Democratic elections are sometimes thought to create accountability by (a) selecting competent types for office and (b) inducing reelection-minded incumbents to exert effort on behalf of voters. Many raise doubts about both claims. We compile a new dataset of more than 350,000 bills introduced in U.S. state legislatures with term limits, and we use it to measure legislator productivity. We establish two key findings consistent with models of electoral accountability. First, legislators who are reelected until they reach the binding term limit have higher average productivity, indicating that elections select for productive legislators. Second, using an individual-level difference-in-differences design, we show that termed-out legislators sponsor fewer bills and serve on fewer committees in their final term, on average, indicating that reelection incentives induce legislator effort. Classifying bills based on textual summaries, we show that the effect of the removal of reelection incentives on bill sponsorship appears to be driven by decreases in substantive, rather than symbolic, legislation. Moreover, this effect is largest for legislators who never seek office again in the future, and is concentrated in states with high legislative salaries.
Professor Hall's research focuses on the nature of democratic government. How effective are elections in controlling the behavior of political representatives? What factors make elections more or less effective, and why? He combines modern statistical techniques with wide-ranging quantitative and text-based datasets on American political activity to attempt to answer these questions. He received his Ph.D. in Political Science and his A.M. in Statistics from Harvard University in 2015. He graduated from Stanford University in 2009 where he majored in Economics and Classics. For more information, see his website at www.andrewbenjaminhall.com.