Failing to Learn in the Laboratories of Democracy

• In Event: Political Economy of Policy Making in the United States

Abstract

We estimate a model of partisan U.S. state policymakers who rely on their own and other states’ past histories to learn about the causal effect of gun control policies on violent crime. We allow partisanship to influence not only policymakers’ preferences and initial beliefs but also the manner in which they take in new information. In particular, we are interested in uncovering systematic biases relative to perfectly-rational Bayesian learning. Do partisan policymakers disregard or overreact to new data? Anecdotally, scholars and the popular press have suggested that, on the contentious issue of gun control, partisan considerations may bias objective policy evaluation. We contribute to this debate by bringing data about state-level gun control legislation and violent crime to the model. Importantly, distinguishing between under- or overreactions to empirical evidence has normative implications, as existing theoretical results suggest that, while underreacting can simply delay identifying the right policy, overreacting can prevent learning the truth altogether.

Authors

• Mary Kroeger, University of Rochester

• Sergio Montero, University of Rochester

• Scott Abramson, University of Rochester