Trading Behavior, Performance, and Stock Preference of Foreigners, Local Institutions, and Individual Investors: Evidence from the Korean Stock Market

Authors


  • Acknowledgments: The authors wish to thank Larry Bajor, Joon Chae, Lucy Chernyk, Sung Wook Joh, Bong Chan Kho, Daniel P. Klein, Frank Laatsch, Christopher Ting, session participants at the Financial Management Association meetings and the Annual International Conference on Asia-Pacific Financial Markets, seminar participants at Seoul National University, an anonymous reviewer and the Special Issue Editor (S. Ghon Rhee) of the Journal for many valuable comments, and Hana Bae for extensive editorial help. Bae and Min acknowledge the financial support from the College of Business Administration Summer Research Grant program at Bowling Green State University and from Seowon University, respectively.

Corresponding author: Department of Finance, College of Business Administration, Bowling Green State University, Bowling Green, OH 43403, USA. Tel: 1-419-372-8714, Fax: 1-419-372-2527, email: bae@bgsu.edu.

Abstract

We examine the trading behavior and performance of foreigners, local institutions, and individual investors in the Korean stock market. The key research issue is whether the commonly-documented information disadvantage of foreign investors translates into their underperformance relative to local institutional and individual investors. Our results show the opposite, that the stocks foreigners buy significantly outperform the stocks they sell in terms of both stock returns and operating profitability, leading to the significant outperformance of foreigners’ trading strategies over those of local investors. Our results provide strong evidence that the superior performance of foreigners is attributed to their ability to discern between company stocks with good versus bad, at least short-term, prospects. Our findings on the trading behavior of investors in the Korean market are, in general, consistent with those for other markets documented in the published literature. Foreigners behave like short-term momentum traders pursuing a growth strategy. Local institutions also trade like momentum traders but tend to buy value stocks. In contrast, individual investors trade like contrarians who buy past losers and sell past winners. Our findings show that foreigners prefer large-cap stocks with high dividends. In sharp contrast, individual investors have a strong preference for small-cap, high-leverage, low dividend paying stocks, whereas local institutions tend to buy small-cap, low leveraged stocks.

1. Introduction

In the present study, we provide new empirical evidence on the trading behavior and performance of investors in one of the fastest growing stock markets. In particular, we investigate whether foreign investors trade and perform differently relative to domestic institutional and individual investors in the Korean stock market. Given the increasing stock ownership of foreign investors and their significant influence on both individual stock prices and overall market movements, the Korean stock market provides a unique opportunity to investigate the different trading behavior and performance of foreign investors relative to local investors.

In general, institutional investors follow momentum or positive feedback trading strategies by buying past winners and selling past losers,1 and prefer larger and more liquid stocks, mainly because of liquidity and transaction costs (Falkenstein, 1996; Gompers and Metrick, 2001). In contrast, individual investors behave like anti-momentum traders or contrarian investors and have a general disposition to sell winners too early and hold losers too long (Shefrin and Statman, 1985; Odean, 1998; Barber and Odean, 2000; Griffin et al., 2003).2

The existing literature also provides evidence on the different trading behavior of foreign investors from local individual investors. Brennan and Cao (1997) show that foreign investors follow momentum trading strategies because of their information disadvantage in the local markets. Similar evidence is documented by Kang and Stulz (1997) on the Japanese market, Choe et al. (1999) on the Korean market during the 1997 Asian financial crisis, Froot et al. (2001) on the US market, Dahlquist and Robertsson (2001)3 on the Swedish market, and Kalev et al. (2008) for stocks traded on the Helsinki stock exchange. Consistent with their information disadvantage in the local markets, foreign investors are also found to prefer large and well-disclosed firms’ stocks.

With regard to the performance of foreigners relative to domestic investors, however, the existing studies provide inconclusive evidence. On the one hand, domestic investors are shown to have higher profits because of their information advantage inherent with geographical proximity, and linguistic and cultural barriers. Malloy (2005) finds that the information advantage of geographically proximate analysts in the analysis of remotely located small firms in the US market results in better performance for local investors. Dvorak (2005) also shows similar evidence for the Indonesian stock market.4Choe et al. (2005) find that prices tend to move more unfavorably for foreign investors than for domestic investors immediately before trading and that this difference is partly explained by the return-chasing behavior of foreigners due to their information disadvantage. Kalev et al. (2008) find that local investors on the Helsinki stock exchange have information advantages in their home market both in the short term and in the long term, which leads local investors to outperform foreign investors.5Agarwal et al. (2009) show that although foreign investors generally underperform domestic investors in the Indonesian stock market, their performance depends on the type of orders (initiated or non-initiated) and their counterparts. The inferior performance of foreign investors is explained by their aggressive trading behavior, rather than by their information disadvantage or poor trading timing.

On the other hand, foreign investors are found to generate higher profits primarily owing to their information-processing advantages associated with sophisticated analytical skills and greater experience. Grinblatt and Keloharju (2000) find that foreign investors act like momentum investors, with high levels of sophistication, and that they outperform local investors in the Finnish stock market. Bae et al. (2006) report that foreign investors in the Japanese stock market consistently generate large trading gains from good market timing. Their findings suggest that although foreign investors invest like less informed investors, they possess greater ability to analyze and forecast stock returns. Oh et al. (2008) compare the trading behavior and performance of online equity investors to non-online equity investors in Korea. They show that online investors perform poorly compared to non-online investors, and that, among investor types, foreigners have better results than local investors.

Although there is some evidence on these issues for the Korean stock market, the evidence is hardly conclusive. Furthermore, the evidence is, to a large extent, inconsistent with the common belief that foreigners generate substantial profits from stock trading at the expense of domestic investors in the Korean stock market.6 Our study aims to document new and robust evidence on these two issues in the Korean stock market, which have not been explored fully in the existing published literature. First, we examine whether foreigners do, in fact, achieve better stock performance than domestic institutions and individuals in the Korean stock market, even in the face of information disadvantage. We intend to test this common belief in the Korean stock market over an extended period. For this purpose, we first construct “Buy” and “Sell” portfolios based on trade imbalances of each investor group from January 1996 to December 2002 and compare their stock return performances. Second, we investigate whether foreign investors in the Korean stock market use different trading strategies and portfolio management techniques to domestic institutional and individual investors. In particular, we test whether foreign investors are more skilled with respect to stock selection and trading timing. Our analysis will identify trading strategies and patterns peculiar to each investor group and compare them to trading patterns observed in other markets, such as momentum trading by institutional investors and contrarian trading by individual investors. Our analysis will also shed light on the characteristics of individual stocks that each investor group has a preference for in the Korean market.

The key research issue in our study is whether the commonly-documented information disadvantage of foreign investors in the Korean stock market translates into underperformance of foreign investors relative to domestic institutional and individual investors.7 If this hypothesis is not supported, this will lead us to test our alternative hypothesis that foreign investors have the superior ability to gather and process new information on performances of individual Korean companies through sophisticated portfolio management techniques and their worldwide information networks, and, therefore, can better transform this information into profitable stock transactions, as noted in Dvorak (2005) and Bae et al. (2006). Although local institutions in Korea might be in a better position to access and interpret firm-specific information than foreigners, they might not outperform foreigners due to several inherent weaknesses, such as the relatively short history of the Korean stock market and the lack of modern asset management techniques and experienced stock analysts. Similarly, individual investors as a whole are presumed to achieve the worst performance among the three investor groups, mainly due to their diverse opinions toward firm-specific information and their inferiority in information processing.

Our empirical findings generally support the common belief and the alternative hypothesis regarding foreign investors in the Korean stock market. We find that foreigners in the Korean stock market behave like positive feedback traders: they tend to buy stocks that have outperformed and sell stocks that have underperformed relative to the market. Similar to foreigners, local institutional investors show a tendency to buy more stocks that have risen in price. In contrast, individual investors trade like contrarians: they tend to buy the past losers and sell the past winners. These findings are consistent with those in the existing published literature (see e.g. Brennan and Cao, 1997; Grinblatt and Keloharju, 2000; Froot et al., 2001).

Regarding return performance following the portfolio formation, we find that foreigners, on average, significantly outperform both domestic institutions and individual investors, as supported by our results that the stocks foreigners buy most significantly outperform the stocks they sell most in a subsequent year. The contrarian strategy followed by individual investors, however, does not lead to a meaningful return performance, and the stock selection of local institutions results in only a mediocre return performance. We also find that foreigners prefer large-cap stocks with high dividend and high operational efficiency, whereas individuals tend to invest in small-cap, high-leveraged stocks with relatively low profitability. Interestingly, a firm’s leverage level is not an important factor in stock selection by foreigners in the Korean market. Unlike the practice in the US market, local institutions in Korea tend to buy small-cap stocks with low leverage and sound operational performance. Overall, our results provide strong evidence supporting the common belief that foreign investors have been consistent winners in the Korean stock market.

2. Related Studies and Characteristics of Foreign Investors in the Korean Stock Market

2.1. Related Studies

Institutional transactions have drawn much interest in the published literature over the past years due to the notion that their alleged momentum (positive feedback) trading and herding might destabilize the stock market or make the market more efficient. On the one hand, momentum trading can overshoot the stock price above and beyond its intrinsic value and, in turn, cause excessive volatility through price reversal in subsequent periods. On the other hand, if positive feedback trading reflects information concerning corporate fundamentals, it can help speed up price adjustments to new information.

There have been extensive studies published on the trading behavior of institutional investors and the profitability of momentum strategies. Jegadeesh and Titman (1993) first report that short-run momentum strategies from buying past winners and selling past losers produce positive abnormal returns over the following 3–12 months after controlling for size and risk. Chan et al. (1996) show that the probable source of profitability of momentum strategies is a delayed price reaction to firm-specific information, such as earnings announcements. According to their study, market risk, size, and book-to-market effects do not explain the drift in future returns, which is economically meaningful and lasts for at least 6 months. Rouwenhorst (1998) finds that short-term momentum is observed not only in the US market but also prevails in other developed stock markets. He also reports that systematic risk and size cannot explain this worldwide phenomenon of short-term return continuation.

Several previous studies provide differing evidence regarding the positive feedback trading by institutional investors. Using quarterly data from 155 mutual funds over 1975–1984, Grinblatt et al. (1995) document that mutual funds with prior superior quarterly returns tend to outperform in the following quarters. They find that the momentum strategy, especially buying past winners, is more effective than the contrarian strategy in the US market. Bohn and Tesar (1996) find that net purchases by US investors are significantly positively correlated with local capital gains and excess returns in most of the large international equity markets, suggesting that US investors tend to be return chasers in international equity markets. They report, however, that this positive feedback trading does not bring in future excess returns. Nofsinger and Sias (1999) show that institutional investors adjust their stock holdings based on past return performance. Because of this positive feedback trading, they find strong positive relations between institutional trading and subsequent returns by which the stocks with net institutional buy outperform the stocks with net institutional sell. They find no return reversals (no overreaction) within up to 2 years after institutional trading. Badrinath and Wahal (2002) also document that institutional investors adjust their portfolios based on past stock return performance. Interestingly, they find that institutional investors act like momentum traders when they purchase but trade like contrarians when they dispose of their holding positions, indicating that institutional investors tend to purchase past winners and simultaneously sell overperforming stocks. They also show, however, that the momentum trading by institutional investors does not generate excess returns because trading behavior and cycles vary across different types of institutions. Based on their study on the tax exempt (pension) equity funds over 1985–1989, Lakonishok et al. (1992) show that institutional investors invest primarily in large-cap stocks, but positive feedback trading and herding prevails in small stocks rather than large-cap stocks. In addition, they document that future quarterly returns are higher for stocks with net institutional selling than those with net institutional buying.

Several published studies attempt to relate the trading behavior of institutional investors to their preference for stock characteristics. In a cross-sectional analysis of New York Stock Exchange and American Stock Exchange stock holdings of US open-end mutual funds over 1991–1992, Falkenstein (1996) shows that US mutual funds have a strong preference for stocks with large market capitalization and high trading volume. Gompers and Metrick (2001) find that as representative investors in the US market shift from individuals to institutional investors, the institutional demand for large-company stocks has gradually increased, mainly due to liquidity and transaction costs, whereas the relative demand for small-company stocks has decreased in the US market over the past two decades. They also show that large-company stocks significantly outperform small-company stocks, resulting in the disappearance of the small-firm stock premium since 1980. They further find that the level of institutional ownership is useful in forecasting individual stock returns. Based on their findings of a negative correlation between institutional ownership and past stock returns, however, they argue that institutions are not momentum investors.

Under informational asymmetry, foreign investors might not invest in mean–variance efficient or well-diversified portfolios. Merton (1987) posits that due to high information costs, rational investors would prefer firms about which they have better information. Kang and Stulz (1997) find that in the Japanese stock market, foreign investors prefer to invest in large manufacturing firms with good operating performance, low unsystematic risk, and low debt to total assets. Even after controlling for firm size, they find that foreign investors tend to invest in stocks of firms with export-oriented sales structures, high liquidity, and high visibility, such as American Depository Receipt-issuing firms. In their study of the Swedish stock market, Dahlquist and Robertsson (2001) find that foreign investors are typically institutional investors, and that the stock ownership of institutional investors is positively associated with firm characteristics such as firm size, cash flows, market liquidity, and firm recognition in the international markets. In contrast, institutional foreign investors show an aversion toward firms with high risk, high leverage, high dividend yields (low growth) and high ownership concentration. As a result, Dahlquist and Robertsson insist on the presence of an institutional investor bias rather than a foreign investor bias.

In their study of the Korean stock market, Choe et al. (1999) find strong evidence of positive feedback trading and herding by foreign investors and contrarian trading by Korean individual investors before the financial crisis in 1997. Their results suggest that the positive feedback trading by foreigners is driven mainly by individual stock returns rather than market returns. They further show that during the financial crisis, foreign investors remain as positive feedback traders at the individual stock level but become significantly contrarian at the market level, and that individuals are net buyers with more of the stocks that have performed poorly in the crisis period. The authors conclude that foreign trades are not destabilizing the market because large trades by foreigners are matched by trades by Korean individuals in the reverse direction. In an analysis of US mutual fund flows into 44 countries from 1994–1998, Froot et al. (2001) find evidence of positive feedback trading by international investors. They show that international portfolio flows are strongly influenced by past returns and are highly persistent in time, and that fund inflows have positive forecasting power for future equity returns, especially in the emerging markets.

Malloy (2005) shows that geographically proximate analysts in the US equity markets outperform their distant counterparts in analyzing small firms located in remote areas, and that this information advantage of proximate analysts results in better performance for local investors. Dvorak (2005) finds that domestic investors in the Indonesian market have, on average, higher profits than foreign investors. His results also suggest that the combination of local information and global expertise leads to higher profits. Choe et al. (2005) find that stock prices tend to move more unfavorably for foreign investors than for domestic investors immediately before trading. They suggest that this difference is partly explained by the return-chasing behavior of foreign investors due to their information disadvantage. Examining the top 35 stocks on the Helsinki stock exchange, Kalev et al. (2008) show that local investors have an information advantage in their home market, both in the short and long term, which results in local investors outperforming foreign investors, with the exception of stocks that are well-known internationally well-known, such as Nokia. In contrast, Bae et al. (2006) report that although foreign investors in the Japanese stock market trade like less informed investors, they possess greater ability to analyze and forecast stock returns, which allows foreign investors to consistently generate large trading gains from good market timing. Agarwal et al. (2009) report mixed performances of foreign and domestic investors, and state that performance depends on whether initiated or non-intiated orders are placed and who investors’ counterparts are. They further show that the mixed performance can be explained by the degree of aggressiveness of investors, not by information advantage or disadvantage or by poor trade timing.

Several prior studies have examined the trading behavior of individual investors in comparison with institutional investors. Shefrin and Statman (1985) propose the disposition effect hypothesis that due to aversion to loss realization under uncertainty, individual investors have a general disposition to sell winners too early and to hold losers too long. They demonstrate that tax consideration alone does not explain this effect. Odean (1998) investigates 162 948 trades of 10 000 accounts from nationwide discount brokerage houses over 1987–1993 and finds that individual investors have a strong preference for realizing winners rather than losers. According to his study, individual investors are presumed to have an unwarranted belief that their current losers will outperform their current winners in the future; however, the winning investments that investors choose to sell continue to outperform the losers they keep in subsequent months. Using transaction data for 66 465 households with large discount brokers over 1991–1996, Barber and Odean (2000) find that individuals, on average, earn an annual return of 16.4% compared to the annual market return of 17.9%. They show that individual investors’ poor performance results not from portfolio selection, but from the cost and frequency of trading: individuals who trade most frequently underperform most severely due to high transaction costs. They further show that individual investors are anti-momentum traders, who tend to hold stocks that have recently underperformed the market. According to their study, individual investors tend to prefer high-beta, small-cap, value (high book to market equity) stocks.

Based on daily and intradaily returns of NASDAQ 100 securities, Griffin et al. (2003) find that, on the daily basis, institutional investors are primarily momentum traders, while individual investors are typically contrarian investors. They find, however, little evidence that the typical trading behavior of either institutional or individual investors can predict future returns in a systematic manner. In contrast to other published studies, Bange (2000) shows that individuals are positive feedback traders. Based on survey data by small individual investors over 1987–1994, she finds that individuals increase their equity holdings after market run-ups and decrease their holdings after market downturns, suggesting a positive relation between changes in investor sentiment and changes in individual equity holdings. She reports, however, that individuals are unable to time the stock market successfully and, even worse, predict the market systematically in the wrong direction. She also finds that stocks owned by individual investors are biased toward small companies.

2.2. Foreign Investors in the Korean Stock Market

Korea is one of the fastest-growing emerging stock markets, with its share trading value amounting to $1,607bn in 2010, next to Japan, and China in the Asia-Pacific region.8 Since the stock market liberalization in 1992, the cross-border investment by foreigners has gradually increased in the Korean stock market. Table 1 shows this upward trend in stock transactions on the Korea Stock Exchange (KSE) by foreign investors. In 2002 alone, foreigners accounted for approximately 12% of total trading volume, which is comparable to the trading volume by domestic institutions. Moreover, the market value of total shares that foreigners held in 2002 accounted for 36% of the total market capitalization of listed companies on the KSE. According to the Korean Financial Supervisory Service Annual Report, approximately 66% of the registered foreign investors in Korea in 2002 were institutional investors, such as mutual funds and pension funds, and more than half are from the USA, the UK and Japan.

Table 1.   Trends in transactions in the Korean stock exchange by investor type
All transaction amounts are in billion Korean won. The proportions of transaction amount by each investor type relative to total transaction amount are in parentheses. Source: The Korea Stock Exchange.
YearForeign investorsLocal institutionsIndividual investorsOthersTotal
199617 173 (6.0%)62 259 (21.8%)200 003 (70.1%)5816 (2.1%)285 251 (100%)
199721 698 (6.7%)56 575 (17.4%)239 960 (73.9%)6329 (2.0%)324 562 (100%)
199828 816 (7.5%)47 349 (12.3%)298 480 (77.3%)11 045 (2.9%)385 690 (100%)
199989 414 (5.2%)278 514 (16.1%)1 320 282 (76.1%)45 637 (2.6%)1 733 847 (100%)
2000114 951 (9.2%)205 079 (16.4%)902 166 (71.9%)32 070 (2.5%)1 254 266 (100%)
2001102 970 (10.5%)138 305 (14.1%)719 478 (73.2%)21 978 (2.2%)982 731 (100%)
2002170 532 (11.5%)204 161 (13.8%)1 065 646 (71.8%)43 962 (2.9%)1 484 401 (100%)

Trades by foreigners have a substantial influence on both individual stock prices and overall market movements, sometimes more than domestic institutions and individual investors do. Furthermore, their great influence in the Korean stock market has fostered an unwarranted belief that foreigners are always ultimate winners, while individual investors and, sometimes even domestic institutions, are losers. Foreigners reportedly have different transaction styles to local institutions and individuals; in fact, it is observed that foreign investors trade in the opposite way to domestic institutions or individuals. Excerpts of news briefings from Korean financial authorities support this conjecture:

Foreign investors’ unrealized capital gains are estimated to add up to over KRW 1.2 trillion from Dec. 11, 1997 to Jan. 31, 1998 shortly after raising the foreign ownership ceiling up to 50%. Foreign investors bought 265 stocks and 102 million shares in net, while domestic institutions and individuals lost the profit opportunities by focusing on selling shares during the same time (Korea Stock Exchange Briefing, 3 February 1998).

Domestic investors have been putting more and more money in the equity related financial products. Net cash inflows of about KRW 12 trillion are estimated to be invested in the equity funds from July 1 to July 19, 1999. However, foreigners are gradually disposing of their stock holdings and are estimated to repatriate U.S. 387 million dollars over the same period (Bank of Korea, Weekly Economic Bulletin, 21 July 1999).

Those stocks bought in net by foreigners for consecutive 30 days rose, on average, 17%, beating the market index by 6% in this year. Those stocks bought in net by domestic institutions underperformed the market, gaining only 7%. Individuals even lost 5% from the stocks they invested (Korea Stock Exchange News, 8 March 2004).

3. Analysis of Portfolio Return Behavior by Investor Type

3.1. Return Data

Our study uses daily stock transaction data of Korean companies listed on the KSE from January 1996 to December 2002. This sample period covers the financial crisis that occurred at the end of 1997 and the boom and bust periods of high-technology venture companies in the years of 2000 and 2001. The KSE data provide daily transactions by investor type, including the number of shares and market values of KSE-listed companies that are bought and sold daily by each of nine investor groups. The nine investor groups include securities companies, investment trust companies, banks, insurance companies, short-term finances and savings, and pension funds, all of which are classified as local institutional investors. In addition, individual investors, foreigners, and other non-institutional firms constitute the rest of the investor groups. We use monthly stock returns of each listed company supplied by the Korea Securities Research Institute.

3.2. Measurement of Stock Return Performance

Following the portfolio construction approach similar to that used in Jegadeesh and Titman (1993), we first compute daily trade imbalances of each listed common stock for each of the three investor groups of foreigners, local institutions, and individuals starting from the first trading day of January 1996. Daily trade imbalances are net buy trades measured by subtracting daily sell trade volumes from daily buy trade volumes. We then sum up daily net buy trade volumes for each stock by each investor group over a 6-month period to compute 6-month cumulative net buy volumes. The 6-month period is selected because it coincides with the earnings announcement interval in Korea.9 Starting from June 1996, all stocks are ranked based on the 6-month cumulative net buy volumes in terms of market value at the end of each month. Through this process, we select 70 stocks with the highest net buy volumes and construct an equally weighted portfolio consisting of these 70 stocks. We call this portfolio a “Buy” portfolio. We also choose 70 stocks with the lowest net buy volumes and construct the “Sell” portfolios. Because the total number of KSE-listed firms is approximately 700, our sample stocks belong to roughly the top and bottom deciles of the total listed stocks. Each subsequent month, we construct new “Buy” and “Sell” portfolios based on the 6-month cumulative net buy volumes over the previous 6 months. The first “Buy” and “Sell” portfolios are formed in June 1996, the second in July 1996, and so on in every subsequent month until the last and 79th in December 2002, which gives us a total of 79 “Buy” and “Sell” portfolios constructed for our analysis.10

To identify the existence of any peculiar trading patterns, such as positive feedback trading and contrarian trading by each investor group, it is necessary to examine the return performance before and after “Buy” and “Sell” portfolios are constructed. Therefore, we measure pre-portfolio formation returns up to a year prior to portfolio construction and post-portfolio formation returns over 3, 6, and 12 months following portfolio construction. We compare post-portfolio formation returns among the three investor groups to examine whether differences exist in portfolio performances and portfolio management skills among the three investor groups. By comparing pre-portfolio and post-portfolio formation returns for each investor group, we aim to identify the existence of return continuation or reversal of each portfolio constructed by each investor group.

In the present study, we measure pre-portfolio and post-portfolio formation returns as abnormal returns (AR) by using three reference pricing models. The first model is the market-adjusted model using both the value-weighted market index and the equally-weighted market index. The second model is the risk-adjusted model using the standard one factor capital asset pricing model (CAPM) as given below:

image(1)

where Rpt = monthly portfolio return; Rft = monthly return on the 5-year Korean Treasury note; Rmt = monthly return on the value-weighted market index, βi = portfolio beta; and εit = regression error terms. The third model, the Fama and French three-factor model, is as given below:

image(2)

where SMBt is the return on the value-weighted portfolio of small stocks minus the return on the value-weighted portfolio of large stocks, and HMLt is the return on the value-weighted portfolio of high book to market stocks minus the return on the value-weighted portfolio of low book to market stocks. Because of the potential change in the firms’ risk characteristics before and after the portfolio formation, we estimate equations (1) and (2) over a total of 36 months: 18 months before and 18 months after the portfolio formation.11

Using the estimated parameters of α and β from equations (1) and (2), we then compute expected returns (E(R)) of portfolio i, which, in turn, are used to measure AR and cumulative abnormal returns (CAR) for each portfolio over the post-portfolio formation period, T, as follows:

image(3)
image(4)

where Rit = the (realized) monthly return of portfolio i.

Finally, using expected returns of each portfolio, we measure buy-and-hold abnormal returns (BHAR) over the post-portfolio formation period, T, in the following manner:12

image(5)

Our primary focus is to test whether “Buy” portfolios outperform “Sell” portfolios for each of the three investor groups over the pre-portfolio and post-portfolio formation periods and whether “Buy (Sell)” portfolios by foreigners perform better (worse) than those by local institutions and individual investors during the post-portfolio formation periods.

3.3. Return Performance for Pre-portfolio Formation Periods

Table 2 reports mean and median CAR of “Buy” and “Sell” portfolios by three investor groups for 6 and 12 months before each portfolio is constructed. CAR are measured using the market-adjusted model with both the value-weighted and equally-weighted market indices. Regardless of the market index used, both mean and median CAR of “Buy” portfolios for foreigners are positive and statistically significant at the 1% level over the 6- and 12-month pre-portfolio formation periods. In contrast, mean and median CAR of “Sell” portfolios for foreigners are all negative and significant at the 1% level in most cases. Most importantly, the differences in mean and median CAR between “Buy” and “Sell” portfolios for foreigners are all positive and significant at the 1% level. These findings indicate that foreigners buy stocks that have performed well and sell stocks that have performed poorly. Similar results are found for local institutional investors, except that mean and median CAR of their “Sell” portfolios are all positive and significant at least at the 5% level.

Table 2.   Cumulative abnormal returns for pre-portfolio formation periods by investor type
The figures (the figures in brackets) represent mean (median) of 79 “Buy” and “Sell” portfolios constructed during the period 1996–2002. VWMI (EWMI) CAR(−t,0) represents cumulative abnormal returns (CAR) using the value-weighted (equally-weighted) market index for t months prior to the portfolio formation. The t-test and the Brown–Mood median test are conducted on the mean and median differences. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
 ForeignersLocal institutionsIndividuals
BuySellDifferenceBuySellDifferenceBuySellDifference
VWMI
CAR
(−12,0)
0.3444*** [0.3221]***−0.0113 [0.0361]0.3557*** [0.2860]***0.3273*** [0.3576]***0.1700*** [0.2000]***0.1573*** [0.1576]***0.0477** [0.0631]**0.4295*** [0.4697]***−0.3818*** [−0.4066]***
VWMI
CAR
(−6,0)
0.1778*** [0.1618]***−0.0458*** [−0.0210]***0.2237*** [0.1828]***0.1809*** [0.1709]***0.0359** [0.0486]***0.1450*** [0.1223]***−0.0514** [−0.0628]**0.2514*** [0.2432]***−0.3028*** [−0.3060]***
EWMI
CAR
(−12,0)
0.2062*** [0.1884]***−0.1495*** [−0.1228]***0.3557*** [0.3112]***0.1892*** [0.1869]***0.0319 [−0.0228]0.1573*** [0.2097]**−0.0904*** [−0.0676]***0.2913*** [0.2412]***−0.3818*** [−0.3088]***
EWMI
CAR
(−6,0)
0.1086*** [0.1075]***−0.1151*** [−0.0985]***0.2237*** [0.2060]***0.1117*** [0.1378]***−0.0333* [−0.0346]**0.1450*** [0.1740]***−0.1206*** [−0.1030]***0.1822*** [0.1645]***−0.3028*** [−0.2675]***

In sharp contrast to the results for foreigners and local institutional investors, we obtain strikingly different results for individual investors. Mean and median CAR of “Buy” portfolios for individual investors are generally negative and significant at least at the 5% level, whereas those of “Sell” portfolios for individuals are all positive and significant at the 1% level. Furthermore, the differences in CAR between “Buy” and “Sell” portfolios are negative and significant at the 1% level. Hence, individuals buy stocks that have performed poorly and sell stocks that have performed well, and the stocks that they buy have significantly underperformed the stocks that they sell, whose results are exactly opposite to those for foreigners and local institutions.

The overall results in Table 2 indicate that while foreigners and local institutions sell underperforming stocks and buy outperforming stocks, individuals tend to trade in the opposite way. Therefore, individual investors show a tendency to dispose of stocks that have appreciated and to buy stocks that have depreciated. Although preliminary, these findings are, in general, consistent with those documented in previous studies that foreigners trade like trend-chasers, whereas individuals trade like contrarians.

3.4. Estimation of Factor Loading Coefficients in the Capital Asset Pricing Model and Fama–French Three-factor Model

Table 3 reports the average factor loading coefficients of two risk-adjusted models, the standard one-factor CAPM and the Fama–French three-factor model, which are used to compute AR during the post-portfolio formation periods. The risk factor loading coefficients are estimated over 36 months around the portfolio formation. As shown in Panel 1, the coefficients of both the size factor loading, Sj (small cap–large cap), and the distress factor loading, hj (high BE/ME − low BE/ME), for foreigners are greater for “Sell” portfolios than for “Buy” portfolios. These findings suggest that foreigners have an aversion toward smaller firms and distressed firms. In contrast, the coefficients of the two factor loadings for individuals are, as shown in Panel 3, greater for “Buy” portfolios than for “Sell” portfolios, indicating that individuals tend to buy stocks of small and distressed firms. Local institutions stand in the middle as they tend to buy small-cap stocks but to sell the stocks of distressed firms.

Table 3.   Estimation of factor loadings in the CAPM and Fama–French three-factor model
The values in the table are estimated average coefficients of 79 “Buy” and “Sell” portfolios constructed during January 1996–December 2002. The capital asset pricing model (CAPM) is defined as follows: inline image where Rpt = monthly portfolio return, Rft = monthly return on 5-year Korean treasury note, Rmt = monthly return on the value-weighted market index, and βi = market beta. The Fama–French three-factor model is defined as follows: inline image where SMBt is the return on a value-weighted portfolio of small stocks minus the return on a value-weighted portfolio of large stocks and HMLt is the return on a value-weighted portfolio of high book to market stocks minus the return on a value-weighted portfolio of low book to market stocks. t-statistics are in parentheses. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
 βj (t-statistics)Sj (t-statistics)hj (t-statistics)
Panel A: Foreigners
Buy
 CAPM1.0126 (104.03)***0.0969 (5.70)***0.1265 (4.51)***
 Fama–French three factor model1.0397 (113.89)***  
Sell
 CAPM1.0144 (105.27)***0.2354 (7.94)***0.1610 (3.13)***
 Fama–French three factor model1.0301 (131.92)***  
Panel B: Local institutions
Buy
 CAPM0.9733 (77.39)***0.2730 (10.93)***0.0653 (2.24)**
 Fama–French three factor model1.0124 (95.49)***
Sell
 CAPM1.0664 (132.49)***0.1816 (6.41)***0.1491 (2.77)***
 Fama–French three factor model1.0663 (141.81)***  
Panel C: Individuals
Buy
 CAPM1.0971 (118.07)***0.3687 (7.73)***0.1653 (2.12)**
 Fama–French three factor model1.1033 (114.96)***
Sell
 CAPM0.9993 (77.03)***0.1930 (7.97)***0.0383 (1.09)
 Fama–French three factor model 1.0341 (88.66)***

3.5. Return Performance for Post-portfolio Formation Periods

Tables 4 and 5 present the return behavior, measured by CAR and BHAR, respectively, of “Buy” and “Sell” portfolios by investor type over three different post-portfolio formation periods. The analysis of CAR reported in Table 4 generally supports the hypothesis that foreigners trade more wisely than local institutions and individuals, at least in the short term. Regardless of the estimation model used, the differences in mean and median CAR between “Buy” stocks and “Sell” stocks for foreigners are positive and significant at least at the 10% level, except in two cases. Hence, these findings strongly indicate that the stocks foreigners buy significantly outperform the stocks that they sell by almost all CAR measures. In contrast, individuals suffer from poor trading decisions. The differences in mean and median CAR for individuals are negative and significant at least at the 10% level in many cases. Therefore, the stocks that individuals buy significantly underperform the stocks that they previously sold. As for local institutions, the differences in post-portfolio formation CAR between “Buy” and “Sell” portfolios have mixed signs and are not significant at the 10% level. These findings for local institutions suggest that local institutional investors in Korea do not possess good stock selection ability.

Table 4.   Cumulative abnormal returns for post-portfolio formation periods by investor type
The values (the values in brackets) in the table indicate the mean (median) of 79 “Buy” and “Sell” portfolios constructed during January 1996–December 2002. VWMI, EWMI, CAPM, and FF represent the value-weighted market index for the market-adjusted model, equally-weighted market index for the market-adjusted model, risk-adjusted single factor CAPM, and risk-adjusted Fama–French three-factor model, respectively. CAR(0,t) represents cumulative abnormal returns for t months following portfolio formation.
 ForeignersLocal institutionsIndividuals
BuySellDifferenceBuySellDifferenceBuySellDifference
EWMI
 CAR(0,3)−0.0156 [−0.0093]−0.0468*** [−0.0330]***0.0312 [0.0237]*−0.0412*** [−0.0297]**−0.0425*** [−0.0318]***0.0013 [0.0021]−0.0618*** [−0.0507]***−0.0326* [−0.0178]*−0.0292 [−0.0329]**
 CAR(0,6)−0.0316 [−0.0176]−0.0744*** [−0.0692]***0.0428 [0.0516]**−0.0692*** [−0.0366]**−0.0818*** [−0.0648]***0.0126 [0.0282]**−0.1137*** [−0.1137]***−0.0563** [−0.0076]−0.0574** [−0.1061]***
 CAR(0,12)−0.0707** [0.0149]−0.1104*** [−0.1217]***0.0398 [0.1366]**−0.1320*** [−0.0388]***−0.1268*** [−0.1170]***−0.0052 [0.0782]−0.1700*** [−0.1458]***−0.1172*** [0.0131]**−0.0528 [−0.1589]**
VWMI
 CAR(0,3)0.0135* [0.0154]**−0.0177 [−0.0025]0.0312** [0.0179]*−0.0120 [0.0004]−0.0134 [−0.0182]0.0013 [0.0186]−0.0326** [−0.0168]*−0.0034 [0.0013]−0.0292* [−0.0181]
 CAR(0,6)0.0218* [0.0329]**−0.0211 [−0.0093]0.0428** [0.0422]−0.0158 [−0.0055]−0.0284* [0.0029]0.0126 [−0.0084]−0.0604*** [−0.0334]**−0.0029 [−0.0016]−0.0574** [−0.0318]
 CAR(0,12)0.0178 [0.0275]−0.0220 [0.0587]0.0398 [−0.0312]−0.0435* [−0.0267]−0.0384 [0.0063]−0.0052 [−0.0330]−0.0816** [−0.0401]**−0.0288 [−0.0139]−0.0528 [−0.0262]
CAPM
 CAR(0,3)0.0194** [0.0233]***−0.0167 [0.0002]0.0361** [0.0231]***−0.0075 [0.0036]−0.0106 [−0.0093]0.0031 [0.0129]−0.0305** [−0.0201]*0.0014 [0.0042]−0.0319* [−0.0243]
 CAR(0,6)0.0325*** [0.0571]***−0.0203 [−0.0081]0.0528** [0.0652]**−0.0054 [0.0130]−0.0216 [−0.0032]0.0163 [0.0162]−0.0567*** [−0.0267]**0.0065 [−0.0026]−0.0632** [−0.0241]
 CAR(0,12)0.0422** [0.0498]**−0.0303 [0.0526]0.0724** [−0.0028]−0.0134 [−0.0010]−0.0309 [0.0324]0.0175 [-0.0334]−0.0820** [−0.0135]*0.0069 [0.0316]−0.0889** [−0.0451]
FF
 CAR(0,3)0.0297*** [0.0296]***−0.0032 [0.0116]0.0329** [0.0180]*0.0053 [0.0132]0.0040 [0.0084]0.0014 [0.0048]−0.0057 [0.0007]0.0124 [0.0175]**−0.0181 [−0.0168]
 CAR(0,6)0.0552*** [0.0613]***0.0044 [0.0264]0.0508*** [0.0349]0.0211 [0.0364]0.0136 [0.0196]0.0075 [0.0168]−0.0005 [0.0201]0.0318** [0.0437]**−0.0323 [−0.0236]
 CAR(0,12)0.0926*** [0.1012]***0.0310 [0.0544]**0.0616** [0.0468]*0.0495** [0.0620]**0.0489** [0.0736]**0.0005 [-0.0116]0.0402 [0.0627]**0.0587*** [0.0781]***−0.0185 [−0.0154]
Table 5.   Buy-and-hold abnormal returns for post-portfolio formation periods
The values (the values in brackets) in the table indicate the mean (median) of 79 “Buy” and “Sell” portfolios constructed during January 1996–December 2002. VWMI, EWMI, CAPM, and FF represent the value-weighted market index for the market-adjusted model, the equally-weighted market index for the market-adjusted model, the risk-adjusted single factor capital asset pricing model, and the risk-adjusted Fama–French three-factor model, respectively. BHAR(0,t) represents buy-and-hold abnormal returns for t months following portfolio formation. The t-test and the Brown–Mood median test are conducted on the mean and median differences. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
 ForeignersLocal institutionsIndividuals
BuySellDifferenceBuySellDifferenceBuySellDifference
EWMI
 BHAR(0,3)−0.0200 [−0.0035]−0.0511*** [−0.0409]***0.0311 [0.0374]−0.0459** [−0.0270]**−0.0470*** [−0.0324]***0.0011 [0.0054]−0.0665*** [−0.0511]***−0.0385* [−0.0132]−0.0279 [−0.0379]***
 BHAR(0,6)−0.0528* [−0.0078]−0.0894*** [−0.0607]***0.0366 [0.0529]**−0.0884*** [−0.0214]**−0.0962*** [−0.0617]***0.0078 [0.0403]**−0.1274*** [−0.1055]***−0.0815** [−0.0034]−0.0460 [−0.1021]***
 BHAR(0,12)−0.1370*** [0.0283]−0.1389*** [−0.1120]***0.0019 [0.1403]**−0.1898*** [−0.0213]***−0.1589*** [−0.1137]***−0.0308 [0.0924]−0.1915*** [−0.1668]***−0.1878*** [0.0123]**−0.0036 [−0.1791]***
VWMI
 BHAR(0,3)0.0152* [0.0134]**−0.0159 [−0.0036]0.0311** [0.0170]**−0.0107 [−0.0043]−0.0118 [−0.0168]0.0011 [0.0125]−0.0313** [−0.0248]**−0.0034 [0.0028]−0.0279 [−0.0276]*
 BHAR(0,6)0.0240** [0.0285]**−0.0126 [−0.0207]0.0366* [0.0492]−0.0116 [−0.0087]−0.0194 [−0.0131]0.0078 [0.0044]−0.0507** [−0.0393]***−0.0047 [−0.0090]−0.0460* [−0.0303]
 BHAR(0,12)−0.0019 [0.0140]−0.0038 [0.0373]0.0019 [−0.0233]−0.0547* [−0.0319]−0.0239 [−0.0187]−0.0308 [−0.0132]−0.0564 [−0.0557]**−0.0528 [−0.0260]−0.0036 [−0.0297]
CAPM
 BHAR(0,3)0.0211** [0.0226]***−0.0150 [−0.0046]0.0361** [0.0272]***−0.0061 [0.0062]−0.0095 [−0.0068]0.0034 [0.0130]−0.0302** [−0.0247]*0.0017 [0.0024]−0.0319* [−0.271]*
 BHAR(0,6)0.0366*** [0.0477]***−0.0126 [−0.0146]0.0492** [0.0623]*0.0003 [0.0094]−0.0130 [−0.0073]0.0133 [0.0167]−0.0491** [−0.0283]**0.0073 [−0.0028]−0.0564** [−0.0255]
 BHAR(0,12)0.0268 [0.0209]−0.0080 [0.0322]0.0348 [−0.0113]−0.0153 [−0.0140]−0.0191 [0.0022]0.0038 [−0.0162]−0.0593* [−0.0371]*−0.0110 [0.0134]−0.0483 [0.0505]*
FF
 BHAR(0,3)0.0315*** [0.0339]***−0.0042 [0.0113]0.0357*** [0.0226]**0.0061 [0.0061]0.0034 [0.0084]0.0027 [−0.0023]−0.0096 [−0.0016]0.0126 [0.0167]*−0.0222 [0.0183]
 BHAR(0,6)0.0572*** [0.0581]***0.0018 [0.0239]0.0554*** [0.0342]**0.0210 [0.0348]0.0119 [0.0151]0.0091 [0.0197]−0.0088 [0.0087]0.0290** [0.0333]**−0.0378* [−0.0246]
 BHAR(0,12)0.0728*** [0.0736]***0.0293 [0.0498]**0.0435 [0.0238]0.0290 [0.0496]*0.0375* [0.0712]***−0.0085 [−0.0216]0.0276 [0.0399]*0.0328 [0.0681]**−0.0052 [−0.0282]

As reported in Table 5, the results from post-portfolio formation BHAR are qualitatively identical to those from CAR reported in Table 4. Over most of the three post-portfolio formation periods of 3, 6, and 12 months, BHAR are significantly greater for the stocks that foreigners buy than for the stocks they sell. For example, when BHAR is estimated using the CAPM, “Buy” portfolios for foreigners earn, on average, a positive and significant (at the 1% level) BHAR of 3.66%, while their “Sell” portfolios earn a negative and insignificant BHAR of −1.26% for up to 6 months after portfolio formation. The difference of 4.92% between the two portfolios is significant at the 5% level. In sharp contrast, the mean BHAR for individuals over the same period is a negative and significant (at the 5% level) −4.91% for “Buy” portfolios but a positive but insignificant 0.73% for “Sell” portfolios; the difference of −5.64% is significant at the 5% level. Consequently, the stocks that individuals buy significantly underperform the stocks that they have previously sold. As for local institutions, there is little difference in BHAR between “Buy” portfolios and “Sell” portfolios over the entire post-portfolio formation periods.

3.6. Robustness Test of Return Performance

As a way to check the robustness of our results, we also construct zero investment portfolios by short-selling “Sell” portfolio stocks and purchasing “Buy” portfolio stocks with the proceeds from short sales. Through this analysis, we aim to test whether trades mimicking each investor group’s trading strategies would generate an economically meaningful return performance. Table 6 reports buy-and-hold returns for the 3, 6, and 12 months after the zero investment portfolios are constructed. On average, a positive and significant (at the 1% level) buy-and-hold return of 3.17% can be earned over the 3-month period by following the foreign investors’ trading strategy, while the individuals’ trading strategy generates, on average, a significant loss of −2.98% over the same period. The poor performance of individual investors’ zero investment portfolios appears to be attributed to their contrarian strategy. However, local institutional trades also do not produce any meaningful return performance.

Table 6.   Robustness test: buy-and-hold returns from zero investment (buy–sell) portfolios
The values (the values in brackets) in the table represent the mean (median) of 79 zero investment portfolios constructed during January 1996–December 2002. Zero investment portfolios are constructed by short-selling “Sell” portfolio stocks and purchasing “Buy” portfolio stocks using the proceeds from short sales. BHR(0,t) represents buy-and-hold returns for t months following zero investment portfolio formation. The t-test and the Brown–Mood median test are conducted on the mean and median differences. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
 ForeignersLocal institutionsIndividuals
BHR(0,3)0.0317*** [0.0168]***−0.0002 [0.0050]−0.0298** [−0.0323]**
BHR(0,6)0.0458** [−0.0030]0.0098 [0.0235]−0.0532** [−0.0649]***
BHR(0,12)0.0518* [0.0325]−0.0008 [0.0169]−0.0285 [−0.1317]**

Overall, the short-term return behavior around portfolio formation suggests that foreigners and local institutions in the Korean stock market adopt momentum trading strategies, whereas individuals tend to trade like contrarians. The post-portfolio formation return performance between foreigners and individuals is strikingly different. The stocks that foreigners buy significantly outperform the stocks that they sell, whereas individuals incur a significant loss from the stocks that they buy. However, the stocks that local institutions buy produce almost the same returns as the stocks that they sell, suggesting the mediocre stock selection capability of local institutions.

4. Analysis of Firms’ Operating and Financial Performance by Investor Type

4.1. Performance Data and Measurement

As we find strong evidence of the superior return performance of foreigners relative to local institutions and individual investors in Korea, we have reason to expect that foreigners’ outperformance might be attributed to their superior ability to analyze firm-specific information regarding the operating and financial performance of KSE-listed companies. To test this conjecture, we examine operating and financial data of KSE-listed companies that are included in either “Buy” or “Sell” portfolios at the end of June and December of each year over the 7-year sample period.13 We collect firms’ annual and semi-annual financial statements for a total of five periods: the contemporaneous period (t) of each portfolio formation and two periods before (t − 2 and t − 1) and after (t + 1 and t + 2) each portfolio formation (compiled by Korea Listed Company Association). Although the quarterly data contain more timely information, it was not mandatory for the KSE-listed firms to report quarterly financial information until 2001.

We use seven performance measures to represent operating and financial performance of each sample company: return on assets (ROA), return on equity (ROE), cash flows from operation divided by sales (CF/Sales), dividend yield (Div. Yield), total debt to asset ratio (Debt/Asset), price earning ratio (PER), and book to market equity ratio (BE/ME). The data for performance measures are collected for up to 12 months before and after the formation of “Buy” and “Sell” portfolios. For those firms whose financial statement release dates do not coincide with the end of June or December, we compute the performance measures using information in their financial statements at the nearest time to the end of June or December.

With individual firms’ operating and financial performance measures, we first perform the univariate analysis to test mean and median differences in performance measures between “Buy” and “Sell” portfolios by each investor group up to 12 months before and after the portfolio formation. We also examine the constituent stocks that are included in the “Buy” and “Sell” portfolios to identify the characteristics of stocks for which each investor group has a preference or aversion. We are particularly interested in exploring any lead or lag relation between buy and sell decisions by each investor group and the operating and financial performance of each individual firm.

4.2. Comparison of Operating Performance of Korea Stock Exchange-listed Firms by Investor Type

Table 7 shows the results from the univariate analysis on financial performance measures, which are computed from the financial report information that becomes publicly available nearest to the time each portfolio is constructed. For instance, ROA(t − 2) is computed based on the financial report information released in the two semi-annual report periods (12 months) prior to the portfolio formation, and ROA(t + 1) based on the financial report released in the one semi-annual report period (6 months) following the portfolio formation. Because the CF/Sales ratio is measured based on information in the annually released cash flow statement and the dividend payment is made on a yearly basis for most KSE-listed companies, CF/Sales and Div. Yield are measured in yearly intervals.14 Due to the skewed distribution of performance measures, we focus on median values.

Table 7.   Comparison of operating performance of Korea stock exchange-listed firms by investor type
The values (the values in brackets) in the table are means (medians) of the constituent listed firms included in each portfolio constructed during January 1996–December 2002. ROA is return on assets, ROE is return on equity, CF/Sales is cash flows from operation to sales, and Div. Yield is dividend yield. Time notations of t − 2, t − 1, t, t + 1, t + 2 represent two and one financial report periods prior to the contemporaneous financial report period, the contemporaneous financial report period, and one and two financial report periods subsequent to portfolio formation, respectively. Caution is needed in interpreting CF/Sales and Div. Yield because these ratios are computed with yearly information; therefore, − 1 data should be compared with + 1 data, while t − 2, t, + 2 data should be compared together. The t-test and the Brown–Mood median test are conducted on the mean and median differences.
 ForeignersLocal institutionsIndividuals
BuySellDifferencep-valueBuySellDifferencep-valueBuySellDifferencep-value
ROA(t − 2)0.0517 [0.0443]0.0438 [0.0374]0.0079 [0.0069]0.002 [0.001]0.0502 [0.0420]0.0444 [0.0398]0.0058 [0.0022]0.040 [0.287]0.0355 [0.0317]0.0511 [0.0443]−0.0160 [−0.0126]0.000 [0.000]
ROA(t − 1)0.0555 [0.0460]0.0388 [0.0368]0.0167 [0.0092]0.000 [0.000]0.0488 [0.0435]0.0457 [0.0401]0.0030 [0.0034]0.228 [0.067]0.0309 [0.0312]0.0535 [0.0465]−0.0230 [−0.0154]0.000 [0.000]
ROA(t)0.0572 [0.0472]0.0356 [0.0327]0.0216 [0.0146]0.000 [0.000]0.0517 [0.0433]0.0413 [0.0374]0.0104 [0.0059]0.001 [0.002]0.0270 [0.0277]0.0545 [0.0460]−0.0280 [−0.0183]0.000 [0.000]
ROA(t + 1)0.0554 [0.0459]0.0299 [0.0304]0.0255 [0.0155]0.000 [0.000]0.0504 [0.0421]0.0392 [0.0374]0.0112 [0.0047]0.000 [0.022]0.0221 [0.0253]0.0545 [0.0466]−0.0320 [−0.0208]0.000 [0.000]
ROA(t + 2)0.0514 [0.0424]0.0333 [0.0385]0.0181 [0.0039]0.000 [0.000]0.0489 [0.0419]0.0391 [0.0351]0.0097 [0.0068]0.000 [0.005]0.0261 [0.0276]0.0514 [0.0441]−0.0250 [−0.0165]0.000 [0.000]
Number of observations974962  967963  955972  
ROE(t − 2)0.0501 [0.0577]0.0389 [0.0470]0.0112 [0.0107]0.246 [0.002]0.0690 [0.0474]0.0294 [0.0460]0.0396 [0.0014]0.118 [0.686]0.0044 [0.0356]0.0710 [0.0527]−0.0670 [−0.0171]0.028 [0.000]
ROE(t − 1)0.0658 [0.0604]0.0357 [0.0451]0.0301 [0.0153]0.191 [0.000]0.0574 [0.0573]0.0374 [0.0481]0.0200 [0.0093]0.230 [0.007]0.0036 [0.0347]0.0705 [0.0608]−0.0670 [−0.0261]0.049 [0.000]
ROE(t)0.0711 [0.0681]−0.0030 [0.0367]0.0743 [0.0314]0.000 [0.000]0.0591 [0.0588]0.0213 [0.0457]0.0378 [0.0131]0.002 [0.000]−0.0490 [0.0295]0.0720 [0.0688]−0.1210 [−0.0393]0.000 [0.000]
ROE(t + 1)0.0744 [0.0627]−0.0410 [0.0361]0.1156 [0.0266]0.000 [0.000]0.0459 [0.0560]0.0209 [0.0441]0.0250 [0.0119]0.079 [0.007]−0.0620 [0.0306]0.0873 [0.0648]−0.1490 [−0.0342]0.000 [0.000]
ROE(t + 2)0.0572 [0.0600]−0.0350 [0.0373]0.0927 [0.0227]0.000 [0.000]0.0574 [0.0519]0.0045 [0.0460]0.0529 [0.0059]0.105 [0.351]−0.0460 [0.0320]0.0705 [0.0650]−0.1160 [−0.0330]0.001 [0.000]
Number of observations970947  965942  918969  
CF/Sales(t − 2)0.0955 [0.0923]0.0638 [0.0723]0.0317 [0.0200]0.049 [0.004]0.0780 [0.0834]0.0836 [0.0740]−0.0060 [0.0094]0.718 [0.201]0.0613 [0.0547]0.0869 [0.0847]−0.0260 [−0.0300]0.180 [0.000]
CF/Sales(t − 1)0.1129 [0.1017]0.0709 [0.0626]0.0420 [0.0391]0.019 [0.000]0.1113 [0.0890]0.0800 [0.0826]0.0313 [0.0063]0.045 [0.476]0.0609 [0.0555]0.0999 [0.0923]−0.0390 [−0.0368]0.031 [0.000]
CF/Sales(t)0.0909 [0.1037]0.0776 [0.0842]0.0133 [0.0195]0.449 [0.007]0.0945 [0.0890]0.0865 [0.0814]0.0081 [0.0076]0.612 [0.460]0.0539 [0.0597]0.0971 [0.1017]−0.0430 [−0.0421]0.032 [0.000]
CF/Sales(t + 1)0.1050 [0.0999]0.0403 [0.0664]0.0647 [0.0335]0.005 [0.000]0.1127 [0.0963]0.0740 [0.0794]0.0387 [0.0169]0.013 [0.105]0.0381 [0.0519]0.1070 [0.1016]−0.0690 [−0.0498]0.000 [0.000]
CF/Sales(t + 2)0.1109 [0.0974]0.0708 [0.0827]0.0401 [0.0147]0.047 [0.065]0.0949 [0.0895]0.0759 [0.0831]0.0189 [0.0064]0.221 [0.434]0.0479 [0.0599]0.1222 [0.0972]−0.0740 [−0.0373]0.000 [0.000]
Number of observations498476  491478  465511  
Div. Yield(t − 2)0.0487 [0.0143]0.1717 [0.0122]−0.1230 [0.0021]0.151 [0.365]0.1466 [0.0151]0.0814 [0.0095]0.0652 [0.0056]0.561 [0.004]0.0851 [0.0054]0.0277 [0.0149]0.0574 [−0.0095]0.320 [0.000]
Div. Yield(t − 1)0.1600 [0.0151]0.1708 [0.0105]−0.0110 [0.0046]0.932 [0.014]0.0776 [0.0159]0.0628 [0.0091]0.0148 [0.0068]0.788 [0.010]0.0192 [0.0025]0.0303 [0.0158]−0.0110 [−0.0133]0.005 [0.000]
Div. Yield(t)0.0456 [0.0133 ]0.1657 [0.0122]−0.1200 [0.0011]0.149 [0.653]0.1577 [0.0185]0.0238 [0.0075]0.1340 [0.0111]0.146 [0.000]0.0713 [0.0006]0.0793 [0.0147]−0.0080 [−0.0141]0.909 [0.000]
Div. Yield(t + 1)0.3747 [0.0152]0.0227 [0.0059]0.3512 [0.0093]0.093 [0.000]0.1902 [0.0185]0.0181 [0.0079]0.1721 [0.0107]0.064 [0.000]0.0717 [0.0000]0.2221 [0.0177]−0.1500 [−0.0177]0.451 [0.000]
Div. Yield(t + 2)0.0632 [0.0142]0.0787 [0.0110]−0.0160 [0.0032]0.821 [0.348]0.0334 [0.0194]0.1130 [0.0075]−0.0800 [0.0119]0.231 [0.000]0.1805 [0.0025]0.0665 [0.0162]0.1140 [−0.0137]0.258 [0.000]
Number of observations581538  556556  529578  

Looking first at the profitability measures of ROA, ROE, and CF/Sales for foreigners, our results show that foreigners are inclined to buy stocks that have previously been highly profitable and to sell stocks that have been less profitable. To be more specific, the median values of ROA(t − 2), ROE(t − 2) and CF/Sales(t − 2) are 4.43%, 5.77%, and 9.23%, respectively, for foreigners’“Buy” stocks, which are all significantly (at the 1% level) greater than 3.74%, 4.70%, and 7.23%, respectively, for their “Sell” stocks. Similar results are obtained for ROA(t − 1), ROE(t − 1), and CF/Sales(t − 1). These findings support our earlier evidence that foreigners follow momentum trading or positive feedback trading. Following the portfolio formation, the stocks that foreigners buy continue to outperform the stocks that they sell in all three measures of profitability. For example, the median ROA(t + 1), ROE (t + 1), and CF/Sales(t + 1) are 4.59%, 6.27%, and 9.99%, respectively, for foreigners’“Buy” stocks, which are all significantly (at the 1% level) greater than 3.04%, 3.61%, and 6.64%, respectively, for their “Sell” stocks. Hence, foreign investors seem to be able to discern company stocks with good versus not-so-good prospects. The good or better performance of the stocks that foreigners buy in the post-portfolio formation period might be, at least in part, attributed to the notion that foreigners play an effective role in monitoring firms’ management and contributing to the enhancement of corporate performance.

The results for local institutions indicate that local institutions in Korea are also able to discern and to predict an individual firm’s operational efficiency, but to a much lesser degree than foreign investors. For up to 12 months (periods t − 1 and t − 2) prior to the portfolio formation, the median values of ROA, ROE, and CF/Sales are generally greater for “Buy” portfolios than for “Sell” portfolios, but the differences in these measures between the two portfolios are not significant at the 10% level in most cases. Therefore, although institutional investors tend to buy stocks that have previously been profitable, they also tend to sell stocks that have been equally profitable. We further observe in Table 7 that changes in the firm’s profitability from the pre-portfolio to post-portfolio formation period are not monotonic. The median values of ROA and ROE decrease following both net buy trades and net sell trades, whereas those of CF/Sales increase following net buy trades and decrease following net sell trades over the time period from t − 1 to t + 1. Furthermore, the median values of profitability measures are all greater for “Buy” stocks than for “Sell” stocks following the portfolio formation (periods t + 1 and t + 2), but their differences between “Buy” stocks and “Sell” stocks are significant at the 10% level in only a few cases. These results suggest that local institutional investors in Korea fail to discern the profitability of KSE-listed firms with good versus bad prospects.

Finally, our results for individual investors provide strong evidence that individuals are incapable of discerning firms with good versus bad prospects and, therefore, tend to make errors in predicting the operating performance of the stocks that they trade. For up to 12 months prior to the pre-portfolio formation, the median values of ROA, ROE, and CF/Sales are all significantly less for “Buy” stocks than for “Sell” stocks; thus, individuals have a strong tendency to buy stocks that have been less profitable and to sell stocks that have been more profitable. These results are consistent with our earlier findings that individuals follow a contrarian trading strategy. Following the portfolio formation, the stocks that individuals buy experience a sizable deterioration in profitability, whereas the stocks that they sell improve or at least sustain their profitability. Over the period − 1 to t + 1, the median values of all three profitability measures decline for “Buy” stocks but increase for “Sell” stocks; for example, the median value of ROE declines from 3.47% to 3.06% for “Buy” stocks but increases from 6.08% to 6.48% for “Sell” stocks. Furthermore, individuals’“Buy” stocks significantly underperform their “Sell” stocks, as evidenced by significantly negative differences in three profitability measures between the two portfolios. For instance, the median ROA(t + 1), ROE(t + 1), and CF/Sales(t + 1) are 2.53%, 3.06%, and 5.19%, respectively, for “Buy” stocks, which are all significantly (at the 1% level) less than 4.66%, 6.48%, and 10.16%, respectively, for “Sell” stocks. These results are in sharp contrast to those for foreigners and strongly suggest that individual investors in the Korean stock market systematically make wrong investment decisions by following a contrarian trading strategy.

The last variable that we examine is a firm’s dividend yield. Our results show that foreigners and local institutions tend to buy stocks with high dividend yields and to sell stocks with low dividend yields; thus, both foreigners and local institutions appear to prefer high-dividend paying stocks to low-dividend paying stocks. In contrast, individuals tend to buy low-dividend paying stocks, whereas they sell relatively high-dividend paying stocks. Following trades by individuals, the dividend yields of their “Buy” stocks remain significantly lower than those of their “Sell” stocks. Considering that small company stocks with high growth potential typically pay little or no dividends, our findings suggest that individual investors in Korea have a strong preference for small company stocks.

4.3. Comparison of Leverage and Firm Size of Korea Stock Exchange-listed Firms by Investor Type

In Table 8, we report firms’ leverage measured as the total debt to total asset ratio (Debt/Asset) and firm size as market capitalization (Mkt. Cap.) of stocks included in the “Buy” and “Sell” portfolios by investor group for the periods t − 2 to t + 2. Foreigners do not seem to have a preference for a particular leverage level of the stocks that they trade. In general, the median values of Debt/Asset are greater for “Buy” stocks than for “Sell” stocks over the entire five periods, but the differences in Debt/Asset between “Buy” stocks and “Sell” stocks are not significant at the 10% level. Consequently, a firm’s leverage level does not appear to be an important factor to foreigners in their stock selections to buy or sell.

Table 8.   Comparison of leverage and firm size of Korea stock exchange–listed firms by investor type
The values (the values in brackets) in the table are means (medians) of the constituent listed firms included in each portfolio constructed during January 1996–December 2002. Debt/Asset is total debt to assets, Mkt. Cap is market capitalization (denominated by million Korean won). Time notations of t − 2, t − 1, t, t + 1, t + 2 represent two and one financial report periods prior to the contemporaneous financial report period, the contemporaneous financial report period, and one and two financial report periods subsequent to portfolio formation, respectively. The t-test and the Brown–Mood median test are conducted on the mean and median differences.
 ForeignersLocal institutionsIndividuals
BuySellDifferencep-valueBuySellDifferencep-valueBuySellDifferencep-value
Debt/Assets(t − 2)5.2747 [1.9656]4.4934 [1.8196]0.7813 [0.1459]0.187 [0.205]5.3537 [1.8500]5.0725 [2.1631]0.2812 [−0.3131]0.734 [0.000]5.9710 [2.3121]5.8580 [1.8533]0.1130 [0.4588]0.906 [0.000]
Debt/Assets(t − 1)4.9411 [1.8802]5.1617 [1.7674]−0.2210 [0.1147]0.785 [0.170]4.1167 [1.7494]5.8954 [2.0489]−1.7790 [−0.2995]0.037 [0.000]7.0763 [2.2906]4.3463 [1.7314]2.7301 [0.5592]0.008 [0.000]
Debt/Assets(t)4.5038 [1.7739]4.5776 [1.6571]−0.0740 [0.1168]0.862 [0.293]4.8908 [1.6649]6.2166 [1.9195]−1.3260 [−0.2547]0.329 [0.004]7.9192 [2.1030]4.9660 [1.6623]2.9532 [0.4407]0.060 [0.000]
Debt/Assets(t + 1)6.4985 [1.6264]5.9900 [1.5977]0.5085 [0.0288]0.735 [0.819]4.0983 [1.5313]4.8406 [1.8321]−0.7420 [−0.3008]0.146 [0.000]6.8741 [1.9855]4.7384 [1.5913]2.1357 [0.3942]0.088 [0.000]
Debt/Assets(t + 2)4.3471 [1.5572]5.6197 [1.5184]−1.2730 [0.0388]0.137 [0.561]6.5297 [1.4704]5.6920 [1.7603]0.8376 [−0.2900]0.631 [0.000]6.4253 [1.8775]6.7220 [1.4568]−0.2970 [0.4207]0.867 [0.000]
Number of observations969947  965942  918968  
Mkt. Cap.(t − 2)1162 [260]800 [176]362 [84]0.024 [0.000]676 [137]1269 [248]−593 [−111]0.561 [0.004]805 [168]1150 [211]−344 [−43]0.035 [0.004]
Mkt. Cap.(t − 1)1244 [304]871 [172]373 [132]0.027 [0.000]774 [165]1348 [274]−573 [−109]0.788 [0.010]894 [168]1222 [243]−328 [−75]0.055 [0.000]
Mkt. Cap.(t)1408 [362]860 [160]548 [202]0.002 [0.000]915 [170]1336 [283]−421 [−113]0.146 [0.000]768 [157]1487 [302]−719 [−145]0.000 [0.000]
Mkt. Cap.(t + 1)1491 [373]882 [151]609 [222]0.001 [0.000]924 [163]1455 [292]−531 [−129]0.064 [0.000]774 [152]1601 [297]−827 [−145]0.000 [0.000]
Mkt. Cap.(t + 2)1554 [374]993 [152]562 [222]0.005 [0.000]985 [165]1580 [295]−595 [−130]0.231 [0.000]897 [149]1671 [294]−773 [−145]0.000 [0.000]
Number of observations974962  967963  955972  

In contrast, local institutions and individual investors are found to have a sharply different preference for the leverage level of the stocks that they buy and sell. The “Buy” stocks of local institutions have significantly (at the 1% level) lower median debt ratios than their “Sell” stocks, while the individuals’“Buy” stocks have significantly (at the 1% level) higher median debt ratios than their “Sell” stocks over the entire five periods. These results strongly indicate that while local institutions prefer low leveraged firms to high leveraged firms, individuals prefer the opposite.

With respect to firm size measured by total market capitalization, foreigners tend to buy large firms (or firms with recent price appreciation) and to sell small firms (or firms with recent price depreciation). These findings are consistent with the firm recognition hypothesis by Merton (1987). In contrast, individual investors have a tendency to sell large-cap firms and to buy small-cap firms instead. Local institutions also tend to switch large-cap stocks with small-cap stocks. Our findings of local institutions’ preference for small-cap stocks in the Korean stock market differ with the evidence documented for the US market.

4.4. Comparison of Growth Potential of Korea Stock Exchange-listed Firms by Investor Type

Table 9 reports the results from the univariate analysis of two market indicators of price–earnings ratio (PER) and book to market equity (BE/ME) ratio as measures of a firm’s growth potential. It is clear that over the entire periods of t − 2 to t + 2, the median PER of both “Buy” and “Sell” stocks generally follow a downward trend and that the median BE/ME ratios for both “Buy” and “Sell” stocks are greater than unity. These findings confirm the price depression phenomenon observed in the Korean stock market during our sample period of January 1996–December 2002.15

Table 9.   Comparison of PER and BE/ME of Korea stock exchange–listed firms by investor type
The values (the values in brackets) in the table are means (medians) of the constituent listed firms included in each portfolio constructed during January 1996–December 2002. PER is price to earnings ratio, BE/ME is book to market equity ratio. Time notations of t − 2, t − 1, t, t + 1, t + 2 represent two and one financial report periods prior to the contemporaneous financial report period, the contemporaneous financial report period, and one and two financial report periods subsequent to portfolio formation, respectively. The t-test and the Brown–Mood median test are conducted on the mean and median differences.
 ForeignersLocal institutionsIndividuals
BuySellDifferencep-valueBuySellDifferencep-valueBuySellDifferencep-value
PER(− 2)25.08 [10.90]28.68 [11.60]−3.60 [−0.70]0.601 [0.271]24.55 [10.74]28.53 [11.81]−3.98 [−1.07]0.566 [0.145]29.01 [10.94]25.08 [11.07]3.93 [−0.13]0.578 [0.926]
PER(t − 1)23.44 [10.47]24.28 [10.75]−0.84 [−0.28]0.905 [0.716]22.09 [9.67]28.83 [11.31]−6.74 [−1.64]0.332 [0.039]27.54 [10.10]23.61 [10.80]3.93 [−0.70]0.580 [0.424]
PER(t)18.40 [10.55]27.41 [8.31]−9.01 [2.24]0.199 [0.004]16.37 [9.19]30.13 [10.37]−13.76 [−1.18]0.186 [0.050]24.86 [7.47]19.31 [10.79]5.55 [−3.32]0.429 [0.000]
PER(t + 1)24.04 [10.07]18.71 [8.10]5.33 [1.97]0.422 [0.002]15.89 [8.82]25.32 [9.87]−9.43 [−1.05]0.155 [0.057]25.21 [7.75]18.74 [10.02]6.47 [−2.27]0.339 [0.007]
PER(t + 2)13.49 [9.80]22.85 [6.99]−9.36 [2.81]0.162 [0.000]21.16 [8.55]21.48 [8.78]−0.32 [−0.23]0.966 [0.712]21.53 [6.58]15.74 [9.94]5.80 [−3.37]0.399 [0.000]
Number of observations974962  967963  955972  
BE/ME(t − 2)0.892 [1.270]0.690 [1.231]0.202 [0.039]0.374 [0.646]0.933 [1.361]1.032 [1.298]−0.099 [0.063]0.472 [0.203]1.073 [1.376]0.926 [1.267]0.147 [0.109]0.184 [0.042]
BE/ME(t − 1)0.885 [1.216]1.067 [1.379]−0.182 [−0.163]0.035 [0.011]0.981 [1.429]0.862 [1.247]0.119 [0.182]0.313 [0.005]0.844 [1.343]0.923 [1.175]−0.079 [0.168]0.551 [0.104]
BE/ME(t)0.895 [1.212 ]1.145 [1.433]−0.250 [−0.221]0.008 [0.000]0.826 [1.361]0.956 [1.310]−0.130 [0.051]0.161 [0.255]1.139 [1.457]0.756 [1.188]0.383 [0.269]0.002 [0.000]
BE/ME(t + 1)0.950 [1.234]1.100 [1.591]−0.150 [−0.357]0.160 [0.000]1.249 [1.483]0.960 [1.370]0.289 [0.113]0.008 [0.070]0.978 [1.576]1.002 [1.256]−0.024 [0.320]0.864 [0.000]
BE/ME(t + 2)1.094 [1.352]1.135 [1.599]−0.041 [−0.247]0.730 [0.000]1.015 [1.508]1.086 [1.404]−0.071 [0.104]0.601 [0.042]1.145 [1.598]0.872 [1.367]−0.273 [0.231]0.025 [0.000]
Number of observations974962  967963  955972  

Looking first at the results for foreigners, we observe no significant difference in either PER or BE/ME between “Buy” and “Sell” stocks up to 12 months prior to the portfolio formation (periods t − 2 and t − 1). During the contemporaneous period (period t), however, foreigners’“Buy” stocks have both significantly greater median PER (10.55 versus 8.31) and significantly less median BE/ME ratios (1.212 versus 1.433) than their “Sell” stocks. Furthermore, over the subsequent 12-month period following the portfolio formation (periods t + 1 and t + 2), the same trends are maintained. For example, for the period t + 1, the median PER is 10.07 for “Buy” stocks and 8.10 for “Sell” stocks, whose difference is significant at the 1% level. For the period t + 2, the difference in PER between the two portfolios widens and remains significant at the 1% level. Similarly, for the period t + 1, the median BE/ME ratio is 1.234 for “Buy” stocks and 1.591 for “Sell” stocks, and the difference is significant at the 1% level. These results provide strong evidence that foreigners buy stocks with a high PER and a lower BE/ME ratio and sell stocks with a low PER and a high BE/ME ratio. Following their trades, the prices of foreigners’“Buy” stocks appreciate significantly more than the prices of their “Sell” stocks. This evidence is consistent with the notion that foreigners prefer to invest in growth stocks than value stocks. According to Badrinath and Wahal (2002), the institutions following growth strategies are more likely momentum traders. In Korea, foreigners appear to behave like momentum traders pursuing growth strategies. It is worthwhile to note that PER (BE/ME ratios) of “Sell” stocks for foreigners decline (increase) monotonically over the periods t − 2 to t + 2, indicating that the stocks that foreigners sell experience a gradual erosion in their prices.

In contrast, local institutions tend to buy relatively low PER stocks and to sell high PER stocks; for the contemporaneous period (period t), the median PER is 9.19 for “Buy” stocks and 10.37 for “Sell” stocks, whose difference is significant at the 5% level. Following the portfolio formation, PER of “Buy” stocks remain below those of “Sell” stocks, although their differences are not significant at the 5% level. Local institutions also tend to buy high BE/ME stocks and sell low BE/ME stocks, as evidenced by higher median BE/ME ratios for “Buy” stocks than for “Sell” stocks for the three periods of t − 2, t − 1, and t. In the subsequent periods of t + 1 and t + 2, the “Buy” stocks sustain higher BE/ME ratios than the “Sell” stocks. The results for local institutions suggest that Korean institutional investors pursue value strategies.

Individuals also tend to buy stocks with low PER and high BE/ME ratios and to sell stocks with high PER and low BE/ME ratios, but to a much greater degree than local institutions. For the contemporaneous period, the median PER and BE/ME ratios are 7.47 and 1.457, respectively, for “Buy” stocks and 10.79 and 1.188, respectively, for “Sell” stocks; their differences are all significant at the 1% level. Over the subsequent periods of t + 1 and t + 2, the “Buy” stocks continue to command significantly (at the 1% level) low PER and significantly (at the 1% level) higher BE/ME ratios than the “Sell” stocks. The results for individuals are consistent with contrarian trading strategies.

Overall, foreigner investors have contrasting trading patterns compared to local institutional and individual investors in the Korean stock market. Foreigners tend to prefer large-cap, high-dividend paying, growth stocks that have recently shown good operating and return performance. Our findings for foreigners in the Korean market are consistent with the behavior of foreigners in the Japanese market documented by Kang and Stulz (1997), in the Korean market by Choe et al. (1999) and Ko et al. (2004), and in the Swedish market by Dahlquist and Robertsson (2001). However, our finding that a firm’s debt ratio is not an important characteristic of their trading stocks differs from the findings in previous published studies.

Our findings on individual investors’ trading behavior also support the disposition effect hypothesis by Shefrin and Statman (1985), and are in line with the evidence in Odean (1998), Barber and Odean (2000), and Griffin et al. (2003). Individuals in the Korean stock market tend to buy past losers and to dispose of past winners. They also have a strong preference for small-cap, value (high book to market equity) stocks with a high debt ratio and low dividend yield.

Trading patterns of local institutions in Korea seem to be in the middle of those of foreigners and individuals. Local institutions trade like trend chasers: they tend to buy (sell) stocks with relatively good (poor) performance in stock return and profitability. They also prefer low-leveraged, high-dividend paying firms. However, they tend to invest in small-cap stocks with a low PER and a high BE/ME ratio, like individual investors. Our findings on the local institutions’ preference for small-cap stocks in the Korean stock market are in sharp contrast to those found in the previous studies of the US markets that show institutional preference for large-cap stocks (see e.g. Lakonishok et al., 1992; Falkenstein, 1996; Gompers and Metrick, 2001).

5. Logistic Regression Analysis by Investor Type

Although our results from the univariate tests show different operating and financial performance of the stocks that foreigners prefer relative to local institutions and individual investors, the results do not reveal the marginal contribution of each performance measure to the buy and sell decisions by each investor group after controlling for other firm characteristics. For this purpose, we conduct multivariate regression analysis on the performance measures. To locate any lead and lag relations between an individual firm’s performance measures and the buy and sell decisions of each investor group, we estimate the following logistic regression models using contemporaneous and lagged variables of performance measures of the sample firms:

image(6)

In equation (6), pi is the conditional probability of each investor group to decide to buy a given firm’s stock in net in time t based on Xi values, where Xi are the contemporaneous and lagged financial performance measures of ROA, ROE, CF/Sales, and Div. Yield. We also include the total debt to asset ratio (Debt/Asset), the natural log of market capitalization (Ln(Mkt. Cap)), and a dummy variable (Dummy Fin) with a value of 1 if a given firm belongs to the financial industry and 0 otherwise as additional explanatory variables to control for firm size, leverage, and industry effect. β is the regression coefficient vector; a positive value of βi would indicate that the probability of buying a certain company stock increases with an increase in an explanatory variable Xi.

Table 10 shows the logistic regression results of the probability of buying a given firm’s stock given the firm’s ROA and ROE. Consistent with the results from the univariate tests, the estimated regression coefficients of ROA(t), ROA(t + 1), ROA(t + 2) and ROE(t), ROE(t + 1), ROE(t + 2) in regressions (1-A) and (1-B), respectively, take on a positive value and are all statistically significant at least at the 10% level. Hence, the stocks that foreigners buy not only have a higher profitability measured by ROA and ROE than the stocks that foreigners sell in the current period, but the “Buy” stocks also continue to command higher profitability than the “Sell” stocks in the subsequent periods. Mkt. Cap. carries a positive and significant (at the 1% level) coefficient, indicating that firm size is positively related to the probability of foreign investors’ buy decision.

Table 10.   Results from logistic regression analysis of return on assets and return on equity
The values in the table are the coefficient estimates of logistic regressions between independent variables and the probability of buying a given firm. ROA is return on assets, ROE is return on equity, Debt/Asset is total debt to assets, Ln(Mkt. Cap) is a natural log of market capitalization. Time notations of t − 2, t − 1, t, t + 1, t + 2 represent two and one financial report periods prior to the contemporaneous financial report period, the contemporaneous financial report period, and one and two financial report periods subsequent to portfolio formation, respectively. χ2-statistics are in parentheses. The likelihood ratio (LR) χ2-statistic indicates a joint test statistic testing for all the regression coefficients being equal to zero. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
 ForeignersLocal institutionsIndividuals
(1-A)(1-B)(2-A)(2-B)(3-A)(3-B)
Intercept−7.5186*** (71.336)−6.8926*** (59.348)6.2446*** (55.226)6.9268*** (62.216)5.3138*** (37.375)4.4559*** (25.325)
ROA(− 2)−1.7658 (1.924) 1.5802 (2.059) −0.8650 (0.440) 
ROA(t − 1)2.0861 (2.088) −3.1977** (5.721) −2.0863 (2.396) 
ROA(t)3.4880** (6.459) −0.6469 (0.205) −1.9063 (1.753) 
ROA(t + 1)5.3028*** (13.285) 5.8748*** (16.594) −7.3230*** (23.452) 
ROA(t + 2)2.2997* (3.248) 1.4727 (1.443) −3.4029*** (6.889) 
ROE(− 2) 0.1305 (0.237) −0.0989 (0.608) −0.0287 (0.030)
ROE(t − 1) −0.1396 (1.643) 0.1914 (0.478) 0.1539 (2.115)
ROE(t) 1.3669*** (14.808) 1.1996*** (10.414) −2.1693*** (33.460)
ROE(t + 1) 1.1608*** (12.121) 0.5972** (4.103) −1.9453*** (26.021)
ROE(t + 2) 0.8571*** (9.453) 0.1316 (0.381) −0.4628* (3.379)
Debt/Asset(t)0.0098 (2.596)0.0197** (4.737)−0.0023 (1.194)0.0020 (0.413)0.0012 (0.441)−0.0050 (1.501)
Ln(Mkt. Cap(t))0.2620*** (60.119)0.2548*** (55.688)−0.2475*** (58.697)−0.2665*** (63.0979)−0.1761*** (28.283)−0.1673*** (24.438)
Dummy Fin0.3667** (6.246)−0.0283 (0.036)0.3052** (4.684)0.1442 (1.157)−0.4007*** (8.218)0.1339 (0.324)
LR χ2-statistic182.42***156.78***87.87**91.57***186.25***175.16***
Number of observations182817841782173117631689

In regression (2-A), the estimated coefficient of ROA(t − 1) is negative and statistically significant at the 5% level, indicating that local institutional investors are likely to sell stocks with a high ROA prevailing in the previous 6-month period. This finding seems to underscore the underperformance of stocks that local institutions trade compared to foreigners with respect to post-portfolio formation returns. In contrast, ROA(t + 1) has a positive and significant (at the 1% level) coefficient of 5.8748; as a result, institutional buy still seems to lead to an increase in ROA, although it is limited only to the subsequent 6-month period. Similar results are obtained for ROE. As shown in regression (2-B), both ROE(t) and ROE(t + 1) have a positive and significant coefficient at least at the 5% level. Thus, local institutions are inclined to buy stocks with a high ROE in the current period, and these stocks’ROE rise further over the next 6 months following the institutional buy. The estimated coefficients of Mkt. Cap.(t) in regressions (2-A) and (2-B) are negative and significant at the 1% level. Hence, as expected from the univariate analyses, local institutions have a strong tendency to purchase small-cap stocks rather than large-cap stocks.

In regressions (3-A) and (3-B), the regression coefficients of ROA(t + 1), ROA(t + 2), ROE(t + 1), and ROE(t + 2) are all negative and significant at least at the 10% level, implying a strong negative relation between a firm’s future profitability and the individual investors’ propensity to buy such a stock. Hence, the stocks that individual investors buy suffer from deterioration in profitability, leading to significantly lower ROA and ROE than the stocks that they sell over the subsequent 6- and 12-month periods following the individuals’ trades. These results suggest that individuals do not possess a predictive ability to make good stock selections. Like local institutional investors, Korean individual investors prefer to buy small-cap stocks, as evidenced by the significant negative coefficients of Mkt. Cap(t).

Tables 11 and 12 report the logistic regression results using CF/Sales and Div. Yield as independent variables. Because only the yearly data of these two variables are available, we perform logistic regression analysis for two subsamples. The first subsample includes firms whose portfolio formations take place within 6 months from the dividend payment. The results using the first subsample are reported in Table 11. The second subsample is comprised of firms whose portfolio formations are carried out in the more than 6 month interval from the dividend payment. The results using the second subsample are presented in Table 12.

Table 11.   Results from logistic regression analysis of semi-annual CF/Sales and Div. Yield
The values in the table are the coefficient estimates of logistic regressions between independent variables and the probability of buying a given firm. CF/Sales is cash flows from operation to sales, Div. Yield is dividend yield, Debt/Asset is total debt to assets, Ln(Mkt. Cap.) is a natural log of market capitalization. Time notations of t − 2, t − 1, t, t + 1, t + 2 represent two and one financial report periods prior to the contemporaneous financial report period, the contemporaneous financial report period, and one and two financial report periods subsequent to portfolio formation, respectively. Caution is needed to interpret CF/Sales and Div. Yield ratios because these ratios are computed with yearly information; therefore, t−1 data should be compared with t + 1 data, while t − 2, t, and t + 2 data should be compared together. χ2-statistics are in parentheses. The likelihood ratio (LR) χ2 statistic indicates a joint test statistic testing for all the regression coefficients being equal to zero. ***, **, and * denote statistical significance at the 1, 5, and 10% level, respectively.
 ForeignersLocal institutionsIndividuals
(4-A)(4-B)(5-A)(5-B)(6-A)(6-B)
Intercept−11.7701*** (67.460)−11.6590*** (75.331)6.4475*** (26.351)3.9610*** (10.995)11.3369*** (63.878)12.0083*** (79.019)
CF/Sales(− 1)0.4356 (2.350) 0.6473** (4.082) −0.1561 (0.298) 
CF/Sales(t + 1)0.5783** (4.149) 0.8794*** (7.736) −0.7979*** (7.804) 
Div. Yield(t − 1) −0.0000 (0.000) −0.0042 (0.004) −4.8729*** (10.865)
Div. Yield(t + 1) 2.6404* (3.546) 9.5012*** (19.440) −0.0136 (0.315)
Debt/Asset(t)0.0172 (2.063)0.0165 (1.979)0.0003 (0.008)0.0026 (0.563)0.0002 (0.004)−0.0012 (0.196)
Ln(Mkt. Cap(t))0.4418*** (65.410)0.4393*** (74.781)−0.2486*** (26.922)−0.1578*** (12.289)−0.4311*** (63.219)−0.4550*** (78.953)
Dummy Fin−0.3033 (2.059)−0.2652 (1.729)−0.2439 (1.769)−0.2015 (1.306)0.5301*** (8.397)0.4894*** (7.696)
LR χ2-statistic95.04***96.57***39.72***51.08***92.81***102.26***
Number of observations933107592410429071038
Table 12.   Results from logistic regression analysis of annual CF/Sales and Div. Yield
The values in the table are the coefficient estimates of logistic regressions between independent variables and the probability of buying a given firm. CF/Sales is cash flows from operation to sales, Div. Yield is dividend yield, Debt/Asset is total debt to assets, Ln(Mkt. Cap) is a natural log of market capitalization. Time notations of t − 2, t − 1, t, t + 1, t + 2 represent two and one financial report periods prior to portfolio formation, the contemporaneous financial report period, and one and two financial report periods subsequent to portfolio formation, respectively. Caution is needed in interpreting CF/Sales and Div. Yield ratios because these ratios are computed with yearly information; therefore, t − 1 data should be compared with t + 1 data, while t − 2, t, and t + 2 data should be compared together. χ2-statistics are in parentheses. The likelihood ratio (LR) χ2 statistic indicates a joint test statistic testing for all the regression coefficients being equal to zero. *, **, and *** denote statistical significance at the 10, 5, and 1% level, respectively.
 ForeignersLocal institutionsIndividuals
(4-A′)(4-B′)(5-A′)(5-B′)(6-A′)(6-B′)
Intercept−4.6060*** (18.591)−5.0724*** (22.781)5.8784*** (27.616)5.6748*** (27.757)1.4271 (1.938)1.8925** (3.548)
CF/Sales(− 2)0.2423 (0.694) −0.1903 (0.348) −0.1855 (0.456) 
CF/Sales(t)−0.1175 (0.195) 0.6107* (3.116) −0.5008* (3.476) 
CF/Sales(t + 2)0.2395 (0.892) 0.4188 (1.569) −0.9042*** (7.953) 
Div. Yield(− 2) −0.0619 (0.744) 0.2727 (0.413) −0.0139 (0.000)
Div. Yield(t) 0.0298 (0.158) −0.2618 (0.218) 0.0906 (0.245)
Div. Yield(t + 2) −0.0150 (0.076) −0.1172 (0.7449) 0.0326 (0.413)
Debt/Asset(t)−0.0090 (1.484)−0.0094 (1.670)−0.0193** (4.106)−0.0218** (5.110)0.0172** (4.576)0.0159** (4.636)
Ln(Mkt. Cap(t))0.1770*** (18.567)     
 0.1970*** (23.743)−0.2259*** (27.391)−0.2152*** (27.533)−0.0557 (1.981)−0.0784** (4.239) 
Dummy Fin0.1308 (0.456)0.0853 (0.645)0.5164** (5.691)0.5351** (6.462)−0.1081 (0.295)−0.1520 (0.670)
LR χ2-statistic28.20***31.02***39.56***41.79***31.24***16.46**
Number of observations905105087510128771002

Looking first at the regression results for CF/Sales, the estimated coefficient of CF/Sales(t + 1) in regression (4-A) is positive and significant at the 5% level. Therefore, foreigners’ buy results in higher cash flows to sales than foreigners’ sell over the 6-month period subsequent to foreigners’ trades. As shown in regressions (5-A) and (5-A′), the regression coefficients of CF/Sales(t − 1), CF/Sales(t), and CF/Sales(t + 1) for local institutions are all positive (0.6473, 0.6107, and 0.8794, respectively) and statistically significant at least at the 10% level. These findings suggest that local institutions also tend to buy stocks with a high CF/Sales ratio during the previous 6-month period, and their buy trades lead to a significantly higher CF/Sales ratio than their sell trades in the subsequent 6-month period. In sharp contrast to foreigners and local institutions, Korean individual investors tend to sell stocks with a high CF/Sales ratio and to buy stocks with a low CF/Sales ratio. Furthermore, the CF/Sales ratio is likely to decline more significantly for the stocks that individuals buy than for the stocks that they sell for at least 1 year following their transactions. This is clearly evidenced by negative and significant (at least at the 10% level) coefficients of −0.5008, −0.7979, and −0.9042 for CF/Sales(t), CF/Sales(t + 1), and CF/Sales(t + 2), respectively, in regressions (6-A) and (6-A′).

With respect to the dividend yield, the regression results are similar to those for CF/Sales. As shown in regressions (4-B) and (5-B), the estimated coefficient of Div. Yield(t + 1) is a positive value of 2.6404 for foreigners and 9.5012 for local institutions, both of which are statistically significant at least at the 10% level. As a result, the stocks that foreigners and local institutions buy tend to generate a higher dividend yield than the stocks that they sell in the subsequent 6-month period following their trades. Unlike the other two investor groups, individuals tend to sell high dividend paying stocks, as evidenced by the negative and significant (at the 1% level) regression coefficient of −4.8729 for Div. Yield(− 1) in regression (6-B).

It is interesting to note that in all regressions (1)–(6), the intercept terms take on negative values for foreigners and positive values for local institutions and individuals. These findings suggest that, other things being held constant, foreigners are inclined to sell stocks more often, whereas domestic individuals and institutions tend to buy stocks more often over our sample period of 1996–2002.

In sum, the results from the logistic regression analysis confirm our earlier findings from the univariate analyses. Foreigners and local institutions generally prefer to buy stocks with high potential for improvement in operational performance. However, individual investors tend to systematically make poor investment decisions; therefore, they are more likely to buy stocks with inferior operational efficiency and to sell stocks with high profitability potential.

6. Summary and Conclusion

We investigate foreign investors’ trading behavior and performance relative to local institutions and individual investors in the Korean stock market, which the existing literature has not explored fully. Using daily transaction data for Korean companies listed on the KSE from January 1996 to December 2002, we find that foreigners tend to buy stocks that have outperformed previously relative to the market and to sell stocks that have underperformed, suggesting that foreigners behave like positive feedback traders. Similar to foreigners, local institutions show a tendency to buy more of the stocks that have relatively risen in price. In contrast, individual investors tend to buy past losers and to sell past winners, implying that individual investors trade like contrarians.

With regard to return performances in the post-portfolio formation period, our results provide strong evidence that foreigners outperform both local institutions and individual investors. The stocks that foreign investors buy significantly outperform the stocks that they sell over the 12-month period following their trades. In contrast, the contrarian strategy of individual investors does not appear to be effective in the Korean market. The stocks that individuals buy produce significantly lower returns than the stocks that they sell in the year following the portfolio formation. Similarly, the stock selection of local institutional investors results in a mediocre return performance in the post-portfolio formation period.

We then test whether the superior return performance by foreigners is attributable to their superior ability to predict the operating and financial performance of an individual firm. Applying both univariate and logistic regression analyses, we find that foreigners tend to buy stocks that have previously been highly profitable and to sell stocks that have been less profitable. Following the portfolio formation, the stocks that foreigners buy continue to outperform the stocks that they sell in terms of profitability. Hence, foreigners seem to be able to discern stocks with good versus bad prospects. In contrast, for institutional investors, there is little difference in profitability between the stocks that they buy and the stocks that they sell both before and after the portfolio formation. These findings suggest that institutional investors in Korea fail to discern the profitability of individual firms. Individual investors are found to have a strong tendency to buy stocks that have been less profitable and to sell stocks that have been more profitable. Therefore, Korean individual investors tend to buy past losers and to dispose of past winners. Furthermore, following the portfolio formation, individuals’ buy stocks significantly underperform their sell stocks.

Consistent with the evidence documented in previous studies, we find that foreigners prefer large-cap stocks with high dividends. However, a firm’s leverage level is not an important characteristic of stocks that foreigners trade, which is contrary to the findings documented in previous published studies. In contrast, individuals have a strong preference for small-cap, high-leveraged, low dividend paying stocks, while local institutions tend to buy small-cap, low-leveraged stocks. Our findings on the local institutions’ preference of small-cap stocks in the Korean stock market are in sharp contrast to the institutional preference for large-cap stocks observed in the US markets.

Footnotes

  • 1

    See, for example, Jegadeesh and Titman (1993), Grinblatt et al. (1995), Bohn and Tesar (1996), Chan et al. (1996), Rouwenhorst (1998), Nofsinger and Sias (1999), Badrinath and Wahal (2002), and Griffin et al. (2003).

  • 2

    One notable exception is a study by Bange (2000). She argues that individual investors are positive feedback traders based on her finding that individuals increase their equity holdings after market run-ups and decrease their holdings after downturns.

  • 3

    Based on their findings that foreign investors are typically institutional investors, Dahlquist and Robertsson (2001) insist on the presence of an institutional investor bias rather than a foreign investor bias.

  • 4

    Dvorak’s finding also suggests that the combination of local information advantage and global expertise leads to higher profits.

  • 5

    Kalev et al. also show that foreign investors can only utilize their superior investment skills to their advantage in the long term for internationally well-known stocks such as Nokia.

  • 6

    See some excerpts of news briefings from Korean financial authorities presented in the next section.

  • 7

    Choe et al. (2005) find no evidence that foreign investors are better informed than domestic institutions in the Korean stock market.

  • 8

    Source: the World Federation of Exchanges, 2010 Market Highlights.

  • 9

    The quarterly sales and earnings reports for KSE listed companies were not available until 2001.

  • 10

    This rollover approach to form portfolios might create a problem of autocorrelation in returns between two consecutive periods. To check the robustness of the results, we also constructed portfolios using the non-overlapping approach, which yields results qualitatively similar to those reported in the present paper.

  • 11

    Chan (1988) shows that in the estimation of abnormal returns of market strategies, the risks of losers and winners are not constant. That is, the estimation of abnormal returns might be sensitive to how the risks are estimated.

  • 12

    Barber and Lyon (1997) note that buy-and-hold abnormal returns and cumulative abnormal returns might yield downward and upward, respectively, biased results to measure long-run return performance.

  • 13

    Approximately 75% of KSE-listed companies have a fiscal year ending on 31 December. Only a small number of financial firms such as securities and insurance companies have a fiscal year ending on 31 March.

  • 14

    Caution is needed in interpreting CF/Sales and Div. Yield ratios because these ratios are computed using yearly information. Hence, t − 1 data should be compared with t + 1 data, while t − 2, t, t + 2 data should be compared together.

  • 15

    The Korea Stock Price Index, a value-weighted market index of all KSE-listed stocks, dropped by 29.4%, from 888.85 in January 1996 to 627.55 in December 2002.

Ancillary