I am a Postdoctoral Researcher at the Haas School of Business, University of California, Berkeley.
I am on the 2026–2027 academic job market.
I am an industrial organization economist studying how competition, market structure, and asymmetric information shape firms’ decisions. My research focuses on the interaction between product and labor markets and on pricing and access in insurance markets.
arubinstein [at] berkeley [dot] edu
Review of Economic Studies, forthcoming
Does competition always mitigate the impact of discrimination on racial wage gaps? This paper examines the impact of product market competition on customer-driven taste-based discrimination in a service economy characterized by homophily. I develop a novel equilibrium model of both product and labor markets in which heightened product market competition tends to exacerbate labor market discrimination in customer-facing occupations. Intensified competition compels firms to reduce prices while also prioritizing the preferences of wealthier demographic groups when the service product is a normal good. This results in widening wage and employment gaps between the privileged demographic groups and others in the service sector. I validate the predictions of the model in the context of intensified competition in the US banking sector, triggered by bank deregulation. I find that wage and employment gaps between Black and White workers in the banking sector widened post-deregulation precisely in client-facing occupations and in states with higher levels of racial animosity.
Geographic expansion can intensify competition for consumers while increasing employers’ power over some workers. This paper develops a spatial competition model with multi-branch firms, endogenous entry, and industry-specific workers with limited outside employment opportunities. Technologies that facilitate geographic expansion can intensify product-market competition while inducing exit, increasing employer concentration, and raising wage markdowns. I examine this mechanism using intrastate branching deregulation in U.S. banking, which relaxed geographic restrictions constraining banks’ ability to exploit these technologies. Following deregulation, branch networks expand, the number of banks declines, and deposit and payroll concentration rise. Despite rising concentration, deposit terms improve: estimated deposit markdowns decline by approximately 2.5%. Workers in occupations more dependent on banking experience larger declines in relative wages and banking employment shares. Combining occupational employment and earnings responses with local employer concentration, the structural analysis attributes approximately half of the exposure-related earnings response to a concentration-related wage-setting component. Counterfactual analysis implies that increased wage markdowns for financial managers and specialists following deregulation reduce their earnings by approximately 4%. These findings support a mechanism through which technology-enabled expansion can strengthen competition for consumers while increasing employers’ wage-setting power and reducing the labor share within surviving firms.
This paper studies when insurers respond to differences in expected costs through premiums and when they instead restrict coverage. Using proprietary data from a large commercial auto insurer, including internal renewal recommendations and market quotes from competing insurers, I document that premiums respond only weakly to expected costs, while coverage denials rise sharply with risk. I develop a model of imperfect competition in which demand-driven market power links these two margins of adjustment. When markups are high, insurers absorb cost differences without fully adjusting premiums; as expected costs rise and markups compress, adjustment shifts toward coverage denial. Consistent with the model, accounting for demand-driven markups restores near-unit cost pass-through, while expected profitability predicts whether insurers adjust premiums or restrict coverage. Adverse selection matters precisely where markups are compressed, but it amplifies rather than generates rationing.
with Rebecca L. Goldfinger Chanin and Yair A. Lev
Health inequity describes avoidable differences in health outcomes across patient groups. These differences are often studied as failures of access to clinical resources needed for effective care. Yet access is not the only source of inequity. It may also emerge within care itself, after patients have reached clinicians, received a diagnosis, undergone a procedure, and become eligible for the same evidence-based treatment. We test whether insurance coverage contributes to treatment inequity in acute coronary syndrome patients treated with percutaneous coronary intervention and drug-eluting stent placement. This is a stringent setting because, among medically eligible patients, treatment selection should be insurance blind. Using electronic health record and claims data from 2011-2025, we showed that guideline-recommended prescribing differed systematically by insurance coverage. Among 8,480 medically eligible patients, physicians prescribed guideline-recommended therapy to 70.4% of commercially insured patients, 60.5% of Medicare Advantage patients, and 50.6% of Medicaid patients. At the Medicare eligibility threshold, where coverage shifted sharply from Medicaid to Medicare Advantage, while clinical indications remain similar, guideline-preferred prescribing increased by 15.7–23.0 percentage-points. These findings suggest that emergency prescribing is not insurance blind: coverage continues to shape whether patients receive evidence-based treatment, even after access has been achieved.
Insurance for durable goods is typically modeled as providing complete protection against loss. This paper shows that this benchmark fails when compensation is based on market prices formed in adversely selected secondary markets. Because resale prices reflect the average quality of assets offered for sale rather than the privately known quality of the insured asset, market-value compensation need not restore the original asset. I introduce the lemons gap, the difference between the quality of the lost asset and the expected quality of its replacement. Embedding this replacement-market failure in a standard insurance-demand model, I show that even modest undercompensation can substantially increase willingness to pay for lower deductibles, causing standard approaches to overstate risk aversion. Coverage choices then also reflect privately known replacement losses, weakening the link between coverage and risk. The same mechanism distorts real allocations: because the lemons gap is largest for newer, higher-value assets, households may shift risky use away from safer assets and toward older, less safe ones.
with Eran Rubinstein
Economic theory highlights the central role of transport costs: distance to markets is a major barrier to exchange, and one of the most effective impediments to trade. This paper asks whether distance similarly acted as a barrier in the implementation of mass murder during the Holocaust. I develop a simple framework to conceptualize, identify, and empirically evaluate the causal impact of railway proximity and transport facilities on Jewish mortality following the Nazis’ “Final Solution.” Drawing on microdata from more than 300 Jewish communities across six European countries—covering approximately 1.5 million victims— matched with data on the 1940s European railway network. Railway proximity significantly increased mortality: distance and station access account for about one-third of cross-community variation in survival.