Retrospective Voting on Crime Rates in the US

• In Event: Electoral Effects of Scandals, Crime, and Punishment

Abstract

Crime rates are a major concern among the American public, but systematic exploration of their impact on US voting behavior remains scarce. Although there is a considerable body of literature on retrospective voting behavior, these studies predominantly focus on economic voting, namely voters' tendency to vote based on the state of the economy. In this study, we undertake a comprehensive and systematic examination of how crime rates influence voting in the US. At the county level, we analyze presidential, US House of Representatives, US Senate, gubernatorial, and state attorney elections. At the city level, we explore mayoral elections. For both county and city level crime rates, we utilize the Uniform Crime Reporting (UCR) data. To address endogeneity and attenuation bias arising from measurement errors in the crime data, we employ an instrumental variable (IV) approach, leveraging the timing of grants provided to law enforcement agencies through the Community Oriented Policing Service (COPS) program. Our data spans from 1995, the inception of the COPS program, to 2014, coinciding with the release of the most recent county-level UCR data. Our preliminary findings indicate that crime rates significantly impact votes for incumbents, particularly in presidential elections.

Authors

• BK Song, Sogang University

• Heesun Yoo, Emory University