Regardless of size, book-to-market ratio, momentum, and to a lesser extent, illiquidity and volatility, local firms outperform geographically dispersed firms in the United States. The authors find that this variation in average returns is particularly pronounced for smaller firms, less-liquid firms, and firms with high idiosyncratic volatility. Furthermore, local firms headquartered in areas where the competition for investor attention is fierce outperform firms headquartered in areas where fewer firms compete for attention.

In the United States, local firms, which the authors define as operating in two or fewer states, have average monthly returns that are 70 bps higher (after the authors control for risks related to size, book-to-market ratio, momentum, and liquidity) than those of firms that operate in more than 20 states. This difference appears to be priced into firms within a year because the returns of firms that become local (dispersed) within the year are similar to the returns of firms that already were local (dispersed). The effect, however, has notably diminished over time, and although returns to dispersion-based strategies were significantly positive from 1994 to 2004, the anomaly vanishes altogether after 2004. The authors conjecture that when the type of textual analysis used in this study became popular in 2004, arbitrageurs and other informed investors acted quickly to eliminate the mispricing.

The authors extend prior research on the investor recognition hypothesis, which argues that stocks with lower investor recognition offer higher expected returns to compensate investors for insufficient diversification. Other researchers have shown that firms with scant media coverage have higher returns than more visible, but otherwise comparable, firms—a result relating equilibrium risk premiums to the presence of informationally constrained investors. Past researchers have also shown that U.S. money managers are more likely to invest in firms headquartered in their own city than in firms headquartered elsewhere. The authors’ research supports these prior findings.

This research is particularly useful for investors who are focused on uncovering value from firms that do not attract broad attention because of their limited geographic scope. The authors conclude that investors tend to prefer firms with local operations rather than dispersed operations, despite the lack of meaningful excess returns from 2004 to 2008.

The authors gather data from the U.S. SEC’s EDGAR database, CRSP, and Compustat for 1994–2008. The sample consists of publicly traded U.S. firms that include the names of U.S. states in certain sections of their annual reports. The authors classify firms as local if the firms mention only one or two states, a result that places them in the lowest quintile when all firms are ranked according to the number of states mentioned. Firms that mention 20 or more states fall in the highest quintile and are classified as geographically dispersed.

To study the relationship between stock returns and the degree of geographic dispersion, the authors sort the portfolios and perform Fama–MacBeth cross-sectional regressions. The sorted portfolios are evaluated using equal-weighted returns and value-weighted returns; the cross-sectional regressions are evaluated for cross sections grouped by size (micro, small, and large capitalization).

The authors study the relationship between geographic dispersion and the investor recognition hypothesis using several methods, including the ratio of the number of listed firms in a state to its population and the difference between mutual fund capital and market capitalization of firms within a small radius (to measure the balance between available capital and investable opportunities).

The authors’ findings support the conclusion that geographic dispersion, measured by the number of states mentioned in an annual report, is related to average monthly returns.

The authors identify an investment relationship that has shown a pattern of positive excess returns in the past and may be useful in supplementing existing investment analyses. I have one area of concern: The research relies on the content of SEC 10-K filings, and the accuracy or completeness with which firms list the states in which they operate is not discussed. To the extent that firms include only a partial listing of their most important locations, the results would be affected by increasing the state count to its accurate level, which could lower the significance of the results. Another likely question for further study is whether investor preferences for local firms vary by locale. For example, do investors in certain parts of the country—say, Texas—show stronger local bias than investors in other parts of the country?