Nick Langston

This paper presents causal evidence on how regulatory compliance burdens affect academic research. I exploit the 1991 Common Rule, which differentially increased researchers' exposure to institutional review board oversight. I classify publications as human subjects (HS) or non-human subjects at scale with supervised machine-learning, and I use these classifications to measure researchers' exposure to the policy shift in terms of baseline propensity to produce HS work. Matched difference-in-differences results indicate that exposed US researchers' probability of publishing any HS work in a given year declines by 5.6 percentage points (compared to a pre-period treatment group mean of 37.5%), and the effect is nearly twice as large for researchers at R2 institutions as at R1 institutions, consistent with greater researcher-borne administrative burdens at less-resourced institutions. The effect is larger among researchers more specialized in HS research at baseline and substantially weaker in a non-US placebo sample. I find little evidence of effects on HS publication counts or overall publication counts, implying that the increase in regulatory compliance requirements led to HS research becoming more concentrated among a smaller set of researchers. More broadly, the results suggest that increased regulatory compliance burdens can change who participates in a regulated activity even when aggregate measures of that activity see little change. Download the working paper from Dropbox.

This paper studies whether reducing barriers to accessing the social safety net at release from jail affects criminal recidivism. I examine an Orange County, California, program that allowed incarcerated individuals to apply for SNAP benefits before release. Using individual-level jail records obtained through public records requests, I estimate difference-in-differences models comparing Orange County with comparison counties before and after the introduction of pre-release enrollment. I estimate that the program reduced one-year recidivism by 2.9 percentage points. The setting extends prior evidence on SNAP and recidivism in several ways: the program applied to individuals incarcerated for a broad range of offenses in county jail, rather than only drug-related offenders in state or federal prison, and it changed the timing and ease of benefit access rather than eligibility itself. The results therefore provide evidence that reducing administrative frictions and delays in accessing food assistance during reentry can meaningfully reduce subsequent criminal justice involvement, with potential benefits for both affected individuals and local governments that bear the costs of county incarceration.

Contributing to the literature of empirical tests of Becker (1968)'s economic model of the decision to commit criminal acts, I exploit a plausibly exogenous ban on a publicly-known market for stolen goods in San Francisco to test whether access to a physical market where stolen goods can be discreetly offloaded is an important driver of criminal behavior. Despite the City of San Francisco claiming a large decrease in robbery and assault incidents, I find no evidence of any favorable effects of the policy change using a difference-in-differences design and data from the San Francisco Police Department.