The Consequences of Risk Adjustment in the Medicare Advantage Program

The Medicare Advantage program both increased total Medicare spending and transferred Medicare resources from the relatively sick to the relatively healthy.

Since the 1980s, people eligible for Medicare have been able to choose between the regular fee-for-service plan, under which the federal government pays a set fee to health care providers for each service provided, and Medicare Advantage (MA), whereby the government pays private health plans a fee for each individual they enroll. Almost one quarter of Medicare beneficiaries are currently enrolled in Medicare Advantage plans.

Paying the same amount for every person enrolled in a health plan encourages plans to enroll low-cost people and to avoid high-cost ones. Because of this, the federal government historically overpaid for MA enrollees relative to their costs in traditional Medicare. So, in 2004 the Medicare program began to adjust its payments to private plans for enrollees' health status. As a result, a plan would, for example, receive a higher "risk-adjusted" payment for a recipient with diabetes or heart disease than for an otherwise identical person without these conditions.

In How Does Risk Selection Respond to Risk Adjustment? Evidence for the Medicare Advantage Program (NBER Working Paper No. 16977), Jason Brown, Mark Duggan, Ilyana Kuziemko, and William Woolston study individual-level data for 55,000 people in the Medicare Current Beneficiary Survey (MCBS) from the period 1994 to 2006. Prior to risk adjustment, insurers simply had an incentive to enroll individuals with low costs. After risk adjustment, insurers instead had an incentive to enroll individuals with low costs conditional on their medical conditions. The main reason for this is that the risk adjustment formula pays the plans the average cost of the average person in a particular risk category. The authors demonstrate that, because individuals with less costly cases of diabetes and other health conditions enrolled in MA plans after the move to risk adjustment, overpayments to these plans actually increased.

The risk adjustment formula that is used also explains only 11 percent of an individual's fee-for-service costs in the year after risk is assessed. The formula systematically over-predicts costs for those with below average costs, and systematically under-predicts costs for those with above average costs. The authors find that individuals who are more expensive than the average person to insure are less likely to enroll in Medicare Advantage plans. So on balance, the government ends up paying the average cost for people who, had they stayed in fee-for-service Medicare, would have cost the government much less.

Before risk-adjustment began in 2004, switching from fee-for-service Medicare to Medicare Advantage increased average individual Medicare spending by $1,800. The authors calculate that using risk adjustment formulas on the population that enrolled before 2004 would have reduced Medicare Advantage overpayments by more than $800 a person. But when the reimbursement formula changed, so did the pattern of enrollment in Medicare Advantage plans. After 2004, switching from fee-for-service to Medicare Advantage increased Medicare spending by approximately $3,000 per person. Thus the shift to risk adjustment actually increased Medicare spending.

Although Medicare Advantage plans did enroll people with higher "risk scores" after risk adjustment was instituted, those people still tended to be significantly below the average cost in their risk category. Furthermore, both before and after risk adjustment, MA enrollees in poor health expressed greater dissatisfaction with their medical care relative to their counterparts in traditional Medicare. This pattern suggests that MA plans invest more resources in their relatively healthy enrollees, perhaps to differentially retain them. Thus the authors conclude that the Medicare Advantage program both increased total Medicare spending and transferred Medicare resources from the relatively sick to the relatively healthy, and that risk-adjustment was not able to address either of these problems.

--Linda Gorman