Research

Working Papers

Evaluation of Targeted Subsidies and Their Impact on Neighborhoods (Job Market Paper)

Residing in low-poverty neighborhoods is suggested to be beneficial for a household's well-being. However, rental assistance recipients are currently disproportionately located in high-poverty neighborhoods and it is unclear whether efforts to promote their relocation will be successful. In addition, providing access to rental assistance recipients in relatively affluent neighborhoods raises concerns about inducing adverse effects on neighborhoods. Focusing on a reform to the largest rental assistance program in the US that increased the share of a neighborhood's rental stock that can be subsidized in high-rent neighborhoods with a commensurate decrease in low-rent neighborhoods, this paper analyzes its causal effect on both the number of rental assistance recipients residing in each neighborhood and neighborhood average house prices. Exploiting a novel triple difference design, that controls for trends in high-rent neighborhoods relative to low-rent neighborhoods and changes in these trends across treatment periods, this paper finds that changing this share significantly influences the number of rental assistance recipients within a neighborhood, with considerable heterogeneity across broad geographical areas. An event study analysis shows this effect takes place only after implementation, peaks $2$ years later, and remains significant after $4$ years. Furthermore, this paper finds no effect on average neighborhood house prices. Finally, synthetic control methods confirm the main conclusions and further suggest the policy reform would be equally effective in areas not implemented.

Local Labor Market Composition Changes and Their Effect on Major Choice

This paper seeks to give a better understanding about the relationship between the major choice process and the labor market structure by examining the responsiveness of major specific completions to the share of the local labor market that is comprised by the graduates of a given major. The dynamic effect is identified by exploiting quasi-random variation that is constructed using shift-share instruments which leverage both changes in the occupational composition and the composition of majors within occupations at the national level. Data on major completions for the time period of 2009 ¡ 2018 are obtained from the Integrated Post-secondary Education Data System (IPEDS) of the National Center for Education Statistics (NCES) whereas the major shares of the local labor markets are constructed using the American Community Survey (ACS) for the same time period. The results of the estimation show that students are in deed responsive to changes in the local labor market with the period between freshman and sophomore years having the biggest effect. A decomposition of the effect shows that most of the effect is comprised of an over-response to the changing occupational composition and a moderate response to changes coming from the major composition within occupations.

Occupational Sorting with Heterogeneous Employer Learning Rates

The employer learning process is an important aspect of wage dynamics and is more pronounced at the beginning of one’s career. Due to asymmetric information, individuals who are more (less) productive than what their observables reveal, will be underpaid (overpaid) at the beginning of their careers. In this paper, I develop a two-period, two-occupation model which analyzes the occupational decisions in an asymmetric information game where workers differ on whether their productivity was underestimated or overestimated, which is called productivity assessment error, and occupations are heterogeneous with respect to the extent at which output is associated with a worker’s productivity assessment error, which is argued to be correlated with the employer learning rates. Two predictions are derived from the model. First, workers with positive (negative) productivity assessment errors are more likely to sort into occupations where employer learning is faster (slower), and second, the incentives to sort in this manner are diminishing with experience. I use data from the National Longitudinal Survey of Youth 1979 (NLSY79) and Current Population Survey (CPS) Merged Outgoing Rotation Group(MORG) to directly test these two predictions separately for workers who have less than college degree education and for college graduates. The findings from the former educational group align with the two predictions, whereas the findings from the latter do not.