Health Care

November 4, 2016
Amy Finkelstein of MIT; Raj Chetty of Stanford University; and Jonathan Gruber of MIT, Organizers

Vilsa Curto, Stanford University; Liran Einav, Stanford University and NBER; Amy Finkelstein; and Jonathan D. Levin and Jay Bhattacharya, Stanford University and NBER

Health Care Spending and Utilization in Public and Private Medicare

Curto, Einav, Finkelstein, Levin, and Bhattacharya compare healthcare spending in public and private Medicare using newly available private claims data from Medicare Advantage (MA) insurers. The researchers find that health care spending is 27 percent lower in MA than for individuals in the same county and same risk score enrolled in public, traditional Medicare (TM). Spending differences between MA and TM are similar across sub-populations of enrollees and subcategories of care. They primarily reflect differences in health care utilization. Average prices for an admission to a given hospital for a given diagnosis are virtually identical in MA and TM. The researchers present evidence consistent with MA employing various types of utilization management and encouraging substitution to relatively less expensive modes of care, such as use of primary care instead of specialists, and outpatient rather than inpatient surgery. Geographic variation in healthcare spending is larger in MA than in TM, although geographic variation in hospital prices is lower in MA than in TM.

Michael D. Frakes, Duke University and NBER, and Jonathan Gruber

Defensive Medicine: Evidence from Military Immunity

Nathaniel Hendren, Harvard University and NBER

Measuring Ex-Ante Welfare in Insurance Markets

Insurance has value by insuring against the realization of risk. Adverse selection occurs when a portion of this risk is already known at the time of contracting. This suggests demand estimates in adversely selected markets tend to understate the ex-ante willingness to pay for insurance. This paper provides new reduced-form methods to infer the ex-ante value of insurance from observed demand and cost curve estimates in markets with adverse selection. The slope of the demand and cost curves measure the quantity of information revealed; by combining with internal or external measures of risk aversion, one obtains ex-ante measures of willingness to pay. Applying the model to existing estimates in health and unemployment insurance contexts, Hendren shows an ex-ante welfare perspective can generate different conclusions about the optimal size of an insurance market, the size and desirability of insurance subsidies, and the desirability of mandates relative to competitive markets. The framework also provides guidance on optimal open-enrollment periods and characterizes when it is optimal to prefer contracting in the presence of adverse selection as opposed to behind the veil of ignorance. In general, policies that are optimal from an ex-ante perspective involve deadweight loss.

Jason Abaluck, Yale University and NBER; Leila Agha, Dartmouth College and NBER; and David C. Chan, Jr, Stanford University and NBER

Discretion and Guidelines: Evidence from Warfarin Administration

Sonia P. Jaffe, Becker Friedman Institute at the University of Chicago, and Mark Shepard, Harvard University and NBER

Price-Linked Subsidies and Health Insurance Markups

Subsidies in the Affordable Care Act exchanges and other health insurance programs depend on prices set by insurers — as prices rise, so do subsidies. Jaffe and Shepard study the economics of these "price-linked" subsidies compared to "fixed subsidies" set independently of market prices. The researchers show that price-linked subsidies encourage higher prices, increasing the markups due to imperfect competition and raising subsidy costs for the government. Using a structural model estimated with administrative data from Massachusetts' health insurance exchange, the researchers find a non-trivial upward price distortion up to 5% of prices. They also simulate the market with only two insurers, and find distortions of 5-10%. Despite this cost, price-linked subsidies do have two advantages when the government is uncertain about health care cost. First, they insure low-income consumers against this cost risk, transferring it to the government. Second, they may indirectly link subsidies to the cost of uncompensated care, which is a key rationale for subsidizing insurance. The researchers analyze these tradeoffs empirically under different cost shocks and find that the losses from higher prices outweigh the uncertainty benefits of price-linked subsidies for reasonable levels of cost uncertainty. They discuss an alternate policy that would eliminate the pricing distortion while maintaining the guaranteed affordability property of price-linking.

Emily Oster, Brown University and NBER

Does Disease Cause Vaccination? Disease Outbreaks and Vaccination Response (NBER Working Paper No. 22464)

Childhood vaccination rates have recently declined in the U.S. due to parental fear of vaccines. Oster tests whether disease outbreaks increase vaccination using a new dataset of county-level disease and vaccination data. A large pertussis outbreak in a county decreases the share of unvaccinated children entering kindergarten by 30% (1.2 percentage points). These responses do not reflect changes in the future disease risk. Oster argues these facts may reflect a model in which perceived risk of disease is influenced by whether a household is aware of any cases of disease. This suggests better promotion of outbreaks could enhance the response.

NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us