本研究透過基金流量、群集行為、與報酬間的關係,驗證基金群集行為的原因。首先,針對前期流量與基金群集指標的關係,本研究推論:如果經理人行為符合聲譽風險假說(Scharfstein and Stein, 1990),則前期淨流出基金的群集指標會高於前期淨流入基金的群集指標;反之,如果經理人行為符合放手一博假說,則淨流出基金的群集指標低於淨流入基金的群集指標,實證結果符合聲譽風險假說。其次,針對基金群集指標與報酬的關係,本研究推論:如果基金群集行為是基於分析與報酬攸關的私有資訊,則群集指標與期後續報酬正相關;反之,如果是基於揣摩其他基金所擁有的資訊,但是該資訊與報酬無關(Froot, Scharfstein, and Stein, 1992),則群集指標與報酬負相關。整體的結果較為支持基金的群集行為是基於揣測他人已知的資訊。第三,針對基金是否從先前擁有較多資訊的領先者的身上而推得資訊,進而形成群集的行為(Bikhchandani, Hirshleifer, and Welch, 1992),本研究推論:基金因追逐前期優勝者的持股,或是規避前期失敗者的持股,因而有較高的群集指標,實證證明以過去一個月報酬分群的結果符合上述預期。最後,本研究探討基金是否因偏好或規避某種特徵的類股,因而造成群集(Falkenstein, 1996)。結果發現基金因偏好交易大規模、高成長、高報酬與高股價的股票,而造成這些特徵股票有較高的群集指標。
The purpose of this study is to investigate the possible explanations for the herding behavior of mutual fund by connecting fund flows, herding measures, and returns. Firstly, we hypothesize that the stocks of funds with net outflow will have a higher herding measure when the fund manager is concerned with his reputation risk (Scharfstein and Stein, 1990), and have a lower herding measure when he engages in gaming the incentive scheme derived from the asymmetric performance-flow relation. The results support the reputation-risk hypothesis. Second, regarding the herding measure and return, we hypothesize positive herding-return relation if fund managers herd on the same private information that is related to fundamentals, and a negative herd-return relation if they herd on information that is completely unrelated to fundamentals (Froot, Scharfstein, and Stein, 1992). In general, the empirical results support the latter argument. Moreover, we also investigate whether managers infer private information from the prior trades of better-informed managers (Bikhchandani, Hirshleifer, and Welch, 1992). We hypothesize that managers might follow previous winners or avoid previous losers and both would result in higher herding measures. Empirical results support the aforementioned argument. Finally, in the investigation of manager's preference for certain stock characteristics (Falkenstein, 1996) we find that fund managers prefer to trade on large-cap, growth, high-return, and high-price stocks, which leads to higher herding measures on these stocks.