Relations between mutual fund flows and stock market returns in Korea

Transcription

1 Int. Fin. Markets, Inst. and Money 17 (2007) Relations between mutual fund flows and stock market returns in Korea Natalie Y. Oh a,, Jerry T. Parwada b,1 a Department of Accounting and Finance, Monash University, PO Box 197 Caulfield East, Vic 3145, Australia b School of Banking and Finance, University of New South Wales, UNSW Sydney, NSW 2052, Australia Received 21 April 2005; accepted 3 October 2005 Available online 11 January 2006 Abstract This paper analyses relations between stock market returns and mutual fund flows in Korea. A positive relationship exists between stock market returns and mutual fund flows, measured as stock purchases and sales and net trading volumes. In aggregate, mutual funds are negative feedback traders. Standard causality tests suggest that it is predominantly returns that drive flows, while stock sales may contain information about returns. After controlling for declining markets, the results suggest Korean equity fund managers tend to increase stock purchases in times of rising market volatility, possibly disregarding fundamental information, and to sell in times of wide dispersion in investor beliefs Elsevier B.V. All rights reserved. JEL classification: G10; G20; G23 Keywords: Mutual funds; Investment flows; Stock markets 1. Introduction Mutual funds have experienced exponential growth in most countries with well-functioning stock markets over the last two decades. In some cases the expansion rate of the funds management industry has sparked speculation in the popular press about its adverse effects on stock markets. 2 Corresponding author. Tel.: ; fax: addresses: natalie.oh@buseco.monash.edu.au (N.Y. Oh), j.parwada@unsw.edu.au (J.T. Parwada). 1 Tel.: ; fax: It is debatable whether equity fund flows are fundamentals, i.e., information motivated volume (see, for example, Warther, 1995). Rapid expansion of volume may be due to price pressure or herding in which prices may deviate from the fundamentals /$ see front matter 2005 Elsevier B.V. All rights reserved. doi: /j.intfin

2 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) We examine flows between the funds and the stock market in Korea between 1997 and Taking advantage of our ability to observe daily stock trading volumes for equity mutual funds, we employ a series of tests to examine the reciprocal effects of mutual fund and stock market activities. As Khorana et al. (2005) point out, there has been relatively little research performed on mutual funds outside the U.S. This paper contributes to the literature by investigating the newlyestablished Korean mutual fund market that has a history commencing in earnest in the early 1970s. 3 Institutional arrangements in the Korean mutual fund industry are different from the more extensively researched markets. For example, the dominant type of mutual fund is of the contractual rather than the corporate type widely found in the U.S. The results of our study can be summarized as follows. We find that a positive relationship exists between stock market returns and mutual fund flows, defined as stock purchases and stock sales. However, net flow displays a negative relationship with market returns which suggests negative feedback trading by the mutual fund industry. 4 Tests on the direction of causality suggest that it is predominantly returns that contain information on flows, although flows measured as stock purchases may also contain information about returns. However, in regressions containing market fundamentals, the most robust finding in terms of explanatory power is that returns drive flows. We also document a significant positive relation between volatility as a proxy for uncertainty and disaggregated mutual fund flows, i.e., stock purchases and stock sales flow. Aggregate flows do not show a significant relation with risk. The rest of the paper proceeds as follows. Section 2 briefly reviews the literature related to the current study. Section 3 describes the size of the Korean mutual fund industry. Section 4 examines the relationship between mutual fund flows and stock market returns. Section 5 summarizes and concludes. 2. Related literature The simultaneous growth of institutional investors and the market index has inspired academics and practitioners to focus on the impact of mutual fund flows on the market index. According to the efficient markets hypothesis, information moves prices but anecdotal evidence suggests supply and demand shocks for stocks may also move prices. A growing literature interprets relations between money flows and returns (or asset prices) in a behavioral finance framework. For example, one important finding is that of momentum trading. Positive feedback ( negative feedback ) traders buy (sell) after market rises (declines) and sell (buy) after market declines (rises) (DeLong et al., 1990). Warther (1995) and Zheng (1999) suggest these demand and supply effects are not confined to the firm level alone; they extend to the aggregated level as well. 5 3 In Korea s neighborhood, Japanese mutual funds have been more extensively researched (see, for example Cai et al., 1997 and Brown et al., 2001). Although the mutual fund industry in Korea has its regulatory origins in the late 1960s when the Securities Investment Trust Business Act (SITBA) introduced investment trusts similar to contractual-type unit trusts found in the United Kingdom and Australia, the first equity investment trust was only introduced in The first full-scale securities investment trust company (ITC) specializing exclusively in mutual fund products, Korea Investment Trust Company, was established in According to Grinblatt et al. (1995), a net flow by an investor-type that is correlated with past returns can be considered feedback trading. 5 For more examples of fund flow stock return studies in the U.S. see Warther (1995); Fant (1999); Edwards and Zhang (1998); Cha and Lee (2001); Goetzmann and Massa (2003).

3 142 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) Table 1 Descriptive statistics on KSE investor profiles ( ) Year ITMCs Sec Co Ins Co Banks Pension Indiv Foreign Others Panel A: purchases Panel B: sales This table provides descriptive statistics on the profile of different types of investors on the Korean Stock Exchange. The figures represent the average percentage volume each type of investor contributes to the market in comparison to the total volume. The average volume is partitioned into stock purchases and sales. ITMCs are Investment Trust Management Companies or funds management companies; the rest of the investor-types are securities companies (Sec Co), insurance companies (Ins Co), banks, pension companies, individuals (Indiv), foreign investors (Foreign), and the rest (Others). 3. The size of the Korean mutual fund industry Khorana et al. (2005) report that Korea, with France, rank second among countries with the highest number of mutual fund offerings, after the U.S. Assets held in Korean mutual funds have outstripped the nation s Gross Domestic Product since 1998 and now represent about half the stock market capitalization. 4. Data and empirical design The data used in this section were provided by the Korean Stock Exchange (KSE). The database, covering the period, contains daily volume and value traded by different types of investors partitioned into sales and purchases. The investor categories are securities companies, insurance companies, investment trust companies, banks, pension companies, individuals, foreign investors and others. 6 Table 1 shows the percentage volume traded by different types of investors in the Korean stock market. Individual investors are the dominant traders, accounting for more than two thirds of both sales and purchases volume, followed by foreign investors and then investment trust companies (ITMCs) or mutual funds. 6 The term banks encompasses commercial, merchant and mutual banks.

4 4.1. Flow return relationship N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) To analyze the relationship between mutual fund flows and stock market returns this paper utilizes methods described in the literature on the relationship between fund inflows and market returns (see, e.g., Warther, 1995; Cha and Lee, 2001; Edelen and Warner, 2001; Goetzmann and Massa, 2003). As our first task, we assess the impact of mutual fund flow on the market index using a vector autoregressive regression (VAR) method (see Seasholes, 2000; Froot et al., 2001). Up to five flow lags are used to detect positive feedback trading. As such, the specification addressing the intertemporal relation between mutual fund flows and KOSPI index returns and the possibility of feedback trading is based on the following model: [ ] [ ] [ ] [ ] Rt αr βλ RR (L) γλ RF (L) ε R = + αf δλ FR (L) ϑλ FF + t, (1) (L) F t where λ(l) are distributed lag operators on lagged flow and returns, and F and R are, respectively, index fund flows and market returns. 7 The three flow variables used are: standardized purchase flow denoted as Purchases; standardized sales flow denoted as Sales; and Net computed as the difference between total purchases and total sales volumes. All flows (Sales, Purchases and Net) are normalized by the trailing 90-day moving average of the KOSPI market capitalization to control for market and fund growth as per Warther (1995); Goetzmann and Massa (2003). Thus, STDFLOWS = RAWFLOWS/ROLLMKT, where STDFLOWS are standardized flows, RAWFLOWS are the raw flows before standardization, and ROLLMKT is the rolling moving average of market capitalization in the previous 90 days. 8 ε F t 4.2. Causality tests Standard causality tests are used to determine the direction of the impact, i.e., whether it is the market index that determines flows or vice versa. Identifying the direction supplements the price pressure hypothesis tests. If flows move the market index then price pressure exists between mutual funds and the stock market. 9 Do flows contain information about returns? As an improvement on the causality tests and to analyze whether equity fund flows affect market returns in the presence of market fundamentals, tests for the effect of equity fund flows in the presence of variables such as dividends, the interest rate and the risk premium are performed in the spirit of Cha and Lee (2001). The usefulness of including market fundamentals lies in the fact that, should causality in this context only be in the direction of stock returns to flows, and not otherwise, then this would prove that it is only market returns that drive mutual fund flows. The following regression equations incorporating market 7 Similar to Goetzmann and Massa s (2003) model, the five flow lags represent the full trading week plus the last day of the preceding week to retain the parsimonious qualities of the specification, while incorporating the salient lags. 8 As an alternative the predictable component of flows is separated from the unpredictable component and tests carried out on how the unexpected flows affect market returns. The results are found not to be sensitive to the two methods and the results reported are based on a standardization method utilizing a 90-day moving average. 9 In this context it should be noted that a Granger causality test (due to Granger, 1969) targets precedence, linear precedence in particular.

5 144 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) fundamentals are used: m m Ret t = α + βret t 1 + χflow t 1 + Ddiv + ΦAxrate + πaspr + ρdint + ε (2) and, i=1 i=1 i=1 m m Flow t = η + μret t 1 + πflow t 1 + θddiv + ωaxrate + ψaspr + ζdint + ε. (3) i=1 In the specifications, Flow t is Purchases, Sales or Net (as defined earlier on); Ret t the market return calculated by the natural logarithm of the difference between today s and yesterday s market index; the market fundamentals are Ddiv, the dividend yield; Axrate, the change in exchange rate (USD/Won); Aspr proxied by the difference between the 3-year government bond and 3- year corporate bond rates; and Dint, the 90-day commercial paper rate. All the fundamental variables are averaged over the preceding 5 days. Ddiv and Dint are differenced to meet stationarity requirements. The null hypothesis that equity fund flows do not Granger-cause stock market returns in the presence of market fundamentals is tested by H 0 :β i = 0, for all i in (2). Similarly, the null hypothesis that stock market returns do not Granger-cause equity fund flows in the presence of market fundamentals is tested by H 0 :μ = 0, for all i in (3) Incorporating risk As the mutual fund managers decisions are based on expectations of risk as much as market returns (Goetzmann and Massa, 2003), we examine the determinants of flow after incorporating risk variables. In addition to being risk measures per se, the flow measures are correlated to diverse beliefs among investors. Indeed, according to Harris and Raviv (1993); Shalen (1993), the wider the dispersion of beliefs among investors, the higher the returns and volumes in excess of the equilibrium. To estimate how the growth of Korean mutual funds is affected by the measures of uncertainty, following Goetzmann and Massa (2003), the following model is used: F t = α + βunc t + γinfv t + δf t 1 + ε t, (4) where F t denotes flow measurements as being Purchases, Sales and Net (measured as the difference between Purchases and Sales). InfV t is a vector of information variables, ε t the error term, and Unc t represents the measure of uncertainty under consideration. Two proxies for uncertainty are considered. First, volatility is measured as the square of the natural logarithm of return. 10 Since anecdotal evidence suggests Korean investors are mostly daytraders, intra-day volatility based on the Garman and Klass (1980) measure is also used as the volatility measure for robustness. 11 Second, as an estimate of the dispersion of investor beliefs, 10 Bae et al. (2004) use this measure in their study of returns and volatility, including Korea in their sample. 11 Garman and Klass (1980) investigate the relative efficiency of various measures of volatility and identify a volatility measure with the highest efficiency. In the so-called GK measure, written as ˆσ t = VAR(GK) = 0.5[LN(High) LN(Low)] 2 [2LN(2) 1][LN(Open) LN(Close)] 2, VAR(GK) is the variance using the Garman Klass (1980) method, LN denotes the natural logarithm, and High, Low, Open, Close are the high, low, open, and closing prices of the day to determine the volatility.

6 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) Table 2 Vector autoregressive regression analysis of flows and returns Purchase Sales Net Return Flow Return Flow Return Flow C Return ( 1) *** 7.515*** 3.873*** *** 3.729*** 3.607*** ( 2) *** ** ** ( 3) ** ( 4) ( 5) *** *** 2.075** 2.959*** Flow ( 1) ** *** *** *** ( 2) *** *** ( 3) *** *** ( 4) * ** ( 5) *** *** R This table summarizes the results from a vector autoregressive regression (VAR) of flows and returns for mutual fund flows in Korea. The three flows are Purchases, Sales and Net. Net is calculated by subtracting Sales from Purchases. All volumes are standardized by dividing each flow measure by the 90-day moving average of the KOSPI index market capitalization. The Return is the daily return on the KOSPI index. Data frequency is daily and the sample period is from 1997 to *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively. open interest, standardized by dividing daily open interest of KOSPI 200 futures by the trading volume on KOSPI 200 futures market on the same day, is utilized. 5. Results 5.1. Relationship between fund flows and market index returns Table 2 shows results from the bivariate VAR model. Purchases have a significant impact on the market return but not other flow measures. Since buying stocks has a significant impact on return this finding could be due to price pressure or a result of information being impounded into prices.

7 146 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) The R 2 is low, however, at about 1%, which implies that flows capacity to explain the market return is only marginal. Return exhibits autocorrelations up to a 2-day lag for all three flows, contrary to the expectation that prices should be random under the neoclassical theory. There is significant positive correlation between returns and both Purchases and Sales but a significant negative correlation is observed in the case of Net. This finding suggests that at an aggregate level, negative feedback trading is indicated, which is inconsistent with the U.S. mutual fund findings (Edelen and Warner, 2001), but similar to Japanese institutions (Kim and Nofsinger, 2005). Beyond 1 day, lagged market returns do not show a significant relationship with any of the flow variables. Strong positive autocorrelation is detected for all three flows up to a 5-day lag. This result implies that an increase or decrease in mutual fund flows tends to spur other mutual fund investors to act in the same direction. On the explanatory power of the regressions, Purchases perform best and Net worst which confirms the finding that purchasing rather than selling bears a stronger relationship with market returns. In summary, the results are broadly consistent with the behavioral finance story. Goetzmann and Massa (2002) interpret contrarian (negative feedback) and momentum (positive feedback) investing as signs of the influence of behavioral factors in investors trading practices Direction of the relationship As a preliminary test, the results of Granger causality tests between flows and returns are reported in Table 3. The hypothesis that return does not Granger-cause flow is rejected for all measures of flows at high levels of statistical significance, suggesting that market returns move flows in Korean equity mutual funds. However, by failing to reject the hypothesis that flow does not Granger-cause return for Purchases but affirming it for Sales or Net, the tests reveal that standardized purchase flows may contain information about returns. Purchases likely exert price pressure. On whether flows Granger-cause returns, institutional buying by Korean mutual funds may have an impact on market return but not on selling activity (F-statistics on Purchases are significant at the 1% level). Detecting flows that seem to affect returns is consistent with the behavioral finance framework that predicts that under certain conditions the actions of investors Table 3 Tests of causality between flows and returns Flows Return does not Granger-cause Flow Flow does not Ganger-cause Return F-Stat. Sig Level F-Stat. Sig Level Five lags Purchases Sales Net Ten lags Purchases Sales Net Granger-causality tests of equity fund flows represented by Purchases, Sales, Net and stock market returns are performed using 5 and 10 lags. Number of observations for 5 lags = 1,649 and 10 lags = 1,644. Net is calculated by subtracting Sales from Purchases. All volumes are standardized by dividing each flow measure by the 90-day moving average of the KOSPI index market capitalization.

8 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) Table 4 Granger causality tests in the presence of instrumental variables Variable Purchases Sales Net Coefficient t-statistic Coefficient t-statistic Coefficient t-statistic Panel A: dependent variable return Constant Ret t *** Ret t *** *** *** Ret t *** *** Flow t ** Flow t Flow t *** Ddiv *** ** Dint Aspr ** Axrate ** R Panel B: dependent variable flow Constant *** Ret t *** *** *** Ret t Ret t ** *** Flow t *** *** *** Flow t *** *** *** Flow t *** *** Ddiv * Dint Aspr Axrate R This table reports the results of Granger-causality tests of equity fund flows and stock market returns performed by incorporating fundamental variables (dividends, interest rate, spread and exchange rate). Ret is market return; Flow is fund flows proxied by Purchases, Sales and Net. Ddiv is dividend yield; Dint is the 90-day commercial paper rate; Aspr is proxied by the difference between the 3-year government bond and 3-year corporate bond rate; Axrate is the change in exchange rate (Won/USD). All the fundamental variables are averages over the preceding 5 days. Ddiv and Dint are differenced to meet stationarity requirements. The null hypothesis for panel A is that equity fund flows do not Granger-cause stock market returns in the presence of market fundamentals. The null hypothesis for panel B is that stock market returns do not Granger-cause equity fund flows in the presence of market fundamentals. The figures are Newey- West heteroskedasticity and autocorrelation adjusted. *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively. can cause asset values to deviate from those that would be suggested by the discounted cash-flow approach. This finding is explored further in the Granger causality tests designed to detect whether flows contain information on stock returns after incorporating market fundamentals. The results reporting the direction of the equity fund flows-stock market returns relationship in the presence of market fundamentals are presented in Table 4. The findings in Panel A of Table 4 reject the hypothesis that equity fund flows do not Granger-cause KOSPI market returns but only for Purchases with a 1-day lag. This result again suggests price pressure from Purchases. In Panel B of Table 4 the hypothesis that returns do not Granger-cause flows in the presence of market fundamentals is more robustly rejected with respect to all the flow measures for market return lagged by 1 day. The flows-stock market returns relationship in the

9 148 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) presence of market fundamentals consistently suggests negative feedback trading by mutual fund investors. If investor sentiment is an important force in the markets and if mutual fund flows are a good measure of that sentiment (Warther, 1995), then mutual fund flows may be expected to have a significant effect on security returns. That the findings in this paper point to institutional investors that appear to be dominated by the effects of market returns may reflect the fact that the mutual fund market in Korea is still at nascent levels of sophistication. This is in keeping with the indicators of market sophistication in general the relationship between the instrumental variables, equity fund flows and the market index. The coefficients on the instrumental variables reported in Panel A generally show low significance levels, which further suggests Korean equity fund managers return chasing behavior when deciding to buy or sell Determinants of flows the role of market uncertainty The final set of tests on the relationship between Korean equity fund flows and stock market returns is based on concurrent uncertainty and information measures designed to shed further light on the results reported above. The results reported in Table 5 Panel A exhibit a significant positive relationship between uncertainty measured as volatility and flows at a disaggregated level (Purchases and Sales) which is consistent with the results of Goetzmann and Massa (2003). However, at a net level (Net) no significant relationship is observed. This implies that when there are volatile periods in the market, the mutual fund managers increase their presence by Table 5 Determinants of flows: the role of market uncertainty Purchases Sales Net Coefficient t-statistic Coefficient t-statistic Coefficient t-statistic Panel A: results without the down market dummy Uncertainty variable volatility Constant *** *** Uncertainty *** *** Ddiv *** *** Axrate Dint Aspr Flow t *** *** *** Flow t *** *** *** Flow t *** *** *** R Uncertainty variable open interest Constant Uncertainty * Ddiv *** *** Axrate Dint Aspr Flow t *** *** *** Flow t *** *** *** Flow t *** *** *** R

10 Table 5 (Continued ) N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) Purchases Sales Net Coefficient t-statistic Coefficient t-statistic Coefficient t-statistic Panel B: results including down market dummy Uncertainty variable volatility Constant *** *** ** Uncertainty *** *** Ddiv ** * Axrate Dint Aspr Flow t *** *** *** Flow t *** *** *** Flow t *** *** *** Downmkt *** *** *** R Uncertainty variable open interest Constant *** Uncertainty ** Ddiv ** * Axrate Dint Aspr Flow t *** *** *** Flow t *** *** *** Flow t *** *** *** Downmkt *** *** R This table presents the regression analysis of the determinants of flows including the uncertainty variable as an independent variable. The functional specification is: F t = α + βunc t + γinfv t + δf t 1 + ε t, where F t represents Purchases, Sales and Net. To standardize flows each flow measure is divided by the 90-day moving average of KOSPI market capitalization. Unc t is one of two uncertainty variables volatility proxied by log of return squared [ln(r 2 )], or dispersion of beliefs, proxied by standardized open interest for KOSPI 200 futures. InfV t is a vector of information variables proxied by dividend yield (Ddiv), change in daily USD/Won exchange rate (Axrate), 90-day commercial paper rate (Dint), the risk premium proxied by the difference in the 3-year corporate and government bond yields (Aspr). All the information variables are averages of the preceding 5 days. Ddiv and Dint are differenced to meet stationarity requirements. Figures are Newey-West heteroskedasticity and autocorrelation adjusted. All observations are daily and the sample period is from 1997 to Panel A of the table reports results without a down market dummy (Downmkt), included in Panel B. Downmkt equals 1 on days when the return on the market index is negative on a close-to-close basis, and 0, otherwise. *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively. raising levels of both buy and sell trades in the market, hoping to reap profit, but apparently without distinguishing between buy or sell signals from the market. The coefficient on uncertainty measured as open interest is not statistically significant for Purchases and Net. However, Sales show a weak significant negative relationship. One interpretation of these results, in line with Goetzmann and Massa (2003), is that Korean mutual funds perceive volatility as an opportunity to profit from the market, whereas the dispersion of beliefs is not perceived as such an opportunity. It is perhaps interesting to consider how these results hold up against an indicator of the direction of the market. We introduce the DOWNMKT dummy variable, taking a value of 1 on a day when the return on the market index is negative on a close-to-close return basis, and 0, otherwise. From Table 5 Panel B it is apparent that the inclusion of the DOWNMKT dummy

11 150 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) variable in the regressions results in volatility being confirmed as positively related to concurrent flows. This result could be interpreted as meaning that after controlling for down markets, the behavior of Korean equity fund managers Purchases and Sales exposes them as institutions that tend to associate volatility with an opportunity to profit, a trait associated with a behavioral finance interpretation. This finding is consistent with U.S. index mutual fund investors that appear to defy the mean-variance framework that predicts investors should retract from the markets as volatility increases (see Goetzmann and Massa, 2003). After controlling for declining markets, the significant weak negative relationship between uncertainty and Sales disappears. This finding suggests that investors are not affected by times of wide dispersion of beliefs which is contradictory to findings by Goetzmann and Massa (2003) whereby U.S. mutual fund investors withdraw from the market in that situation. Together, these results suggest that investors react differently to volatility and the dispersion of beliefs. Finally, also of note is that, after controlling for declining markets, the role of the information variable is not significant. This result raises the possibility that equity fund managers in Korea tend to ignore market fundamentals such as the risk premium in pursuit of high returns during rising markets and is consistent with an increase in stock purchases with rising volatility. 6. Summary and conclusions In this paper we study the relation between fund flows and stock market activity in Korea. Based on data on portfolio flows from fund managers into the stock market, a positive relationship is found to exist between stock market returns and mutual fund flows measured as stock purchases and sales. However, in terms of net trading flows, mutual fund investors are negative feedback traders. It is predominantly returns that drive flows. Stock purchases Granger-cause returns, raising the specter of price pressure. Flows do not respond to uncertainty measures, although, after controlling for declining markets, Korean equity fund managers tend to increase their presence indiscriminately in times of rising market volatility. While this paper examines aspects of Korean mutual fund managers behavior in aggregate, a potentially fruitful research agenda remains at the individual fund level. Ongoing developments, for example, the recent entry of international fund research agencies and professional managers, point to a growing profile for the Korean mutual fund industry, and with it, outstanding issues for investigation. Acknowledgments The authors thank the Securities Industry Research Centre of Asia-Pacific for financial support and participants at the Australasian Finance and Banking Conference 2003 and an anonymous referee for helpful comments. References Bae, K., Chan, K., Ng, A., Investibility and return volatility. Journal of Financial Economics 74, Brown, S.J., Goetzmann, W.N., Hiraki, T., Otsuki, T., Shiraishi, N., The Japanese open-end fund puzzle. Journal of Business 74, Cai, J., Chan, K., Yamada, T., The performance of Japanese mutual funds. Review of Financial Studies 10, Cha, H., Lee, B., The market demand curve for common stocks: evidence from equity mutual fund flows. Journal of Financial and Quantitative Analysis 36,

12 N.Y. Oh, J.T. Parwada / Int. Fin. Markets, Inst. and Money 17 (2007) DeLong, J., Shleifer, A., Summers, L., Waldman, R., Noise trader risk in financial markets. Journal of Political Economy 98, Edelen, R., Warner, J., Aggregate price effects of institutional trading: a study of mutual fund flow and market returns. Journal of Financial Economics 59, Edwards, F.R., Zhang, X., Mutual funds and stock and bond market stability. Journal of Financial Services Research 13, Fant, L.F., Investment behavior of mutual fund shareholders: the evidence from aggregate fund flows. Journal of Financial Markets 2, Froot, K., O Connell, P., Seasholes, M., The portfolio flows of international investors. Journal of Financial Economics 59, Garman, M., Klass, M., On the estimation of security price volatilities from historical data. Journal of Business 53, Goetzmann, W., Massa, M., Daily momentum and contrarian behavior of index fund managers. Journal of Financial and Quantitative Analysis 37, Goetzmann, W., Massa, M., Index funds and stock market growth. Journal of Business 76, Granger, C.W., Investigating causal relations by econometric methods and cross spectral methods. Econometrica 37, Grinblatt, M., Titman, S., Wermers, R., Momentum investment strategies, portfolio performance, and herding: a study of mutual fund behavior. American Economic Review 85, Harris, M., Raviv, A., Differences of opinion make a horse race. Review of Financial Studies 6, Kim, K., Nofsinger, J., Institutional herding, business groups, and economic regimes: evidence from Japan. Journal of Business 78, Khorana, A., Servaes, H., Tufano, P., Explaining the size of the mutual fund industry around the world. Journal of Financial Economics 78, Seasholes, M., Smart Foreign Traders in Emerging Markets. Harvard Business School, Mimeo. Shalen, C.T., Volume, volatility, and the dispersion of beliefs. Review of Financial Studies 6, Warther, V., Aggregate mutual fund flows and security returns. Journal of Financial Economics 39, Zheng, L., Is money smart? A study of mutual fund investors fund selection ability. Journal of Finance 54,