The home bias in international investment portfolios is well documented and is often referred to as the “home bias puzzle.” Coval and Moskowitz contribute to solving this puzzle by investigating the bias among U.S. managers of domestic portfolios for investing in companies that are located close to the managers' own locations.

Coval and Moskowitz's sample consists of U.S. investment managers from Nelson's 1996 Directory of Investment Managers. To limit the research sample to managers who invested primarily in U.S. equities, they consider only those managers who had at least 5 of their top 10 holdings in U.S. companies. Companies located in Alaska, Hawaii, or Puerto Rico are excluded so that outliers do not unduly influence the results. Coval and Moskowitz also exclude all index funds and ignore investments made by one manager in another manager's fund. The final research sample consists of 1,189 investment managers running 2,183 different funds. To get the 1995 investment holdings data and location of each manager, Coval and Moskowitz use the Nelson data set. The investment managers had holdings in 2,736 individual U.S. companies, and data on these companies are from the 1995 Compustat tapes and the 1995 Compact Disclosure database.

For each investment manager, Coval and Moskowitz calculate the average distance from his or her own location to the locations of the headquarters of the companies in the manager's investment portfolio. They compare this number with the average distance from the manager's own location to the locations of the headquarters of all the available companies (all the companies in which the managers in the research sample invested). Coval and Moskowitz find that U.S. fund managers invest in companies located, on average, between 160 and 184 kilometers closer to themselves than the average company available to them. This difference represents 9–10 percent of the average distance and is statistically significant. They further find that this preference for local companies is related to company size, leverage, and output tradability. The locally held companies tend to be small and highly levered and to produce goods not traded internationally. These are the companies with variable earnings, for which information locally available would be the most valuable. The results thus suggest an information-based explanation for the local preference and are common across a variety of manager types and fund classes. Furthermore, size and leverage are linked to previously documented price anomalies. The relationship between geographical proximity and these variables may shed light on these anomalies.

Coval and Moskowitz study the bias for geographical proximity in domestic portfolios but also extrapolate their results to international investment portfolios. They find that about one-third of the home bias phenomenon in international portfolios could be explained by the bias for geographical proximity, with the remaining two-thirds associated with a preference for investment in the domestic economy.