Enrollment Brokers Did Not Increase Medicaid Enrollment, 2008-2018, with Becky Staiger, Diane Alexander, and Molly Schnell. Health Affairs. 2022.

Between 2008 and 2018, six states and Washington, D.C., began contracting with enrollment brokers to facilitate enrollment into Medicaid, joining the eighteen states that already had such contracts in place as of 2008. Using newly collected data covering all contracts between state Medicaid agencies and independent enrollment brokers during this period, we compared changes in Medicaid participation following the initiation of contracts with enrollment brokers with contemporaneous changes in Medicaid participation in states that never contracted with brokers. We find that contract initiation had no statistically significant effects on state-level Medicaid participation. We further find no evidence of other enrollment-related benefits, such as improved application processing times.

Preparing for the Work of the Future: A Research Agenda, with David Autor and Matthew Notowidigdo. Prepared for J-PAL North America’s “Work of the Future” Initiative, 2019.

Breakthroughs in artificial intelligence and robotics are rapidly changing the terms of comparative advantage between humans and machines in the workplace, with potentially profound implications for labor market operation and for the opportunities available to human workers of different skill and education levels. This overview paper discusses four areas where research is needed to identify strategies for managing these changes in ways that benefit workers, firms, and the broader economy: (1) skills training, (2) postsecondary education, (3) alternative work arrangements and the “gig economy,” and (4) management practices. In each area, we highlight key findings from recent studies and discuss opportunities for further innovative experimentation to evaluate program and policy options, pilot novel interventions, and help workers, firms, and governments prepare for the “Work of the Future.”

Work-in-Progress

The Welfare Effects of Intermediation in the Employer-sponsored Insurance Market, with Tong Liu, Anthony LoSasso, and Nicholas Tilipman.

We examine the welfare effects of insurance brokers in the small-group employer-sponsored health insurance market.  Almost all employers work with insurance brokers to choose products, affecting 54% of the US population. However, there are limited empirical studies on how brokers affect insurance access, choice quality, or prices. We collect novel data on broker-insurer contracting networks, commission schedules, and employers’ broker and insurer choices. We find that brokers reduce search costs and increase insurance offering rates of small-group employers. We also find exclusive broker-insurer contracts soften the upstream price competition of insurers, and commissions distort brokers’ product recommendations, generating an agency problem for employers.  To quantify the tradeoff between access, price, and quality, we develop an equilibrium model of individuals’ demand for insurance products, employers’ choice of brokers and insurers, brokers’ product recommendations, and insurers’ pricing and commission decisions. We plan to simulate the welfare effects of policies, including a ban on commission payments or exclusive contracts between insurers and brokers.

Information Frictions in Physician Referral Markets, with Linda Ouyang.

We examine whether primary care physicians (PCPs) respond to information disclosures on specialists’ quality.  More than a third of patients in the US are referred to a specialist each year, but there are limited studies on the determinants of physician referral patterns. We employ the exogenous information shock of state-level publication of malpractice records in an event-study design. Preliminary analysis shows a modest decrease in specialists’ services provisions following the disclosure of malpractice events. To further disentangle the roles of preferences and information in shaping PCPs’ referral patterns, we develop a discrete choice model of specialist choice with Bayesian beliefs of specialists’ quality, exploiting the variations in malpractice labels and specialists’ underlying value-added measured with Medicare claims records. We will compare referral and specialist choice patterns for shoppable and un-shoppable medical services (e.g., imaging or coronary artery bypass grafting surgery, separately), to inform whether the provision of specialist quality information should be targeted to patients or PCPs. We plan to simulate the effects of regulations on referral networks and information disclosure.

Elections and Mental Health, with Janet Currie and Molly Schnell.

We examine the effects of a shift in political power on mental health, exploiting the 2008, 2012, and 2016 presidential elections. Exploiting an event-study design, we find an increase in anti-anxiety and anti-depressant prescriptions in counties that support the losing party. In contrast, the number of prescriptions stays unchanged in counties that support the winning party. These results indicate that political leadership, or social factors apart from medical care, could shape population health in powerful ways.