Research Papers

“A Dirichlet Ordered Logit Analysis of Women’s Labor Supply After Childbirth” 

(Job Market Paper)

This paper investigates the effect of job continuity on the supply of labor of new mothers for the five-year period after their first childbirth. Using the National Longitudinal Survey of Youth 1997 (NLSY97), we extend the beta-logistic model of Heckman and Willis (1977) to a higher-dimensional Dirichlet generalized ordered logit (DGOL) model. This approach considers individual-specific state dependence and heterogeneity in employment choices. The DGOL model nearly doubles the predictive accuracy of the standard generalized ordered logit model. The study also highlights the pivotal roles of job continuity and job protection: working full-time in any given period increases the likelihood of continuing full-time work in the next year by 64.8 percent. The findings suggest that a higher job continuity probability within a year after childbirth significantly boosts full-time employment rates by over 10 percent, offsetting potential drawbacks of lower education and older age. Simulated policy counterfactuals indicate that policies targeting a higher rate of return to previous employers may benefit employment.

“The Price of Short-term Housing: A study of Airbnb on 26 regions in the United States” (joint with Wenzhen Lin)

(Under review at Journal of Housing Economics)

This study offers an in-depth analysis of the factors affecting Airbnb listing prices, using a comprehensive dataset that spans 100,275 listings across 26 U.S. regions and 7,695,428 review comments. By integrating text-mining techniques like sentiment analysis and topic modeling with traditional hedonic and quantile regression models, the paper identifies key attributes that resonate with Airbnb users. It confirms the significance of well-known numerical pricing determinants, while shedding light on the impact of positive and negative customer reviews and review topics. Specifically, negative reviews have twice the influence on listing prices as positive reviews, a ratio that increases to fourfold for lower-priced listings. Furthermore, reviews related to the host and location are found to have the most significant impact on prices. Notably, the influence of reviews diminishes for listings situated farther from the city center. The study also uncovers the existence of a lag effect of review, showing that reviews require time to significantly influence listing prices. Employing New York City data, the study tracks changes in review impact during the COVID-19 pandemic and finds a temporary decline followed by a rebound in influence. Overall, the study enriches existing literature on short-term rental pricing by taking into account the content and type of customer reviews, offering a more comprehensive understanding of pricing dynamics in the sharing economy.

“Estimating the Determinants of Return to Work Durations after Childbirth: A Control Function Approach” (joint with Jan Ondrich)

Maternity leaves allow mothers to return to the labor force after giving birth without having to find a new job at a potentially lower wage rate. Contrary to expectations, recent research, such as Bailey et al. (2019), suggests that California's paid leave program has had limited impact on women's labor market outcomes. Traditional labor market models often suffer from bias due to the endogeneity of wages and unobserved variables. Building on the methodological framework set by Petrin and Train (2005, 2006), our paper introduces an innovative duration analysis technique. This approach incorporates instrumental variables via control functions to examine job attachment after childbirth using U.S. data from NLSY97. In addition, We prove the consistency of our estimation. Our empirical findings reveal a that the effect of wages on job attachment and the baseline hazard for leaving the job both increase substantially when we use information about maternity leave as an instrument for missing employment conditions

“Disability and Financial Satisfaction” (joint with Jan Ondrich)

We examine the effect of disability in the family on financial satisfaction of the female head or spouse using the Panel Study of Income Dynamics. This empirical work is based on the theoretical work on the capability approach developed primarily by Amartya Sen and Martha Nussbaum, which emphasizes the presence of capabilities as opposed to actual outcomes. Following the empirical work of Kuklys (2005), we estimate a probit of financial satisfaction on log income, disability variables, and variables representing other capabilities. We find that the loss in financial satisfaction due to disability in the United States is substantially greater than the corresponding loss in Britain. Point estimates suggest that conversion factors for disability reduce effective income by as much as 60 percent for male heads and 70 percent for female heads. These figures are over and above the direct income loss due to reduced labor market opportunity.