Research

I estimate the effects of pro-natalist cash transfers on fertility, sex ratio at birth, and infant health in South Korea, using registry data from 2000–2015. I find that the total fertility rate in 2015 would have been 4.7 percent lower without the transfers. Cash transfers also lowered the sex ratio at birth, historically skewed toward boys in South Korea. Additionally, I show that the cash transfers led to reductions in birth weight and gestational length, which appear to be driven by negative selection into fertility, and that the cash transfers may have had positive impacts on birth weight among low-income families.

In this paper, we exploit a series of relocations of public-sector entities in South Korea as an exogenous source of variation in public sector employment to estimate the local employment multiplier. We find that an introduction of 1 public sector employment increases private sector employment by 1 unit, almost completely driven by the service sector. In line with the literature, we document that the effect of public employment on private employment is highly localized. In addition to changes in private employment, we also find that the relocation led to a positive net inflow of residents into the treated neighborhood; this effect is also localized. Lastly, we estimate heterogeneous local employment multipliers and provide evidence that this heterogeneity is shaped by the size of public sector shocks, different types of relocation, and the extent of migratory responses.

We study optimal dynamic lockdowns against COVID-19 within a commuting network. Our framework integrates canonical spatial epidemiology and trade models, and is applied to cities with varying initial viral spread: Seoul, Daegu, and NYC-Metro. Spatial lockdowns achieve substantially smaller income losses than uniform lockdowns, and are not easily approximated by simple centrality-based rules. In NYC-Metro and Daegu—with large initial shocks—the optimal lockdown restricts inflows to central districts before gradual relaxation, while in Seoul it imposes low temporal but large spatial variation. Actual commuting responses were too weak in central locations in Daegu and NYC-Metro, and too strong across Seoul.

We study how industrial robots, which can substitute for routine production tasks, affect human capital investment during high school. Using administrative records from Texas for academic years 2004/05-2018/19 and an instrument based on European robot adoption, we find that greater local robot exposure reduces dropout, raises observed high school completion, and shifts coursework toward college preparation and away from vocational subjects. Advanced course enrollment rises throughout high school, while dropout reductions and additional math, science, and technology coursework are concentrated in the upper grades. Both boys and girls increase academic investment. Boys' larger reductions in vocational coursework narrow the gender gap in vocational course taking, while girls' larger gains in AP/IB enrollment widen their advantage in advanced preparation. Adjustment to automation thus begins before entry into sustained employment, through continued schooling and increased investment in preparation for further education.

We study the impacts of childbirth on maternal mental health and the role of pro-natalist cash transfers. Using claims-level data from South Korea's universal healthcare system, we find that mental health diagnoses rise by 34.8% (198.7%) after the first (second) birth. We find little evidence that cash transfers mitigate these effects. As potential mechanisms, we examine liquidity constraints, labor market changes, time use, and social stigma. Lastly, we document that poor mental health after childbirth is negatively associated with the likelihood of having another child.

We study how automation affects internal migration in a labor-scarce economy. A stylized framework shows that task-level substitution need not imply local contraction: when routine labor is scarce, automation's expansionary effects on local labor demand can outweigh displacement. Using South Korean data and a long-difference design with shift-share robot exposure, we find that greater robot exposure increases job-related net migration but not non-job-related migration. Robot exposure is also associated with higher wages, expansion in employment categories associated with complementary activities, and manufacturing establishment growth. Consistent with the framework, the job-related migration response is substantially larger in tighter baseline manufacturing labor markets. We then compare automation with an independently generated immigration shock. Both shocks increase job-related native migration, remain similar when included jointly, and show little evidence of systematic amplification or attenuation. The findings suggest that both automation and immigration can facilitate local adjustment to labor scarcity by attracting workers rather than displacing them.

We study the spatial dispersion of gender and racial wage gaps across U.S. cities. First, we document substantial geographic dispersion in both wage gap measures across U.S. metropolitan areas, and show that cities with larger gender wage gaps also tend to have larger racial wage gaps. Second, we build a model of worker and firm location featuring group-specific Nash bargaining over wages and task bundles in production. The key implication of the model is the log-linear wage structure which motivates a three-way fixed-effects decomposition of wages into worker, firm, and city components, an extension of the standard AKM model, which we estimate using matched employer-employee data (Longitudinal Employer-Household Dynamics). Even after absorbing worker and firm heterogeneity, the same city offers different returns to different groups: group-specific city fixed effects account for 15.8 percent of the spatial variance in the wage gap. Our findings suggest that place is an independent determinant of gender wage inequality, and that understanding spatial variation in local labor market conditions is essential for explaining the persistent wage gap across different groups of workers.

We study the unintended consequences of South Korea's 2008–2010 immigration reforms for marriage markets and native households. The reforms expanded work and residency opportunities for ethnic Korean Chinese migrants, reducing reliance on marriage-based entry. Exploiting pre-existing variation in sex ratios and marriage-migrant networks in a DDD design, we find substantial reductions in marriage migration and marriages between Korean men and foreign women, alongside increased work migration. Among native couples whose marriages predate the reforms, wives' employment-decision autonomy rises by 0.36 standard deviations, employment increases by 9 percentage points, and market work rises by five hours per week. Household income and expenditures increase, subjective well-being improves, and divorce falls by 16 percent, with the decline concentrated in divorces attributed to irreconcilable differences and the share attributed to psychological or physical abuse rising. The results show how immigration policy can reshape marriage-market conditions and affect bargaining, behavior, welfare, and marriage stability.

How much do people value public spending? I answer this question using an extended gravity framework that jointly accounts for migration and commuting. Leveraging South Korean census data and national tax reforms as exogenous variation, I estimate the elasticity of residential choice with respect to local government spending to be about one. Public goods are valued, and congestion effects indicate partial rivalry. Embedding the framework in a general-equilibrium setup, I characterize the welfare-maximizing spatial tax redistribution that balances fiscal externalities shaped by workers' spatial mobility. Ignoring either migration or commuting biases key elasticity estimates and misinforms place-based policy design.

Previously circulated as “The Valuation of Local Government Spending: Gravity Approach and Aggregate Implications.”

Changes in local news ownership can reshape the information environments through which communities encounter political issues and social groups. Yet while research has examined how media ownership and partisan news influence political attitudes and behavior, less is known about their implications for intergroup relations and social conflict. This study examines these questions in the context of the expansion of Sinclair Broadcast Group across U.S. local television markets. Study 1 uses an event-study design to examine whether Sinclair’s entry into local media markets is associated with changes in hate crime. Study 2 provides complementary evidence from local newscast transcripts, examining whether Sinclair-owned and non-Sinclair-owned stations differ in their portrayals of minority groups using sentiment analysis, discrete-emotion measures, and contextual-embedding-based metaphor analysis. Results show that Sinclair’s market entry is associated with an average increase of approximately 0.5 hate crimes per 100,000 residents, with the increase persisting for five years following entry and concentrated among crimes targeting victims based on sexual orientation, race/ethnicity, and religion. The content analysis further shows that Sinclair-owned stations exhibit more negative coverage of Muslims and Islam and greater use of anger-, fear-, and disgust-related language. Sinclair-owned stations also generally exhibit stronger dehumanizing metaphor associations than non-Sinclair-owned stations. Together, the findings demonstrate systematic differences associated with local news ownership in both community-level intergroup outcomes and the representation of social groups in local news.

The U.S. total fertility rate fell during the Great Recession and did not return to its pre-recession level. We examine how this decline varied with the local severity of the recession. Using the universe of U.S. births from restricted-use NCHS Vital Statistics Natality files and cross-commuting-zone variation in the 2007–2009 unemployment rate change, we find that a one-percentage-point larger unemployment shock reduced the total fertility rate by 0.015 children per woman during the recession and by 0.023 during 2010–2019 relative to 2006, with no recovery a decade after the recession ended. The losses were concentrated in the hardest-hit labor markets and among less-educated and married women. First births largely recovered, while second- and higher-order births remained depressed, and the missing births fell disproportionately on intended births, in contrast to the decline in unintended births that dominates the national time series. The persistent losses are thus concentrated among families that had already begun childbearing. These findings suggest that the demographic costs of recessions are geographically concentrated and fall on continued childbearing rather than entry into parenthood.

South Korea experienced an exceptional decline in fertility, from among the highest rates in the world to one of the lowest today. The decline coincided with an aggressive government family-planning campaign and overlapped with Korea’s remarkable economic growth. We study the long-run consequences of this demographic transition using a calibrated overlapping-generations model in which government policy shifts fertility norms, while fertility affects human-capital investment, labor supply, and child-rearing costs. Lower fertility initially creates a demographic dividend by reducing dependency burdens and increasing investment per child. These forces also reshape social norms and future labor supply, allowing low fertility to persist after the policy ends. Over time, however, smaller working-age cohorts increase the burden of supporting a largely elderly population. We quantify the historical gains from Korea’s demographic window and compare them with the emerging and projected costs of demographic drag.

This chapter reviews the literature on the causal effects of policies on fertility. It focuses on evidence from experiments and quasi-experiments in low fertility contexts, including studies from Europe, Northern America, Oceania, and Asia. Making no a priori restrictions on policy type, the review encompasses evaluations of parental leave, childcare, health insurance, and financial incentives such as child transfers. Childcare expansions increase completed fertility. Financial incentives had positive effects on fertility across contexts, both in the short and long run. Expansions of parental leave rights in Central Europe, and introduction of parental leave in the USA, also had positive effects. Distributional effects of these policies are very different, with parental leave compensation benefiting high-earning couples, while expansions of childcare programs have potential to reduce social inequalities. Publicly funded assisted reproduction can increase fertility for women over the age of 35.