I. Introduction Recently, there has been considerable interest in political media bias(Groseclose and Milyo 2004 Baron [2005, and Mullainathan and Shleifer 2005). There is also growing interest in whether the media biases its content to benefit advertisers. For their part, media outlets tend to strongly deny that such a pro-adviser bias exists. For example, a 1996 article in Kiplinger's Personal Finance printed statements from editors at a number of personal finance publications(including the three in our study) claiming that advertisers have no influence over published content. 2 Despite the important role that the media play in generating and disseminating information to consumers and investors, we are aware of few systematic ttempts to test the accuracy of these claims In this paper, we test for advertising bias within the financial media. Specifically, we study mutual fund recommendations published between January 1997 and December 2002 in five of the top six recipients of mutual fund advertising dollars. Controlling for observable fund characteristics and total family advertising expenditures, we document a positive correlation between a family's lagged advertising expenditures and the probability that its funds are recommended in each of the personal finance publications in our sample ( Money Magazine, Kiplinger's Personal Finance, and SmartMoney ). While we consider several alternative explanations below, the robustness of the correlation leads us to conclude that the most plausible explanation is the causal one, namely, that personal finance publications bias their recommendations--either consciously or subconsciously-to favor advertisers. In contrast, we find no such correlation between advertising and content in either national newspaper (the New York Times and Wall Street Journal) For example, Baker [1994] and Hamilton oth argue that the media biases its conte Ellman and Germano [2005] model this bias as from advertisers committing to punish stories. A related bias is posited by Dyck and 2003, who document publications that en the way earnings nnouncements are reported in a press release and the way they are reported in the media and argue that the correlation consistent with reporters biasing articles in exchange for to private information. 2Goldberg, Steven, "Do the ads tempt the editors?(influence of mutual fund advertising on personal finance publication ditors), " Kiplinger's Personal Finance, May 1996. The article was written in response to an earlier article in Fortune accusing Forbes of"turning downbeat stories into upbeat stories in order to keep advertisers happy- even at the risk of misleading their own readers 3One exception is Reuter [2002], who asks whether advertising biases wine ratings. We briefly discuss his findings below nother exception is Miller (2004), who examines a sample of firms that the SEC found guilty of accounting fraud and finds that the media is no less likely to break stories about firms in the 15 indu acknowledges that the use of industry-level advertising data may reduce the power of this test. More generally, our work relates other studies of correlations between expert opinion and business interests. For example, Lin and McNichols [1998 and Michaely and Womack [1999] find that sell-side analysts buy and sell recommendations favor the companies with which their employers do investment banking br Zitzewitz [2005] finds that figure skating judges are nationalistically biased and sell" bias to colleagues by engaging in vote tradingI. Introduction Recently, there has been considerable interest in political media bias (Groseclose and Milyo [2004], Baron [2005], and Mullainathan and Shleifer [2005]). There is also growing interest in whether the media biases its content to benefit advertisers.1 For their part, media outlets tend to strongly deny that such a pro-adviser bias exists. For example, a 1996 article in Kiplinger’s Personal Finance printed statements from editors at a number of personal finance publications (including the three in our study) claiming that advertisers have no influence over published content.2 Despite the important role that the media plays in generating and disseminating information to consumers and investors, we are aware of few systematic attempts to test the accuracy of these claims.3 In this paper, we test for advertising bias within the financial media. Specifically, we study mutual fund recommendations published between January 1997 and December 2002 in five of the top six recipients of mutual fund advertising dollars. Controlling for observable fund characteristics and total family advertising expenditures, we document a positive correlation between a family’s lagged advertising expenditures and the probability that its funds are recommended in each of the personal finance publications in our sample (Money Magazine, Kiplinger’s Personal Finance, and SmartMoney). While we consider several alternative explanations below, the robustness of the correlation leads us to conclude that the most plausible explanation is the causal one, namely, that personal finance publications bias their recommendations—either consciously or subconsciously—to favor advertisers. In contrast, we find no such correlation between advertising and content in either national newspaper (the New York Times and Wall Street Journal). 1For example, Baker [1994] and Hamilton [2004] both argue that the media biases its content to benefit advertisers, and Ellman and Germano [2005] model this bias as arising from advertisers committing to punish publications that run negative stories. A related bias is posited by Dyck and Zingales [2003], who document a positive correlation between the way earnings announcements are reported in a press release and the way they are reported in the media and argue that the correlation is consistent with reporters biasing articles in exchange for access to private information. 2Goldberg, Steven, “Do the ads tempt the editors? (influence of mutual fund advertising on personal finance publication editors),” Kiplinger’s Personal Finance, May 1996. The article was written in response to an earlier article in Fortune accusing Forbes of “turning downbeat stories into upbeat stories in order to keep advertisers happy—even at the risk of misleading their own readers.” 3One exception is Reuter [2002], who asks whether advertising biases wine ratings. We briefly discuss his findings below. Another exception is Miller [2004], who examines a sample of firms that the SEC found guilty of accounting fraud and finds that the media is no less likely to break stories about firms in the 15 industries with the highest propensity to advertise, but acknowledges that the use of industry-level advertising data may reduce the power of this test. More generally, our work relates to other studies of correlations between expert opinion and business interests. For example, Lin and McNichols [1998] and Michaely and Womack [1999] find that sell-side analysts’ buy and sell recommendations favor the companies with which their employers do investment banking business. Zitzewitz [2005] finds that figure skating judges are nationalistically biased and “sell” bias to colleagues by engaging in vote trading. 1