Policy Diffusion and Polarization Across US States; Stefano DellaVigna (UC-Berkeley)

Abstract

Economists have studied the impact of changes in numerous state laws, ranging from welfare rules to voting ID requirements. Yet for all this policy evaluation, what do we know about policy diffusion -- how these policies spread from state to state? We present a series of facts based on a data set of over 700 policies adopted by US states over the past 7 decades. First, we consider innovators, states that are more likely to originate new laws. State capacity seems to have a small role, in that larger and richer states are only slightly more likely to innovate policy. Second, the diffusion of policies from 1950 to 1990 is best predicted by proximity, in that a state is more likely to adopt a policy if nearby states have already done so, consistent with localized learning, geographical correlated preferences or competition. Third, this pattern has changed in the last two decades, with political alignment outperforming geographic proximity in predicting diffusion. Fourth, while the diffusion results are descriptive, event studies from changes in partisan control of state governments suggest a causal impact of political polarization since the 1990s. We discuss the implications of these patterns in light of a simple model, along with a case study of Medicaid adoptions. Finally, we highlight how the changes in policy diffusion patterns imply different threats to identification in difference-in-difference policy papers.