Insurance without Commitment: Evidence from the ACA Marketplaces (with Rebecca Diamond, Tim McQuade, Petra Persson)

Abstract: We study the dynamics of participation and health care consumption in the Affordable Care Act’s health insurance marketplaces. Unlike other health insurance contexts, we find individuals commonly drop coverage midyear; roughly 30% of enrollees exit within nine months of sign-up. These dropouts spend more on health care while covered than in the months before sign-up or after exit. We model the consequences of drop-out on equilibrium premiums and consumer welfare. While dropouts generate a type of adverse selection, the welfare effect from their participation is ambiguous and depends on the relative spending per month of part-year vs. full-year enrollees. Using our empirical model, we quantify changes in premiums and welfare after the imposition of penalties targeting drop-out. We find that overall welfare declines with a ban on drop-out: young and healthy consumers---those who can more easily re-time their health spending, as well as those who value the option to exit---choose to forego coverage entirely, leading to higher average costs among the insured population and thus higher premiums.

WORKING PAPER

NBER Working Paper No. 24668

Incorporating Wait Times in Health Insurance Design (with Pierre Bodéré and Guillaume Fréchette)

Abstract: Wait time is a key factor of health-care services, yet data is rare, often unreliable, and inconsistently measured across systems. The scarcity of data prevents systematic analyses of allocative inefficiency and inequality in access to health. We propose a measure of wait times - detection to treatment (DTT) - solely based on medical variables, which are both widely available and standardized. DTT records the time elapsed between the detection of a patient as being high-risk of receiving a surgery, and the date of the procedure. We use recurrent neural networks to represent patients’ high-dimensional medical trajectories as a risk profile over time. As expected for a measure of wait times, we find that DTT increases with supply constraints. Patients enrolled in more restrictive insurance plans experience longer DTT and an increase in the load of medical providers increases the wait time. Using provider loads as exogenous variation in wait times, we show that an increase in DTT results in higher medical expenditures, longer hospitalization, and increased use of addictive drugs.

Efficient Provision of Experience Goods: Evidence from Antidepressant Choice

Abstract: In the market for medical care, physicians often face uncertainty about how a newly diagnosed patient will respond to available treatments. I design a framework to analyze how price and promotion influence the learning process as the patient and physician jointly search for the most effective treatment. The dynamic model I employ accommodates large choice sets and permits learning to be correlated within clusters of choices. Applying this model to depression care, I ask how the design of a health insurance plan, including the required patient out-of-pocket costs by drug, might interact with the physician’s learning process. In the data, patient costs largely correspond to the drug’s wholesale cost. In contrast, I design a new drug pricing schedule that lowers the patient cost for those drugs that the model suggests are best to sample early in the search process. By using these price incentives to redirect the search process, I find physicians identify the optimal treatment faster, leading to lower overall costs, improved adherence, and ultimately better patient health.

DOWNLOAD

Physician vs. Patient Incentives in Prescription Drug Choice

Revise and resubmit, American Economic Journal: Economic Policy

Abstract: In response to rising health spending, public and private insurers use two mechanisms to direct spending toward more valuable treatments: “demand-side” incentives, which impose costs on the patient to limit moral hazard, and “supply-side” incentives, which adjust the physician’s compensation to discourage spending. Using variation in patients’ and physicians’ exposure to incentives, I identify important differences in cost and health outcomes under these two mechanisms. Demand-side cost-sharing discourages both initial treatment and later adherence. Payment reforms drive physicians to substitute drug care and specialist referrals for office visits. I discuss the implications of these outcomes for optimal insurance design.

DOWNLOAD

The Role of Experience in Physician Treatment Decisions: Evidence from the Introduction of Medicare Part D (with Marissa King and Tanja Saxell)

Abstract: How do physicians learn about new treatments? Using the setting of antipsychotic treatment choice, we measure the relative importance of two key pathways: observational learning, in which physicians update their knowledge from public signals, and learning-by-doing, in which the physician relies on her own private experience treating patients. To do so, we exploit two sources of exogenous shocks to physicians’ information. First, in 2007, regulators issued new guidance in the antipsychotic market, approving one drug as a secondary treatment for depression and warning that another posed a substantial risk of side effects. Second, in 2006, the introduction of Medicare Part D shocked the typical physician’s patient composition, with more patients obtaining private insurance coverage. Examining the time periods surrounding the drug advisories, we find physicians with greater patient volume and with more specialized training learn about product quality sooner. Public warnings primarily affect the decisions of the least experienced and least specialized physicians. Importantly, among physicians seeing few patients, recent graduates react more quickly and robustly to the advisories following their publication. We further show that exploiting variation in experience stemming from Medi- care’s insurance expansion is necessary to distinguish the effect of volume from unobserved factors, such as physician quality.

Accounting for Structural and Measurement Error in Binary Choice Problems: A Moment Inequality Approach (with Eduardo Morales)

Abstract: Many economic decisions involve a binary choice - for example, when consumers decide to purchase a good or when firms decide to enter a new market. In such settings, agents’ choices often depend on imperfect expectations of the future payoffs from their decision (expectational error) as well as factors that the econometrician does not observe (structural error). In this paper, we show that expectational error, under an assumption of rational expectations, is a source of classical measurement error, and we propose a novel moment inequality estimator that accounts for both expectational error and structural error in a binary choice model. With simulated data and Chilean firm-level customs data, we illustrate the identifying power of our inequalities and show the biases that arise when one ignores either source of error. We use the customs data to estimate the fixed costs exporters face when entering a new market.

DOWNLOAD
Online Appendix
Matlab code for simulation exercise