We study how market design choices exacerbate or mitigate pre-existing inequalities among participants. We introduce outside options in a well-known school choice model, and show that students always prefer manipulable over strategy-proof mechanisms if and only if they have an outside option. We test for the proposed relationship between outside options and manipulability in a setting where we can identify students' outside options and observe applications under two mechanisms. Consistent with theory, students with an outside option are more likely to list popular, highly-rated schools under the Boston mechanism, and this gap disappears after switching to a Deferred Acceptance mechanism.
More than two million U.S. households have an eviction case filed against them each year. Policymakers at the federal, state, and local levels are increasingly pursuing policies to reduce the number of evictions, citing harm to tenants and high public expenditures related to homelessness. We study the consequences of eviction for tenants using newly linked administrative data from two large cities. We document that prior to housing court, tenants experience declines in earnings and employment and increases in financial distress and hospital visits. These pre-trends are more pronounced for tenants who are evicted, which poses a challenge for disentangling correlation and causation. To address this problem, we use an instrumental variables approach based on cases randomly assigned to judges of varying leniency. We find that an eviction order increases homelessness, and reduces earnings, durable consumption, and access to credit. Effects on housing and labor market outcomes are driven by impacts for female and Black tenants.
This project is supported by the National Science Foundation, the Laura and John Arnold Foundation, the Spencer Foundation, the Kreisman Initiative on Housing Law and Policy, the Horowitz Foundation for Social Policy, the Robert Wood Johnson Foundation, the Becker Friedman Institute, and the Tobin Center for Economic Policy. It is part of the "Using Linked Data to Advance Evidence-Based Policymaking" initiative, a collaboration between Chapin Hall and the Census Bureau. The project was referenced in the Economist and in the New York Times.
We evaluate how nonresponse affects conclusions drawn from survey data and consider how researchers can reliably test and correct for nonresponse bias. To do so, we examine a survey on labor market conditions during the COVID-19 pandemic that used randomly assigned financial incentives to encourage participation. We link the survey data to administrative data sources, allowing us to observe a ground truth for participants and nonparticipants. We find evidence of large nonresponse bias, even after correcting for observable differences between participants and nonparticipants. We apply a range of existing methods that account for nonresponse bias due to unobserved differences, including worst-case bounds, bounds that incorporate monotonicity assumptions, and approaches based on parametric and nonparametric selection models. These methods produce bounds (or point estimates) that are either too wide to be useful or far from the ground truth. We show how these shortcomings can be addressed by modeling how nonparticipation can be both active (declining to participate) and passive (not seeing the survey invitation). The model makes use of variation from the randomly assigned financial incentives, as well as the timing of reminder emails. Applying the model to our data produces bounds (or point estimates) that are narrower and closer to the ground truth than the other methods.
This paper analyzes the effect of Europe’s largest public housing program on socio-economic outcomes for low-income households. Using lotteries for housing units in the Netherlands and data linking national registers to application choices, I show that the average move into public housing negatively affects labor market outcomes and proxies for neighborhood quality, and increases public assistance receipt. However, consistent with a model of labor supply responses to conditional in-kind transfers, average impacts miss substantial heterogeneity both across neighborhoods and, within neighborhood, across recipients. Moves into high-income neighborhoods generate positive effects, which are driven by ‘upward’ moves made by individuals previously living in low- or middle-income neighborhoods. Lateral and ‘downward’ moves have the opposite effect. To evaluate whether these results generalize to non-recipients, I develop a model of application behavior that utilizes panel data on application choices and exploits variation induced by the housing allocation mechanism. Using the model, I recover the distribution of heterogeneity that drives selection into and returns from lotteries, and estimate that selection on gains is limited. This suggests that targeting public housing in high-income neighborhoods based on observable characteristics can increase economic self-sufficiency.
WORK IN PROGRESS
This paper examines the effects of conviction without incarceration -- a common outcome of criminal court proceedings -- and of incarceration on recidivism. We study felony cases in Virginia that are quasi-randomly assigned to judges, and make three contributions. First, we present estimates of the impact of conviction on recidivism based on a 2SLS regression with judge stringency instruments. If given a causal interpretation, our estimates would imply large and sustained increases in recidivism from receiving a conviction relative to dismissal. Using a similar research design, we find that incarceration reduces recidivism in the first year, likely due to incapacitation, with no longer-term effects. These conclusions about incarceration are further supported by analysis based on discontinuities in sentencing guidelines. Second, we discuss how, in multiple-treatment settings, some models of judge decision making facilitate the interpretation of 2SLS estimates as well-defined treatment effects, while others do not. In particular, we consider which models of the judge decision process imply that 2SLS estimates interpretable treatment effects for a particular margin, such as conviction vs dismissal, or incarceration vs conviction. Third, we discuss and implement several methods which allow us to recover margin-specific treatment effects under sets of assumptions where 2SLS estimates do not. Most of these yield conclusions similar in sign and magnitude to those drawn based on the 2SLS estimates, although they are sometimes less precise. We conclude that conviction may be an important and potentially overlooked driver of recidivism, while incarceration mainly has shorter-term incapacitation effects.