State Earned Income Tax Credit Generosity and Crime: Evidence from Staggered Adoption

Abstract

Does state-level income support reduce crime? I exploit the staggered adoption of state Earned Income Tax Credits (EITC) across 29 US jurisdictions (28 states plus the District of Columbia) between 1987 and 2019 to estimate the causal effect of income support on crime rates. Unlike prior work using panels beginning in the late 1990s, this study employs an extended 1987–2019 panel that provides pre-treatment observations for nearly all adoption cohorts, including the earliest adopters (Vermont 1988, Wisconsin 1989). Maryland (adopted 1987, the first year of the panel) is the only cohort without pre-treatment observations. Using a difference-in-differences framework with multiple modern estimators robust to heterogeneous treatment effects—including Callaway-Sant'Anna and Sun-Abraham interaction-weighted estimation, with Goodman-Bacon decomposition to diagnose TWFE bias—I find no statistically significant effect of state EITC adoption on property crime. The two-way fixed effects estimate indicates a small reduction (coefficient: $-$0.8%, 95% CI: [$-$5.9%, 4.3%]), while the Callaway-Sant'Anna ATT is $-$2.1% (SE: 2.4%). I also examine time-varying EITC generosity as a continuous treatment, finding that a 10 percentage point increase in state EITC match rates is associated with a statistically insignificant 1.2% reduction in property crime. Wild cluster bootstrap inference confirms these null results. Event study analysis reveals no significant pre-trends, supporting the parallel trends assumption. These findings suggest that the EITC's income support mechanism does not substantially reduce economically-motivated property crime at the state level, though effects may exist at finer geographic scales or for specific subpopulations.