Working Papers
Marital Stability and Intrahousehold Inequality with Jacob Penglase and Tomoki Fujii, March 2025
We examine which factors are predictive of unstable marriages using a unique panel data set of Japanese couples. We employ several machine learning and econometric techniques to identify characteristics of the couple pertaining to their consumption allocations, labor supply, savings decisions, and stated satisfaction that are associated with a higher divorce probability. We find time-varying characteristics of the couple, such as the wife's income and labor supply, are most predictive, while characteristics of the couple at the time of marriage are less so. We further show that marriage market conditions are highly predictive of divorce. We relate these findings to the theoretical literature on the drivers of divorce.
Child Penalty and Mothers' Age at First Birth, revised August 2025
I study ``child penalties'' -- the negative impacts of motherhood on women's earnings and labor force participation, accounting for the heterogeneity in mothers' age at first childbirth. By estimating a dynamic and heterogeneous treatment effects model developed by Sun and Abraham (2021) using the NLSY79 data, I find substantial heterogeneity in women's earnings, weeks employed, working hours, and hourly rate after the first childbirth both in the short- and long-term (retirement age). Overall, the child penalty expands over the lifecycle between mothers and childless women. Younger first-time mothers suffer larger child penalty compared to older ones in terms of total earnings. Older ones lose more in terms of labor force participation, but are able to smooth more the birth shock by climbing faster and higher in the occupational rank over the lifecycle by delaying fertility. The results highlight the potential heterogeneity in the effects of maternity leave policies by mothers' age at first birth.
The Demand for Soft Drinks: Evidence from Purchases At-Home and Away-From-Home with Linqi Zhang, revised April 2025
Using a novel dataset that includes at-home and away-from-home food purchases, we study who is affected by soda taxes. We nonparametrically estimate a random coefficient nested logit model to exploit the rich heterogeneity in preferences and price elasticities across households, including SNAP participants and non-SNAP-participant poor. By simulating its impacts, we find that soda taxes are less effective away-from-home while more effective at-home, especially by targeting the total sugar intake of the poor, those with high total dietary sugar, and households without children. Our results suggest that ignoring either segment can lead to biased policy implications.
Publications
Scanner Data, Food Consumption, and SNAP, forthcoming in the Elgar Encyclopedia of Consumption, 2025, edited by José M. Labeaga and José Alberto Molina.
Food Demand and Cash Transfers: A Collective Household Approach with Homescan Data, 2023 Journal of Economic Behavior and Organization.
Identification of Semiparametric Model Coefficients, With an Application to Collective Households with Arthur Lewbel, 2022 Journal of Econometrics.