and selling(or avoiding) funds that receive negative mentions As such, we contribute to a large literature that seeks to predict future mutual fund performance using lagged fund and fund manager characteristics, and measures of past fund performance. Beg ginning with Jensen 1968, this literature generally finds it difficult to identify funds that will persistently outperform others. That said, recent academic research provides some evidence of predictability, suggesting that full-time industry experts should be able to identify better than average future performers. For example funds with the highest expenses persistently underperform funds with low expenses Elton, Gruber, Das Hlavka, 1993. Chevalier and Ellison [(1999 find that some fund manager characteristics are associated with consistently higher risk-adjusted returns. Carhart [1997 finds some persistence of past performance especially of bad past performance, and Kacperczyk, Sialm, and Zheng 2005 find that their ability to predict underperformance increases if they focus on holdings-adjusted returns (i.e, the difference between actual returns and the returns of the fund s initial holdings). Therefore, it is an open question whether the media mentions in our sample help investors to choose funds with above average future returns We begin by calculating the relative return of fund i in month t as its monthly return minus the equal weighted average monthly return of funds with the same investment objective. This measure is positive hen fund i outperforms the average fund in its asset class and negative when it does not. For each type of media mention, we then regress monthly relative returns on a dummy variable that indicates whether fund i received the specified media mention at least once over the past twelve months. Specifically, for the personal finance publications and Consumer Reports, which are typically available shortly before the month listed on their cover, we assume that investors purchase funds on the first trading day of the month the issue is dated (for example, June 1, 1998 in the case of the first Money 100 list )and sell them on the last trading day of the month twelve months later(May 31, 1999 ). For mentions in the Wall Street Journal and New York Times, which occur throughout the month, we assume that investors purchase funds on the first trading day of the month after the mention and hold them for twelve months In Table Vi, we report estimated coefficients for the media mention holding period dummy variables l6For example, Jain and Wu [2000] find little evidence that funds which advertise in Barrons or Money earn higher future returns than non-advertisers, and Blake and Morey [2000] find little evidence that Morningstar ratings help predict future fundand selling (or avoiding) funds that receive negative mentions. As such, we contribute to a large literature that seeks to predict future mutual fund performance using lagged fund and fund manager characteristics, and measures of past fund performance. Beginning with Jensen [1968], this literature generally finds it difficult to identify funds that will persistently outperform others.16 That said, recent academic research provides some evidence of predictability, suggesting that full-time industry experts should be able to identify better than average future performers. For example, funds with the highest expenses persistently underperform funds with low expenses [Elton, Gruber, Das, Hlavka, 1993]. Chevalier and Ellison [1999] find that some fund manager characteristics are associated with consistently higher risk-adjusted returns. Carhart [1997] finds some persistence of past performance, especially of bad past performance, and Kacperczyk, Sialm, and Zheng [2005] find that their ability to predict underperformance increases if they focus on holdings-adjusted returns (i.e., the difference between actual returns and the returns of the fund’s initial holdings). Therefore, it is an open question whether the media mentions in our sample help investors to choose funds with above average future returns. We begin by calculating the relative return of fund i in month t as its monthly return minus the equal￾weighted average monthly return of funds with the same investment objective. This measure is positive when fund i outperforms the average fund in its asset class and negative when it does not. For each type of media mention, we then regress monthly relative returns on a dummy variable that indicates whether fund i received the specified media mention at least once over the past twelve months. Specifically, for the personal finance publications and Consumer Reports, which are typically available shortly before the month listed on their cover, we assume that investors purchase funds on the first trading day of the month the issue is dated (for example, June 1, 1998 in the case of the first Money 100 list) and sell them on the last trading day of the month twelve months later (May 31, 1999). For mentions in the Wall Street Journal and New York Times, which occur throughout the month, we assume that investors purchase funds on the first trading day of the month after the mention and hold them for twelve months. In Table VI, we report estimated coefficients for the media mention holding period dummy variables 16For example, Jain and Wu [2000] find little evidence that funds which advertise in Barron’s or Money earn higher future returns than non-advertisers, and Blake and Morey [2000] find little evidence that Morningstar ratings help predict future fund performance. 14