- I study the connection between employment, earnings, and crime for a large sample of felony offenders in Washington State. Using administrative records on arrests, charges, and incarceration linked to unemployment insurance earnings data, I show that ex-offenders' risk of arrest increases by 50% immediately after job loss during a mass layoff and remains elevated for 2-3 additional quarters. Effects are concentrated in property crimes and domestic violence offenses. I then show that panel regressions estimated on the full dataset and using coworkers' outcomes as instruments yield similar results for job loss and an elasticity of offending to income of about -0.1. The causal mechanism appears to be unanticipated income shocks rather than incapacitation effects of employment. Individuals exposed to highly seasonal income, such as construction workers, do not exhibit seasonal patterns in offending. And a regression kink analysis of offending rates around maximum weekly unemployment insurance benefits implies a similar income elasticity to the panel estimates. While the evidence supports Becker (1968) rational models of criminal behavior, the results also highlight the limited scope for income shocks to explain changes in aggregate crime rates. The majority of serious crime is committed by individuals with no formal labor market attachment and the effects of layoffs appear to be pro-cyclical, suggesting the marginal job loser in a recession is a lower crime risk.
- In 2013, Seattle, WA, implemented a "ban the box" (BTB) ordinance prohibiting public and private employers in the city from asking prospective hires about their criminal history until after an initial screening. I evaluate the law theoretically and empirically. I first use a classic statistical discrimination model to show that while BTB must have opposite effects on individuals with and without criminal records, respectively, its impacts on specific demographic groups' average employment rates are theoretically ambiguous. I then test whether Seattle's law did affect individuals with records using employment and earnings data from the state unemployment insurance system for 300,000 ex-offenders. These results show that BTB had no detectable impact on ex-offenders' employment or wages across multiple difference-in-differences research designs. Given the zero effect of BTB, other policies that target the overall employability of ex-offenders are likely to be more effective, especially since this population's employment rates and earnings remain extremely low even before initial conviction.
The Rise and Fall of Female Labor Force Participation During World War II (revision in progress)
Revise and Resubmit at The Journal of Economic History
- I use newly digitized and existing data on military manpower mobilization, employment in war-related industries, and public employment office activities to revisit the impact of World War II on female labor force participation (FLFP). First, I show that geographic variation in female employment during the war was largely driven by the allocation of military contracts. Manpower mobilization, which several recent studies implicitly use as an instrument for female labor demand, has a limited connection to wartime work. Second, I show that the wartime boom in female employment had no detectable effect on FLFP by 1950 beyond modest effects in manufacturing industries for white women. Finally, I provide evidence using data and narratives from the U.S. Employment Service (USES) that the decline in female employment immediately following the war was the result of displacement by returning veterans and sharp cutbacks in war-related industries, rather than drops in female labor supply. The results suggest that the rapid rise of female employment during WWII was made possible by temporary changes in labor demand that quickly reverted at the war's conclusion. Permanent shifts in opportunities for women in the labor force would continue to evolve gradually over the ensuing decades, despite the sudden and short-lived WWII shock.