Optimal Risk Adjustment in Markets with Adverse Selection: An Application to Managed Care

It is well known that adverse selection causes distortions in contracts in markets with asymmetric information. Taxing inefficient contracts and subsidizing the efficient ones can improve market outcomes (Bruce C. Greenwald and Joseph E. Stiglitz, 1986), although regulators rarely seem to implement tax and subsidy schemes with adverse-selection motives in mind. Contracts are often complex and “incomplete,” and it is the “inefficient” elements of the contract that are difficult to verify and hence tax or subsidize. This is precisely the reason that in health insurance markets, rather than subsidizing contracts, regulators and payers contend with adverse selection by taxing and subsidizing the price paid to insuring health plans on the basis of observable characteristics of the persons joining the plan—a practice known as “risk adjustment.” Risk-adjusted premiums are paid to “managed-care” plans—plans that ration care by management, rather than by conventional approaches like coinsurance and deductibles, and offer a bundle of characteristics (quality, access for many services) that is fundamentally outside the scope of direct regulation. Selection-related incentives threaten the efficiency and fairness of this organization of health insurance markets by inducing plans to distort the quality of the services they offer to discourage high-cost persons from joining the plan. As managed care becomes the predominant source of health care for residents of the United States and many other countries, payers attempt to address this incentive by setting a risk-adjusted price that pays more for more-expensive enrollees. As it is conventionally practiced, risk adjustment sets prices for people proportional to their expected cost based on observable characteristics. The federal Medicare program, for example, has used age, sex, welfare status, and county-of-residence adjusters to set prices to managed-care plans. To convey how what we term “conventional” risk adjustment works, suppose age is the risk adjuster for a Medicare population over 65. If it is determined that the 75to 84-year-old population costs 20 percent more than the overall average in Medicare, the assumption in conventional risk adjustment is that the premium paid to plans for someone in this group should be 20 percent above the average. We fundamentally disagree that this is the right way to think about and do risk adjustment.