Turnout and the Labor Market

Using county level data from 1969-200 and various OLS and TSLS models, we find that increases in local per capita earnings and employment lowers voter turnout in gubernatorial and Senate elections but has not effect on Presidential turnout. We present a model in which risk-averse agents vote only if sufficiently informed about political candidates. When agents work more, as they do in periods of high local wages and employment, they devote less time to information acquisition, are less informed and thus vote less. This negative effect should be smaller in elections, such as that for the President, where information is so ubiquitous that reductions in political attentiveness have little effect on uncertainty. Consistent with the model’s predictions, we find using individual data from several waves of the American National Election Study (ANES) we find that: less informed voters, and especially political moderates, are less likely to vote; uncertainty is smaller in Presidential as opposed to other elections; and that, voters' accessing of media, political knowledge and interest in politics vary negatively with changes in employment. Preliminary. Please Do not Cite without Permission. We thank seminar participants at Cornell University, the University of Chicago, MIT, Princeton University for their comments. We have also benefited from discussions with Erik Hurst, Matthew Gentzkow, Chris Berry and Jesse Shapiro.