在股市中，機構投資人的交易扮演著舉足輕重的角色，過去的研究發現機構投資人似乎傾向於跟隨其他投資者的交易，其動機則有許多種的考量或因素，本篇研究檢視Sias (2004)的文章，針對其衡量從眾行為的指標，指出其不足之處，並提出更佳的衡量指標，並輔以佐證，研究方法為先計算兩種方法下的從眾指標，再透過迴歸分析方法，觀察兩種從眾指標與後期股票報酬之間的關聯性，亦即探討是否有報酬反轉的現象產生，該篇文章的研究期間為1983年至1997年，本篇則將樣本期間延伸至2010年，以觀察近期的變化，研究的對象為持有美國三大證交所普通股的機構投資人；最後，研究機構投資人產生從眾行為的原因，分析機構投資人交易行為是否與資訊有關，此部分的研究依據Bennett et al. (2003)的方法，選取幾項過去許多文獻認定的股票特性，觀察機構投資人是否產生特性從眾(Characteristic herding)的現象。 This articleexamines the herding behaviors of institutional investors using Sias (2004) approach. Moreover, I find that Sias (2004) may ignore the impact of company events on the herding measures and I bring up a more appropriate method to solve this problem. In this article, I use my method as well as Sias’ method to see if the herding results will be the same using the quarterly data during the 1983-1997 period and 1998-2010 period. The findings indicate that institutional herding exists in both methods. However, to test whether institutional investors’ trades are related to information or not, I examine the return reversals. Evidence shows that Sias’ method is affected by the company events and mine is not, showing that my method would be a better measure of institutional herding. I also investigate the possible reason of institutional herding by examining the existence of characteristic herding, following the methodology of Bennett, Sias, and Starks (2003). By testing the preference of firm characteristics of institutional investors, I further prove that institutional herding result from something other than information.
