Job Market Paper

Does "Welfare-to-Work" Work? Evaluating Long-Run Effects over a Generation

This version: January 2023

Work requirements remain a popular yet controversial policy in welfare programs around the world. This paper provides a comprehensive evaluation of welfare-to-work reforms by estimating their long-run effects on multiple outcomes for a generation of birth cohorts. I apply recent advances in difference-in-differences methods and large-scale administrative data to study the introduction of mandatory, publicly-managed work programs in Denmark's social assistance program. Effects are highly heterogeneous across cohorts based on the time the reforms were introduced in the life-cycle. Adults already eligible for welfare at the time of the reforms incur null or modest negative effects on income and substitute toward crime and alternative welfare programs. Meanwhile, not-yet-eligible children experience significant gains in schooling and income. This heterogeneity is consistent with a model where younger cohorts invest in their human capital in anticipation of future work requirements while older cohorts adjust along margins with high social costs. Evidence suggests that heterogeneity over birth cohorts can persist for decades over the life-cycle and spill over to the next generation. Cost-benefit analyses reveal that welfare-to-work is cost-effective in the long-run, but is likely driven by anticipatory behavioral responses of younger cohorts aging into the population rather than the effect of participating in work requirements. This sheds light on the interpretation of aggregate effects of welfare-to-work over time and alternative, more efficient policy designs.

Working Papers

Intergenerational Mobility with Steven N. Durlauf

NBER Working Paper #29760

Prepared for The Inequality Reader, Fifth Edition, D. Grusky, N. Dahir and C. Daviss, eds.

This version: February 2022

This essay reviews the theory and empirics of intergenerational mobility. Our review draws on models and empirical analyses of classic and more recent work from both economics and sociology. We summarize models and the surrounding empirical evidence of two key sets of mechanisms: family factors (income, education, credit constraints, household composition, and genes) and social factors (schools, neighborhood sorting, racial segregation, and peer and role model effects). We then discuss and evaluate current methods used to measure intergenerational mobility, including linear regressions and Markov chains. Theoretical models imply nonlinear relationships between parent and child status that are often ignored in practice and offer potentially different interpretations of the evidence of heterogeneity in mobility across locations, groups, and time. We conclude that the next generation of studies would benefit from a closer integration of theory with empirics.

Understanding Heterogeneity in Intergenerational Mobility across Neighborhoods: From Excavation to Refinement with Steven N. Durlauf, Rasmus Landersø, and Salvador Navarro

Draft coming soon

Recent research has uncovered large spatial heterogeneity in intergenerational mobility across neighborhoods in countries around the world. Yet there is nothing close to a consensus on the reasons why mobility is high in some neighborhoods and low in others. Our paper examines the roles that sampling error, families’ self-selection into neighborhoods, and locational characteristics play in generating this spatial heterogeneity. We use administrative data from Denmark to estimate a sequence of richer models that “excavate” these distinct factors across 276 municipalities and nearly 2,000 parishes. Families’ self-selection into neighborhoods and sampling error explain the majority of observed heterogeneity across neighborhoods, yet an irreducible neighborhood effect remains that cannot be explained by these factors. Using cluster analysis, we “refine” the structure of this irreducible component and uncover distinct types of neighborhoods. Types tend to be spatially segregated across rural and urban lines and suggest the presence of multiple equilibria predicted by theory.

Works in Progress

Long-run Effects of Two-Generation Social Policy at Scale

This paper provides one of the first evaluations of social policies that simultaneously target parents and children of disadvantaged families at a population scale. I exploit welfare-to-work reforms and the roll-out of public daycare in Denmark during the 1990s to analyze the effects of work program requirements and daycare programs on the socioeconomic outcomes of single mothers and their children twenty years later. To understand the behavioral mechanisms behind these effects, I develop a selection model of work programs and daycare participation where mothers account for her own and her child's long-run outcomes. I link model identification with policy variation through a multiple instrumental variables framework. This analysis will demonstrate whether welfare-to-work and childcare policies are an effective dual-pronged approach for improving intergenerational mobility among the disadvantaged.

Child and Parent Agency in College Attendance

What role do parents play in their child's decision to attend college and their subsequent outcomes? I use the National Education Longitudinal Survey of 1988 to develop a unique measure of a child's degree of agency relative to parents in the decision of whether to attend college. I find that college attendance, college completion, and earnings at age 25 significantly decrease in the degree of child agency. Heterogeneous patterns emerge for different demographic groups, and post-secondary school attendance strongly depends on family background and the parent's preferred schooling decisions. To rationalize these patterns, I develop a two-agent discrete choice model describing the college decision-making process between children and parents that distinguishes between a parent’s paternalistic and altruistic motives. With survey data on parent preferences and observed decisions, I identify child and parent preferences for attending college and the intra-family mechanisms behind the college attendance decision. I find that parents exhibit strong paternalism, as well as evidence that some parents possess agency to cause their children to attend college even when their expected returns from attending were negative or to not attend when their expected returns were positive.

Mapping Nonlinearities in Intergenerational Mobility to Neighborhoods with Steven N. Durlauf

Two empirical facts have been uncovered in the intergenerational income mobility literature: (1) there is large heterogeneity in income mobility across neighborhoods and (2) the mobility process at the population level is nonlinear in parental income. Are these two facts interlinked? This paper develops statistical tests that distinguish between these two phenomena that account for neighborhood sorting in parental income and other background characteristics. We employ these tests on mobility estimates from the basic linear mobility model and from enriched models that allow for heterogeneity in other family background characteristics in Denmark. We find evidence that neighborhood heterogeneity measured in the basic model exists after accounting for the presence of nonlinearities, but is substantially reduced after accounting for the role of other family characteristics.

Research Notes

Semiparametric Identification of Triple Difference Designs

This version: June 2022

Email for draft

Comparing the Heterogeneity of Neighborhood Mobility between Denmark and the United States

This version: August 2021

Email for draft