Junk Social Science in the Medical Bankruptcy Debate

Josh Wright —  26 July 2007

My GMU colleague Todd Zywicki and Gail Heriot (USD) have an op-ed in the Washington Times exposing Harvard Professors David Himmelstein and Elizabeth Warren’s study on medical debt and bankruptcy, presented to Congress earlier this week, as “one of the most misleading pieces of research ever placed before Congress — no small dishonor.” 

The punchline of the study is that over 50% of bankruptcies have a medical cause, a conclusion that Zywicki and Heriot say is reached only because the researchers have a fairly odd definition of “medical cause” which includes uncontrolled gambling, drug or alcohol addiction, the birth or adoption of a child, or $1,000 or more in out-of-pocket medical expenses over the two years prior to bankruptcy.  The study has been criticized extensively elsewhere (see, e.g. here). 

Maybe Congress isn’t interested in discovering the “true” causal relationship between medical debt and bankruptcy and is simply looking for a study that is consistent with its priors.  I don’t know.  But I don’t expect much more from a Congress that is apparently willing to pass price gouging legislation in the face of all theory and empirical evidence suggesting that this is a horrible idea.  One potential solution to preventing junk social science from influencing policy decisions is to allow for an adversial process in presenting the data so that spurious correlations can be brought to the surface.  Apparently, this solution was undone by some political manuevering.  From Gail Heriot’s post at the Right Coast:

The agenda as originally prepared called for Donna Smith, who was featured in Michael Moore’s Sicko, to testify first, followed by the experts witnesses on both sides, so that the witnesses invited by the minority would have a chance to respond to the study co-authors.  Minutes before the hearing began, the order of witnesses was re-arranged, so that Zywicki & and Clifford J. White III, Director of the Executive Office of United States Trustees, the other witness invited by the minority, would directly follow Ms. Smith’s emotional testimony.  The co-authors of the study, who were invited by the majority would both go later and thus be unrebutted. 

For TOTM readers interesting in seeing the primary sources themselves: Professor Warren’s testimony is available here, the study is available here, and Professor Zywicki’s testimony is available here.

12 responses to Junk Social Science in the Medical Bankruptcy Debate

    market failure, right here 30 August 2007 at 6:54 am

    I give you points for consistency, but serious demerits for a complete lack of a soul. Are you really saying that you place a higher value on a supposedly “truly free market,” than in protecting childrens’ lives? That’s cold.


    Mr. Market Failure —

    since no one else responded to your question I will: the answer is clearly yes. “the amount of bankruptcy-due-to-debt might actually be a lot lower than it is today” – and then again, it “might actually” be a lot higher. and since the primary concern is for a “truly free market,” one is entitled to be agnostic as to the consequences for children of poverty. That is their problem — we are just concerned with getting ourselves to a truly free market.

    market failure, right here 6 August 2007 at 11:23 am

    Hodak–Are you saying that if a “market economy” reaches an equilibrium where children of poverty die of easily and cheaply preventable causes at a certain rate, then that rate is not only acceptable, but preferrable? The “invisible hand” determines who lives and who dies? The rich receive multiple organ transplants and sex meds, while the poor can’t find innoculations for polio?

    It’s time to stop thinking about healthcare as a consumption good like Josh’s power sucking bigscreen TV.


    “What is the optimal level of bankruptcy-due-to-medical-debt in a market economy? Could it ever get to a level that would lead to concern among those who oppose universally guaranteed coverage?”

    Two questions:

    The answer to the first is, obviously: whatever level of bankruptcy-due-to-debt level you would observe in a market economy. Unfortunately, we don’t have anything approaching a market economy in the highly regulated health insurance and health care markets, so we can’t know what that level actually is. One should allow the possibility that in a truly free market, the amount of bankruptcy-due-to-debt might actually be a lot lower than it is today.

    The second question is flawed in its framing, as partly exposed by Ryan. This question implicitly presupposes that universal guaranteed coverage would substantially eliminate the excess costs of care for all individuals such that bankruptcy would no longer pose a significant risk for a large number of people. This premise, however, is highly doubtful, even aside from ignoring the implicit trade-off between universal coverage, actual access to health care, and innovation in treatments.


    > what is the optimal level of
    > bankruptcy-due-to-medical-debt in a market
    > economy? Could it ever get to a level that would
    > lead to concern among those who oppose universally
    > guaranteed coverage?

    Well, you might have the answer to that question, if those who presented the study had resisted the temptation to obviously and shamelessly inflate the results.

    A counterquestion: could taxation levels and expanding queues ever get to a level that would lead to concern among those who propose socializing the healthcare system?

    Just askin’, is all…


    Regardless of the merits of particular studies, one interesting question this raises is: what is the optimal level of bankruptcy-due-to-medical-debt in a market economy? Could it ever get to a level that would lead to concern among those who oppose universally guaranteed coverage?

    This page provides several reports on the problems that medical debt poses.

    Also, check out this study on how even small levels of medical debt can cause a “downward spiral” for low-income individuals.


    My thoughts on this dust-up, extracted from my post on the Knowledge Mosaic Blogwatch today:

    If one reads the Harvard study, it is clear that the claims of Zywicki and Heriot are malicious and unfair. The Harvard study’s authors come clean about the significant limitations of the survey data they are using. They draw only the most general, tentative, and circumspect conclusions. Their policy prescriptions – paid disability and sick leave, stopgap insurance coverage between jobs, and catastrophic insurance coverage – are sufficiently bromidic to leave one wondering what all the fuss is about.

    Regarding the idea that uncontrolled gambling is a major and legitimate cause of medical bankruptcy, the authors merely state that uncontrolled gambling is defined as a psychiatric disorder and that it contributes (not causes, mind you) to 1 percent of bankruptcies. They admit that out-of-pocket expenses for debtors were often below levels typically considered catastrophic, and attribute causal impact more to concomitant income loss or the ongoing need for costly medical care over a period years. Medical causes of bankruptcy related to childbirth-related conditions that led, among other things, to one parent leaving their job to stay home with a child. Hardly a “lifestyle” decision.


    Ted: Good question. I’ll defer to anybody else who has more experience with this than I. Perhaps because the rebuttal is already in the written testimony, it is not worth it for the minority members to post additional questions now — they can just read the record.


    From my experience in testifying before a Congressional subcommittee, I don’t understand why the speaking order matters. Why couldn’t a minority member use the questioning period to ask Zywicki and Heriot whether they had any problems with the Warren study, and thus permit rebuttal then? Indeed, why can’t a minority member *now* pose additional questions to Zywicki and Heriot in writing, and give them a chance to rebut the Warren claims in the record?


    Why should we be surprised at this? In light of Gonzalez v. Carhart and the Court’s overwhelming deference to Congressional fact finding, I think we can expect a lot more of this in the future.

Trackbacks and Pingbacks:

  1. A Stitch in Haste - July 30, 2007


    –Is it a proper function of government to declare the Titanic “an international maritime memorial”?

    –Finding true love in …

  2. The Volokh Conspiracy - July 27, 2007

    More Junk Social Science:…

    In a Washington Times op-ed yesterday Gail Heriot and I noted that Congress “has been on a binge diet of junk social science.”

    Josh W……