Health insurers increasingly compete on their networks of medical providers. Using data from Massachusetts’s insurance exchange, I find substantial adverse selection against plans covering the most prestigious and expensive “star” hospitals. I highlight a theoretically distinct selection channel: consumers loyal to star hospitals incur high spending, conditional on their medical state, because they use these hospitals’ expensive care. This implies heterogeneity in consumers’ incremental costs of gaining access to star hospitals, posing a challenge for standard selection policies. Along with selection on unobserved sickness, I find this creates strong incentives to exclude star hospitals, even with risk adjustment in place. (JEL D82, G22, H75, I11, I13, I18)

Are application hassles, or “ordeals,” an effective way to limit public program enrollment? We provide new evidence by studying (removal of) an auto-enrollment policy for health insurance, adding an extra step to enroll. This minor ordeal has a major impact, reducing enrollment by 33 percent and differentially excluding young, healthy, and economically disadvantaged people. Using a simple model, we show adverse selection—a classic feature of insurance markets—undermines ordeals’ standard rationale of excluding low-value individuals since they are also low-cost and may not be inefficient. Our analysis illustrates why ordeals targeting is unlikely to work well in selection markets. (JEL D82, G22, H75, I13, I18)

How much are low- income individuals willing to pay for health insurance, and what are the implications for insurance markets? Using administrative data from Massachusetts’ subsidized insurance exchange, we exploit discontinuities in the subsidy schedule to estimate willingness to pay and costs of insurance among low- income adults. As subsidies decline, insurance take- up falls rapidly, dropping about 25 percent for each $40 increase in monthly enrollee premiums. Marginal enrollees tend to be lower- cost, indicating adverse selection into insurance. But across the entire distribution we can observe (approximately the bottom 70 percent of the willingness to pay distribution) enrollees’ willingness to pay is always less than half of their own expected costs that they impose on the insurer. As a result, we estimate that take- up will be highly incomplete even with generous subsidies. If enrollee premiums were 25 percent of insurers’ average costs, at most half of potential enrollees would buy insurance; even premiums subsidized to 10 percent of average costs would still leave at least 20 percent uninsured. We briefly consider potential explanations for these findings and their normative implications.

Insurance markets often feature consumer sorting along both an extensive margin (whether to buy) and an intensive margin (which plan to buy). We present a new graphical theoretical framework that extends a workhorse model to incorporate both selection margins simultaneously. A key insight from our framework is that policies aimed at addressing one margin of selection often involve an economically meaningful trade-off on the other margin in terms of prices, enrollment, and welfare. Using data from Massachusetts, we illustrate these trade-offs in an empirical sufficient statistics approach that is tightly linked to the graphical framework we develop.