Research

To help unemployed low-income individuals find work, is it better to directly match them with job openings, or first have them complete remedial coursework? I reexamine the original data from the GAIN experiment of the early 1990s, utilizing unexplored details about program intake that created quasi-random variation in treatment status within each county. Using a regression discontinuity design, I find support for the notion that at the margin of treatment categorization defined by an intake exam, those induced directly into the labor force saw much larger increases in earnings relative to those who completed coursework first. Effects are strongest along the extensive margin; neither type of programming differentially induces those already in the labor force to earn more. Finally, results grow in magnitude over time. This evidence leads to the conclusion that direct labor force strategies may be superior to remedial coursework for welfare participants who already have baseline cognitive skills.

We show that natural disasters affect a region's aggregate human capital through at least four channels. In addition to causing out-migration, natural disasters reduce student achievement, lower high school graduation rates, and decrease post-secondary attendance. We estimate that disasters that cause at least $500 in per-capita property damage reduce the net present value (NPV) of an affected county's human capital by an average of $505 per person. These negative effects on human capital are not restricted to large disasters: less severe events – disasters with property damages of $100-$500 per capita – also cause significant and persistent reductions in student achievement and post-secondary attendance.

This paper asks how individuals across the income distribution value access to public transit and how this shapes the rental housing market. To study this question, we use the announced closure of the L train in NYC to study how rents were affected in the housing market and the mobility decisions of individuals across the income distribution. We find that low-income households value the subway more than high-income households, while high-income households likely value other neighborhood amenities more. We show evidence of landlord market power and find that landlords give existing low-income households a higher discount on their rent to incentivize them to stay in the affected neighborhoods. From a policy perspective, one must keep the dynamics of other neighborhood amenities in mind while evaluating whether intended beneficiaries benefit in the long run.