• When politics meets econometrics

    From today’s Wonkbook preamble:

    Obamacare’s first year will be full of glitches and hiccups and mistakes and misfires. That’s true for every big, complicated law (remember when Boehner called Medicare Part D’s implementation “horrendous”?). The expectation is that Republicans will be able to take advantage of those problems. But if the early implementation comes in the context of an extended government shutdown, Republicans might well get blamed for implementation glitches as the media and voters ask whether the law wouldn’t work more smoothly if the GOP hadn’t turned out the lights.

    The political squabble and implications are interesting (or infuriating) enough. But that’s not why I’m posting. I’m posting because we should absolutely expect implementation glitches, hiccups, mistakes, and misfires to (a) vary by state and (b) depend, in part, on the type of political squabbling that keeps Obamacare at the top of Wonkbook.

    Of course organizational and financial support matter for implementation. The whole point of encouraging or withholding (obstructing) them is to affect implementation. However you feel about these tactics, you probably believe — know — they’ll work to some extent. You’re right. They will.

    Moreover, the degree of support for the law varies by state. Consequently, enrollment will vary, as will the relative mix of enrollment by the healthy vs. sick. Healthier and younger people are, famously, the target of both pro- and anti-Obamacare outreach, after all. Thus, premiums, because they depend on risk selection, which depends on enrollment, will also vary due to institutional and political support for the law, among other reasons.

    And that brings me to my main point. Researchers who, in years to come, study consumers’ response to the new insurance marketplaces had better build into their econometric models some controls for degree of support for the program. If they don’t, those results will be biased.

    Without such controls, one might misattribute the degree to which relatively higher premiums are the reason consumers are not enrolling in this state as much as that one. Premiums won’t be the full cause. They’ll be driven by selection, which will be driven, in part, by support for the law, which is a deeply, passionately political issue. In other words, people won’t be deterred from enrolling due to (or only because of) higher premiums. Higher premiums will be partly a result of active efforts at deterring enrollment. Conversely, they’ll be lower where active efforts at encouraging it succeed.

    I’m telling you now, if in a few years you review a paper on the functioning of insurance marketplaces in the Obamacare era and you do not see controls for degree of implementation support, you should reject it. (OK, that’s harsh, but you should at least ask for a revision.)

    (Related post on similar issues and research with respect to Part D here.)

    @afrakt

    Share
    Comments closed
     
    • Perhaps more to the point, such state differences are something econometricians will exploit to say more about the effects of the ACA.

    • What are some possible controls that would be informative & have some degree of validity? Is there anyway to quantify degree of acceptance of ACA?

      Dummy codes for states or party governership?
      State spending per capita on implementation or marketing?
      Amount spent on disinformation?