Media Coverage and Hedge Fund Returns

The authors classified news items about equity hedge funds over 1999–2008 into three source groups—general newspapers, specialized magazines, and corporate communications—and found that corporate-covered funds outperformed general-covered funds by about 11 percentage points annually. Investor fund flows, however, were not related to information sources, which suggests that the return spread is not fully understood by investors. The magnitude of this return spread reflects the extensive costs of processing information to generate alpha.Our primary research interest was whether media coverage contains information about future fund performance. We ask the following questions: Do different media sources systematically differ in the manner in which they cover hedge funds? Are they overly negative or too lenient? Do media contain information about future fund performance? And to the extent that we found such a relationship, do investors understand this information and act on it? We found supporting evidence that media coverage contains information about fund performance; however, this information does not seem to affect investor fund flows. The results suggest that managers of alternative portfolios should incorporate media-based information in the investment management process to generate abnormal returns.We accessed various media sources and devised a measure of the information they report. We collected news articles on a sample of U.S. long–short equity hedge funds over 1999–2008. The sample includes about 3,600 unique media sources, which we separated into three categories: general, which typically includes daily newspapers, such as the New York Times and the Washington Post; specialized, which includes such industry venues as Pensions & Investments, BusinessWeek, and other investment magazines; and corporate communications, which includes press releases and wire services, such as PR Newswire and Business Wire. We then applied a textual analysis to the title of each news item to gauge a measure of news item sentiment. Essentially, this procedure classifies each word as positive, negative, or undefined, according to the Harvard IV-4 psychosocial dictionary. We then measured the (positive) sentiment of a news item title as the ratio of its positive words to the sum of its positive and negative words.The main result of the article is that corporate-covered funds outperformed and general-covered funds underperformed, with a performance difference of about 11 pps annually. Such a result may be consistent with sentiment-related biases, such as the reporting style bias and editorial selection bias. To control for such biases, we computed sentiment-adjusted returns. We found that the outperformance of corporate-covered funds and the underperformance of general-covered funds remained unchanged when we used sentiment-adjusted returns. In fact, unreported results show that sentiment alone does not predict fund performance. Therefore, sentiment-related biases do not explain the inter-source return spread. We further investigated whether investors respond to the return information embedded in media coverage. Despite the economically significant inter-source return spread documented in this article, we found that investors do not seem to differentially respond to the media coverage of hedge funds by different sources. That is, although hedge fund investor flows generally respond to media coverage (e.g., Ozik and Sadka 2010b), regressing hedge fund flows on the different media sources results in no differential response across the different type of media sources. Therefore, investors do not seem to fully distinguish between the different sources.The fact that media coverage predicts sentiment-adjusted returns suggests that although investors may understand biases stemming from differences in sentiment, they may not fully understand inter-source biases. The power of our tests stems from comparing the media coverage of almost 1,000 funds from thousands of different media sources. In addition, our findings mainly pertain to funds exclusively covered by one of the source groups. Therefore, to gauge whether a fund is, for example, exclusively covered by general media requires one to verify not only that the fund is covered by at least one source classified as general but also that the fund is not covered by any source classified as corporate or specialized. The extensive gathering and processing of such information is fairly costly. Thus, although media coverage is public information, investors may not be able to process it, and therefore, the return spread can be interpreted as a pure alpha, reflecting the costs of obtaining information.