earticle

영어

Since the introduction of the Act of Financial Investment Services and Capital Markets in 2009, the fund markets in Korea have enormously grown. In this study, we examine three salient composites: fund style, performance persistency, and market timing ability of fund managers. To this end we use all equity funds survived during the period of 2001 through 2009 (all 139 funds). In addition, we look into the relation between the persistency of fund style and the fund performance. We also study how the fund performance has been changed since the financial crisis in 2007. To classify fund styles and examine fund performance, we use a return-based style analysis as in Sharpe (1992). The return-based style analysis sort funds according to the magnitude of the coefficient estimates. Unlike Sharpe who uses simply a regression model comprised of several indexes, we use the Fama-French (1993) five-factor model composed of three equity factors and two bond market factors. The equity factors are the market factor and two factors related to size and book-to-market, namely SMB and HML, respectively; the bond market factors are term spread and default spread. The spread in yields between 5-year Korean Treasury Bonds (KTB) and 91-day Certificate of Deposit (CD) is used as term spread, and the spread in yields between BBB3-rated and AA3-rated corporate bonds is used as default spread. The reason we select the Fama-French 3 factor model to analyze equity funds in Korea is that, as recently reported by Kim, Kim, and Shin (2012), the Fama-French 3 factor model explains Korean stock returns best among many factor models considered. Our paper is the first paper that uses the bond market factors in analyzing equity mutual funds in Korea. Following Sharpe (1992), we classify fund style according to the estimated coefficients (or factor loadings) on each of the five factors. For example, if the magnitude of the factor loading of a fund on SMB is in the top (bottom) 30%, this fund is classified as small (large) fund. If the magnitude of the factor loading of a fund on HML is in the top (bottom) 30%, this fund is classified as value (growth) fund. If the magnitude of the factor loading of a fund on term spread is in the top (bottom) 30%, this fund is classified as business sensitive (insensitive) fund. If the magnitude of the factor loading of a fund on default spread is in the top (bottom) 30%, this fund is classified as high (low) credit risk fund. Of course, one fund can be classified into multiple funds. If the factor loading of a factor falls between top 30% and bottom 30%, this fund is not classified. The main results of this study are as follows: First, we confirm that the Fama-French 5-factor model shows the best goodness of fit and the lowest tracking errors for Korean equity mutual funds among many factor models considered, such as the Capital Asset Pricing Model (CAPM), the Fama-French (1993) three-factor model, and Carhart (1997) four-factor model. To compare the performance of the factor models, we estimated the models by observing the year-by-year rolling over for the three-year period for its monthly returns. We also find that the two bond factors are important for better understanding the performance of Korean mutual funds. In particular, since fund performance is affected by business cycles, the use of the bond factors would be essential in analyzing fund performance. Further, the bond factors are good proxies for business cycle. In fact, we confirm that mutual funds returns in Korea are significantly positively correlated with the Korea business-leading indicator and are sensitive to business cycles, which are also determined by the same indicator. Another important finding is that large and growth funds, rather than small, value funds, perform better and take more fund inflows in Korea. Also, Korean mutual fund managers tend to build their investment strategies toward value, large, business cycle-and creditsensitive fund styles. Fourth, only fund managers of a fraction of mutual funds show significant selection ability while most fund managers rarely show market timing ability. For some fund managers equipped with market timing ability, their market timing ability tends to be higher during the recent financial crisis (2007~2009) than during the other times. To analyze the market timing ability of fund managers, we use the conditional model on macroeconomic variables by following Ferson and Schadt (1996) to adjust for time-varying of the factor loadings. Fifth, while only a fraction of mutual funds show positive abnormal returns, they are shown to last just briefly. Rather, low performance (i.e., negative abnormal returns) tends to be strongly persistent. These results are somewhat consistent with the performance of U.S. mutual funds. Among the funds showing positive abnormal returns, growth and large funds show stronger persistence of good performance than do value and small funds. This result implies that the persistency of fund style could be related to good fund performance. Finally, there also exists in Korea the smart money effect, indicating that investors have an ability to identify superior fund managers and invest accordingly; this attests to the fact that fund inflows improve fund performance. Our study indicates, however, that this effect lasts only for three years.

한국어

본 연구는 채권시장 요인이 포함된 Fama-French(1993) 5요인 모형을 사용하여 국내 주식형펀드의 스타일분석, 스타일에 따른 성과분석, 펀드매니저의 시장예측능력, 스타일의 지속성과 펀드성과 간의 관계를 분석하였다. 논문의 주요 실증분석 결과는 다음과 같다. 첫째, 2001년부터 2009년까지 생존한 펀드의 수익률을 바탕으로 추정한 결과, Fama-French 5요인 모형은 다른 모형에 비해 높은 적합도와 낮은 추적오차를 나타내고, Fama-French 5요인 모형에서 채권시장요인은 펀드 수익률 설명에 중요한 역할을 하였다. 둘째, 국내 시장에서 대형 및 성장형 펀드의 수익률이 소형 및 가치형에 비해 높고, 평균 자금흐름 규모도 큰 것으로 나타났다. 셋째, 대부분의 국내 펀드는 일반적인 시장 포트폴리오에 비하여, 가치형, 대형, 경기민감형 및 고신용형 전략에 집중하는 공격적 스타일 특성을 지니는 것으로 확인되었다. 넷째, 국내 시장에서 초과수익률이 존재하는 펀드가 일부 존재하였으나, 시장예측능력이 존재하는 펀드는 매우 적었다. 다섯째, 우수한 초과수익률을 보인 성장형 및 대형 펀드의 스타일 지속성은 가치형 및 소형 펀드보다 높은 것을 확인할 수 있었다. 여섯째, 국내 시장에서 자금흐름 증대로 펀드 성과 개선되는 ‘스마트 머니 효과’가 존재하지만, 동 효과의 지속기간은 3년 전후에 국한되는 것으로 분석되었다.