Offshore Investment Funds

Offshore investment funds are alleged to have engaged in trading behavior that is different from their onshore counterparts because they may be subject to less supervision and regulation. In particular, they may trade more intensely. They could also pursue more aggressively certain trading strategies such as positive feedback trading or herding that could contribute to a greater volatility in the market. Using a unique data set, this chapter compares the trading behavior in the Korean stock market between offshore investment funds with their onshore counterparts registered in the US and UK. There are a number of interesting findings. First, there is indeed evidence suggesting that the offshore funds trade more intensely than their onshore counterparts. Second, however, there is no evidenOffshore investment funds are alleged to have engaged in trading behavior that is different from their onshore counterparts because they may be subject to less supervision and regulation. In particular, they may trade more intensely. They could also pursue more aggressively certain trading strategies such as positive feedback trading or herding that could contribute to a greater volatility in the market. Using a unique data set, this chapter compares the trading behavior in the Korean stock market between offshore investment funds with their onshore counterparts registered in the US and UK. There are a number of interesting findings. First, there is indeed evidence suggesting that the offshore funds trade more intensely than their onshore counterparts. Second, however, there is no evidence that the offshore funds engage in positive feedback trading. In contrast, there is strong evidence that the funds from the U.S. and U.K. do. Third, while offshore funds do herd, they do so far less than onshore funds in the U.S. or UK. Fourth, offshore funds hold less glamour stocks (e.g. stocks with high P/E) in their portfolio than funds in the U.S. or U.K. do. Moreover, flight to glamour stocks during the in-crisis period is less evident in the case of offshore funds. In sum, offshore funds are no especially worrisome monsters.


Introduction
Offshore funds are collective investment funds registered in tax havens, typically small islands in the Caribbean, Europe and Asia Pacific.The host countries/territories not only do not tax the funds, they typically do not forward the financial information to other tax and financial authorities.Furthermore, the regulation on these funds in the tax havens is often less stringent than that of major industrialized countries where most of the onshore investment funds are located.Helm (1997, p414) listed seven areas in which offshore funds face less regulations as compared with their counterparts in the U.S. For example, offshore funds would have greater flexibility and less procedural delays in changing the nature, structure, or operation of their products, and they would face fewer investment restrictions, short-term trading limitations, capital structure requirements, governance provisions, and restrictions on performance-based fees.
As a consequence, offshore funds may engage in trading behaviors that are different from their onshore counterparts.For example, it has been alleged that foreign portfolio investors may engage in positive feedback trading (e.g., rushing to buy when the market is booming and rushing to sell when the market is declining), and eager to mimic each other¡¯s behavior while ignoring information about the fundamentals.There is concern that offshore funds may be more prone to this kind of trading pattern than their onshore counterparts either due to the nature of their investment styles or due to lower regulatory constraints they face at home.Behaviors such as these by offshore funds could exacerbate a financial crisis in a country to an extent not otherwise warranted by economic fundamentals.
A better understanding of the offshore funds¡¯ behavior is highly relevant for th renewed debate on capital controls on short-term portfolio capital flows.Aside from outright capital controls imposed by capital receiving countries, one may imagine better supervision and risk regulation by the governments of the capital-exporting countries as another way to regulate international capital flows.Indeed, many may prefer this approach to outright capital controls imposed by capital-importing countries.However, the presence of offshore funds adds challenges to this approach.Even when the G7 governments can agree on a particular regulatory structure, it may not apply to the offshore centers.Moreover, many currently onshore funds could migrate offshore as a result of changes in the regulations in their onshore domiciles.
The hypothesis that offshore funds may pursue destabilizing trading strategies can be connected with an emerging literature on behavioral finance, mostly in the domestic finance context.For example, using evidence from domestic market data, it has been argued that institutional investors often exhibit herding behavior, though the tendency is quantitatively small (see Lakonishok, Shleifer and Vishny, 1992).There are also theoretical models in which rational investors may pursue positive feedback strategies, destabilizing prices in the process (De Long, Shleifer, Summers, and Waldmann, 1990).
A number of authors have empirically examined the behavior of foreign investors in emerging markets.They include Frankel andSchmukler (1996, 1998), who have investigated closed-end country funds; Choe, Kho, Stulz (1998), who have examined the effects of foreign investor as a whole on the Korean stock prices; Froot, O¡¯Connell an Seasholes (1998) who have examined the aggregate portfolio flows into various countries; and Kim and Wei (1999), who have looked into the differences as well as similarities in trading behavior between individual versus institutional foreign investors, and foreign investors who reside in Korea versus those outside.None of these papers has compared the behavior between offshore and onshore funds.
Fung and Hsieh (1997), Brown, Goetzmann andIbbotson (1999) andBrown Goetzmann andPark (1999) pioneered the examination of trading strategies of hedge funds, many of them located offshore.They find that hedge funds appear to shift weights on different assets very frequently.The last paper finds that the currency hedge funds were unlikely to have triggered the Asian currency crisis.Lacking the data on actual position holdings of the funds, these papers utilize return information to infer trading strategies a la Sharpe¡¯s (1992) style analysis.This is clever and very useful, but ther can be errors if certain assets that the funds have actually traded on are not included in the analysis by the econometricians, and the omitted and included assets have correlated returns.
In this paper, we utilize a unique data set on actual month-end trading positions of foreign funds in Korea to study the behavior of offshore funds.To put the results in context, we compare them with those funds that are registered in the United States and United Kingdom (and also Singapore and Hong Kong as a supplementary group), where the relevant regulations and regulators are well-respected, and where most onshore funds are located.The data covers the period from the end of 1996 to June 30, 1998, which allows us to see if the behavior of the funds changes during a financial crisis.
The paper is organized as follows.Section 2 describes our data sets.Sections 3, 4, and 5 examine three aspects of foreign investor behavior, respectively: turnover, feedback trading, and herding.Section 6 offers some concluding remarks.

Data
Offshore and onshore funds and their positions Our investor position data set identifies each foreign investor by a unique ID code, and reports the domicile of each fund, and its month-end holding of every stock listed in the Korean stock exchange.Our sample covers the period from the end of 1996 to June 30, 1998.This proprietary data set was kindly provided to us by the Korea Securities Computer Corporation (KOSCOM), an affiliate to the Korea Stock Exchange (KSE).
Our set of offshore funds are mutual funds or unit trusts that report their domicile to the Korean government as either Bahamas, Bermuda, Cayman Islands, Channel Islands, Guernsey, Jersey, Liechtenstein, Panama, or the British Virgin Islands.There are 77 such funds that own some stocks at least sometime during the sample.It is interesting to note that almost every single such domicile has a current or historical Anglo-Saxon connection.
According to anecdotal evidence, many of the investors in the offshore funds are current or past nationals of the United States, United Kingdom or other G7 countries.
For comparison, we also look at mutual funds or unit trusts that are registered in the United States and United Kingdom (as a group), two largest homes of the onshore investment funds, and those in Singapore and Hong Kong (as another group).All of the four have well-regarded securities and mutual fund laws and competent regulatory agencies.There are a maximum of 783 funds in the US/UK group, and 36 funds in the Singapore/HK group in the sample.
We exclude funds from many other domiciles such as Luxembourg from the analysis because we cannot separate offshore from onshore funds registered in the same country.We also exclude pension funds, commercial banks, investment banks, or insurance companies from our analysis, because none of them active in Korea except for one commercial bank comes from an offshore center on our list.30,1998).
November 1997 was the month when the foreign exchange crisis occurred in Korea.On November 18, the Bank of Korea gave up defending the Korean Won.And on November 21, the Korean government asked the IMF for a bail out.In some of our analyses, we break the sample into two: a pre-crisis period before and including October 1997 (ten months in our sample), and an in-crisis period from November 1997 to June 1998.

Intensity of Trading
Not having to pay capital gains tax, and facing less supervision and regulation from home governments may induce offshore funds to trade more intensively than their onshore counterparts 3 .In addition, investment funds that prefer to trade more actively may self-select to locate in the offshore centers.
In this section, we examine whether offshore funds actually trade more intensely or not.Because our data does not record within-month transactions, we cannot compute an accurate measure of turnover.However, we observe the total changes in the weights allocated to different stocks on a monthly basis.Our presumption is that, across investor groups, the total changes in the month-to-month weights are highly correlated with the true turnovers.We will use the term ¡°trading intensity¡± in subsequent discussions denote the changes in the weights on all the stocks.
Let w(j, k, t) denote the market value of the position in stock k held by investor j at the end of month t, divided by the total value of all stocks held by the same investor at the same time.We compute the sum of the absolute values of the changes in the weights across all stocks for investor j at time t using the following definition: 2 www.bog.frb.fed.us/release/H10/hist/ 3 While the offshore funds may not pay taxes in their domiciles, they may still need to pay taxes in Korea, in particular, 25% withholding tax on dividend and interest, and 10% of the gross proceeds realized from the sale for capital gains.In cases where the purchasing price is available, the tax is the lesser of 25% of the capital gains and 10% of the gross proceeds.See the Korea Stock Exchange Website, www.kse.org/kr/stat/index.html.These tax rates are typically lower than what the onshore funds have to pay to their home taxing authorities.
The average trading intensity (weight changes) for investor j defined as: where T is the total number of months in the sample.The average trading intensity for investors in a given group is then the average of all TN(j) over investor j in the group i (subscript-i omitted): Under the central limit theory, the TN measure is asymptotically normal.
Panel A of Table 2 reports, for each of the three groups of the funds, the trading intensity measured in this way.For the whole sample, we see that the average trading intensity for the offshore funds is 45% bigger than that for the US/UK funds.Using a difference-in-mean test, we can see that the difference between the two is statistically significant at the five percent level (Column 4).On the other hand, the offshore funds¡ trading intensity is not statistically different from the Singapore/Hong Kong funds (Column 5).
If we break the sample into pre-crisis and in-crisis sub-periods, we see an interesting pattern.The average trading intensity increases for each of the three groups of funds in the crisis period relative to the pre-crisis period (and significant for the US/UK funds).The offshore funds¡¯ average trading intensity continues to be bigger than th onshore funds from the US/UK.
As a robustness check, we also experiment with defining the trading intensity in terms of the physical shares of stocks instead of the market value of the stocks.To be more precise, we let w(j, k, t) be the number of stock k held by investor j at the end of month t, divided by the total number of all stocks that she held at the same time.Then, TN(j) and TN are defined in the same way as before.The results are reported in Panel B of Table 2.We can see clearly that all the qualitative results from Panel A remain to be true here.Thus, the offshore funds do trade more intensely than onshore funds (from the US and UK) both before the crisis, and even more so during the crisis.

Positive Feedback Trading
There are concerns that offshore funds may engage in positive-feedback trading more aggressively than onshore funds, and that positive feedback trading could destabilize the market.Positive feedback trading pattern is when one buys securities when the prices rise and sells when the prices fall.This trading pattern can result from extrapolative expectations about prices, from stop-loss orders --automatically selling when the price falls below a certain point, from forced liquidations when an investor is unable to meet her margin calls, or from a portfolio insurance investment strategy which calls for selling a stock when the price falls and buying it when the price rises.
Positive feedback trading can destabilize the market by moving asset prices away from the fundamentals.At least since Friedman (1953), many economists believe that positive feedback traders cannot be important in market equilibrium as they are likely to lose money on average.This view has been challenged in the last decade or so.De Long, Shleifer, Summers, and Waldmann (1990) argued that in the presence of noise traders, even rational investors may want to engage in positive feedback trading, and in the process destabilize the market.
Empirical examination of this issue has emerged recently.Using quarterly data on U.S. pension funds in the U.S. market, Lakonishok, Shleifer, and Vishny (1992, LSV for short in later reference) did not find strong evidence of significant feedback trading.
Using transaction-level data, Choe, Kho, and Stulz (1998) also find evidence that foreign investors as a group engage in positive feedback trading in Korea.No paper that we are aware of compares the positive trading tendencies of offshore versus onshore trading strategies.

Methodology
The objective is to examine the connection between the trading behaviors of the investors (within a given sub-group) and the previous month performance of the stocks.
Within The first measure describes the fraction of active traders that is a net buyer.It is constructed to minimize the dominance of a few large traders in the statistics.The second measure describes the net purchase (scaled by the total trading).The denominator (the scale adjustment) makes sure that a large purchase does not receive more weight than a small purchase To avoid possible biases in quantifying the trading behavior, we exclude certain observations (investors or stock-month).First, investors who are registered after December 31, 1996 are dropped because their entrance to the market could show up only as a buy.Second, stock-months for which a stock has reached the foreign ownership limit are dropped because any change in the net position of the foreign investors as a whole has to be a sell to Korean investors.

Results and Interpretations
Table 3 reports the basic finding using buyer¡¯s ratio as a measure of tradin direction.Let us look at the US/UK funds first.For the entire sample period (97.1-98.6)(reported in Column 4 of the top panel), 39% of active traders buy the worst performing stocks (in terms of last month returns), compared to more than 50% of active traders who buy the recent best performing stocks.Indeed, in the sixth row, we report a formal t-test on difference between the two buyers¡¯ ratios.The standard errors are reported i parenthesis4 .We see the difference is positive and statistically significant.This is consistent with the view the US/UK funds are positive feedback traders.
In contrast, for the offshore funds (reported in Column 3), the buyer¡¯s ratios fo the recent worst and best performing stocks are 41% and 46%, respectively.The difference between the two ratios is smaller than for the US/UK funds.In fact, a formal ttest indicates that the difference is not statistically significant at the ten percent level5 .
The same is true for funds from Singapore/Hong Kong.
When we look into pre-and in-crisis sub-samples (middle and lower panels of Table 3), we see that the propensity to engage in positive feedback trading by American and British funds is stronger during the crisis than before it.There is still no statistically significant evidence that offshore funds engage in positive feedback trading.
In Table 4, we use the scale-adjusted net purchase as an alternative measure of trading patterns.Onshore funds from the US and UK sell recent losers more aggressively than recent winners, a pattern consistent with positive feedback trading.In comparison, the offshore funds do not exhibit statistically significant difference in the net purchase of the recent worst and best performing stocks.Hence, we reach the same qualitative conclusion as before: no evidence to support the hypothesis that offshore funds engage in positive feedback trading more aggressively than onshore funds from the US or UK.If anything, the contrary is true.

11
In Table 5, we decompose the stocks along a second dimension, the market capitalization at the beginning of the month, into small, medium and large stocks.So within a sample period, the stocks are now classified into nine categories.We observe that the offshore funds tend to hold mostly medium and large stocks relative to the U.S./UK funds.Moreover, for the US/UK funds, the positive feedback trading pattern is most visible for large stocks in the pre-crisis period, but most visible for small or medium stocks during the crisis.
A possible defense of positive feedback trading is that foreign investors (residing abroad) may be informationally disadvantaged relative to domestic investors.They may take a (relatively greater) decline in the price of a particular stock as unfavorable news revealed by domestic investors, and may therefore rationally choose to sell it (more aggressively relative to other stocks) (See Brennan and Cao, 1997, for such a model).It may be useful to check if the positive-feedback-trading pattern in our sample is ex post profitable.We do it in two steps.First, in each month, we form an equally-weighted portfolio of ten best performing stocks, and another equally-weighted portfolio of ten worst performing stocks, based on the previous month¡¯s return as defined above fo Tables 3 and 4.
The average returns of the two portfolios in the previous months are reported in the first row of each of the three panels (representing three different periods) in Table 6 (labeled as ¡°horizon -1¡±).Second, we track their performances over the subsequent s months.The results are reported in the other rows of Table 6 (labeled as ¡°horizons 1 6¡±).We perform a difference in mean test (mean return of the past winners minus that o the past losers) and find that the difference is negative for all six horizons under investigation.The difference is statistically significant for the one-to five-month horizons at the ten percent level.In other words, the data suggest that the relative ranking of stock performance reverses itself in the sample.On average, if one has to choose between a negative and a positive feedback trading strategy, the former would have been superior, at least at the one-or two-month horizon.The excess return is quantitatively large at 8% monthly rate.Of course, in this down market, selling both the best and worst performing portfolios would be ex post more profitable (and one should sell recent winners more aggressively).
As a robustness check, we also form equally weighted portfolios of 30 best performing and 30 worst performing (based on previous-month¡¯s returns) stocks.Th results are reported in the right half of Table 6.For these enlarged portfolios, again, there is reversal in the ranking of relative performance.In fact, the recent past losers outperform the recent winners, in a statistically significant and quantitatively large way, over one-month, two-month, and so on, all the way to five-month horizons.Again, a contrarian trading strategy rather than a positive feedback one would have been profitable.
As qualifications, we note that our thought experiments have not adjusted for risk levels of the stocks, and do not preclude the possibility that a positive feedback trading strategy could be profitable within a day or for horizons longer than six months.

Correlated Trading
Herding is the tendency that investors of a particular group mimic each other¡t rading.Portfolio investors may herd rationally or irrationally.Informational asymmetry may cause uninformed but rational speculators to choose to trade in the same way as informed traders (Bikhchandani, Hirshleifer and Welch, 1992;and Banerjee, 1992).
Since informational problem may be more serious when it comes to investing in a foreign market than the domestic one, herding may be more severe correspondingly.Whether offshore funds herd more or less than the onshore funds depends on their relative capacity in collecting and processing information about the emerging market in question.
There is an alternative explanation for herding among institutional investors.
Unlike individual investors, fund managers face regular reviews (e.g., quarterly for mutual funds, and annually for pension funds) on their performance relative to a benchmark and/or to each other.This may induce them to mimic each other¡¯s trading t a greater extent than they otherwise would (See Scharfstein and Stein, 1990).By this logic, whether the offshore funds herd more or less than the onshore funds depends on whether informational asymmetry is greater or less for them.By this logic, there might be less herding among offshore funds if they are subject to either fewer or less frequent performance reviews.
There have been several empirical papers that quantify herding behavior.Using data on institutional investors, the pioneering paper by Lakonishok, Shleifer, and Vishny (or LSV, 1992), followed by Grinblatt, Titman, and Wermers (1995) ? ?To avoid any possible bias in computing the herding indices, we exclude certain investors and observations (stock-month) from our sample.Like the sample we have constructed to examine positive feedback trading, we exclude here (1) investors that are registered after December 31, 1996, (2) stock-months for which the foreign ownership limit is reached, and (3) stock-months for which the stocks are not owned by foreign investors in the previous month.The last exclusion is motivated by the short-selling constraint.When short selling is not allowed, any trade on that stock would have to first show up as a buy, thus biasing the herding index upward (Wylie, 1997).Finally, if a stock in a given month is traded by only one foreign investor in that group, that observation is dropped.

Results and Interpretations
The basic results are presented in Table 7a.For each investor group i and sample period, we report the corresponding herding statistics, H(i), with standard errors in the parenthesis below.Then we perform a sequence of difference-in-mean tests between offshore and onshore funds (reported in Columns 4 and 5), and between pre-crisis and incrisis periods for any given group of investors (reported in Row 4).
The most important findings are the following.First, for both offshore funds as well onshore funds from the US and UK, their positive herding statistics are statistically significant.The only possible exception is the set of funds from Singapore and Hong Kong.Second, most importantly, the evidence suggests that, to the extent investment funds herd, the US/UK funds herd significantly more than their offshore counterparts (for the whole sample and for the pre-crisis period).
One may worry that a firm that issues new stocks or buys back its stocks could artificially inflate the herding measure even there is no herding.In Table 7b, we drop all the observations that involve changes in the quantity of outstanding shares6 .We find that, aside from some minor differences, the results we have reached from Table 7b are essentially the same as those in Table 7a.

Ex post Profitability
What we label as ¡°herding statistics¡± (following LSV, 1992) is actually a measu of correlated trading.A bigger value of the ¡°herding¡± measure for the US/UK fun could result from the fact that they are more likely to respond to common signals than the offshore funds.In other words, the herding measures do not distinguish between two possibilities: that investors intentionally (rationally or not) mimic each other¡¯s trading versus that investors respond to common information about the fundamentals.
To distinguish between the two is difficult which is probably why previous empirical papers do not do this.We decide to provide some suggestive evidence here by examining ex post rationality of the herding behavior in our sample.Under the joint hypotheses that the funds respond to common signals and that the signals are payoff-relevant, we would expect that those stocks that the investors herd more aggressively should yield abnormal returns (relative to those stocks they do not herd as much).
Let 1 + jt R denote the return of stock j from t to t+1 in excess of the KOSPI return minus the won exchange rate depreciation.Let jt H denote LSV herding index for stock j in month t , and jt NP the (scale-adjusted) net purchase of stock j in month t .All three variables are defined for a given investor group, i , which we omit from the subscripts for simplicity.For each investor group, we run the following fixed effects regression: (7) where t α and k α are time and industry dummies7 .If those stocks that the funds herd to buy appreciate faster than others, and/or if those that the funds herd to sell depreciate faster than others, we would expect 1 β to be positive.We perform this regression for both the one-month and three-month investment horizons.The results are reported in Table 8.
In overwhelming number of groups, we see that the estimates of 1 β are not different from zero, and in the two instances when they are significant, they have a negative sign.This is true for both the one-month and three-month horizons.Hence, the joint hypotheses are rejected.

Concluding Remarks
In this paper, we study the behavior of offshore investment funds as compared with their onshore counterparts in the US, UK, Singapore and Hong Kong.This is made possible by a unique data set that details the monthly stock positions of foreign investors.
There are a number of findings that are worth highlighting here.First, there is evidence that offshore funds indeed trade more aggressively than their onshore counterparts, judging from the average turnover (or more precisely, monthly average value of changes in the month-to-month positions, scaled by the funds¡¯ size).Second there is no significant evidence to support the allegation that the offshore funds engage in positive feeding trading.In contrast, there is strong evidence that funds from the US and UK do exhibit a tendency to do so.Third, while offshore funds do herd, they do so far less than onshore funds from the US or UK.
In sum, the offshore funds are not especially worrisome monsters.Notes: (1) Stock-months non-resident foreign institutions invest are divided into five groups according to prior-month return, defined as return in excess of the KOSPI return minus the won depreciation against the US dollar.For each return-group, the (equally-weighted) mean value of buyers¡¯ ratio [=(no.o buyers -no. of sellers) / (no. of traders] is reported. (2) Within each investor group and sample period, difference in mean t-test is performed on the (equally-weighted) mean value of buyers¡¯ ratio stocks that are best and worst performers in th previous month.Standard errors are in the parentheses.** and * indicate significant at the 5% and 10% levels, respectively.Notes: (1) We form portfolios of best and worst performers based on previous month excess returns (reported in the rows labeled as ¡°horizon -1¡±), and then track their relati performances in the subsequent six months (reported in rows labeled as ¡°horizons 1-6¡± We constrain the sample to those that three investor groups trade on. (2) The return (for a given stock) is defined as (lnP t -lnP    1997Ä 2 1997Ä 3 1997Ä 4 1997Ä 5 1997Ä 6 1997Ä 7 1997Ä 8 1997Ä 9 1997Ä 10 1997Ä 11 1997Ä 12 1998Ä 1 1998Ä 2 1998Ä 3 1998Ä 4 1998Ä 5 1998Ä 6 Period G-B Residents Non-residents  1997Ä 2 1997Ä 3 1997Ä 4 1997Ä 5 1997Ä 6 1997Ä 7 1997Ä 8 1997Ä 9 1997Ä 10 1997Ä 11 1997Ä 12 1998Ä 1 1998Ä 2 1998Ä 3 1998Ä 4 1998Ä 5 1998Ä 6 period G-B

Figures of 1 Figure 5 :
Figures of 1-3, ex rate, kospi, and value of $100 Figure 4: evolution of market capitalization Figure 5: Size distribution by categories

Figure
Figure 4. Total Market Value of Position by Domocile (Million U.S. Dollars)

Figure
Figure 5. News on the Korean Economy: Asians vs. Non-Asians

Figure
Figure 6.News Gap: Residents vs. Non-residents

Figure
Figure 5. News on the Korean Economy: Asians vs. Non-Asians

Figure
Figure 6.News Gap: Residents vs. Non-residents

Table 1
reports the number of funds in each category.We see that the average position of an offshore fund in Korea is a lot smaller than the average of an American or British fund, though slightly larger than that of a Singapore or Hong Kong fund.There is no category labeled as hedge funds in our sample.Our understanding from communicating with KOSCOM is that they would register themselves either as mutual funds, unit trusts, or as ¡°others¡±.Notice that a hedge fund can either be an onshore each time period, we form five approximately equally sized (in terms of stock-months) portfolios based on the previous month performance of the stocks.The performance of a stock is defined as the return of the stock in excess of the market return, Lakonishok, Shleifer and Vishny (1992)on exchange rate against the U.S. dollar.That is, the return for a particular stock from month t-1 to month t is [ln(P t ) -ln(P t-1 )] -[ln(KOSPI t ) -ln(KOSPI t-1 )] -[ln(S t ) -ln(S t-1 )], where P t , KOSPI t , and S t are the price of the stock (stock subscript omitted), KOSPI index, and Won/$ exchange rate at time t.FollowingLakonishok, Shleifer and Vishny (1992), we employ two measures of investors¡¯ trading direction: a buyers¡¯ ratio and a scale-adjusted net purchas

Table 3 :
Chasing the past returns (buyer¡¯s ratioTable 4: scale-adjusted net purchases Table 5: net purchase, stocks also divided by market cap or share price Table 6: ex post profitability of chasing the past returns

Table 7 :
Herding (chasing each other¡¯s positions >>> change # investors to be in square bracket [ ]

Table 8 :
ex post profitability of herding

Table 1 : Number of Foreign Investors by Origin
The investors in the table include only portfolio investors who had registered with the Korea Securities Supervisory Board (KSSB) by December 31, 1996.

Table 4 . Positive Feedback Trading
Note: Please see the footnotes to Table3.

Table 8 . Ex-Post Profitability on Herding (Net Purchase)
is the return from t to t + 1 on stock j ; α t , Month dummy; R k α , Industry dummy; jt H , Herding index at time t for stock j ; jt NP , Scale adjusted net purchase at time t for stock j .