In a recent paper (Coval, Joshua D., and Tobias J. Moskowitz, 2001, ``The Geography of Investment: Informed Trading and Asset Prices,'' Journal of Political Economy, forthcoming), we identify a strong geographic link between mutual fund investment and performance. Analyzing the equity holdings of a large sample of actively managed mutual funds, we find that fund managers appear to earn substantial abnormal returns in their geographically proximate investments. The average fund manager generates an additional return of 2.67% per year from his local investments (defined as holdings within 100 kilometers of the fund headquarters), relative to his non-local holdings. Adjusting for risk, his local investments earn 1.84% per year above passive benchmark portfolios and earn 1.18% per annum more than his distant holdings. Furthermore, local stocks avoided by the manager underperform local stocks actually held by a risk-adjusted 3% per year. The results are robust across a wide array of fund types, to various local definitions, and to a variety of alternative specifications.

Managers appear to earn abnormal returns in their local holdings as compensation for information they may acquire about local companies. This information may be the result of improved monitoring capabilities or access to private information of geographically proximate firms. Consistent with this view, the gains from local investment are larger among funds better able to exploit local knowledge.

For instance, funds that are more agile, including smaller funds, funds with relatively few holdings and older funds, generate higher local returns and invest more heavily in local stocks. In addition, areas where local information may be most valuable and more difficult for outsiders to obtain--such as small cities and remote locations--offer larger profits for local investors.

Moreover, funds better able to select local stocks concentrate their holdings more locally. Despite the apparent gains from investing locally, the average fund in our sample exhibits only a modest bias toward local stocks. Certain managers, on the other hand, exhibit a strong local bias, tilting their portfolios by as much as 20% to 25% toward local stocks. These funds also generate the largest gains from local investment, outperforming their distant holdings by as much as 3% per year and outpacing local stocks not held by 4% to 5% per year, on a risk-adjusted basis. Conversely, funds that do not exhibit a local bias generate no abnormal performance in their local holdings. These results suggest an informational link between geography and investment.

INFORMED TRADING
Since investors located near potential investments exhibit significant stock selection ability, geographic proximity may be useful in identifying informed trading. With this in mind, we shift the focus of our analysis from the perspective of the fund manager to that of the stock, examining the relation between local investor ownership and asset prices.

The risk-adjusted premium for locally held stocks over firms held by the market is 1.1% per year, but increases to over 3% per annum among the smallest firms and among firms with high levels of mutual fund ownership. These findings suggest that geographic location, informed trading and asset prices are closely related.

Finally, the local abnormal returns we document do not disappear immediately, but persist for several months after portfolio formation. Therefore, if uninformed or distant investors could obtain timely local portfolio positions of funds, they may be able to replicate some of the local gains achieved by our managers.

Hence, our results, which at least appear inconsistent with the strongest notion of market efficiency, may also violate a semi-strong efficient market if portfolio positions become publicly accessible within several months of Securities Exchange Commission filings (upon which our holdings data are based). At the same time, once investors are aware of this possibility, the gains from local investing might disappear or become shorter-lived.