Asiasn review of Financial research

Past Issues

HOMEPast Issues Past Issues

Asian Review of Financial Research Vol.24 No.2 pp.411-453
Fund Managers' Skill and Herding in the Korean Active Fund Market
Hyuk Choe* Professor, College of Business Administration, Seoul National University
Jee-Hyun Kim Ph.D. Candidate, College of Business Administration, Seoul National University
Key Words : Active Fund,Managers' Skill,Luck,Bootstrap Simulation,Herding Phenomenon


As fund markets have experienced a tremendous growth internationally, a lot of studies on the performance of funds have been conducted in recent years. At the same time, academic interest to investigate the impact of managerial skills on a fund's performance has also grown. However, due to the short history and the lack of data on the Korean fund market, only a few studies have actually been conducted on the performance of the Korean active fund market. In this study, therefore, we analyze whether and to what extent the performance of Korean funds is influenced by the skill of managers or simply by luck. For this investigation, the bootstrap simulation method is applied. Previous studies traditionally use the persistence test to find out whether the performance of a fund is based on a manager's skill. These studies test whether a past winner continues to show good performance. Since this performance test is based on the short-term performance of funds, we cannot rule out the possibility of the intervention of noise. To solve this problem, recent studies, such as Kosowski, Timmermann, Wermers, and White (2006) and Fama and French (2010) use the bootstrap simulation. Thus, we have decided to apply this methodology to the active fund market in Korea. A simulation run produces a random sample within the period of 95 months, from February 2001 to December 2008. We examine 3,000 simulation runs. Using this method, we derive a distribution of excess returns of funds compared to a return of a benchmark portfolio under the assumption that managers are unskilled. Comparing the distribution from the simulation and the real distribution of excess returns of funds, we are able to estimate if real excess returns are extreme cases under the simulation distribution. In addition, we divide our sample into two groups : one group of funds in the early 2000s and the other group of funds in the late 2000s. We also subdivided the sample into two groups as long- and short-term groups funds that have existed for 5 or more years and those that have existed for less than 1 year, respectively. Fund groups are also sorted out according to the size. In fact, the sample funds are ranked based on their size: if a fund's size is below the median value, the fund is included in the small-size group while the rest is included in the large-size group. While most previous studies have focused on the fund performance itself or the characteristics affecting the performance, our study aims to investigate the impact of fund managers' skills on the fund performance in the Korean active fund market. As far as we are aware, our study is the first study of its kind to ever show also the effect of the herding phenomenon among managers on the possibility of the linkage of fund performance and the skill of managers. Furthermore, we use the methodology that can reduce biases such as survivorship bias and incubation bias, which could appear in the examination of funds. Our major findings are as follows. First, real excess returns exist in a narrow range as compared to the simulation distribution of excess returns, which are derived under the assumption of unskilled managers. This finding suggests that Korean fund managers possess even levels of fund managing skills, neither too superior nor too inferior compared to one another. Second, fund performance is more highly influenced by managers' skill among the funds that belong to the late 2000s group than those of the early 2000s group. This finding suggests that the fund managers' skill has improved over the years. Third, the performance of funds is more likely to be affected by managers' skill in the long-term fund group than in the short-term fund group. This finding implies that fund managers'skill is more likely to affect the performance of funds that are running on a long-term basis. Fourth, the performance of funds based on managers' skill seems similar between a small-size group and a large-size group. At this point, it is important to observe a notable feature of Korea's fund market that is an extremely narrow range of the distribution of real excess returns especially compared to that of the corresponding U.S. market. To identify some potential causes of this difference, we investigate whether the herding phenomenon can be found among fund managers. If managers copy one another in making investment decisions regardless of their access to private information, indeed, the performance among funds is highly likely to be similar. For this investigation, we apply the LSV herding measure developed by Lakonishok, Shleifer, and Vishny (1992). Using this measure, we find that managers in the Korean active fund market herd one another more strongly than in the U.S fund market. We also find that when mangers herd severely, then the effect of managers' skill on the performance of funds is expectedly very limited. This finding suggests that the stronger the herding phenomenon is, the less relevant the skill of fund managers is in the fund performance.
Export citation