Research


Job Market Paper:

Foster Family Characteristics and Children's Educational Outcomes


Roughly 6% of US children are removed from their home at least once to be placed with a foster family or in a group home. Foster care aims to prevent some of the worst disinvestments in children’s human capital, such as abuse and neglect, but is an inherently destabilizing intervention. Improving stability for children in foster care through productive placements is therefore of policy interest. I examine the impact of foster care placements on children’s human capital accumulation as measured by short-run educational outcomes such as school attendance and suspensions. The placements I consider include group homes and foster families, either relatives (kinship placements) or unrelated foster families. While previous research has focused on the extensive margin of foster care placement, I include multiple treatments to identify causal effects of placement types, with a focus the differential impacts of unrelated families who are of the same race versus a different race. I use administrative data from Wisconsin containing unique information on providers and children and identify the model with caseworker propensity instruments. I find that while both foster homes and kinship placements improve outcomes relative to group care, kinship care leads to the largest reductions in suspensions and increases in attendance. However, considering all foster placements together obscures important heterogeneity. Placement with a same-race foster provider increases attendance and decrease suspensions relative to both non-race-matched foster care and kinship care for non-white children. To investigate channels, I study the effects of provider income and school stability. I provide suggestive evidence that same-race families tend to be those that lead to less upheaval in a child’s life. These findings suggest that demographic match is one potential way to improve match quality in the foster care program.

Other Dissertation Essays:

Medical Innovation and Health Disparities (with Bart Hamilton, Andres Hincapie, and Nicholas Papageorge). Under Review. NBER working paper 28864.

We examine how medical innovation can reinforce existing health disparities by disproportionately benefiting socioeconomically advantaged patients. Previous work has generally focussed on price and access. We propose an additional channel: differences across sociodemographic groups in how health and work interact. To investigate the interplay between innovation and health disparities, we develop a lifecycle model in which the effect of medical treatment on labor supply varies across sociodemographic groups. We estimate the model using rich data on treatment choices and employment decisions of men infected with HIV. In the model, treatments can improve long-run health, but can also cause immediate side effects that interact with the utility cost of work. Estimates indicate that HIV-infected men often forego medication to avoid side effects, in part to remain employed. This effect is stronger for people with fewer years of education, leading to lower use of treatment and worse health outcomes. As a result, while a breakthrough HIV treatment---known as HAART---improved lifetime utility for all patients, it disproportionately benefitted those with higher levels of completed education, thereby reinforcing existing inequality. A simulation mandating HAART use worsens inequality, pushing low-education individuals out of the labor force. However, a counterfactual subsidy that increases non-labor income reduces employment for all education groups, but only increases adoption of HAART and improves health among lower-education individuals, who face a starker health-work tradeoff.


Working the System: Unemployment and Foster Care


I study the relationship between foster care spell lengths and local economic conditions. I expand on the literature by focusing on spell duration rather than point-in- time caseloads. Previous work finds that caseloads are higher when unemployment is higher, which could be due to additional children in foster care or because of longer spells. I find that spells are, on average, shorter when unemployment is higher, suggesting that caseloads increase on the extensive, rather than intensive, margin. This finding has implications for policy, as the literature suggests that children on the margin may be better off out of foster care.

Other Work in Progress:

Optimal Taxation and the Great MTR Pivot in U.S. Transfer Programs (with Robert Moffitt)


A great transformation in the structure of benefits and marginal tax rates in US transfer programs has occurred in the last 40 years, called the Great MTR Pivot because it twisted, or pivoted, the budget constraint around its central point, lowering benefits at low levels of income and raising them at high levels of income. We examine the implications of this transformation for optimal tax rates. We set up a traditional static labor supply model and calibrate it to the existing evidence on the effects of the major US transfer programs on labor supply. We include the EITC, all positive tax programs, SNAP, Medicaid, subsidized housing, and AFDC/TANF. We allow the labor supply effects of in-kind transfers to differ from those of cash and for incomplete take-up, since all transfer programs have take-up rates of less than 100 percent of eligible. Using our calibrated model, we posit a traditional social welfare function to estimate the tradeoffs involved. We assume a set of different parameters to represent inequality aversion and redistributional preferences. We then use our calibrated model to first determine whether the MTR Pivot transformation is likely to have increased or decreased social welfare for different social welfare parameters, and then investigate alternative MTR structures that could improve social welfare.

Expectations and Aspirations: What Matters (with Jodi Miller and Nicholas Papageorge)

This paper examines the differential effect of parental aspirations and expectations for their children's education. We find that the two concepts are distinct and have separate implications for parental behavior and child outcomes.