Due to the regulatory constraints placed on hedge funds’ ability to market themselves to the public, the media’s interpretation of events takes on a significant importance. While the mass media often takes a dreary view of hedge funds in general, what about the media’s bias in reporting developments surrounding specific funds? Are they overly negative or too lenient? And if they do show a bias, can you use that information to make money?
These are some of the questions addressed in a study by Gideon Ozik of Edhec and Ronnie Sadka of Boston College. The duo uses Google News to collect nearly 70,000 articles containing references to about 750 long/short equity funds in the Lipper-Tass hedge fund database (categorized into: mass media like the WSJ, specialized media like Pensions & Investments, and “corporate media” such as press releases appearing mainly on sites like PR Newswire) . Then they analyze the words contained in these articles looking for positive and negative words (as defined in the widely-used Harvard psychosocial dictionary).
Many of the stories are carried by only one of the three sources (mass media, specialists, or press release). You cam imagine circumstances when the mass media reported on a fund, but that fund did not issue a press release. You can also imagine the opposite occurring. And specialist media might report a story that the mass media finds too specific.
Things have been changing, however. As it turns out, over the past decade the number of stories that only made the pages of specialist media has decreased (as the mass media took an interest in reporting on hedge funds), and the number of press releases effectively ignored by all other media has increased between 2001 ad 2003 (perhaps as hedge funds began to issue useless press releases – no names now).
Ozik and Sadka develop a metric to measure the sentiment contained in these stories (the ratio of positive words to all positive and negative words). As you might guess, the corporate communications tended to be more positive and the mass media sentiment was more “balanced”. The chart below contains data from Table 3 in the study and shows the sentiment scores of stories that were reported exclusively by each category of media (stories that were reported by more than one type of media – non-exclusively – showed basically the same pattern).
Okay. So far, no big whoop, right? But can you trade on the information contained in these biases? Apparently, yes, you can.
In general, funds covered exclusively by their own marketing department (i.e. ones with no outside media coverage) did better than funds covered exclusively by outside media. This might occur, for example, if the fund took a bath on one of their positions. Not surprisingly, there aren’t a lot of press releases announcing “We just lost our shirt on XYX Co”. The general media loves a train wreck though, and they would likely be chomping at the bit to report how the rich hedge fund investors will be crying themselves to sleep that night. The authors call this the “editorial bias“.
Looking backward, the authors created portfolios of funds each month: “undefined” sentiment funds, “positive” sentiment funds (whose media stories had an above average sentiment the previous month) and “negative” sentiment funds (whose stories had a below average sentiment the previous month). They then adjusted returns for Fung & Hsieh factors plus a beta for the sentiment of their (positive or negative) sentiment group for that month. The result is a “sentiment adjusted” returns for the monthly portfolios. As you can see from the chart below, those returns were largely positive over the ensuing 18 months – especially for funds whose media stories were exclusively delivered by their own PR departments.
But if corporate PR material is usually quite positive (see chart above), then how do funds that issue lots of press releases still manage to outperform those covered only by the general media? As the authors point out, this would imply that those press releases contain relatively pessimistic views – not really standard fare from the marketing department you’d guess. The answer to this paradox may be that despite the funds’ propensity to describe their situation in the best possible light (or to only issue press releases when they’re certain they are about to hit a home run), they are still a little gun-shy about over-promising. The authors call this the “content bias” and say:
“…to the extent that Corporate coverage reflects the incentives of fund managers, it is quite reasonable to expect them to express moderate, conservative views about their performance (for example, concerns about litigation risk). Therefore, even though managers would choose to disclose favorable information, they might not want to fully reveal it…In sum, a content bias suggests that given the choice to cover a particular news event, Corporate media sources will tend to express pessimistic views, while General will express optimistic views.”
This study provides some tasty food for thought. But we still wonder about the methodology, although 10 points for innovation as traditional Lexis Nexis searches apparently don’t return press releases. For example, maybe the fund managers referenced in the 70,000 Google News stories were not talking about the prospects for their own funds, but were instead delivering their opinion on the economy or some particular short position they had (“This stock a bad, terrible, the worst! That’s why our fund is going to make money shorting it…”). Indeed why would any fund use “negative” language to describe itself in its own corporate communications? The use of such negative language in press releases may not always (if ever) have to do with the fund itself, but may in fact describe the markets.
Nonetheless, as the authors point out, this study may be of interest to those who wonder how the media affects demand for stocks. It has always been very difficult to disentangle the effect of negative media coverage on a stock from the negative business development that precipitated that very coverage. But since the value per unit of investment funds does not reflect anything about the fund itself, the effect of changes to business fundamentals is mitigated.
The bottom line seems to be that corporate communication may be totally biased to the upside , but maybe not biased enough to the upside in order to fully capture the good news brewing deep within a fund. And investors , it appears, may be oblivious to the information contained in media word choices. As Ozik and Sadka conclude:
“…investor fund flow does not react to this information, which suggests that investors do not seem to exploit valuable information embedded in media coverage.”