How Well Does Risk Adjustment Work?

Medicare uses an elaborate risk adjustment model incorporating individual demographic characteristics and medical conditions into 70 different hierarchical condition categories in order to decide how much to pay Medicare Advantage plans for each individual enrollee. The goal is to predict each enrollee’s medical expenses as accurately as possible and to pay a premium that reflects that expectation. These models will also be used in the new ObamaCare exchanges in order to “tax” plans with healthier enrollees and subsidize those with “sicker” enrollees. Private insurers have historically set their premiums using experience rating rather than risk adjustment.

So how well does risk adjustment work? According to the June 2012 MedPAC Report, the current model is a “much better predictor of a beneficiary’s costliness” than the previous model. The previous model explained “only about 1 percent” of an individual’s costliness. The current model predicts “about 11 percent.”

Comments (10)

Trackback URL | Comments RSS Feed

  1. Alex says:

    That’s so startlingly accurate I can hardly beleive it. [/sarcasm]

    Thanks for the post Linda!

  2. david says:

    10% may not seem like a lot, but a .1 jump in a coefficient of determination is huge, especially given how much money we’re dealing with.

    Linda also doesn’t say how well “experience” is working for private insurers; there’s certainly nothing scientific about experience and she doesn’t show the private sector is doing a better job, but seems to be making that assumption and suggesting the public sector follow suit.

    If using 70 factors only returns 11% explanatory power, all this proves is that expected health risk is too volatile to determine.

  3. Devon Herrick says:

    Look at the bright side: they’ve improve their risk-rating model 1000%! However, I am a little surprised the model only explains 11% of differences in expenditures.

  4. Plebian says:

    (somewhere, a bureaucrat-researcher is furiously drafting a new model that will predict 14 percent of an individual’s costliness)

  5. Linda Gorman says:

    If expected health risk is “too volatile to determine,” how do companies that write health insurance, and have to predict future payouts that are based in part on claims generated by individual health, stay in business?

  6. david says:

    Apparently from experience, but I have no idea what that means… maybe you don’t either? That’s certainly not “advantage private insurers.”

  7. Bob Hertz says:

    I do not have technical expertise, but I believe that the countries which have full-blown risk adjustment operate retrospectively.

    In other words, the insurers or sickness funds who make a profit at the end of year due to lower claims are forced to share most of that profit with the carriers that have lost money.

    No insurer makes any money on basic coverage. Their only profit comes from supplemental policies.

    If America did this, stockholders would desert the health insurance industry. Huge salaries and option grants for CEO’s would disappear.

    Maybe not the end of the world.

    But Linda is correct, our timid versions so far of risk adjustment will make very little impact.

    If the history of Medicare Advantage is any indication, our government will essentially have to throw money at private insurers to keep them in the Exchanges at all.

  8. Ambrose Lee says:

    @Linda, they stay in business because they don’t have to determine expected health risk (or at least, they don’t have to determine all of it). That’s why premiums and deductibles are set to offer a substantial cushion for the policy writer. If they think I’ll incur $1,000 in costs this year, I’ll probably be saddled with a $1,200 premium, just in case. Of course, the more accurately they can predict my health consumption, the more risk they can eliminate, allowing them to undercut competitors by lowering my rates. This is why some companies have acturial tables with hundreds (not 70) factors.

  9. John R. Graham says:

    Linda Gorman has written some good posts critical of Medicare Advantage’s risk adjustment. As a fan of Medicare Advantage, they dishearten me. However, I’m still convinced that Medicare Advantage is superior to fee-for-service Medicare.

    Perhaps one problem is that Medicare Advantage contracts last for only one year. I’ve recommended blending the best characteristics of Medicare Advantage with the best of Medigap. I.e. underwriting upon initial enrolment, and then guaranteed renewability like Prof. Cochrane or Prof. Pauly recommend. If a senior changes health status, the “health-status” insurance kicks in (as proposed by Prof. Cochrane), and we resolve the problem of risk adjustment.

  10. Bob Hertz says:

    John Graham is instructive as always, but he kind of dodges my point.

    Prospective risk adjustment can be easily manipulated by insurers, and from my knowledge is being manipulated today.

    Whereas Retrospective adjustment strips out all the profits from covering low risk populations. Health insurance profits disappear. Carriers make enough money to pay decent salaries. Stockholders and CEO’s get nothing basically.

    bring it on, I say!

    Bob Hertz
    The Health Care Crusade