Automatic Voter Registration and Electoral Participation: Evidence from Staggered State Adoption

Abstract

Automatic voter registration (AVR) laws, which register eligible citizens to vote by default during government agency interactions, have been adopted by 20 U.S. states since Oregon's pioneering 2015 legislation. Using staggered difference-in-differences methods and Census CPS Voting Supplement data (2010-2022), I estimate the causal effect of AVR on voter registration and turnout. Despite theoretical predictions and single-state case studies suggesting substantial effects, I find null results: AVR adoption is associated with a statistically insignificant 0.2 percentage point increase in registration rates (SE=0.9pp, p=0.78) and a -0.1 percentage point change in turnout (SE=1.2pp, p=0.94). Event study estimates reveal violations of parallel trends assumptions, with treated states exhibiting declining pre-treatment registration relative to controls. These findings suggest that AVR's effects may be more modest and context-dependent than earlier literature indicated, highlighting the importance of multi-state analyses with heterogeneity-robust methods. I discuss measurement challenges, selection into treatment, and concurrent voting reforms as potential explanations for the null findings.