Publications and Acceptances

Peri, G., Rury, D., and Wiltshire, J.C. 2024. “The Economic Impact of Migrants from Hurricane Maria”, The Journal of Human Resources 59 (6): 1795-1829

Minimum Wage Effects and Monopsony Explanations (with Carl McPherson, Michael Reich, and Denis Sosinskiy)

Working Papers

Walmart Supercenters and Monopsony Power: How a Large, Low-Wage Employer Impacts Local Labor Markets

Abstract: This paper considers the extent and impact of monopsony power exercised by Walmart Supercenters. I address the issue of potential bias from endogenous store entry, as well as other identification concerns, by adopting a stacked synthetic control approach to estimate average county-level labor market effects of the Walmart Supercenter roll-out across the U.S. Crucially, I construct the pools of synthetic control donor counties from novel observations of counties where Walmart tried to open a Supercenter but was blocked by local efforts. I first show Supercenter entry sharply increased labor market concentration. Supercenters were able to hire large numbers of retail workers with zero increase in average earnings, indicating Walmart had wage-setting power. I then show Supercenter entry caused large declines in overall local employment and earnings, particularly among local goods-producers, indicating Walmart displaced manufacturing demand away from local producers and to its own national and international suppliers. In counties with a Supercenter, subsequent exogenous minimum wage increases led to significant growth in aggregate and retail employment. These results run counter to predictions for competitive labor markets, and indicate Walmart Supercenters gradually accumulated and exercised monopsony power, with negative consequences for workers.

allsynth: (Stacked) Synthetic Control Bias-Correction Utilities for Stata

Abstract: Synthetic control methods are widely-used for estimating counterfactuals and treatment effects of policy interventions. allsynth adds greatly-enhanced functionality to the user-written synth module for Stata, which is widely used by practitioners to implement the “classic” synthetic control estimation strategy. allsynth automates implementation of several extensions to the classic approach while retaining the syntax of synth. The enhanced functionality includes automation of a bias-correction procedure that adjusts for differences in the predictor variable values between a treated unit and its synthetic control donors, automation of in-space placebo treatment estimation for randomization inference, and automated synthetic control estimation in environments with many treated units and treatment periods (”stacked” synthetic control estimation). allsynth also provides enhanced automated graphing capability and thorough diagnostics to help users with implementation. allsynth version 1.2 can be installed by typing ssc install allsynth, replace all in Stata's command line.

Lifting the Cap on Non-Resident University Enrollment: Evidence from Wisconsin (with Natalia Orlova and Derek Rury)

Abstract: Non-resident students are often accused of negatively affecting academic quality and crowding out resident students. We present new evidence on this relationship by exploiting the removal of an enrollment cap on non-resident students at a highly ranked state flagship university. We find this policy yielded a 29 percent increase in non-resident enrollment (coming almost entirely from domestic—rather than international—students), and a consequent 47 percent increase in tuition revenue which funded large increases in financial aid disbursed at the university, particularly for low-income resident students. We find no evidence of negative effects on several measures of academic quality or resident-student enrollment.

Crosswalks from ‘County Groups’ to Counties for the 1970 and 1980 U.S. Decennial Census Metro Samples

Abstract: County of residence is not observed in the public-use 1970 and 1980 Decennial Census data, to preserve respondent confidentiality. Instead, the metro 1% samples delineate ‘county groups’ for each year. This prevents researchers from conducting research with the these Decennial Census data at the level of any time-consistent geographic unit smaller than commuting zones—and that is only possible using probabilistic crosswalks written by other researchers (Autor and Dorn, 2013). I provide and detail new population-based crosswalks which allow researchers to probabilistically distribute individuals to their counties-of-residence, enabling research with these data at the county level. To demonstrate the accuracy of the crosswalks, I compare the results to a variety of county-level data from the Census Bureau.

Selected Works in Progress

Playing with Fire: The Unintended Consequences of Logging Restrictions in the Northern Spotted Owl Habitat (with Natalia Orlova)

Health Provider Concentration and Medical Debt (with Alaa Abdelfattah, Sergio Pinto, and Marshall Steinbaum)

The Impact of Walmart’s Growth on the U.S. Social Safety Net

Growing Regional Demand Through Cooperative Competition

Labor Market Power and Non-wage Compensation