provide liquidity by acquiring stocks (Coughenour and Saad, 2004). With these liquidity effects, trading volume falls following a drop in stock prices as the cost and risk of trading increase. The impact of the liquidity effect on trading volume appears to be stronger when stock prices drop sharply. Put differently, the liquidity effect implies that medium and large negative returns lead to a drying up of trading volume.

To test the liquidity effect, we divide returns into three regimes: 1) large negative returns, 2) medium and small negative returns, and 3) positive returns. Then, we examine the return-volume relation conditional on these three regimes. The focus of this test is to determine whether decreases in volume due to large negative past returns are larger than those due to medium and small negative past returns. However, we do not find evidence that the sensitivity of volume to large negative past returns is larger in magnitude than that of volume to medium and small negative past returns for our sample countries. (10) 11 As such, the liquidity effect is not an important cause of the return-volume relation for our sample countries.

F. Disposition Effect

Shefrin and Statman (1985) propose the disposition effect, where investors are predisposed to selling winners too soon and holding losers too long. The disposition effect implies that volume follows returns because investors are eager to lock in gains after an increase in the stock price and reluctant to trade after accruing poor returns. More specifically, it implies that volume will be higher after small positive returns than after large positive returns and negative returns. Odean (1998a) and many others provide evidence in support of the disposition effect. We would expect to see that volume will be higher following small price increases than following large price increases because of the disposition effect.

To test the disposition effect, we divide returns into three regimes: (1) small positive returns, (2) medium and large positive returns, and (3) negative returns. Then, the return-volume relation is examined conditional on these three regimes. The observation that small positive returns lead to larger volume increases than medium and large positive returns will provide evidence in support of the disposition effect. However, we find evidence to the contrary. That is, we find that volume increases more so following medium and large positive returns than following small positive returns for our sample countries, providing evidence against the disposition effect.

G. Individual and Institutional Trading

Institutional investors do not participate in emerging markets as actively as they do in developed markets. Therefore, their trading is expected to be lower than individual trading in Asian stock markets (Bae et al., 2002). Although our results using the market-level data may imply that individual investors display overconfident trading behavior, they do not necessarily imply that this is the case for institutional investors. Moreover, using the market-level data may conceal institutional and individual investors' overconfident trading behavior to a higher and lower degree, respectively.

In this subsection, we attempt to find evidence of overconfident trading behavior in Asian stock markets by looking at institutional and individual investors' trading behavior separately. In addition, our investigation in this subsection has two additional purposes. First, if we find that institutional investors also display overconfident trading behavior, we further compare the degree of overconfident trading of individual versus institutional investors. Additionally, we combine individual and institutional investors' overconfident trading behavior to check the robustness of our results in Section II. For these purposes, we collect the data on trading volume of individual and institutional investors for the six sample countries: Japan, Malaysia, Thailand, Indonesia, Korea, and Taiwan from their stock exchanges. (11) Due to the different lengths in the sample period, if we use the multivariate SUR model across the six sample countries, the sample period will be trimmed from October 2009 to December 2010, leading to a substantial loss of observations. (12) Therefore, we decide to employ the bivariate SUR model for each of the six sample countries using their longest sample period and utilize the multivariate SUR model across the five sample countries, excluding Malaysia, to test the degree of ss versus sc investors' trading after domestic and US market gains from December 2003 to December 2010. (13) Since there are only five sample countries available to conduct such a test, it is expected to see that the results from this test will not be as consistent as the results using the market-level data of the ten sample countries. Moreover, the results from the multivariate SUR model may be slightly different from those obtained from the bivariate SUR model due to different model specifications and sample periods.

To conserve space, we report and discuss only the summary results of the tests using individual and institutional trading data in terms of "Yes" and "No," but the detailed results are available upon request. (14) Table VI reports the summary results of the causal relations in Panel A and those of the causal relations conditional on the market states, on the extremely high market returns, and on the high investor sentiment in Panels B, C, and D, respectively. Each panel contains the summary results of the ten tests, and each test is denoted by a capital letter with the number in the corresponding panel.

Some important results are noted as follows. First, we find that individual as well as institutional investors trade more in bull markets, in periods of high investor sentiment, and in periods of extremely high market returns. For example, the B3 test indicates that individual investors in all six sample countries, except for Thailand, trade more after US market gains in US bull markets than in nonbull markets, and the B4 test finds that institutional investors exhibit a similar trading behavior in Indonesia and Korea. The D3 test shows that individual investors trade more when US investor sentiment is high than when it is not in Korea and Taiwan, while the D3 and D4 tests show that both individual and institutional investors display a similar trading behavior in Japan and Thailand. However, we also find from the C4 test that none of the institutional investors in the six sample countries trade more after the US market experiences extremely high returns than when it does not.

Second, we find that individual investors tend to trade more than institutional investors. For example, the B5 (B6) test shows that domestic (US) market gains push individual investors to trade more than institutional investors in Japan, Thailand, and Taiwan (in Japan and Taiwan) in domestic (US) bull markets. The D5 (D6) tests indicate that domestic (US) market gains encourage individual investors to trade more than institutional investors when domestic (US) investor sentiment is high in Indonesia, Korea, and Taiwan. However, we also find a few cases where institutional investors trade more than individual investors. For example, in Malaysia and Indonesia institutional investors trade more than individual investors after US market gains and after the US market experiences extremely high returns, as indicated by the A8 and C6 tests, respectively.

Third, we find that most results using the individual and institutional trading data are consistent with the results using the market-level data. For example, consistent with the results of Table II, the A5 and A6 tests show that individual and institutional investors trade more after domestic market gains in the six sample countries and after US market gains in Malaysia and Korea, respectively. The B7 test reports that individual and institutional investors trade more after domestic market gains in domestic bull markets than in domestic nonbull markets in the six sample countries, except for Malaysia, which is consistent with the findings in Table IV. The results of the C7 (C8) test regarding whether individual and institution investors trade more after the domestic (US) market experiences extremely high returns than when it does not are consistent with what we observe in Table V in all cases, except for the case of Thailand (Malaysia). Also consistent with the results documented in Sections II.C and II.D, the D7 test shows that individual and institutional investors trade more after domestic market gains when domestic investor sentiment is high than when it is not in Japan, Thailand, and Indonesia.

Yet, we find a few cases that are inconsistent with the previous results. For example, the A6 test shows that individual and institutional investors trade more after US market gains in Indonesia, which is inconsistent with the results of Table II. We note that this result is due to the fact that institutional investors also engage in overconfident trading after US market gains, as revealed in the A4 test. The B8 test reports that individual and institutional investors trade more after US market gains in US bull markets than in US nonbull markets in Indonesia and Taiwan. This is because in Table VI, the B1 and B2 tests and the B1 test show, respectively, that both individual and institutional investors in Indonesia and individual investors in Taiwan trade more after US market gains in US bull markets than in US nonbull markets.

Fourth, as for the degree of ss versus sc investors' trading conditional on specific events, the test statistics of the A9, B9, and C9 tests are all significant at least at the 5% level, indicating that sc investors, on average, trade more than ss investors after domestic market gains, after domestic market gains conditional on the market states, and after domestic market gains conditional on the extremely high market returns, respectively. These results are consistent with those in Tables II, IV, and V Also consistent with the results in Table II and in Section II.C, the test statistics of the BIO and DIO tests are not significant at conventional levels indicating that sc investors, on average, do not trade more than ss investors after US market gains in US bull markets and after US market gains when US investor sentiment is high, respectively.

Inconsistent with the results in Tables II and V, however, the significant test statistics of the A10 and CIO tests indicate that sc investors, on average, trade more than ss investors after US market gains and after the US market experiences extremely high returns, respectively. These two results arise from the fact that sc investors, but not ss investors, trade more in these two situations. In Section II.D, we do not perform the test of the degree of ss versus sc investors' trading due to the use of the regression model for each sample country. Based on the significant test statistic of the D9 test, we now find that sc investors typically trade more than ss investors after domestic market gains when domestic investor sentiment is high. However, we find no evidence that sc investors, on average, trade more overconfidently than ss investors after US market gains when US investor sentiment is high since the test statistic of the DIO test is not statistically significant.

In sum, using the individual and institutional trading data, we find evidence that both US and domestic market gains encourage Asian individual and institutional investors to trade with more overconfidence in bull markets, in periods of high investor sentiment, and in periods of extremely high market returns. We also find evidence that individual investors tend to trade more overconfidently than institutional investors in Asian stock markets. More importantly, we note that most results are consistent with the previous results using the market-level data. Indeed, the results presented in this subsection provide more evidence on overconfident trading behavior in Asian stock markets.

H. The Impact of US Capital Flows

US capital flow may play a dual role in our tests. First, as argued by Baker et al. (2012), US capital flow helps spread the optimistic sentiment from US investors to Asian investors, which further nurtures Asian investors' overconfidence and is reflected in their trading. This lends support to our hypothesis. Second, prior studies find that US investors tend to move their capital into markets that have recently experienced good returns (Froot, O'Connell, and Seasholes, 2001). In this case, if US investors adopt this return-chasing strategy after US market gains, then the observed relation between past US returns and the current Asian trading volume may be driven by capital flows from the US to Asian countries, which is unrelated to Asian investors' overconfident trading.

To evaluate whether US capital flow provides an alternative explanation for our findings, we collect the monthly data of portfolio equity flow from the US to our sample Asian countries from the US TIC reporting system. Following Edison and Warnock (2008), the capital flow from the US to an Asian country is defined as the gross purchases of foreign securities by US residents from residents of that country minus the gross sales of foreign securities from US residents to residents of that country. As before, we replicate our tests in Section II, using lagged capital flows as control variables. (15) Using the monthly data, we find that lagged capital flows have an impact on domestic trading volume for only a few sample stock markets. More importantly, all of our conclusions drawn from the results in Section II remain unchanged. This implies that US capital flow is not a driving force in our findings in Section II. To conserve space, we do not report the results of these robustness tests, but they are available upon request.

IV. Concluding Remarks

Evidence on investors' overconfident trading motivated by domestic market gains is documented in previous studies. Yet, to date, there is little empirical evidence that investors' trading in domestic markets may also be prompted by foreign market gains, in spite of the fact that Asian investors' trading in US stocks has been increasing recently. This article takes a step toward filling this void in the literature by showing that in addition to domestic market gains, US market gains also encourage Asian investors to trade more. The results of our subperiod analysis suggest that the stronger integration of Asian stock markets with US stock markets after the Asian financial crisis is an important reason for US market gains encouraging Asian investors' overconfident trading behavior in their own domestic markets.

Moreover, our study is the first to empirically document evidence that investors' trading increases with short-sale constraints when their trading is induced by their domestic market gains. This finding has important policy implications in that investors' irrational trading can be reduced, to a lower degree, by removing short-sale constraints. Indeed, we find that our sample stock markets without short-sale constraints are more stable than those with short-sale constraints.

The theory suggests that investors' overconfident trading will be more pronounced when investors are more optimistic and when the market is in a bull market and experiences extremely high returns. We are the first to examine the impact of investor sentiment and extremely high market performance on investors' trading. We test Asian investors' trading conditional on these events. In general, we find evidence that domestic and US market gains encourage Asian individual and institutional investors to trade with more overconfidence in these situations. This sheds additional light on understanding the causes and complexities of investors' trading and their corresponding overconfident trading behavior. We also find that individual investors tend to trade more than institutional investors in Asian stock markets.

Considerable attention has been focused on the impact of local factors on investors' sentiment and trading behavior. Until recently, there has been little emphasis on whether investors' sentiment and trading behavior are affected by factors outside the country. This issue is important as it helps build a more comprehensive understanding of the consequence of investors' sentiment and trading behavior. Baker et al. (2012) provide evidence that sentiment spreads across markets; that is, the sentiment of investors in one country can be affected by that of investors in another country. We believe the issues concerning how cross-market factors affect investors' sentiment, as well as trading behavior, remain important topics for future research.

References

Admati, A.R. and P. Pfleiderer, 1988, "A Theory of Intraday Patterns: Volume and Price Variability," Review of Financial Studies 1, 3-40.

Akaike, H., 1974, "A New Look at the Statistical Identification Model," IEEE Transactions on Automatic Control 19, 716-723.

Bae, K.H., J.K. Kang, and J.M. Kim, 2002, "Tunneling or Value Added? Evidence from Mergers by Korean Business Groups," Journal of Finance 57, 2695-2740.

Baker, M. and J.C. Stein, 2004, "Market Liquidity as a Sentiment Indicator," Journal of Financial Markets 7, 271-299.

Baker, M. and J. Wurgler, 2006, "Investor Sentiment and the Cross-Section of Stock Returns," Journal of Finance 61, 1645-1680.

Baker, M., J. Wurgler, and Y. Yuan, 2012, "Global, Local, and Contagious Investor Sentiment," Journal of Financial Economics 104, 272-287.

Barber, B.M. and T. Odean, 2008, "All That Glitters: The Effect of Attention and News on the Buying Behavior of Individual and Institutional Investors," Review of Financial Studies 21, 785-818.

Bekaert, G. and M.S. Urias, 1996, "Diversification, Integration and Emerging Market Closed-End Funds," Journal of Finance 51, 835-869.

Bernardo, A. and I. Welch, 2004, "Liquidity and Financial Market Runs," Quarterly Journal of Economics 119, 135-158.

Brennan, M.J. and H.H. Cao, 1997, "International Portfolio Investment Flows," Journal of Finance 52, 1851-1880.

Bris, A., W. Goetzmann, and N. Zhu, 2007, "Efficiency and the Bear: Short Sales and Markets Around the World," Journal of Finance 62, 1029-1079.

Brown, W. and M. Cliff, 2005, "Investor Sentiment and Asset Valuation," Journal of Business 78, 405-439.

Chen, G., K.A. Kim, J.R. Nofsinger, and O.M. Rui, 2007, "Trading Performance, Disposition Effect, Overconfidence, Representativeness Bias, and Experience of Emerging Market Investors," Journal of Behavioral Decision Making 20, 425-451.

Chuang, W.I. and B.S. Lee, 2006, "An Empirical Evaluation of the Overconfidence Hypothesis," Journal of Banking and Finance 30, 2489-2515.

Chuang, W.I. and R. Susmel, 2011, "Who is the More Overconfident Trader? Individual vs. Institutional Investors," Journal of Banking and Finance 35, 1626-1644.

Copeland, T.E., 1976, "A Model of Asset Trading Under the Assumption of Sequential Information Arrival," Journal of Finance 31, 1149-1168.

Coughenour, J.F. and M.M. Saad, 2004, "Common Market Makers and Commonality in Liquidity," Journal of Financial Economics 73, 37-69.

Daniel, K., D. Hirshleifer, and A. Subrahmanyam, 1998, "Investor Psychology and Security Market Under-and Overreactions," Journal of Finance 53, 1839-1886.

Daniel, K., D. Hirshleifer, and A. Subrahmanyam, 2001, "Overconfidence, Arbitrage, and Equilibrium Asset Pricing," Journal of Finance 56, 921-965.

De Long, J.B., A. Shleifer, L. Summers, and R.J. Waldmann, 1990, "Positive Feedback Investment Strategies and Destabilizing Speculation," Journal of Finance 45, 379-395.

Diamond, D.W. and R.E. Verrecchia, 1987, "Constraints on Short-Selling and Asset Price Adjustment to Private Information," Journal of Financial Economics 18, 277-312.

Dickey, D.A. and W.A. Fuller, 1979, "Distribution of the Estimators for Autoregressive Time Series with a Unit Root," Journal of the American Statistical Association 74, 427-431.

Diebold, F.X. and K. Yilmaz, 2009, "Measuring Financial Asset Return and Volatility Spillovers, with Application to Global Equity Markets," Economic Journal 119, 158-171.

Ding, Z, C.W.J. Granger, and R.F. Engle, 1993, "A Long Memory Property of Stock Market Returns and a New Model," Journal of Empirical Finance 1, 83-106.

Edison, H.J. and F.E. Warnock, 2008, "Cross-Border Listings, Capital Controls, and Equity Flows to Emerging Markets," Journal of International Money and Finance 27, 1013-1027.

Ericsson, N.R. and J. Marquez, 1993, "Encompassing the Forecasts of US Trade Balance Models," Review of Economics and Statistics 75, 19-31.

French, K.R. and J.M. Poterba, 1991, "Investor Diversification and International Equity Markets," American Economic Review 81, 222-226.

Froot, K., P. O'Connell, and M. Seasholes, 2001, "The Portfolio Flows of International Investors," Journal of Financial Economics 59, 151-193.

Fuller, W.A., 1976, Introduction to Statistical Time Series, New York, NY, John Wiley.

Gallant, A.R., P.E. Rossi, and G. Tauchen, 1992, "Stock Prices and Volume," Review of Financial Studies 5, 199-242.

Gervais, S. and T. Odean, 2001, "Learning to be Overconfident," Review of Financial Studies 14, 1-27.

Glosten, L.R., R. Jagannathan, and D. Runkle, 1993, "On the Relation Between the Expected Value and the Volatility of the Nominal Excess Return on Stocks," Journal of Finance 48, 1779-1801.

Granger, C.W.J., 1969, "Investigating Causal Relations by Econometric Models and Cross-Spectral Methods," Econometrica 37, 424-438.

Griffin, J.M., F. Nardari, and R.M. Stulz, 2007, "Do Investors Trade More When Stocks Have Performed Well? Evidence from 46 Countries," Review of Financial Studies 20, 905-951.

Grossman, S. and J. Stiglitz, 1980, "On the Impossibility of Informationally Efficient Markets," American Economic Review 70, 393-408.

Hamao, Y., R. Masulis, and V Ng, 1990, "Correlations in Price Changes and Volatility Across International Stock Markets," Review of Financial Studies 3, 281-308.

Hameed, A., W. Kang, and S. Viswanathan, 2010, "Stock Market Declines and Liquidity," Journal of Finance 65, 257-293.

Hardouvelis, G.A. and P. Theodossiou, 2002, "The Asymmetric Relation Between Initial Margin Requirements and Stock Market Volatility Across Bull and Bear Markets," Review of Financial Studies 15, 1525-1559.

Hirshleifer, D., 2001, "Investor Psychology and Asset Pricing," Journal of Finance 56, 1533-1597.

Hoffmann, A.O.I., T. Post, and J.M.E. Pennings, 2012, "Individual Investor Perceptions and Behavior During the Financial Crisis," Journal of Banking and Finance 37, 60-74.

Hong, H., J. Scheinkman, and W. Xiong, 2006, "Asset Float and Speculative Bubbles," Journal of Finance 61, 1073-1117.

Jennings, R., L. Starks, and J. Fellingham, 1981, "An Equilibrium Model of Asset Trading with Sequential Information Arrival," Journal of Finance 36, 143-161.

Jeon, J., Y. Oh, and D.Y. Yang, 2006, "Financial Market Integration in East Asia: Regional or Global?" Asian Economic Papers 5, 73-89.

Jones, C. and M. Lipson, 2001, "Sixteenths: Direct Evidence on Institutional Execution Costs," Journal of Financial Economics 59, 253-278.

Kim, K.A. and R. Nofsinger, 2007, "Behavior of Japanese Individual Investors During Bull and Bear Markets," Journal of Behavioral Finance 8, 138-153.

Kyle, A.S. and F.A. Wang, 1997, "Speculation Duopoly with Agreement to Disagree: Can Overconfidence Survive the Market Test?" Journal of Finance 52, 2073-2090.

Lemmon, M., and E. Portniaguina, 2006, "Consumer Confidence and Asset Prices: Some Empirical Evidence," Review of Financial Studies 19, 1499-1529.

Lin, W.L., R.F. Engle, and T. Ito, 1994, "Do Bulls and Bears Moves Across Borders? International Transmission of Stock Returns and Volatility," Review of Financial Studies 7, 507-538.

Lo, A.W. and J. Wang, 2000, "Trading Volume: Definition, Data Analysis, and Implications of Portfolio Theory," Review of Financial Studies 13, 257-300.

Miller, D.T. and M. Ross, 1975, "Self-Serving Bias in Attribution of Causality: Fact or Fiction?" Psychological Bulletin 82, 213-225.

Odean, T., 1998a, "Are Investors Reluctant to Realize Their Losses?" Journal of Finance 53, 1775-1798.

Odean, T., 1998b, "Volume, Volatility, Price, and Profit When All Traders Are Above Average," Journal of Finance 53, 1887-1934.

Orosel, G.O., 1998, "Participation Costs, Trend Chasing, and Volatility of Stock Prices," Review of Financial Studies 11, 521-557.

Pagan, A.R., and G.W. Schwert, 1990, "Alternative Models for Conditional Stock Volatility," Journal of Econometrics 45, 267-290.

Ross, S., 1989, "Information and Volatility: The No-Arbitrage Martingale Approach to Timing and Resolution Irrelevancy," Journal of Finance 44, 1-18.

Scheinkman, J.A., and W. Xiong, 2003, "Overconfidence and Speculative Bubbles," Journal of Political Economy 111, 1183-1219.

Seasholes, M. and G. Wu, 2007, "Predictable Behavior, Profits, and Attention," Journal of Empirical Finance 14, 590-610.

Shefrin, H. and M. Statman, 1985, "The Disposition to Sell Winners Too Early and Ride Losers Too Long: Theory and Evidence," Journal of Finance 40, 777-791.

Statman, M, S. Thorley, and K. Vorkink, 2006, "Investor Overconfidence and Trading Volume," Review of Financial Studies 19, 1531-1565.

Wang, I, 1994, "A Model of Competitive Trading Volume," Journal of Political Economy 102, 127-168.

Yates, J.F., J.W. Lee, and J.G. Bush, 1997, "General Knowledge Overconfidence: Cross-National Variations, Response Style, and "Reality," Organizational Behavior and Human Decision Process 70, 87-94.

Zellner, A., 1962, "An Efficient Method for Estimating Seemingly Unrelated Regressions and Tests for Aggregate Bias," Journal of the American Statistical Association 57, 348-368.

The authors would like to thank Marc Lipson (Editor), an anonymous referee, Yuanchen Chang, Robin K. Chou, Nicole Choi, and conference participants at 2011 FMA Asian Conference and 2012 Asian Finance Association (Asian FA) Annual Meeting for helpful comments. Wen-I Chuang and Kai-Li Wang gratefully acknowledge the financial support from the National Science Council of the Republic of China (NSC 92-2416-H-029-007-EF). All remaining errors are our own responsibility.

Wen-I Chuang *, Bong-Soo Lee, and Kai-Li Wang

* Wen-I Chuang is an Associate Professor of Finance in the College of Management at National Taiwan University, Taipei, Taiwan. Bong-Soo Lee is a Professor of Finance in the College of Business at Florida State University, Tallahassee, FL, USA. Kai-Li Wang is a Professor of Finance in the College of Management at Tunghai University, Taichung, Taiwan.

(1) The five alternative theories include theories based on the participation effect, the liquidity effect, the disposition effect, and informed and momentum trading. We discuss these theories in detail and their tests in Section III.

(2) In our trivariate Granger-causality tests, we allow for the different lag lengths of the right-hand side variables in each equation. Specifically, the tests of whether domestic and U.S. market returns Granger-cause domestic trading volume are performed after controlling for the impact of information flows emanating from domestic and US stock markets on domestic trading volume (see our discussion in Section II.A for details). Our Granger-causality tests examine not only for the causal relations, but also for their signs. Specifically, the examination of whether the sum of estimated lagged coefficients is greater than zero is used to identify the sign of the causal relations. The causality tests and the sign of the causality are performed using a Wald test.

(3) As a robustness check, we also estimate the models without deleting the insignificant estimated coefficients and conduct the tests with only the significant estimated coefficients. Overall, we find that evidence on Asian investors' overconfident trading is even slightly stronger than what we present in the paper. To conserve space, we do not report the results of this robustness check, but they are available from the authors upon request.

(4) For ease of discussion, in Table II and the tables that follow, we do not report the results of the beta coefficients on lagged domestic returns conditional on positive market returns with high volatility for Hong Kong and Singapore and those on lagged US returns associated with high market volatility for Taiwan since they cannot be interpreted as investors' overconfident trading. However, if we find the results of the return-volume relation that can be interpreted as investors' overconfident trading, we will report them in the tables.

(5) The data on the monthly US orthogonalized sentiment index are obtained from Jeffrey Wurgler's website at www.stem.nyu.edu/~jwurgler. We are grateful to Jeffrey Wurgler and Malcolm Baker for making their sentiment index publicly available.

(6) The sample period begins from January 1995 for China, from the first quarter of 2000 for Hong Kong, from June 2000 for Indonesia, from November 2001 for Japan, from December 1998 for Korea, from July 1999 for Thailand, and from January 1999 for Taiwan. All sample periods end in December 2010.

(7) Alternatively, we also define an extremely high market return as one in the top 5% and 15% of its distribution and find that although the results using these two alternative definitions are slightly different from those using the definition presented in this paper, the conclusions remain the same as those drawn from Table V.

(8) The conditional event is devised as a dummy variable that takes a value of one on week I when the return for the sample stock markets and the US stock market is positive and the conditional market volatility estimated from the Glosten, Jagannathan, and Runkle (1993) GJR-GARCH model is included in the top 30% of its distribution, and zero otherwise.

(9) To capture high market volatility, we define two dummy variables that take a value of one on week I when the conditional market volatility estimated from the GJR-GARCH model for the sample stock markets and the U.S. stock market are in the top 30% and between 30% and 70% of its distribution, and zero otherwise.

(10) Alternatively, we also test the relation between current volume and past negative returns in which negative returns are divided into three regimes: (1) large, (2) medium, and (3) small negative returns. Still, we do not find evidence that large negative past returns decrease volume more than medium and small negative past returns.

(11) In unreported results, we find that, with the exception of Japan, individual trading, as expected, is higher than institutional trading in the other five sample countries. Specifically, institutional trading accounts for 62.50% of the total trading volume in Japan, 43.30% in Malaysia, 14.57% in Thailand, 15.13% in Indonesia, 13.27% in Korea, and 27.46% in Taiwan. Institutional trading accounts for a substantial proportion of the total trading volume in Japan as its stock market is much more developed than the other sample Asian stock markets. Institutional trading accounts for more than 40% of the total trading volume in Malaysia. This is because institutional trading increases over time in most emerging markets and because the sample period for Malaysia is from October 2009 to May 2012 during which institutional trading is higher than in the 1990s and early 2000s.

(12) The sample period for Japan is from January 2000 to December 2010, for Malaysia from October 2009 to May 2012, for Thailand from January 1995 to December 2010, for Indonesia from May 1995 to December 2010, for Korea from December 2003 to December 2010, and for Taiwan from January 2001 to December 2010. It is noted that the sample period for Malaysia ends in May 2012, rather than December 2010. This is because its sample period begins in October 2009. To make all of the tests associated with Malaysia more powerful, we decide to extend its sample period to the most recent data available when we collect the data.

(13) The bivariate SUR model has two equations with the same independent variables for each market, one with the dependent variable of (detrended) individual trading volume and the other one with that of (detrended) institutional trading volume. Specifically, the independent variables in the bivariate SUR model are the same as those in Equations (2), (4), and (5). The multivariate SUR model across the five sample countries is an extension of the bivariate SUR model and contains ten equations, each market with two equations.

(14) We also test whether the pattern of the return-volume relation using the individual and institutional trading data is consistent with the implications of the five alternative theories discussed above. We find that the return-volume relation for institutional trading is consistent with the implication of the informed trading theory taking into account the short sale costs only for Taiwan. As a consequence, we control for this effect when we investigate Taiwanese institutional investors' trading behavior as we do previously.

(15) We also include the current capital flow as an additional control variable for our robustness tests. We find that Asian domestic trading volume is contemporaneously related to the capital flow for most of the sample stock markets. However, the results of this robustness test do not change our conclusions obtained from the results in Section II. As a further robustness check, we also use NASDAQ returns to replace S&P 500 returns and find that the results are qualitatively similar.

Table I. Descriptive Statistics of Weekly Sample

The table reports descriptive statistics on the time series of
the weekly index returns and detrended trading volume for the 10
Asian stock markets from January 1995 to December 2010. Weekly
returns are denominated in percentage points. SD is the standard
deviation. ADF test denotes the t-statistic of the augmented
Dickey-Fuller (1979) test for a unit root. [[rho].sub.1] and
[[rho].sub.2] are used to measure the sum of the correlations
between the current volume and lagged domestic returns up to
three lags and that between the current volume and lagged US
returns up to three lags, respectively. Causality Tests 1, 2, and
3 are the test results of the trivariate Granger causality tests
of domestic trading volume, domestic market returns, and US
market returns in which all variables are standardized (Granger,
1969). Causality Test 1 is the test result of whether the
domestic market returns Granger-cause domestic trading volume;
Causality Test 2 is the test result of whether the US market
returns Granger-cause domestic trading volume; Causality Test 3
is the test result of whether the US market returns Granger-
cause domestic market returns. Sum is the sum of the lagged
coefficients in each Granger-causality test. Sign test is the
test result of the sign of causality. The causality and the
corresponding sign tests are performed using a Wald test. The lag
length in both the ADF and Granger-causality tests is chosen by
considering AIC(Akaike, 1974).

Market (i) Index           Hong Kong (HK) HSI

Variable              Return                  Volume

Mean                   0.158                   0.000
SD                     3.948                   0.258
Minimum              -33.780                  -0.798
Maximum               18.761                   1.413
ADF test              -8.665 ***             -26.269 ***
[[rho].sub.1]                      0.032
[[rho].sub.2]                      0.116
Causality Test 1                   4.323 **
Sum                                7.689
Sign test                          4.323 **
Causality Test 2                   8.303 ***
Sum                               10.506
Sign test                          8.303 ***
Causality Test 3                   3.401 *
Sum                                0.070
Sign test                          3.401 *

Market (i) Index           Thailand (TH) SET

Variable              Return                  Volume

Mean                  -0.040                   0.000
SD                     4.482                   0.350
Minimum              -34.291                  -0.889
Maximum               19.788                   1.199
ADF test             -15.282 ***             -26.002 ***
[[rho].sub.1]                      0.190
[[rho].sub.2]                      0.058
Causality Test 1                   9.767 ***
Sum                               12.696
Sign test                          9.767 ***
Causality Test 2                   1.358
Sum                                4.278
Sign test                          1.358
Causality Test 3                  14.232 ***
Sum                                0.170
Sign test                          7.076 ***

Market (i) Index        The Philippines (PH) PSECI

Variable              Return                  Volume

Mean                   0.061                   0.000
SD                     3.982                   0.477
Minimum              -25.816                  -1.186
Maximum               15.801                   1.743
ADF test              -7.721 ***             -25.827 ***
[[rho].sub.1]                     0.332
[[rho].sub.2]                     0.068
Causality Test 1                 27.887 ***
Sum                              27.035
Sign test                        25.882 ***
Causality Test 2                  0.629
Sum                               3.003
Sign test                         0.629
Causality Test 3                  7.607 *
Sum                               0.157
Sign test                         5.017 **

Market (i) Index            Japan (JA) TOPIX

Variable              Return                  Volume

Mean                  -0.079                   0.000
SD                     3.160                   0.169
Minimum              -29.335                  -0.583
Maximum                9.315                   0.716
ADF test             -25.612 **              -26.834 ***
[[rho].sub.1]                      0.139
[[rho].sub.2]                      0.020
Causality Test 1                  15.808 ***
Sum                               15.434
Sign test                         15.808 ***
Causality Test 2                   0.644
Sum                                3.307
Sign test                          0.644
Causality Test 3                   8.843 **
Sum                                0.179
Sign test                          6.888 ***

Market (i) Index            China (CH) SSEC

Variable              Return                  Volume

Mean                   0.212                   0.000
SD                     4.370                   0.321
Minimum              -24.670                  -0.974
Maximum               24.126                   2.606
ADF test              -8.665 ***             -26.269 ***
[[rho].sub.1]                      0.200
[[rho].sub.2]                     -0.004
Causality Test 1                  50.588 ***
Sum                               30.181
Sign test                         50.588 ***
Causality Test 2                   0.032
Sum                                0.621
Sign test                          0.032
Causality Test 3                   0.003
Sum                                0.002
Sign test                          0.003

Market (i) Index              Taiwan (TA) TWI

Variable              Return                  Volume

Mean                   0.032                   0.000
SD                     3.947                   0.212
Minimum              -27.375                  -0.584
Maximum               15.369                   0.635
ADF test             -15.547 ***             -25.813 ***
[[rho].sub.1]                     0.239
[[rho].sub.2]                     0.155
Causality Test 1                 20.282 ***
Sum                              24.234
Sign test                        20.191 ***
Causality Test 2                 11.498 ***
Sum                              15.897
Sign test                         5.753 **
Causality Test 3                  5.574 **
Sum                               0.357
Sign test                         5.574 **

Market (i) Index           Malaysia (MA) KLSE

Variable              Return                 Volume

Mean                   0.068                  0.000
SD                     3.699                  0.285
Minimum              -16.143                 -0.810
Maximum               27.966                  1.340
ADF test              -7.922 ***            -25.860 ***
[[rho].sub.1]                      0.087
[[rho].sub.2]                      0.048
Causality Test 1                   9.953 ***
Sum                               16.338
Sign test                          9.753 ***
Causality Test 2                   0.410
Sum                                2.096
Sign test                          0.410
Causality Test 3                   3.192 *
Sum                                0.061
Sign test                          3.192 *

Market (i) Index          Indonesia (IN) JKSE

Variable              Return                 Volume

Mean                   0.302                  0.000
SD                     4.216                  0.364
Minimum              -31.205                 -1.360
Maximum               19.970                  2.151
ADF test              -9.846 ***            -25.945 ***
[[rho].sub.1]                      0.385
[[rho].sub.2]                      0.044
Causality Test 1                  36.259 ***
Sum                               34.105
Sign test                         33.630 ***
Causality Test 2                   0.001
Sum                               -0.092
Sign test                          0.001
Causality Test 3                  10.888 **
Sum                                0.216
Sign test                          9.856 ***

Market (i) Index           U.S. (US) S&P 500

Variable              Return                 Volume

Mean                   0.148                     --
SD                     2.848                     --
Minimum              -27.177                     --
Maximum               14.623                     --
ADF test              -2.980 ***                 --
[[rho].sub.1]             --
[[rho].sub.2]             --
Causality Test 1          --
Sum                       --
Sign test                 --
Causality Test 2          --
Sum                       --
Sign test                 --
Causality Test 3          --
Sum                       --
Sign test                 --

Market (i) Index          Singapore (SI) STI

Variable              Return                  Volume

Mean                   0.053                   0.000
SD                     3.667                   0.401
Minimum              -36.464                  -1.885
Maximum               20.783                   2.631
ADF test              -7.812 ***             -26.186 ***
[[rho].sub.1]                      0.028
[[rho].sub.2]                      0.125
Causality Test 1                   0.078
Sum                                1.076
Sign test                          0.078
Causality Test 2                   6.468 **
Sum                                9.761
Sign test                          6.468 **
Causality Test 3                   3.974 **
Sum                                0.085
Sign test                          3.974 **

Market (i) Index           Korea (KO) KOSPI

Variable              Return                  Volume

Mean                   0.106                   0.000
SD                     4.654                   0.197
Minimum              -25.557                  -0.840
Maximum               16.519                   0.666
ADF test             -10.513 ***             -26.182 ***
[[rho].sub.1]                      0.118
[[rho].sub.2]                      0.092
Causality Test 1                  13.581 ***
Sum                               14.867
Sign test                         13.581 ***
Causality Test 2                   3.406 *
Sum                                7.043
Sign test                          3.406 *
Causality Test 3                   3.624 *
Sum                                0.074
Sign test                          3.624 *

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Notes: Critical values and statistical significance levels for
the ADF unit root statistic with more than 500 observations are:
-2.570 at 10%, -2.860 at 5%, and -3.430 at 1% (Fuller, 1976,
table 8.5.2, p. 373).

Table II. Causal Relations Across Markets

The following multivariate SURs model is estimated using a two-
step procedure to investigate the causal relations from domestic
and US stock returns to domestic trading volume across the 10
Asian stock markets from January 1995 to December 2010. In the
first step, the following SUR model of Equation (2) is estimated.
In the second step, the SUR model estimated from the first stage
procedure is re-estimated by dropping all insignificant
independent variables at the 10% significance level.

[AV.sub.it] = [[alpha].sub.i1] + [[alpha].sub.i2][DAR.sub.it] +
[[alpha].sub.i3][DAR.sub.t] + [[summation].sub.j]
[[beta].sub.ij][AR.sub.it-j] + [[summation].sub.k]
[[gamma].sub.ik][UR.sub.t-k] + [[epsilon].sub.it],

where i represents the cross-sectional unit of the 10 Asian stock
markets; [AV.sub.it] is the trading volume of the Asian stock
market i on week t; [AR.sub.it] ([UR.sub.t]) is the stock return
of market i (US stock market) on week t; and [DAR.sub.it]
([DUR.sub.t]) is the detrended absolute value of [AR.sub.it]
([UR.sub.t]) on week t. All variables are standardized. The
number of lags in Equation (2) is chosen by considering the
Akaike (1974) information criterion (AIC). When i = ss, it
indicates Hong Kong, Japan, Malaysia, Singapore, and Thailand
stock markets where short sales are allowed and practiced. When i
= sc, it indicates China, Indonesia, Korea, Philippines, and
Taiwan stock markets where short sales are either prohibited or
not practiced. The [[chi square].sub.[beta]] statistic is the
chi-squared statistic with one degree of freedom under the null
hypothesis that the average of the three sums of the
[[beta].sub.ssj] coefficients across JA, MA, and TH is equal to
that of the five sums of the [[beta].sub.scj] coefficients across
all sc markets. The [[chi square].sub.[gamma]] statistic is the
chi-squared statistic with one degree of freedom under the null
hypothesis that the average of the three sums of the
[[gamma].sub.ssk] coefficients across HK, MA, and SI is equal to
the sum of the [[gamma].sub.sck] coefficients for KO. The p-
values are reported in parentheses.

Asian Stock Market (i)                   Honq Kong   Japan    Malaysia
                                           (HK)       (JA)      (MA)

[[summation].sub.j] [[beta].sub.ij]                  16.031    17.809

[[summation].sub.k] [[gamma].sub.ik]      16.031                6.180

[[summation].sub.ss]                                           18.407
[[summation].sub.j] [[beta].sub.ssj]/
3 or [[summation].sub.sc]
[[summation].sub.j] [[beta].sub.scj]/5

[[summation].sub.ss]                                           10.345
[[summation].sub.k] [[gamma].sub.ssk]/
3 or [[summation].sub.k]
[[gamma].sub.sck]

Do sc investors, on average, trade more after domestic market
gains than ss investors?

Hypothesis Test: [[summation].sub.ss]
[[summation].sub.j]
[[beta].sub.ssj]/3 =
[[summation].sub.sc]
[[summation].sub.j]
[[beta].sub.scj]/5

[[chi square].sub.[beta]]

Do sc investors, on average, trade more after U.S. market
gains than ss investors?

Hypothesis Test: [[summation].sub.ss]
[[summation].sub.k]
[[gamma].sub.ssk]/3 =
[[summation].sub.k] [[gamma].sub.sck]

[[chi square].sub.[gamma]]

Asian Stock Market (i)                   Singapore   Thailand    China
                                           (SI)        (TH)       (CH)

[[summation].sub.j] [[beta].sub.ij]                   21.381     27.991

[[summation].sub.k] [[gamma].sub.ik]       8.824

[[summation].sub.ss]
[[summation].sub.j] [[beta].sub.ssj]/
3 or [[summation].sub.sc]
[[summation].sub.j] [[beta].sub.scj]/5

[[summation].sub.ss]
[[summation].sub.k] [[gamma].sub.ssk]/
3 or [[summation].sub.k]
[[gamma].sub.sck]

Do sc investors, on average, trade more after domestic market
gains than ss investors?

Hypothesis Test: [[summation].sub.ss]                8.468 ***
[[summation].sub.j]
[[beta].sub.ssj]/3 =
[[summation].sub.sc]
[[summation].sub.j]
[[beta].sub.scj]/5

[[chi square].sub.[beta]]                             (0.004)

Do sc investors, on average, trade more after U.S. market
gains than ss investors?

Hypothesis Test: [[summation].sub.ss]                  0.754
[[summation].sub.k]
[[gamma].sub.ssk]/3 =
[[summation].sub.k] [[gamma].sub.sck]

[[chi square].sub.[gamma]]                            (0.385)

Asian Stock Market (i)                   Indonesia   Korea
                                           (IN)       (KO)

[[summation].sub.j] [[beta].sub.ij]       33.707     14.280

[[summation].sub.k] [[gamma].sub.ik]                 6.683

[[summation].sub.ss]                                 27.833
[[summation].sub.j] [[beta].sub.ssj]/
3 or [[summation].sub.sc]
[[summation].sub.j] [[beta].sub.scj]/5

[[summation].sub.ss]                                 6.683
[[summation].sub.k] [[gamma].sub.ssk]/
3 or [[summation].sub.k]
[[gamma].sub.sck]

Do sc investors, on average, trade more after domestic market
gains than ss investors?

Hypothesis Test: [[summation].sub.ss]
[[summation].sub.j]
[[beta].sub.ssj]/3 =
[[summation].sub.sc]
[[summation].sub.j]
[[beta].sub.scj]/5

[[chi square].sub.[beta]]

Do sc investors, on average, trade more after U.S. market
gains than ss investors?

Hypothesis Test: [[summation].sub.ss]
[[summation].sub.k]
[[gamma].sub.ssk]/3 =
[[summation].sub.k] [[gamma].sub.sck]

[[chi square].sub.[gamma]]

Asian Stock Market (i)                   Philippines   Taiwan
                                            (PH)        (TA)

[[summation].sub.j] [[beta].sub.ij]        25.531      23.294

[[summation].sub.k] [[gamma].sub.ik]

[[summation].sub.ss]
[[summation].sub.j] [[beta].sub.ssj]/
3 or [[summation].sub.sc]
[[summation].sub.j] [[beta].sub.scj]/5

[[summation].sub.ss]
[[summation].sub.k] [[gamma].sub.ssk]/
3 or [[summation].sub.k]
[[gamma].sub.sck]

Do sc investors, on average, trade more after domestic market
gains than ss investors?

Hypothesis Test: [[summation].sub.ss]
[[summation].sub.j]
[[beta].sub.ssj]/3 =
[[summation].sub.sc]
[[summation].sub.j]
[[beta].sub.scj]/5

[[chi square].sub.[beta]]

Do sc investors, on average, trade more after U.S. market
gains than ss investors?

Hypothesis Test: [[summation].sub.ss]
[[summation].sub.k]
[[gamma].sub.ssk]/3 =
[[summation].sub.k] [[gamma].sub.sck]

[[chi square].sub.[gamma]]

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table III. The Effect of Market Integration on the Causal
Relations

The following multivariate SUR model is estimated using the same
two-step procedure as in Equation (2) to investigate the causal
relations from domestic and US stock returns to domestic trading
volume conditional on market conditions across the 10 Asian stock
markets from January 1995 to December 2010.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (3)

where the variables are defined as before, and the dummy variable
[S2.sub.t], ([S3.sub.t]) takes a value of one for the subsample
period from January 1999 to June 2007 (from July 2007 to December
2010), and zero otherwise. All variables are standardized. When i
= ss, it indicates Hong Kong, Japan, Malaysia, Singapore, and
Thailand stock markets where short sales are allowed and
practiced. When i = sc, it indicates China, Indonesia, Korea,
Philippines, and Taiwan stock markets where short sales are
either prohibited or not practiced. The [[chi
square].sub.[beta]](1j vs. 2j + 3j) statistic is the chi-squared
statistic with one degree of freedom under the null
hypothesis that the sum of the [[beta].sub.i1j] coefficients is
equal to that of the ([[beta].sub.i2j] and [[beta].sub.i3j]
coefficients. The [[chi square].sub.[gamma]](1k vs. 2k + 3k)
statistic is the chi-squared statistic with one degree of freedom
under the null hypothesis that the sum of the [[gamma].sub.ilk]
coefficients is equal to that of the [[gamma].sub.i2k] and
[[gamma].sub.i3k] coefficients. The [[chi
square].sub.[beta](1,j)] statistic is the chi-squared statistic
with one degree of freedom under the null hypothesis that the
average of the three sums of the [[beta].sub.ss1j] coefficients
across HK, MA, and SI is equal to that of the three sums of the
[[beta].sub.sc1j] across CH, KO, and TA. The [[chi
square].sub.[beta](2j+3j)] statistic is the chi-squared statistic
with one degree of freedom under the null hypothesis that the
average of the three sums of the [[beta].sub.ss2j] and
[[beta].sub.ss3j] coefficients across JA, MA, and TH is equal to
that of the five sums of the [[beta].sub.sc2j] and
[[beta].sub.sc3j] coefficients across all sc markets. The [[chi
square].sub.[gamma]](1k) statistic is the chi-squared statistic
with one degree of freedom under the null hypothesis that the sum
of the [[gamma].sub.i1k] coefficients for JA is equal to zero.
The [[chi square].sub.[gamma](2k+3k)] statistic is the chi-squared
statistic with one degree of freedom under the null
hypothesis that the average of the four sums of the
[[gamma].sub.ss2k] and [[gamma].sub.ss3k] coefficients across HK,
MA, SI, and TH is equal to the sum of the [[gamma].sub.sc2k] and
[[gamma].sub.sc3k] coefficients for KO. The p-values are reported
in parentheses.

Asian Stock Market (i)                   Hong Kong (HK)   Japan (JA)

[[summation].sub.j] [[beta].sub.i1j]     9.496

[[summation].sub.j] [[beta].sub.i2j] +                    14.276
[[summation].sub.j] [[beta].sub.i3j]

[[summation].sub.k] [[gamma].sub.i1k]                      7.781

[[summation].sub.k] [[gamma].sub.i2k]     9.441
+ [[summation].sub.k]
[[gamma].sub.i3k]

[[chi square].sub.[beta]](1j vs. 2j +     5.969 **        15.729 ***
3j)                                      (0.015)          (0.000)

[[chi square].sub.[gamma]](1k vs. 2k +    7.689 ***        4.803 **
3k)                                      (0.006)          (0.028)

[[summation].sub.ss]
[[summation].sub.j] [[beta].sub.ss1j]/
3 or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc1j]/

[[summation].sub.ss]
[[summation].sub.j] [[beta].sub.ss2j]
+ [[beta].sub.ss3j]/3
or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc2j]
+ [[beta].sub.sc3j]/5

[[summation].sub.k] [[gamma].sub.ss1k]
[[summation].sub.ss]
[[summation].sub.k]
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])/4
or

[[summation].sub.k]
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])

Do sc investors, on average, trade more after domestic market
gains before the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
[[beta].sub.ss1j]/3 = [[summation].sub.sc] [[summation].sub.j]
[[beta].sub.sc1j]/3

[[chi square].sub.[beta](1j)]

Do sc investors, on average, trade more after domestic market
gains after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
([[beta].sub.ss2j] + [[beta].sub.ss3j]) = [[summation].sub.sc]
[[summation].sub.j] ([[beta].sub.sc2j] + [[beta].sub.sc3j])/5

[[chi square].sub.[beta](2j+3j)]

Do sc investors, on average, trade more before US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.k] [[gamma].sub.ss1k] = 0

[[chi square].sub.[gamma](1k)]

Do sc investors, on average, trade more after US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])/4 = [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])

[[chi square].sub.[gamma](2k+3k)]

Asian Stock Market (i)                   Malaysia (MA)   Singapore (SI)

[[summation].sub.j] [[beta].sub.i1j]      9.776           9.736

[[summation].sub.j] [[beta].sub.i2j] +   13.508
[[summation].sub.j] [[beta].sub.i3j]

[[summation].sub.k] [[gamma].sub.i1k]

[[summation].sub.k] [[gamma].sub.i2k]     7.262           8.914
+ [[summation].sub.k]
[[gamma].sub.i3k]

[[chi square].sub.[beta]](1j vs. 2j +     0.616           5.433 **
3j)                                      (0.433)         (0.020)

[[chi square].sub.[gamma]](1k vs. 2k +    4.683 **        5.802 **
3k)                                      (0.030)         (0.016)

[[summation].sub.ss]                      9.670
[[summation].sub.j] [[beta].sub.ss1j]/
3 or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc1j]/

[[summation].sub.ss]                     19.284
[[summation].sub.j] [[beta].sub.ss2j]
+ [[beta].sub.ss3j]/3
or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc2j]
+ [[beta].sub.sc3j]/5

[[summation].sub.k] [[gamma].sub.ss1k]    7.781
[[summation].sub.ss]
[[summation].sub.k]
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])/4
or

[[summation].sub.k]                       7.882
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])

Do sc investors, on average, trade more after domestic market
gains before the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
[[beta].sub.ss1j]/3 = [[summation].sub.sc] [[summation].sub.j]
[[beta].sub.sc1j]/3

[[chi square].sub.[beta](1j)]

Do sc investors, on average, trade more after domestic market
gains after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
([[beta].sub.ss2j] + [[beta].sub.ss3j]) = [[summation].sub.sc]
[[summation].sub.j] ([[beta].sub.sc2j] + [[beta].sub.sc3j])/5

[[chi square].sub.[beta](2j+3j)]

Do sc investors, on average, trade more before US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.k] [[gamma].sub.ss1k] = 0

[[chi square].sub.[gamma](1k)]

Do sc investors, on average, trade more after US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])/4 = [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])

[[chi square].sub.[gamma](2k+3k)]

Asian Stock Market (i)                   Thailand (TH)   China (CH)

[[summation].sub.j] [[beta].sub.i1j]                     18.470

[[summation].sub.j] [[beta].sub.i2j] +   30.068          26.516
[[summation].sub.j] [[beta].sub.i3j]

[[summation].sub.k] [[gamma].sub.i1k]

[[summation].sub.k] [[gamma].sub.i2k]     5.911
+ [[summation].sub.k]
[[gamma].sub.i3k]

[[chi square].sub.[beta]](1j vs. 2j +     4.352 **        1.088
3j)                                      (0.037)         (0.297)

[[chi square].sub.[gamma]](1k vs. 2k +    2.752 *
3k)                                      (0.097)

[[summation].sub.ss]
[[summation].sub.j] [[beta].sub.ss1j]/
3 or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc1j]/

[[summation].sub.ss]
[[summation].sub.j] [[beta].sub.ss2j]
+ [[beta].sub.ss3j]/3
or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc2j]
+ [[beta].sub.sc3j]/5

[[summation].sub.k] [[gamma].sub.ss1k]
[[summation].sub.ss]
[[summation].sub.k]
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])/4
or

[[summation].sub.k]
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])

Do sc investors, on average, trade more after domestic market
gains before the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
[[beta].sub.ss1j]/3 = [[summation].sub.sc] [[summation].sub.j]
[[beta].sub.sc1j]/3

[[chi square].sub.[beta](1j)]             0.272
                                         (0.602)

Do sc investors, on average, trade more after domestic market
gains after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
([[beta].sub.ss2j] + [[beta].sub.ss3j]) = [[summation].sub.sc]
[[summation].sub.j] ([[beta].sub.sc2j] + [[beta].sub.sc3j])/5

[[chi square].sub.[beta](2j+3j)]          4.217 **
                                         (0.040)

Do sc investors, on average, trade more before US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.k] [[gamma].sub.ss1k] = 0

[[chi square].sub.[gamma](1k)]            4.803 **
                                         (0.028)

Do sc investors, on average, trade more after US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])/4 = [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])

[[chi square].sub.[gamma](2k+3k)]         0.031
                                         (0.859)

Asian Stock Market (i)                   Indonesia (IN)   Korea (KO)

[[summation].sub.j] [[beta].sub.i1j]                      13.536

[[summation].sub.j] [[beta].sub.i2j] +   47.255           13.019
[[summation].sub.j] [[beta].sub.i3j]

[[summation].sub.k] [[gamma].sub.i1k]

[[summation].sub.k] [[gamma].sub.i2k]                      7.172
+ [[summation].sub.k]
[[gamma].sub.i3k]

[[chi square].sub.[beta]](1j vs. 2j +     8.777 ***        0.010
3j)                                      (0.003)          (0.921)

[[chi square].sub.[gamma]](1k vs. 2k +                     3.956 **
3k)                                                       (0.047)

[[summation].sub.ss]                                      11.668
[[summation].sub.j] [[beta].sub.ss1j]/
3 or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc1j]/

[[summation].sub.ss]                                      28.899
[[summation].sub.j] [[beta].sub.ss2j]
+ [[beta].sub.ss3j]/3
or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc2j]
+ [[beta].sub.sc3j]/5

[[summation].sub.k] [[gamma].sub.ss1k]
[[summation].sub.ss]
[[summation].sub.k]
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])/4
or

[[summation].sub.k]                                        7.172
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])

Do sc investors, on average, trade more after domestic market
gains before the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
[[beta].sub.ss1j]/3 = [[summation].sub.sc] [[summation].sub.j]
[[beta].sub.sc1j]/3

[[chi square].sub.[beta](1j)]

Do sc investors, on average, trade more after domestic market
gains after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
([[beta].sub.ss2j] + [[beta].sub.ss3j]) = [[summation].sub.sc]
[[summation].sub.j] ([[beta].sub.sc2j] + [[beta].sub.sc3j])/5

[[chi square].sub.[beta](2j+3j)]

Do sc investors, on average, trade more before US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.k] [[gamma].sub.ss1k] = 0

[[chi square].sub.[gamma](1k)]

Do sc investors, on average, trade more after US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])/4 = [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])

[[chi square].sub.[gamma](2k+3k)]

Asian Stock Market (i)                   Philippines (PH)   Taiwan (TA)

[[summation].sub.j] [[beta].sub.i1j]                         2.990

[[summation].sub.j] [[beta].sub.i2j] +   34.079             23.628
[[summation].sub.j] [[beta].sub.i3j]

[[summation].sub.k] [[gamma].sub.i1k]

[[summation].sub.k] [[gamma].sub.i2k]
+ [[summation].sub.k]
[[gamma].sub.i3k]

[[chi square].sub.[beta]](1j vs. 2j +     5.253 **           8.098 **
3j)                                      (0.022)            (0.004)

[[chi square].sub.[gamma]](1k vs. 2k +
3k)

[[summation].sub.ss]
[[summation].sub.j] [[beta].sub.ss1j]/
3 or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc1j]/

[[summation].sub.ss]
[[summation].sub.j] [[beta].sub.ss2j]
+ [[beta].sub.ss3j]/3
or
[[summation].sub.sc]
[[summation].sub.j] [[beta].sub.sc2j]
+ [[beta].sub.sc3j]/5

[[summation].sub.k] [[gamma].sub.ss1k]
[[summation].sub.ss]
[[summation].sub.k]
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])/4
or

[[summation].sub.k]
([[gamma].sub.ss2k] +
[[gamma].sub.ss3k])

Do sc investors, on average, trade more after domestic market
gains before the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
[[beta].sub.ss1j]/3 = [[summation].sub.sc] [[summation].sub.j]
[[beta].sub.sc1j]/3

[[chi square].sub.[beta](1j)]

Do sc investors, on average, trade more after domestic market
gains after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.j]
([[beta].sub.ss2j] + [[beta].sub.ss3j]) = [[summation].sub.sc]
[[summation].sub.j] ([[beta].sub.sc2j] + [[beta].sub.sc3j])/5

[[chi square].sub.[beta](2j+3j)]

Do sc investors, on average, trade more before US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.k] [[gamma].sub.ss1k] = 0

[[chi square].sub.[gamma](1k)]

Do sc investors, on average, trade more after US market gains
after the Asian financial crisis period than ss investors?

Hypothesis Test: [[summation].sub.ss] [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])/4 = [[summation].sub.k]
([[gamma].sub.ss2k] + [[gamma].sub.ss3k])

[[chi square].sub.[gamma](2k+3k)]

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table IV. Causal Relations Conditional on Market Conditions

The following multivariate SUR model is estimated to investigate
the causal relations from domestic and US stock returns to
domestic trading volume conditional on market conditions across
the 10 Asian stock markets from January 1995 to December 2010.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (4)

where the variables are defined as before and the dummy variable
[AB.sub.it]([UB.sub.t]) takes a value of one if week t is
included in the period of a bull market. All variables are
standardized. When i = ss, it indicates Hong Kong, Japan,
Malaysia, Singapore, and Thailand stock markets where short sales
are allowed and practiced. When i = sc, it indicates China,
Indonesia, Korea, Philippines, and Taiwan stock markets where
short sales are either prohibited or not practiced. The [chi
square].sub.[beta]] (1j vs. 2j) statistic tests the null
hypothesis that the sum of [[beta].sub.i1j] is equal to the sum
of [[beta].sub.i2j] for market i. The [[chi square].sub.[gamma]]
(1k vs. 2k) statistic tests the null hypothesis that the sum of
[[gamma].sub.i1k] is equal to the sum of [[gamma].sub.i2k]
coefficients for the market i. The [[chi square].sub.[beta](1j)]
statistic tests the null hypothesis that the average of the five
sums of [[beta].sub.ss1j] across all ss markets is equal to that
of the five sums of [[beta].sub.sc1j] across all sc markets. The
[[chi square].sub.[gamma](1k)] statistic tests the null
hypothesis that the average of the three sums of
[[gamma].sub.ss1k] across HK, MA, and SI equal to the sum of
[[gamma].sub.sc1k] for KO. The p-values are reported in
parentheses.

Asian Stock Market (i)                   Hong Kong (HK)   Japan (JA)

[[summation].sub.j][[beta].sub.i1j]      13.012           20.039

[[summation].sub.j][[beta].sub.i2j]                        9.534

[[summation].sub.k][[gamma].sub.i1k]      9.429
[[summation].sub.k][[gamma].sub.i2k]

[[chi square].sub.[beta]](1j vs. 2j)     13.879 ***        4.441 **
                                         (0.000)          (0.035)

[[chi square].sub.[gamma]](1k vs. 2k)     7.798 ***
                                         (0.005)

[[summation].sub.ss]
[[summation].sub.j][[beta].sub.ss1j/
5 or [[summation].sub.sc]
[[summation].sub.j][[beta].sub.sc1j]/
5

[[summation].sub.ss]
[[summation].sub.k][[gamma].sub.ss1k]/
3 or [[summation].sub.k]
[[gamma].sub.sc1k]/1

Do sc investors trade more after domestic market gains in
domestic bull markets than 55 investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.j] [[beta].sub.ss1j]/5 =
[[summation].sub.sc][[summation].sub.j] [[beta].sub.sc1j]/5

[[chi square].sub.[beta](1j)]

Do sc investors trade more after US market gains in US bull
markets than ss investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.k] [[gamma].sub.ss1k]/3 =
[[summation].sub.k] [[gamma].sub.sc1k]/1

[[chi square].sub.[gamma](1k)]

Asian Stock Market (i)                   Malaysia (MA)   Singapore (SI)

[[summation].sub.j][[beta].sub.i1j]      12.761          15.103

[[summation].sub.j][[beta].sub.i2j]      14.576

[[summation].sub.k][[gamma].sub.i1k]      5.810          10.083
[[summation].sub.k][[gamma].sub.i2k]

[[chi square].sub.[beta]](1j vs. 2j)      0.085          16.863 ***
                                         (0.771)         (0.000)

[[chi square].sub.[gamma]](1k vs. 2k)     2.953 *         7.468 ***
                                         (0.086)         (0.006)

[[summation].sub.ss]                     16.398
[[summation].sub.j][[beta].sub.ss1j/
5 or [[summation].sub.sc]
[[summation].sub.j][[beta].sub.sc1j]/
5

[[summation].sub.ss]                      8.441
[[summation].sub.k][[gamma].sub.ss1k]/
3 or [[summation].sub.k]
[[gamma].sub.sc1k]/1

Do sc investors trade more after domestic market gains in
domestic bull markets than 55 investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.j] [[beta].sub.ss1j]/5 =
[[summation].sub.sc][[summation].sub.j] [[beta].sub.sc1j]/5

[[chi square].sub.[beta](1j)]

Do sc investors trade more after US market gains in US bull
markets than ss investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.k] [[gamma].sub.ss1k]/3 =
[[summation].sub.k] [[gamma].sub.sc1k]/1

[[chi square].sub.[gamma](1k)]

Asian Stock Market (i)                   Thailand (TH)   China (CH)

[[summation].sub.j][[beta].sub.i1j]      21.077          20.902

[[summation].sub.j][[beta].sub.i2j]      10.228

[[summation].sub.k][[gamma].sub.i1k]
[[summation].sub.k][[gamma].sub.i2k]

[[chi square].sub.[beta]](1j vs. 2j)      3.539 *         7 094
                                         (0.060)         (0.008)

[[chi square].sub.[gamma]](1k vs. 2k)

[[summation].sub.ss]
[[summation].sub.j][[beta].sub.ss1j/
5 or [[summation].sub.sc]
[[summation].sub.j][[beta].sub.sc1j]/
5

[[summation].sub.ss]
[[summation].sub.k][[gamma].sub.ss1k]/
3 or [[summation].sub.k]
[[gamma].sub.sc1k]/1

Do sc investors trade more after domestic market gains in
domestic bull markets than 55 investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.j] [[beta].sub.ss1j]/5 =
[[summation].sub.sc][[summation].sub.j] [[beta].sub.sc1j]/5

[[chi square].sub.[beta](1j)]             4.554 **
                                         (0.033)

Do sc investors trade more after US market gains in US bull
markets than ss investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.k] [[gamma].sub.ss1k]/3 =
[[summation].sub.k] [[gamma].sub.sc1k]/1

[[chi square].sub.[gamma](1k)]            0.002
                                         (0.963)

Asian Stock Market (i)                   Indonesia (IN)   Korea (KO)

[[summation].sub.j][[beta].sub.i1j]      29.177           16.770

[[summation].sub.j][[beta].sub.i2j]      15.259            0.638

[[summation].sub.k][[gamma].sub.i1k]                       8.544
[[summation].sub.k][[gamma].sub.i2k]

[[chi square].sub.[beta]](1j vs. 2j)      4.497 **         5.726 **
                                         (0.034)          (0.017)

[[chi square].sub.[gamma]](1k vs. 2k)                      5.605 **
                                                          (0.018)

[[summation].sub.ss]                                      22.917
[[summation].sub.j][[beta].sub.ss1j/
5 or [[summation].sub.sc]
[[summation].sub.j][[beta].sub.sc1j]/
5

[[summation].sub.ss]                                       8.544
[[summation].sub.k][[gamma].sub.ss1k]/
3 or [[summation].sub.k]
[[gamma].sub.sc1k]/1

Do sc investors trade more after domestic market gains in
domestic bull markets than 55 investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.j] [[beta].sub.ss1j]/5 =
[[summation].sub.sc][[summation].sub.j] [[beta].sub.sc1j]/5

[[chi square].sub.[beta](1j)]

Do sc investors trade more after US market gains in US bull
markets than ss investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.k] [[gamma].sub.ss1k]/3 =
[[summation].sub.k] [[gamma].sub.sc1k]/1

[[chi square].sub.[gamma](1k)]

Asian Stock Market (i)                   Philippines (PH)   Taiwan (TA)

[[summation].sub.j][[beta].sub.i1j]      24.952             22.783

[[summation].sub.j][[beta].sub.i2j]      11.416             11.857

[[summation].sub.k][[gamma].sub.i1k]
[[summation].sub.k][[gamma].sub.i2k]

[[chi square].sub.[beta]](1j vs. 2j)      4.262 **           3.121 *
                                         (0.039)            (0.077)

[[chi square].sub.[gamma]](1k vs. 2k)

[[summation].sub.ss]
[[summation].sub.j][[beta].sub.ss1j/
5 or [[summation].sub.sc]
[[summation].sub.j][[beta].sub.sc1j]/
5

[[summation].sub.ss]
[[summation].sub.k][[gamma].sub.ss1k]/
3 or [[summation].sub.k]
[[gamma].sub.sc1k]/1

Do sc investors trade more after domestic market gains in
domestic bull markets than 55 investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.j] [[beta].sub.ss1j]/5 =
[[summation].sub.sc][[summation].sub.j] [[beta].sub.sc1j]/5

[[chi square].sub.[beta](1j)]

Do sc investors trade more after US market gains in US bull
markets than ss investors? [H.sub.0] :
[[summation].sub.ss][[summation].sub.k] [[gamma].sub.ss1k]/3 =
[[summation].sub.k] [[gamma].sub.sc1k]/1

[[chi square].sub.[gamma](1k)]

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table V. Causal Relations Conditional on Extremely High Market Returns

The following multivariate SUR model is estimated to investigate the
causal relations from domestic and US stock returns to domestic
trading volume conditional on market conditions across the 10 Asian
stock markets from January 1995 to December 2010.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)

where the variables are defined as before and the dummy variable
[EAR.sub.it] ([EUR.sub.t]) takes a value of one for week t when the
return for the Asian stock market i (US stock market) belongs to the
top 10% of its distribution, and zero otherwise. All variables are
standardized. When i = ss, it indicates Hong Kong, Japan, Malaysia,
Singapore, and Thailand stock markets where short sales are allowed
and practiced. When i = sc, it indicates China, Indonesia, Korea,
Philippines, and Taiwan stock markets where short sales are either
prohibited or not practiced. The [[chi square].sub.[beta]] (1j vs. 2j)
statistic tests the null hypothesis that the sum of [[beta].sub.i1j]
is equal to that of [[beta].sub.i2j] for market i. The [[chi
square].sub.[gamma]] (1k vs. 2k) statistic tests the null hypothesis
that the sum of [[gamma].sub.i1k] is equal to that of
[[gamma].sub.i2k]. The [[chi square].sub.[beta](1j)] statistic tests
the null hypothesis that the average of the four sums of
[[beta].sub.ss1j] across JA, MA, SI, and TH is equal to that that of
the four sums of [[beta].sub.sc1j]; across all sc markets. The [[chi
square].sub.[gamma](1k)] statistic tests the null hypothesis that the
average of the three sums of [[gamma].sub.ss1k] across HK, MA, and SI
is equal to that of the three sums of [[gamma].sub.sc1k] across KO,
PH, and TA. The p-values are reported in parentheses.

Asian Stock Market (i)                           Hong Kong     Japan
                                                    (HK)       (JA)

[[summation].sub.j][[beta].sub.i1j]                           11.401
[[summation].sub.j][[beta].sub.i2j]                           16.600
[[summation].sub.k][[gamma].sub.i1k]             10.798
[[summation].sub.k][[gamma].sub.i2k]                           6.513
[[chi square].sub.[beta]](1j vs. 2j)                           0.749
                                                              (0.387)
[[chi square].sub.[gamma]](1k vs. 2k)            10.071 ***    3.321 *
                                                 (0.002)      (0.068)
[[summation].sub.ss] [[summation].sub.j]
  [[beta].sub.ss1j]/4 or [[summation].sub.sc]
  [[summation].sub.sc1j] [[beta].sub.ss1j]/5
[[summation].sub.ss] [[summation].sub.k]
  [[gamma].sub.ss1j]/3 or [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1j]/3
Do sc investors trade more after the domestic market experiences
  extremely high returns than ss investors? [H.sub.0] :
  [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ss1j]/4 =
  [[summation].sub.ss] [[summation].sub.sc] [[beta].sub.sc1j]/5
[[chi square].sub.[beta](1j)

Do sc investors trade more after the US market experiences extremely
  high returns than .v.s investors? [H.sub.0] : [[summation].sub.ss]
  [[summation].sub.k] [[gamma].sub.ss1k]/3 = [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1k]/3
[[chi square].sub.[gamma](1k)

Asian Stock Market (i)                         Malaysia     Singapore
                                                 (MA)        (SI)

[[summation].sub.j][[beta].sub.i1j]              8.179       6.739
[[summation].sub.j][[beta].sub.i2j]             13.929
[[summation].sub.k][[gamma].sub.i1k]             6.556      10.756
[[summation].sub.k][[gamma].sub.i2k]
[[chi square].sub.[beta]](1j vs. 2j)             1.089       3.404 *
                                                (0.297)     (0.065)
[[chi square].sub.[gamma]](1k vs. 2k)            4.039 **    8.359 ***
                                                (0.044)     (0.004)
[[summation].sub.ss] [[summation].sub.j]        11.582
  [[beta].sub.ss1j]/4 or [[summation].sub.sc]
  [[summation].sub.sc1j] [[beta].sub.ss1j]/5
[[summation].sub.ss] [[summation].sub.k]         9.370
  [[gamma].sub.ss1j]/3 or [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1j]/3
Do sc investors trade more after the domestic market experiences
  extremely high returns than ss investors? [H.sub.0] :
  [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ss1j]/4 =
  [[summation].sub.ss] [[summation].sub.sc] [[beta].sub.sc1j]/5
[[chi square].sub.[beta](1j)

Do sc investors trade more after the US market experiences extremely
  high returns than .v.s investors? [H.sub.0] : [[summation].sub.ss]
  [[summation].sub.k] [[gamma].sub.ss1k]/3 = [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1k]/3
[[chi square].sub.[gamma](1k)

Asian Stock Market (i)                           Thailand     China
                                                   (TH)       (CH)

[[summation].sub.j][[beta].sub.i1j]              20.008      23.277
[[summation].sub.j][[beta].sub.i2j]               9.498      11.52
[[summation].sub.k][[gamma].sub.i1k]
[[summation].sub.k][[gamma].sub.i2k]
[[chi square].sub.[beta]](1j vs. 2j)              2.806 *     6.157 **
                                                 (0.094)     (0.013)
[[chi square].sub.[gamma]](1k vs. 2k)

[[summation].sub.ss] [[summation].sub.j]
  [[beta].sub.ss1j]/4 or [[summation].sub.sc]
  [[summation].sub.sc1j] [[beta].sub.ss1j]/5
[[summation].sub.ss] [[summation].sub.k]
  [[gamma].sub.ss1j]/3 or [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1j]/3
Do sc investors trade more after the domestic market experiences
  extremely high returns than ss investors? [H.sub.0] :
  [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ss1j]/4 =
  [summation].sub.ss] [[summation].sub.sc] [[beta].sub.sc1j]/5
[[chi square].sub.[beta](1j)                         6.829 ***
                                                    (0.009)
Do sc investors trade more after the US market experiences extremely
  high returns than .v.s investors? [H.sub.0] : [[summation].sub.ss]
  [[summation].sub.k] [[gamma].sub.ss1k]/3 = [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1k]/3
[[chi square].sub.[gamma](1k)                        0.933
                                                    (0.334)

Asian Stock Market (i)                           Indonesia    Korea
                                                   (IN)        (KO)

[[summation].sub.j][[beta].sub.i1j]                21.762     21.021
[[summation].sub.j][[beta].sub.i2j]                12.941      7.749
[[summation].sub.k][[gamma].sub.i1k]                           6.685
[[summation].sub.k][[gamma].sub.i2k]
[[chi square].sub.[beta]](1j vs. 2j)               2.844 *    3.935 **
                                                  (0.092)    (0.047)
[[chi square].sub.[gamma]](1k vs. 2k)                         5.887 **
                                                             (0.015)
[[summation].sub.ss] [[summation].sub.j]                     18.787
  [[beta].sub.ss1j]/4 or [[summation].sub.sc]
  [[summation].sub.sc1j] [[beta].sub.ss1j]/5
[[summation].sub.ss] [[summation].sub.k]                      6.684
  [[gamma].sub.ss1j]/3 or [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1j]/3
Do sc investors trade more after the domestic market experiences
  extremely high returns than ss investors? [H.sub.0] :
  [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ss1j]/4 =
  [[summation].sub.ss] [[summation].sub.sc] [[beta].sub.sc1j]/5
[[chi square].sub.[beta](1j)

Do sc investors trade more after the US market experiences extremely
  high returns than .v.s investors? [H.sub.0] : [[summation].sub.ss]
  [[summation].sub.k] [[gamma].sub.ss1k]/3 = [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1k]/3
[[chi square].sub.[gamma](1k)

Asian Stock Market (i)                           Philippines   Taiwan
                                                    (PH)        (TA)

[[summation].sub.j][[beta].sub.i1j]                16.873     11.002
[[summation].sub.j][[beta].sub.i2j]                17.289     18.407
[[summation].sub.k][[gamma].sub.i1k]                6.785      6.582
[[summation].sub.k][[gamma].sub.i2k]
[[chi square].sub.[beta]](1j vs. 2j)                0.004      1.537
                                                   (0.947)    (0.215)
[[chi square].sub.[gamma]](1k vs. 2k)               3.349 *    3.411 *
                                                   (0.067)    (0.065)
[[summation].sub.ss] [[summation].sub.j]
  [[beta].sub.ss1j]/4 or [[summation].sub.sc]
  [[summation].sub.sc1j] [[beta].sub.ss1j]/5
[[summation].sub.ss] [[summation].sub.k]
  [[gamma].sub.ss1j]/3 or [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1j]/3
Do sc investors trade more after the domestic market experiences
  extremely high returns than ss investors? [H.sub.0] :
  [[summation].sub.ss] [[summation].sub.j] [[beta].sub.ss1j]/4 =
  [[summation].sub.ss] [[summation].sub.sc] [[beta].sub.sc1j]/5
[[chi square].sub.[beta](1j)

Do sc investors trade more after the US market experiences extremely
  high returns than .v.s investors? [H.sub.0] : [[summation].sub.ss]
  [[summation].sub.k] [[gamma].sub.ss1k]/3 = [[summation].sub.sc]
  [[summation].sub.k] [[gamma].sub.sc1k]/3
[[chi square].sub.[gamma](1k)

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table VI. The Summary Results Using the Individual and Institutional
Trading Data

This table reports the summary results of the replicated tests in
Tables II, IV, and V and the causal relations conditional on high
investor sentiment using the individual and institutional trading data.
All tests use the bivariate SUR model for each sample country, except
for the tests of the degree of ss versus sc investors' trading that use
the multivariate SUR model across the five sample countries. The
bivariate SUR model is estimated for Japan from January 2000 to
December 2010, for Malaysia from October 2009 to May 2012, for Thailand
from July 1999 to December 2010, for Indonesia from June 2000 to
December 2010, for Korea from December 2003 to December 2010, and for
Taiwan from January 2001 to December 2010. The multivariate SUR model
across Japan, Thailand, Indonesia, Korea, and Taiwan is estimated from
December 2003 to December 2010. Both the bivariate and multivariate SUR
models are estimated using the same two-step procedure as in Equation
(2). In the cross-equation test in the multivariate SUR model, ss
investors denote investors in Japan and Thailand and sc investors
denote investors in Indonesia, Korea, and Taiwan. "Yes" written in
italics in the A8 test indicates that institutions trade more
frequently after US market gains than individuals in Malaysia and
Indonesia and that in the C6 test indicates that institutions trade
more frequently after the US market experiences extremely high returns
than individuals in Malaysia and Indonesia. The test statistics
reported in parentheses in the last two tests of each panel are used to
test the equality of the degree of sc versus 55 investors' trading. The
corresponding p-values of each test statistic are reported in brackets.

Asian Stock Market (i)                          Japan   Malaysia
                                                (JA)    (MA)
Panel A. Causal Relations

A1 Test: Do individuals trade more after        Yes     Yes
domestic market gains?

A2 Test: Do institutions trade more after       Yes     Yes
domestic market gains?

A3 Test: Do individuals trade more after US     No      No
market gains?

A4 Test: Do institutions trade more after US    No      Yes
market gains?

A5 Test: Do individuals and institutions        Yes     Yes
trade more after domestic market gains?

A6 Test: Do individuals and institutions        No      Yes
trade more after US market gains?

A7 Test: Do individuals trade more after        Yes     No
domestic market gains than institutions?

A8 Test: Do individuals trade more after US     --      Yes
market gains than institutions or vice versa?

A9 Test: Do sc investors, on average, trade
more after domestic market gains than ss
investors?

A10 Test: Do sc investors, on average, trade
more after US market gains than ss investors?

Panel B. Causal Relations Conditional on Market Conditions

B1 Test: Do individuals trade more after        Yes     Yes
domestic market gains in domestic bull
markets than in domestic nonbull markets?

B2 Test: Do institutions trade more after       No      No
domestic market gains in domestic bull
markets than in domestic nonbull markets?

B3 Test: Do individuals trade more after US     Yes     Yes
market gains in US bull markets than in U.S.
nonbull markets?

B4 Test: Do institutions trade more after US    No      No
market gains in US bull markets than in U.S.
nonbull markets?

B5 Test: Do individuals trade more after        Yes     No
domestic market gains in domestic bull
markets than institutions?

B6 Test: Do individuals trade more after US     Yes     No
market gains in US bull markets than
institutions?

B7 Test: Do individuals and institutions        Yes     No
trade more after domestic market gains in
domestic bull markets than in domestic
nonbull markets?

B8 Test: Do individuals and institutions        No      Yes
trade more after US market gains in US bull
markets than in US nonbull markets?

B9 Test: Do sc investors, on average, trade
more after domestic market gains in domestic
bull markets than ss investors?

B1O Test: Do sc investors, on average, trade
more after US market gains in US bull markets
than ss investors?

Panel C. Causal Relations Conditional on Extremely High Market Returns

Cl Test: Do individuals trade more after the    No      Yes
domestic market experiences extremely high
returns than when it does not?

C2 Test: Do institutions trade more after the   Yes     No
domestic market experiences extremely high
returns than when it does not?

C3 Test: Do individuals trade more after the    No      No
US market experiences extremely high returns
than when it does not?

C4 Test: Do institutions trade more after the   No      No
US market experiences extremely high returns
than when it does not?

C5 Test: Do individuals trade more after the    No      No
domestic market experiences extremely high
returns than institutions?

C6 Test: Do individuals trade more after the    --      Yes
US market experiences extremely high returns
than institutions or vice versa?

C7 Test: Do individuals and institutions        No      No
trade more after the domestic market
experiences extremely high returns than when
it does not?

C8 Test: Do individuals and institutions        No      No
trade more after the US market experiences
extremely high returns than when it does not?

C9 Test: Do sc investors, on average, trade
more after the domestic market experiences
extremely high returns than ss investors?

C1O Test: Do sc investors, on average, trade
more after the US market experiences
extremely high returns than ss investors?

Panel D. Causal Relations Conditional on High Investor Sentiment

D1 Test: Do individuals trade more after        No      --
domestic market gains when domestic investor
sentiment is high than it is not?

D2 Test: Do institutions trade more after       Yes     --
domestic market gains when domestic investor
sentiment is high than it is not?

D3 Test: Do individuals trade more after US     Yes     --
market gains when US investor sentiment is
high than it is not?

D4 Test: Do institutions trade more after US    Yes     --
market gains when US investor sentiment is
high than it is not?

D5 Test: Do individuals trade more after        No      --
domestic market gains when domestic investor
sentiment is high than institutions?

D6 Test: Do individuals trade more after US     No      --
market gains when US investor sentiment is
high than institutions?

D7 Test: Do individuals and institutions        Yes     --
trade more after domestic market gains when
domestic investor sentiment is high than it
is not?

D8 Test: Do individuals and institutions        Yes     --
trade more after US market gains when US
investor sentiment is high than it is not?

D9 Test: Do sc investors, on average, trade
more after domestic market gains when
domestic investor sentiment is high than ss
investors?

D1O Test: Do sc investors, on average, trade
more after US market gains when US investor
sentiment is high than ss investors?

Asian Stock Market (i)                          Thailand    Indonesia
                                                (TH)        (IN)
Panel A. Causal Relations

A1 Test: Do individuals trade more after        Yes         Yes
domestic market gains?

A2 Test: Do institutions trade more after       Yes         Yes
domestic market gains?

A3 Test: Do individuals trade more after US     No          No
market gains?

A4 Test: Do institutions trade more after US    No          Yes
market gains?

A5 Test: Do individuals and institutions        Yes         Yes
trade more after domestic market gains?

A6 Test: Do individuals and institutions        No          Yes
trade more after US market gains?

A7 Test: Do individuals trade more after        No          Yes
domestic market gains than institutions?

A8 Test: Do individuals trade more after US     --          Yes
market gains than institutions or vice versa?

A9 Test: Do sc investors, on average, trade     Yes ( = 11.830 [0.001])
more after domestic market gains than ss
investors?

A10 Test: Do sc investors, on average, trade    Yes ( = 10.502 [0.001])
more after US market gains than ss investors?

Panel B. Causal Relations Conditional on Market Conditions

B1 Test: Do individuals trade more after        Yes         Yes
domestic market gains in domestic bull
markets than in domestic nonbull markets?

B2 Test: Do institutions trade more after       No          Yes
domestic market gains in domestic bull
markets than in domestic nonbull markets?

B3 Test: Do individuals trade more after US     --          Yes
market gains in US bull markets than in U.S.
nonbull markets?

B4 Test: Do institutions trade more after US    --          Yes
market gains in US bull markets than in U.S.
nonbull markets?

B5 Test: Do individuals trade more after        No          Yes
domestic market gains in domestic bull
markets than institutions?

B6 Test: Do individuals trade more after US                 No
market gains in US bull markets than
institutions?

B7 Test: Do individuals and institutions        Yes         Yes
trade more after domestic market gains in
domestic bull markets than in domestic
nonbull markets?

B8 Test: Do individuals and institutions        --          Yes
trade more after US market gains in US bull
markets than in US nonbull markets?

B9 Test: Do sc investors, on average, trade     Yes ( = 16.984 [0.000])
more after domestic market gains in domestic
bull markets than ss investors?

B1O Test: Do sc investors, on average, trade    No ( = 0.100 [0.752])
more after US market gains in US bull markets
than ss investors?

Panel C. Causal Relations Conditional on Extremely High Market Returns

Cl Test: Do individuals trade more after the    No          Yes
domestic market experiences extremely high
returns than when it does not?

C2 Test: Do institutions trade more after the   Yes         No
domestic market experiences extremely high
returns than when it does not?

C3 Test: Do individuals trade more after the    --          --
US market experiences extremely high returns
than when it does not?

C4 Test: Do institutions trade more after the   --          No
US market experiences extremely high returns
than when it does not?

C5 Test: Do individuals trade more after the    No          Yes
domestic market experiences extremely high      --          Yes
returns than institutions?

C6 Test: Do individuals trade more after the
US market experiences extremely high returns    No          Yes
than institutions or vice versa?

C7 Test: Do individuals and institutions
trade more after the domestic market            --          No
experiences extremely high returns than when
it does not?

C8 Test: Do individuals and institutions
trade more after the US market experiences      Yes ( = 5.183 [0.023])
extremely high returns than when it does not?

C9 Test: Do sc investors, on average, trade
more after the domestic market experiences      Yes ( = 18.543 [0.000])
extremely high returns than ss investors?

C1O Test: Do sc investors, on average, trade
more after the US market experiences
extremely high returns than ss investors?

Panel D. Causal Relations Conditional on High Investor Sentiment

D1 Test: Do individuals trade more after        Yes         Yes
domestic market gains when domestic investor    No          No
sentiment is high than it is not?

D2 Test: Do institutions trade more after
domestic market gains when domestic investor    Yes         No
sentiment is high than it is not?

D3 Test: Do individuals trade more after US
market gains when US investor sentiment is      Yes         No
high than it is not?

D4 Test: Do institutions trade more after US
market gains when US investor sentiment is      No          Yes
high than it is not?

D5 Test: Do individuals trade more after
domestic market gains when domestic investor    No          Yes
sentiment is high than institutions?

D6 Test: Do individuals trade more after US
market gains when US investor sentiment is      Yes         Yes
high than institutions?

D7 Test: Do individuals and institutions
trade more after domestic market gains when     Yes         No
domestic investor sentiment is high than it
is not?

D8 Test: Do individuals and institutions        Yes ( = 6.776 [0.009])
trade more after US market gains when US
investor sentiment is high than it is not?

D9 Test: Do sc investors, on average, trade     No ( = 2.605 [0.107])
more after domestic market gains when
domestic investor sentiment is high than ss
investors?

D1O Test: Do sc investors, on average, trade
more after US market gains when US investor
sentiment is high than ss investors?

Asian Stock Market (i)                          Korea   Taiwan
                                                (KO)    (TA)
Panel A. Causal Relations

A1 Test: Do individuals trade more after        Yes     Yes
domestic market gains?

A2 Test: Do institutions trade more after       Yes     Yes
domestic market gains?

A3 Test: Do individuals trade more after US     Yes     No
market gains?

A4 Test: Do institutions trade more after US    No      No
market gains?

A5 Test: Do individuals and institutions        Yes     Yes
trade more after domestic market gains?

A6 Test: Do individuals and institutions        Yes     No
trade more after US market gains?

A7 Test: Do individuals trade more after        No      Yes
domestic market gains than institutions?

A8 Test: Do individuals trade more after US     Yes     --
market gains than institutions or vice versa?

A9 Test: Do sc investors, on average, trade
more after domestic market gains than ss
investors?

A10 Test: Do sc investors, on average, trade
more after US market gains than ss investors?

Panel B. Causal Relations Conditional on Market Conditions

B1 Test: Do individuals trade more after        Yes     Yes
domestic market gains in domestic bull
markets than in domestic nonbull markets?

B2 Test: Do institutions trade more after       No      No
domestic market gains in domestic bull
markets than in domestic nonbull markets?

B3 Test: Do individuals trade more after US     Yes     Yes
market gains in US bull markets than in U.S.
nonbull markets?

B4 Test: Do institutions trade more after US    Yes     --
market gains in US bull markets than in U.S.
nonbull markets?

B5 Test: Do individuals trade more after        No      Yes
domestic market gains in domestic bull
markets than institutions?

B6 Test: Do individuals trade more after US     No      Yes
market gains in US bull markets than
institutions?

B7 Test: Do individuals and institutions        Yes     Yes
trade more after domestic market gains in
domestic bull markets than in domestic
nonbull markets?

B8 Test: Do individuals and institutions        Yes     Yes
trade more after US market gains in US bull
markets than in US nonbull markets?

B9 Test: Do sc investors, on average, trade
more after domestic market gains in domestic
bull markets than ss investors?

B1O Test: Do sc investors, on average, trade
more after US market gains in US bull markets
than ss investors?

Panel C. Causal Relations Conditional on Extremely High Market Returns

Cl Test: Do individuals trade more after the    No      No
domestic market experiences extremely high
returns than when it does not?

C2 Test: Do institutions trade more after the   No      No
domestic market experiences extremely high
returns than when it does not?

C3 Test: Do individuals trade more after the    Yes     Yes
US market experiences extremely high returns
than when it does not?

C4 Test: Do institutions trade more after the   --      No
US market experiences extremely high returns
than when it does not?

C5 Test: Do individuals trade more after the    No      No
domestic market experiences extremely high
returns than institutions?

C6 Test: Do individuals trade more after the    Yes     No
US market experiences extremely high returns
than institutions or vice versa?

C7 Test: Do individuals and institutions        Yes     No
trade more after the domestic market
experiences extremely high returns than when
it does not?

C8 Test: Do individuals and institutions        Yes     Yes
trade more after the US market experiences
extremely high returns than when it does not?

C9 Test: Do sc investors, on average, trade
more after the domestic market experiences
extremely high returns than ss investors?

C1O Test: Do sc investors, on average, trade
more after the US market experiences
extremely high returns than ss investors?

Panel D. Causal Relations Conditional on High Investor Sentiment

D1 Test: Do individuals trade more after        No      No
domestic market gains when domestic investor
sentiment is high than it is not?

D2 Test: Do institutions trade more after       No      No
domestic market gains when domestic investor
sentiment is high than it is not?

D3 Test: Do individuals trade more after US     Yes     Yes
market gains when US investor sentiment is
high than it is not?

D4 Test: Do institutions trade more after US    --      No
market gains when US investor sentiment is
high than it is not?

D5 Test: Do individuals trade more after        Yes     Yes
domestic market gains when domestic investor
sentiment is high than institutions?

D6 Test: Do individuals trade more after US     Yes     Yes
market gains when US investor sentiment is
high than institutions?

D7 Test: Do individuals and institutions        No      No
trade more after domestic market gains when
domestic investor sentiment is high than it
is not?

D8 Test: Do individuals and institutions       Yes     Yes
trade more after US market gains when US
investor sentiment is high than it is not?

D9 Test: Do sc investors, on average, trade
more after domestic market gains when
domestic investor sentiment is high than ss
investors?

D1O Test: Do sc investors, on average, trade
more after US market gains when US investor
sentiment is high than ss investors?
COPYRIGHT 2014 Financial Management Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Chuang, Wen-I.; Lee, Bong-Soo; Wang, Kai-Li
Publication:Financial Management
Geographic Code:90ASI
Date:Mar 22, 2014
Words:24824
Previous Article:Institutional shareholder investment horizons and seasoned equity offerings.
Next Article:Emerging market exposures and the predictability of hedge fund returns.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters