Domestic Investors’ Herding Behavior in Reaction To Foreign Trading Yea-Mow Chen Department of Finance San Francisco State University
I. Introduction n n Herding in the stock market is a group of investors’ trading in the same direction over a period of time. Herding could be n n n a rational reaction if it responds to information lagged in the market or an irrational reaction if that follows the trading of a leading group. Few empirical studies have been done on emerging markets where herding behavior could be more significant than in developed markets.
I. Introduction n n Herding could be prompted by international cash flows. International cash flows could bring substantial investment funds to many emerging countries, but make them suddenly vulnerable to international cash outflows. Domestic investors in emerging markets usually take foreign trading as a leading market indicator. If domestic investors were to follow foreign investors’ trading, especially during the period of speculative attacks, local markets might overreact to foreign flows. This phenomenon could explain why domestic investors might have deepened the Asian financial crisis.
I. Introduction n In this study: 1. We design a VAR model to examine how domestic investors react to the foreign investors’ trading in pre- and post-Asian financial crisis periods. 2. We test the causal relationship between foreign investors and domestic investors based on stock market capitalization, average return, volatility of foreign trading volume, and the percentage of foreign investors’ holdings.
II. Literatures on Herding n n II. 1. Herding and Feedback Trading by Individual and Institutional Investors Individual investors who appear ignorant, uninformed, and trading on sentiment are a common theme in herding literature. n n Shiller (1984) and De Long et al. (1990), for example, posit that the influences of fad and fashion are likely to impact the investment decisions of individual investors. Similarly, Shleifer and Summers (1990) suggest that individual investors may herd if they follow the same signals (brokerage house recommendations, popular market gurus, or forecasters) or place greater importance on recent news (overreact).
II. Literatures on Herding Lakonishok, Shleifer, and Vishny (1994) posit that individual investors engage in irrational positive feedback trading because they extrapolate past returns. n Alternatively, Shefrin and Statman (1985) argue that individual investors tend to a negative-feedback trade by selling past winners (the ”disposition effect”). n
II. Literatures on Herding n Much of the empirical evidence focuses on whether individual investors’ herding impacts both closed-end fund discounts and the returns of capitalization stocks. n n The empirical evidence largely supports the hypothesis that there is positive correlation between small firm returns and closed-end fund discounts. There is considerable debate regarding the statistical and economic significance of the correlation (see Lee, Shleifer, and Thaler (1991), Chopra et al. (1993), Chen, Kan, and Miller (1993), Swaminathan (1996), Sias (1997), and Neal and Wheatley (1998)).
II. Literatures on Herding n The empirical evidence also suggests that individual investors’ herding be related to return, that is, individual investors feedback trade. n n n Patel, Zeckhauser, Hendricks (1991), for example, demonstrate that flows into mutual funds are an increasing function of recent market performance. Similarly, Sirri and Tufano (1998) present evidence that individual investors invest proportionately in funds with strong prior performance. Alternatively, Odean (1998) presents evidence that investors are more likely to sell past winners than losers.
II. Literatures on Herding n n n 2. Herding and Feedback Trading by Institutional Investors Institutional herding is primarily responsible for large price movements of individual stocks, and moreover, it destabilizes stock prices. However, Lakonishok, Shleifer, and Vishny (1992) show that institutional herding affecting prices does not necessarily imply that it is destabilizing. If, for example, institutional investors are better informed than individual investors, institutional investors will likely herd to undervalued stocks and away from overvalued stocks. Such herding can move price toward, rather than away from equilibrium value (Froot, Schartstein, and Stein (1992), Bikhchandani, Hirshleifer, and Welch (1992), and Hirshleifer, Subrahmanyam, and Titman (1994)).
II. Literatures on Herding n n n II. 3. Herding in Emerging Markets Kim and Wei (1999), and Choe, Kho and Stulz (1999) suggested that before the Asian crisis, domestic institutional investors in Korea tended to negativefeedback trade by buying recent losers and selling recent winners, but foreign institutional investors tend to positivefeedback trading. However, during the crisis period, Kim and Wei (1999) demonstrated that those foreign or domestic institutional investors were positive feedback traders and that those foreign institutional investors’ herding behavior was more significant.
II. Literatures on Herding n n n In contrast, Choe, Kho and Stulz (1999) found no evidence that foreign investors were positive feedback traders. Therefore, foreign feedback trading couldn’t destabilize the stock prices. Chang, Cheng, and Khorana (1999) demonstrate that macroeconomic information could affect the decision making process of investors because those investors had incomplete firm-specific information in emerging markets where the evidence of herding was found. Wei, Liu, Yi and Su (1997) found that there was a significant herding effect of the institutional investors, especially between qualified foreign institutional investors (QFIIs) and domestic mutual fund companies in Taiwan stock market.
III. Methodology n In the study, we use the Granger VAR model to examine the reaction of local investors to the trading of foreign investors, and to test whether positive feedback trading on the part of local investors as compared to the foreign trading would provide some evidence that local investors deepened the Asian crises.
III. Methodology n n n III. 1. International Cash flows in Taiwan Until 1983, foreign investors were prohibited from directly investing in the Taiwan stock market. They were allowed to invest in Taiwan only through the purchase of beneficial certificates of financial assets. Direct investing into the Taiwan stock market was opened to qualified foreign institutional investors (QFIIs) in December 1990 and to general foreign investors in 1996. In March 1996, QFIIs were able to hold up to 20% of any company listed in the Taiwan stock market. Net international cash flows increased since 1991 and the net cash flows peaked in April 1997, with a single month of total stock purchase/sale of NT $166 billion. When the Asian crisis broke out in April 1997, the foreign trading started to cool off and large outflows began in July 1997, resulting in a single month net foreign sales of NT $19. 3 billion. The peak of foreign outflows was in Oct. 1997, resulting in a single month of net sale NT$30. 4 billion.
III. Methodology n n n III. 2. Data 40 stock that were most frequently traded by foreign investors based on QFII net purchase/sale volume as well as stock prices from January 05, 1995 to November 28, 2000 were obtained from the Taiwan Economic Journal’s Great China Database. We chose the date of July 22, 1997 as the event date to study the impact of the Asian crisis on the Taiwan market. Large capital outflows from the Taiwan Stock Exchange started from this day and on.
III. Methodology n We also divided the sample into 4 groups: n n n (1) stock capitalization; (2) the daily turn of stocks; (3) the volatility of QFII net purchase/sale volume (4) QFII holding percentage. We also used VAR to retest each group as an individual portfolio on 100 days and 500 days pre- or post- The Asian crisis.
III. Methodology n n n n The ten groups are classified: P 1: companies with smaller market capitalization; P 2: companies with larger market capitalization among 36 stock companies; P 3: companies with lower daily average return; P 4: companies with higher daily average return; P 5: companies with lower volatility of QFII net purchase/sale volume based on the absolute volume of QFII net purchase/sales; P 6: companies with higher volatility of net purchase/sale volume; P 7 and P 8: companies with lower and higher volatility based on the standard deviation of QFII net purchase/sale volume, respectively;
III. Methodology n n n III. 3. The Granger Causality Test The Granger approach to the question of whether X causes R is to see how much of the current R can be explained by past values of R and then to see whether adding lagged values of X can improve the explanation. In the context of on study, we test whether the information of QFII net purchase/sale volume could predict stock return or the information of stock return could predict QFII net purchase/sale volume.
III. Methodology n n K Ri, t = a 0 + ak Ri, t-k + bk Xi, t-k + t (1) K=1 n n n K K=1 K Xi, t = c 0 + ck Ri, t-k + dk Xi, t-k + t (2) K=1 Where Ri, t is the return of stocks, X i, t is the QFII net purchase/sale volume and t is the column vector of residuals.
III. Methodology n According to the Akaike Information Criteria (AIC) and Schwarz Criteria (SC) tests, we find that the proper number of lag in this test should be two because the smaller the value of the information criteria—AIC and SC—the better the model.
III. Methodology n n n To verify the Granger VAR results further, we run the same VAR estimation on each individual stock in the sample over four sub-periods— 100 -day and 500 -day precrisis periods, and 100 -day and 500 -day post-crisis periods. In addition, we retest the VAR on 10 sub-groups, classified by market capitalization, average return, the volatility of QFII net purchase/sale and holding percentage of QFII in 100 -day and 500 -day base. We also retest each group as an individual portfolio in 500 day pre-crisis and post-crisis periods.
IV. Result n n n Table 1 shows stock return leads QFII net purchase/sale volume for the whole sample period, 500 -day pre- and post -crisis periods, and 100 -day post-crisis periods; but QFII net purchase/sale volume causes the return for the wholesample period and 500 -day pre-crisis period. For the whole-sample period, returns and foreign investors' trading can affect each other because both of them have significant F tests. When we compare between the pre-crisis and post-crisis periods, we find that stock returns and QFII net purchase/sale volume could lead each other 500 -day before the crisis, but only return cause foreign investors' trading in 500 -day after the crisis period.
IV. Result n In the 100 day pre- and post-crisis tests, the result displays that only the return can lead QFII net purchase/sale volume in 100 days after the crisis period and no significant result to prove that QFII net purchase/sale volume can lead return in 100 days pre- and post-crisis periods.
IV. Result n n n The foreign daily trading is used as sources of information that lead to domestic trading. If the foreign trading leads domestic trading, there should be a causal relationship between foreign trading and domestic trading. If investors are likely to positive-feedback trade, they would be likely to herd because they would buy and sell as group. From Table 1, we know that the impact of return to foreign trading as well as the impact of foreign trading to return are significant in all 500 -day tests. The impact of return is consistent with foreign trading in 100 -day after crisis period.
IV. Result n n n It takes about two days for the local market returns to react on foreign trading (t-test is – 0. 57956) and takes about two days for the foreign trading to react on return (t-test is 0. 59048) in the 500 days before-crisis period. It takes two days for the foreign trading to react on return during the 500 -day and 100 -day after crisis periods (t-test are – 1. 92748 and – 2. 03257, respectively). All positive responses of stock return and QFII net purchase/sale volume in different periods provide that the return does positively impact upon the foreign trading and foreign net purchase/sale volume does positively impact upon the stock return in Taiwan.
IV. Result n When foreign investors buy stocks, domestic investors follow foreign investors to buy stocks as well. As the impact of return to foreign investors has positive responses, foreign investors are likely to be positively feedback trading. When the price increases, foreign investors buy stocks and vice versa. This result coincides with the findings of Chen and Gong (1999).
IV. Result n n n The large-capitalization group has significantly more firms indicating the impact of QFII net purchase/sale volume to the return in 500 -day before crisis period. 50% (4) more companies in P 2 have the impact of foreign trading to return than P 1 in 500 -day before crisis period, but 33% (1) more companies in P 2 has this impact than P 1 in 500 -day after crisis period. 9 out of 18 firms in the largecapitalization group, P 2, has the impact of return to foreign trading and 5 out of 18 firms in P 1 has this impact in the 500 -day pre-crisis period. In post-crisis period, 7 out of 18 firms in P 1 has the impact of return to foreign trading and 1 out of 18 firms in P 2 has this impact. This result proves that the small-capitalization group has significantly more firms indicating the impact of stock return to the foreign trading in the post-crisis period. In
IV. Result n n n We classified the average-return group for P 3 and P 4 to see the impact of return and QFII net purchase/sale volume between high-return firms and low-return firms. In the 500 -day before crisis period, there was not enough difference in the result indicating that the impact of foreign trading would show returns between low-return and highreturn groups; however, the high-return group, P 4, had significantly two more firms indicating the impact of return to foreign trading than the low-return group, P 3. In the 500 -day after crisis period, the high-return group, P 4, had significantly more firms indicating the impact of foreign trading to return and the impact of return to foreign
IV. Result n n n We classify the group based on the volatility of QFII net purchase/sale volume—sum of QFII net purchase/sale absolute volume and standard deviation of QFII net purchase/sale volume—and we get P 5, P 6 and P 7, P 8 groups. In the 500 -day before crisis period, the high-volatility of QFII net purchase/sale volume group had more firms detecting the impact of return to foreign trading and the impact of foreign trading to return. To our surprise, low-volatility of QFII net purchase/sale volume group had more firms indicating the impact of return to foreign trading in the 500 -day after the crisis period. No apparent result shows that either the low or the high-volatility group had more firms indicating the impact of foreign trading to return.
IV. Result n The groups P 9 and P 10 are classified by holding a percentage of foreign investors. Table 4 shows that a highholding percentage of foreign investors, P 10, had more firms indicating the significantly impact of return to foreign trading and the effect of foreign trading to return in 500 -day before crisis period. In 500 -day after crisis period, P 9 and P 10 had almost the same number of firms explaining the impact of return to foreign trading and P 10 had more firms to indicate the effect of foreign trading to return.
IV. Result n n We were concerned about what length of periods could affect the results we found above. We also did the Granger VAR test in the 100 day based as the short-term period. Table 5 shows the result. We found the result of the causal relationship between return and QFII net purchase/sale is almost the same as 500 -day base. However, more firms in each group have the significant causal relationship between return and QFII net purchase/sale in 500 -day before crisis period than the relationship in the 100 -day before crisis. In Table 4, 14 out of 36 firms had the significant impact of return to foreign trading and 20 out of 36 firms had the significant impact of foreign trading to return in the 500 -day before crisis period.
IV. Result n n In Table 5, we found 10 out of 36 firms had the impact of return to foreign trading and 7 out of 36 firms had the significant impact of foreign trading to return in 100 -day before crisis period. When we looked at the causal relationship in the post-crisis period, we found 8 out of 36 firms had the significant impact of return to foreign trading and 7 out of 36 firms had significant impact of foreign trading to return disregarding the 100 -day or 500 -day tests. Therefore, 100 -day test is less accurate than the 500 -day results.
IV. Result n n We retested the Granger VAR on each group as an individual portfolio in the 500 -day before and after crisis periods. Table 6 shows the result of the causal relationship between return and foreign trading. Before the Asian crisis, the return leads QFII net purchase/sale volume in all marketcapitalization portfolios. The high-return portfolio had the significant impact of return to foreign trading, but the small -return portfolio had no compelling results to prove the impact of return caused foreign trading.
IV. Result n n n The low-volatility of foreign trading portfolios, P 5 as well as P 7, had significant impact of return to foreign trading, but the return in highvolatility of foreign trading portfolios were not consistent with foreign trading. P 10, the portfolio with high-holding percentage of foreign investors, had the significant impact of return to QFII net purchase/sale volume, but there was no significant impact of return to foreign trading in P 9 portfolio. The result of QFII net purchase/sale volume to return showed that the p -value of P 2 is 0. 0055%, P 6 is 0. 0073% and of P 8 is 0. 104%. The foreign trading caused the return in large-capitalization, P 2, highvolatility portfolios, P 6 and P 8. There was no significant impact of foreign trading to return on the small-capitalization and the lowvolatility portfolio before the Asian crisis.
IV. Result n n After the Asian crisis, we found that return leads the foreign trading in all portfolios; but only P 7, the lowvolatility portfolio, had slightly significant impact of foreign trading to return. This result showed that domestic trading led the foreign trading after The Asian crisis; however, foreign trading didn’t lead the domestic trading. Since foreign investors can’t get more effective information regarding the Taiwan market than the domestic investors, they herd more significantly than domestic investors. This result coincides with the finding from Chang, Cheng, and Khorana.
V. Conclusion n n This paper studies the impact of foreign trading in regards to the Taiwan stock returns. Empirical study displays that before the Asian crisis, the foreign trading could lead the return as well as be caused by the return; however, there was no compelling result to prove that foreign trading leads the return after the crisis. Domestic investors take two days to positively react to the foreign trading. Foreign investors are positive feedback traders—when the stock price increases, they buy the stocks and when the stock price decreases, they sell the stocks.
V. Conclusion n We classified 36 firms into 10 groups based on market capitalization, average return, the volatility of foreign trading, and holding percentage of foreign investors in 500 -day and 100 -day pre- and post-crisis periods. We found that the result in 500 -day test is more accurate, especially in 500 -day before crisis period. The large-capitalization, high-return, high-volatility and high-holding percentage of foreign investors groups had significantly more firms indicating the impact of return to foreign trading as well as the impact of foreign trading to return before crisis.
V. Conclusion n After the Asian crisis, the small-capitalization, high-return, low-volatility groups have significantly more firms indicating the impact of return to foreign trading. The large-capitalization, high-return, high-volatility and highholding percentage of foreign investor groups have significantly more firms indicating the impact of foreign trading to return after crisis.
V. Conclusion n We also tested the Granger VAR on each group as an individual portfolio. The finding was that return leads the foreign trading pre- and post-crisis periods and foreign trading affects the return before crisis. However, there is no compelling result to prove that foreign trading leads the return. Before the Asian crisis, foreign trading affected the return in the large-capitalization, high-volatility portfolio, but this effect could be seen in the low-volatility portfolio after the crisis.
<Table 1> The result of the Granger VAR in the different periods. n We obtain the daily 40 most frequently traded stocks by foreign investors provided by the TEJ Data Base based on QFII net purchase/sale volume as well as stock price which will be used as source of information that lead to domestic trading in Taiwan from January 05, 1995 to November 28, 2000. If foreign trading leads domestic trading, there should be a casual relationship between foreign trading and domestic trading. After the Asia crisis broke out on July 02, 1997 in Thailand, its baht suddenly depreciated a lot and baht devaluation erupts other countries’ economy crisis. We choose July 22, 1997 as the event day of Asia crisis in Taiwan market because the large capital outflows from the Taiwan Stock Exchange started at this day after July 02, 1997. We use the event day classify the two periods—precrisis and post-crisis in order to compare the effect of foreign investors’ trading to domestic investors’ decisions.
<Table 4> The result of the Granger VAR for each group in the 500 -day pre-crisis and post-crisis periods n The number of firms affected in each of the 10 groups for each of four different categories of exposure. The ten groups are classified as P 1 represents companies with smaller market capitalization; P 2 represents companies with larger market capitalization among 36 firms; P 3 represents companies with lower daily average return; P 4 represents companies with higher daily average return; P 5 represents companies with lower volatility of foreign trading based on the absolute volume of QFII net purchase/sale volume; P 6 represents companies with higher volatility of foreign trading; P 7, P 8 represent companies with lower and higher volatility calculating by the standard deviation of QFII net purchase/sale volume, respectively; P 9 and P 10 separately represent companies having small and large holding percentage of foreign investors.
<Table 5> The result of the Granger VAR for each group in the 100 -day pre-crisis and post-crisis periods n The number of firms affected in each of the 10 groups for each of four different categories of exposure. The ten groups are classified as P 1 represents companies with smaller market capitalization; P 2 represents companies with larger market capitalization among 36 firms; P 3 represents companies with lower daily average return; P 4 represents companies with higher daily average return; P 5 represents companies with lower volatility of foreign trading based on the absolute volume of QFII net purchase/sale volume; P 6 represents companies with higher volatility of foreign trading; P 7, P 8 represent companies with lower and higher volatility calculating by the standard deviation of QFII net purchase/sale volume, respectively; P 9 and P 10 separately represent companies having small and large holding percentage of foreign investors.
<Table 6> The result of the Granger VAR for each portfolio in the 500 -day pre-crisis and post-crisis periods n The portfolios for each of four different categories of exposure. The ten portfolios are classified as P 1 represents companies with smaller market capitalization; P 2 represents companies with larger market capitalization among 36 firms; P 3 represents companies with lower daily average return; P 4 represents companies with higher daily average return; P 5 represents companies with lower volatility of foreign trading based on the absolute volume of QFII net purchase/sale volume; P 6 represents companies with higher volatility of foreign trading; P 7, P 8 represent companies with lower and higher volatility calculating by the standard deviation of QFII net purchase/sale volume, respectively; P 9 and P 10 separately represent companies having small and large holding percentage of foreign investors.
<Table 2> The result of the estimate VAR in the pre-crisis and the post-crisis periods n n Sample (adjusted): 3 500 in the 500 -day period; 3 100 in the 100 -day period included observations: 498 in the 500 -day period; 98 in the 100 -day period after adjusting Endpoints, standard errors & t-statistics in parentheses
<Table 3> The mean, standard deviation and number of stocks in each group n The ten groups are classified as P 1 represents companies with smaller market capitalization; P 2 represents companies with larger market capitalization among 36 firms; P 3 represents companies with lower daily average return; P 4 represents companies with higher daily average return; P 5 represents companies with lower volatility of foreign trading based on the absolute volume of QFII net purchase/sale volume; P 6 represents companies with higher volatility of foreign trading; P 7, P 8 represent companies with lower and higher volatility calculating by the standard deviation of QFII net purchase/sale volume, respectively; P 9 and P 10 separately represent companies having small and large holding percentage of foreign investors.
Скачать презентацию Domestic Investors Herding Behavior in Reaction To Foreign
51709c6eff810be5add1fa755618b5d8.ppt