Mutual Fund Portfolio Contagion
Fund managers' holdings of a particular stock as a percentage of their total portfolios increase by roughly 0.2 percent when fund managers at different firms in the same city increase their holdings in the same stock by a single percent.
Do fund managers at different mutual funds influence each other in terms of what stocks to buy? Yes, according to Harrison Hong, Jeffrey Kubik, and Jeremy Stein, writing in Thy Neighbor's Portfolio: Word-of-Mouth Effects in the Holdings and Trades of Money Managers (NBER Working Paper No. 9711). The researchers show that - controlling for what fund managers in other cities are doing - fund managers are more likely to hold or trade stocks if their neighbors do so. They find that fund managers' holdings of a particular stock as a percentage of their total portfolios increase by roughly 0.2 percent when fund managers at different firms in the same city increase their holdings in the same stock by a single percent.
This is the first large-scale study of whether the common wisdom, that investors who communicate regularly influence one another's thinking, is indeed true. The data on mutual fund holdings come from CDA Spectrum. The researchers augment the data by including the name of the city where the fund manager is located and the city in which each stock in question is located. Their sample contains 1,715 mutual funds, located in 15 cities, and 2,000 stocks owned in those funds. New York is home to 462 funds, the most of any city, but these account for only 16 percent of total funds under management. Boston is home to 340 funds, but accounts for 39 percent of total funds under management. The next largest fund cities are Philadelphia and Los Angeles.
The researchers compare the observed relationship to an epidemic, with the fund managers spreading the information by word-of-mouth when they come into contact with others. One interpretation of the results is that stocks tend to under-react to information when it first becomes public, leading to momentum effects as more and more investors realize its importance. Physical distance - between cities - may be why it takes time for information to disperse across a company's entire investor base.
Hong, Kubik, and Stein purposely control for the distance between the fund managers and the headquarters of the company whose stock is in question. After all, fund managers are better informed about local companies, and therefore more likely to act ahead of the crowd or to have a "home-bias." The authors also control for investment styles, showing that the relationship they find does not simply reflect the fact that fund managers in a particular city might be more likely to invest in growth, or value, or small cap, or technology stocks.
Of course, the researchers cannot reject the possibility that fund managers in the same city read the same newspapers, or watch the same television shows, thus deriving some investment ideas from the same sources. Another possibility is that companies feed fund managers information; for example, a company executive may visit a series of fund managers in the New York area in local conferences or one-on-one meetings, and convey the same information to investors in the same town. However, small companies are less likely to feature in the media, or to have conducted extensive investor relations or put on road shows, yet the researchers show that fund managers in the same city are more likely to hold or trade very small company stocks in tandem. Moreover, both the "town crier" and the "local investor relations" effects are instances of investors getting information from a local source. They are both consistent with the idea that information about stocks diffuses gradually across the investing population, and that this "epidemic effect" leads to momentum effects in stock ownership and trading.
-- Andrew Balls