Research

State policies often rely on local institutions that also provide other public services. This paper studies how local criminal justice systems accommodated Texas's Operation Lone Star, which pursued immigration objectives through state criminal law beginning in 2021. Using administrative data from 2018 through 2024, a difference-in-differences design compares 34 initially designated counties with clean never-treated counties. Filings in the felony category containing program cases roughly double. Courts expand nontrial processing but carry larger backlogs, while ordinary felony resolutions fall by roughly 20 percent and ordinary backlogs accumulate later. County jails devote an additional 6.8 percentage points of rated capacity to people on local charges, and the pretrial population on county charges, including detainees held elsewhere, rises by roughly 43 percent. Comparisons with counties matched on pretreatment characteristics and initial pandemic disruption preserve the main patterns, although several estimates become less precise. Local arrest responses are less uniform, and reported-crime estimates depend on the counterfactual trend. The findings suggest that expanding state enforcement can change the allocation of local criminal justice services, creating potential costs for people beyond the policy's intended targets.

This paper estimates the ridership response to a sustained, system-wide deterioration in transit service quality at the Massachusetts Bay Transit Authority's (MBTA) heavy-rail system. Federal oversight prompted by years of deferred maintenance produced an extended period of slower service across essentially every pair of stations. To measure service quality, the paper introduces a gravity-style accessibility index that weights destinations by walkshed employment and normalizes the contemporaneous gravity sum pair-by-pair against a bin-matched pre-shock baseline; the resulting measure varies only with changes in transit times, isolating supply-side variation. A two-way fixed effects specification on a station-day panel covering 2018–2024 recovers a within-station elasticity of log ridership with respect to log accessibility deviations of 0.336 (s.e. 0.060), implying a transit-time elasticity of approximately –0.44; within the range of pre-pandemic estimates from the transit-demand literature. The result is robust to alternative weighting schemes, fixed-effects structures, sample windows, and a Poisson pseudo-maximum-likelihood specification on the unfiltered panel; a complementary cross-sectional dose-response design built around the March 2023 systemwide order recovers a comparable slope. The findings argue that sharp, well-identified supply shocks offer a viable complement to the long-run time-series variation that has historically anchored estimates of transit-demand elasticities.

The majority of felony cases in the United States involve indigent defendants represented by publicly financed attorneys. We examine the effects of efforts to improve representation for economically disadvantaged defendants through the adoption of managed indigent defense offices (MIDs) in Texas, which impose a layer of oversight and accountability on attorneys who provide indigent defense services. Exploiting the staggered roll-out of MIDs across Texas counties, together with administrative data on attorney employment and caseloads, jail populations, and newly compiled arrest and prison records, we show that introducing management concentrates cases among a smaller set of local lawyers. Management reduces the use of bench warrants and the probability of imprisonment. Sentences among those incarcerated lengthen as imprisonment falls, implying that the defendants diverted from prison would have served relatively short terms. Overall, we find that management matters in the administration of criminal justice.