Abstract Since the open of domestic financial market recent years, the financial products have become more and more various. In six years, there has been 2.7 funds issued every month on average. For investors, how to choose a good fund company and an appropriate fund from plenty of new offered funds is uneasy. Particularly new offered fund has no history data to examine the performance, and also lacks the regular benchmarks like £]and Sharp ratio. This thesis uses the data of the domestic stock funds which were offered over 1 year from 1997 January 1 to 2006 December 31.The data resources is from TEJ. The study will distinguish all data from whether the prospectuses regulate the funds¡¦ stock holding percentage over 70 ¢H or not. Then we want to investigate the behavior in the first year and to discuss if the fund managers¡¦ experiment will result in the abnormal performance. The study results shows that the stock holding percentage of the domestic stock funds have been over 70¢H on average since offered 3 months, but the percentage of the funds with free investment period of 3 months is less than 60¢H. However the percentage of the funds with free investment period of 6 months would rise to be 70¢H early in the second or third month. We try to use a statistic way and set a dummy variable to test whether there is different between the performance of the funds with free period and those with limited period. The result shows that there is no fund performance difference between these two period. From the second month after the fund is offered, turnover rare is close or even over the average of one year. The average level is about 30¢H and it shows the domestic stock funds have changed stocks frequently since it is offered . The tenure of Fund managers¡¦ after domestic stock funds raised is 20 months on average from 95 months to 1 month.13¢Mof fund managers will be taken place less than half year after the fund offered, 39¢Mwill be less than 1 year. 60¢Mof the fund managers have ever managed other funds before the newly one which he manage now, but 40¢Mhave no such experiment. The statistics shows that there is no significant different between fund managers¡¦ experiment and fund¡¦s abnormal returns.

本文利用投信公司公佈之基金持股比例及持股內容明細，分析基金經理人的操作行為以及和股市股價指數的互動關係。研究主題有二： 　　一、群集效果的研究，從持股內容的分析來研究「基金經理人在操作基金時是否有群集行為」。 　　二、研究共同基金對國內股市的影響，探討「共同基金平均持股比例和股價指數間的因果關係及相關性」。 　　實證結果如下： 　　一、國內的基金經理人在操作時有明顯的群集行為（效果）。 　　二、台灣證券交易所股價指數和基金平均持股比例兩者間，不論是當期對當期，或是股價指數落後一週至十週的檢定，皆為顯著的正相關。 　　三、若以八個分類股價指數及對應之基金平均持股來看，塑化、機電、紡織、造紙、營建、金融股的11個檢定值皆為顯著正相關。水泥股及食品股的11個值則皆為顯著負相關。