An Opportunity for Arbitrage?

If markets are efficient at eroding price differentials, why does this phenomenon exist? [h/t to Paul Kedrosky]

The Only Game in Town: Stock-Price Consequences of Local BiasHarrison Hong, Jeffrey D. Kubik, Jeremy C. Stein

Theory suggests that, in the presence of local bias, the price of a stock should be decreasing in the ratio of the aggregate book value of firms in its region to the aggregate risk tolerance of investors in its region. We test this proposition using data on U.S. Census regions and states, and find clear-cut support for it. Most of the variation in the ratio of interest comes from differences across regions in aggregate book value per capita. Regions with low population density--e.g., the Deep South--are home to relatively few firms per capita, which leads to higher stock prices via an "only-game-in-town" effect. This effect is especially pronounced for smaller, less visible firms, where the impact of location on stock prices is roughly 12 percent.

Are information costs or labour costs or some other costs so high that firms don't know to relocate to these areas (especially firms that want to raise capital)? Why don't more start-ups emerge in these areas? How long can these types of differentials persist over time? Or is the efficient markets hypothesis inapplicable in these situations?