Avoiding convictions : regression discontinuity evidence on court deferrals for first-time drug offenders

This paper studies the causal impact of court deferrals, a legal strategy to help defendants avoid a felony conviction record, on the future criminal and labor market outcomes of first-time felony drug offenders. To accomplish this, we exploit two natural experiments in Harris County, Texas, in which defendants appearing in court one day versus the next experienced abruptly different likelihoods of deferral. In 1994 deferral rates dropped by 34 percentage points the day following the implementation of a penal code reform; in 2007 deferral rates increased by 22 percentage points the day after the unexpected failure of a ballot initiative to expand the county jail. Using administrative data and local polynomial regression discontinuity methods, we find robust evidence consistent across both experiments that regimes with expanded use of court deferrals generated substantially lower rates of reoffending and unemployment over a five-year follow-up period. Additional analysis delves further into the timing, nature and incidence of these impacts. Together our results suggest that increasing the use of deferral programs may be an attractive and feasible option for a jurisdiction seeking to reduce the fiscal cost and community impact of its criminal justice system.