How Risk Adjustment Can Backfire

We show that firms reduce selection along dimensions included in the risk-adjustment formula, while increasing selection along excluded dimensions. These responses can actually increase differential payments relative to pre-risk-adjustment levels and thus risk adjustment can raise the total cost to the government of providing the public service. We confirm both selection predictions using individual-level data from Medicare, which in 2004 began risk-adjusting payments to private Medicare Advantage plans. We find that differential payments actually rise after risk adjustment and estimate that they totaled $30 billion in 2006, or nearly eight percent of total Medicare spending.

Full article on how risk adjustment affects selection and differential payments.

Comments (9)

Trackback URL | Comments RSS Feed

  1. Paul H. says:

    Good point. And more important than most people understand.

  2. Devon Herrick says:

    I’ve heard it explained by saying the government uses fewer variables to calculate risk-adjusted payments. By contrast, insurers who are in the business of providing coverage, use far more variables and a more sophisticated algorithm. If the government does a sloppy job of calculating risk-adjusted payments, it runs the risk of enriching insurers unnecessarily; or creating an environment where few insurers want to participate if payments are too low.

  3. Tom H. says:

    This is bad news for the health insurance exchanges.

  4. Greg says:

    I agree with Paul. Imperfect risk adjustment can make things worse than if there had been no risk adjustment to begin with.

  5. Stephen C. says:

    I think this is a very important discovey. Especially considering that all risk adjustment is imperfect.

  6. I had suspected that this was the case, and discussed it as a drawback in a paper I wrote about Medicare Advantage a while back ( As I suggested then, if we don’t think government is competent to determine the prices of individual medical goods and services, why would we think that government is competent to determine the “price” of a patient’s aggregate medical condition?

    I don’t think that this problem can be solved by exchanges or any other government dictated method. However, if the government allowed individually owned, incentive compatible, guaranteed renewable policies (e.g. as proposed by Prof. Mark Pauly and separately by Prof. John Cochrane) then the market would price medical conditions.

  7. Linda Gorman says:

    Private insurers have typically rated on the basis of experience. Government programs have gone to academics for models of “risk adjustment.

  8. Jeff says:

    I definitely agree that risk adjustment is always imperfect. I don’t agree that it always makes things worse than non-risk-adjusted premiums.

  9. Bob Hertz says:

    I believe that in the German and Swiss model of risk adjustment has two huge differences from what we have seen so far in Medicare Advantage:

    1. The adjustments are retrospective based on claims, rather than prospective based on insurer profiling, which can be fudged;

    2. The risk adjustment money does not come from the government, it comes from the more profitable sickness funds.

    This implies to me that if premium support is actually going to save money, the insurance industry will be regulated more than ever. Wyden and Ryan kind of left this out of their proposals.