Published Work

Social Transfers and Spatial Distortions (with Mark Colas)
Journal of Labor Economics (forthcoming)
Previously as: OIG Working Paper # 54

US social transfer programs vary substantially across states, incentivizing households to locate in states with more generous transfer programs. Further, transfer formulas often decrease in income, therefore rewarding low-income households for living in low-paying cities. We quantify these distortions by combining a spatial equilibrium model with a detailed model of transfer programs in the US. The current system leads to locational inefficiency of 4.38% of total transfer spending. A reform that both harmonizes transfer policies across states and indexes household income to local average earnings reduces this inefficiency by over 85 percent while still preserving the programs’ means-tested nature.

Working Papers

The Effect of Violent Video Games on Violent Crime (with Gretchen Gamrat)

We analyze the effect that violent video games have on violent crime in the United States. Using county-level variation in retail sales of ``mature" video games, some of which occur on proprietary platforms, we leverage exogenous variation in exposure to identify corresponding changes in crime outcomes. Especially after high-profile violent crimes, policymakers and the news media frequently argue that increased exposure to violent games leads to increased violent crime. We find no such evidence. If anything, our analysis suggests that short-run decreases in violent crime, specifically violent sexual offenses, follow the release of mature video games.

Combinatorics, mean convergence, and grade-point averages (with Glen Waddell)
Reject and Resubmit: Journal of Human Resources
Previously as: IZA Discussion Paper # 15414

While comparing students across large differences in GPA follows one's intuition that higher GPAs correlate positively with higher-performing students, this need not be the case locally. Grade-point averaging is fundamentally a combinatorics problem, and thereby challenges inference based on local comparisons—this is especially true when students have experienced only small numbers of classes. While the effect of combinatorics diminishes in larger numbers of classes, mean convergence then has us jeopardize local comparability as GPA better delineates students of different ability. Given these two characteristics in decoding GPA, we discuss the advantages of machine-learning approaches to identifying treatment in educational settings.