as past returns, expense ratios, and the fund familys contact information. For Money, we focus on the composition of the Money 100 list, a list of recommended mutual funds published once a year between 1998 and 2002. Similarly, for Consumer Reports we focus on articles listing recommended equity funds, bond funds, or both. For each of these publications, we classify the mutual fund mentions as positive Since Kiplinger's Personal Finance and SmartMoney do not publish lists of recommended funds anal gous to the Money 100 list, for these two publications we analyze every article containing the word"fund. For articles that focus on mutual funds(rather than mention them in passing), we classify the article as making recommendations across investment objectives, within a particular investment objective, or within a particular mutual fund family. We also classify each mutual fund mention as positive or negative. As uggested by the representative article titles reported in Table I, this rarely involved close judgment calls When we could not determine whether a mention was positive or negative, we dropped the mention from our sample: we did this for 8 of the 783 mentions in Kiplinger's and 16 of the 2, 417 mentions in SmartMoney Finally, for the Wall Street Journal, we focus on a daily column titled"Fund Track, "that mentions funds either because they are the subject of news(such as manager turnover) or because their managers are being quoted on an issue. In the (pre-scandal) time period we study, being quoted on an issue in the Wall Street Journal is presumably positive exposure for the managers fund. However, since "Fund Track" rarely rec- ommends a course of action for fund investors, mentions in the Wall Street Journal are potentially different from mentions in the other publications we study. At a minimum, mentions in the Wall Street Journal can be viewed as proxies for the news coverage of specific funds Data on U.S. mutual fund returns and characteristics come from CRSP. The unit of observation is fund i in month t and the sample period is January 1996 through December 2002. Table II reports summary statistics for the full sample of mutual funds and for funds receiving mentions from a given publication in month t. Univariate comparisons indicate that funds receiving positive media mentions tend to be larger, ad bond funds, but excludes money market funds because they are rarely mentioned in the publications we study. For mutual inflows, and other characteristics, and include one observatie per fund per month in our sample. Also, since we merge the media mention data with the CRSP Survivor-Bias Free US Mutual Fund Database using ticker, our sample is limited to funds for which w e able to locate a ticker for at least one share class Because magazines are typically available on the newsstand in the month prior to the month stated on their cover, we code mentions in the month t+ I issue of Money, Kiplinger's, Smart Money, and Consumer Reports as occurring in month t. For mple, we code mentions in the June 1998 issue of Money as occurring in May 1998. In contrast, mentions in a June 1998 issue of the Wall Street Journal or New York Times are coded as occurring in June 1998as past returns, expense ratios, and the fund family’s contact information. For Money, we focus on the composition of the Money 100 list, a list of recommended mutual funds published once a year between 1998 and 2002. Similarly, for Consumer Reports we focus on articles listing recommended equity funds, bond funds, or both. For each of these publications, we classify the mutual fund mentions as positive. Since Kiplinger’s Personal Finance and SmartMoney do not publish lists of recommended funds analogous to the Money 100 list, for these two publications we analyze every article containing the word “fund.” For articles that focus on mutual funds (rather than mention them in passing), we classify the article as making recommendations across investment objectives, within a particular investment objective, or within a particular mutual fund family. We also classify each mutual fund mention as positive or negative. As suggested by the representative article titles reported in Table I, this rarely involved close judgment calls. When we could not determine whether a mention was positive or negative, we dropped the mention from our sample; we did this for 8 of the 783 mentions in Kiplinger’s and 16 of the 2,417 mentions in SmartMoney. Finally, for the Wall Street Journal, we focus on a daily column titled “Fund Track,” that mentions funds either because they are the subject of news (such as manager turnover) or because their managers are being quoted on an issue. In the (pre-scandal) time period we study, being quoted on an issue in the Wall Street Journal is presumably positive exposure for the manager’s fund. However, since “Fund Track” rarely recommends a course of action for fund investors, mentions in the Wall Street Journal are potentially different from mentions in the other publications we study. At a minimum, mentions in the Wall Street Journal can be viewed as proxies for the news coverage of specific funds. Data on U.S. mutual fund returns and characteristics come from CRSP. The unit of observation is fund i in month t and the sample period is January 1996 through December 2002.6 Table II reports summary statistics for the full sample of mutual funds and for funds receiving mentions from a given publication in month t. 7 Univariate comparisons indicate that funds receiving positive media mentions tend to be larger, 6Our sample includes all domestic equity funds, international equity funds, hybrid funds (which invest in debt and equity), and bond funds, but excludes money market funds because they are rarely mentioned in the publications we study. For mutual funds with multiple share classes, we calculate fund-level returns, inflows, and other characteristics, and include one observation per fund per month in our sample. Also, since we merge the media mention data with the CRSP Survivor-Bias Free US Mutual Fund Database using ticker, our sample is limited to funds for which we were able to locate a ticker for at least one share class. 7Because magazines are typically available on the newsstand in the month prior to the month stated on their cover, we code mentions in the month t + 1 issue of Money, Kiplinger’s, SmartMoney, and Consumer Reports as occurring in month t. For example, we code mentions in the June 1998 issue of Money as occurring in May 1998. In contrast, mentions in a June 1998 issue of the Wall Street Journal or New York Times are coded as occurring in June 1998. 4