A new study by economists Carlos Dobkin, Amy Finkelstein, Raymond Kluender, and Matthew J. Notowidigdo -- "Myth and Measurement — The Case of Medical Bankruptcies" [subscription required] -- challenges the conventional wisdom on the effect of medical bills on the rate of personal bankruptcy. From the study:
Policymakers' beliefs about the frequency of medical bankruptcies are based primarily on two high-profile articles that claim that medical events cause approximately 60 percent of all bankruptcies in the United States. In these studies, people who had gone bankrupt were asked whether they'd experienced health-related financial stress such as substantial medical bills or income loss due to illness. People were also asked whether they went bankrupt because of medical bills. People who reported any of these events were described as having experienced a medical bankruptcy...
[But] the existing, widely cited evidence on medical bankruptcy is built on the fallacy that when two things occur together there is necessarily a causal relationship between them.
The study's authors looked instead at people who had a hospitalization to see whether that expensive episode of care increased the probability of filing for bankruptcy. They write, "we estimate that hospitalizations cause only 4 percent of personal bankruptcies among nonelderly U.S. adults." Even among uninsured adults, "hospitalizations are responsible for only 6 percent of personal bankruptcies." While medical bills can still drive someone to bankruptcy even if they don't experience a hospitalization, the authors conclude, "focusing on hospitalized people probably does not lead to vast underestimation of the effect of all illness and injury on bankruptcy rates."
- Always be skeptical of everything you read. (Up to and including this blog!)
- Keep in mind this study does not show the overall personal bankruptcy rate is lower than believed. It shows only that the share attributable to medical expenses is lower than believed. It therefore follows that, to the extent your support for single-payer springs from a desire to reduce bankruptcies, you should shift your energies toward combating whatever is actually causing the 56 percent of bankruptcies you incorrectly believed to be attributable to medical expenses.
- Health care reform should be able to get the medical-bankruptcy rate down even more.
As President Obama gears up to deliver a major address on the supposed successes of the Affordable Care Act, a study by one of the nation's top health economists is pouring cold water on the ACA's main engine for expanding health-insurance coverage: its expansion of Medicaid to cover able-bodied, childless adults.
MIT's Amy Finkelstein has won a slew of awards, including the prestigious John Bates Clark Medal, for her work in health economics. In “The Value of Medicaid: Interpreting Results from the Oregon Health Insurance Experiment,” Finkelstein, Nathaniel Hendren, and Erzo Luttmer, used various econometric methods to quantify the benefits that enrollees receive from Medicaid. They drew from the Oregon Health Insurance Experiment, on which Finkelstein was a lead investigator.
The trio found that Medicaid enrollees receive very little benefit from each dollar spent on Medicaid. The absolute minimum enrollees receive is 15 cents of benefit per dollar spent. The authors' best guess is that enrollees receive somewhere around 20-40 cents of benefit per dollar spent. The maximum is 90 cents--that is, no matter how the authors sliced the data, Medicaid's costs exceed the benefits to enrollees. If the government just gave enrollees the money, Medicaid is such a bad deal that enrollees would not buy Medicaid coverage with it.
Medicaid spends, non-enrollees receive about 60 cents of benefit. The authors don't identify who Medicaid's real beneficiaries are, but they likely include those who receive Medicaid subsidies (hospitals, insurance companies, pharmaceutical companies, doctors, device manufacturers) and people who would otherwise make charitable contributions to provide medical care to enrollees. In other words, Medicaid's actual beneficiaries are different from its intended beneficiaries.
That's something to keep in mind when President Obama says, "There are outcomes we can calculate" like "the number of newly insured families" and that "those numbers add up to success."
A lot of people are writing about the Oregon Health Insurance Experiment results, released yesterday, which found zero evidence that expanding Medicaid to the most vulnerable people targeted by ObamaCare's Medicaid expansion improves their physical health. Here's my take on the study and its implications. Megan McArdle, Shikha Dalmia, Avik Roy, and Peter Suderman are making solid contributions to the debate. Zeke Emanuel gets points for making an admission against interest ("It's disappointing"). Points also to Jennifer Rubin for her take on what the OHIE says about ObamaCare's Medicaid expansion: "If there had been a giant trial of a heart medication with lousy results we wouldn’t proceed in mass-marketing the drug; we might even take it off the shelves." Not a bad idea. Ezra Klein and Evan Soltas call for more such experiments. Yes! Let's have more randomized, controlled trials of the effects of Medicaid, on pre-ObamaCare populations, in big states like California, New York, Texas, Florida, and Illinois, where we can harnass even more statistical power. The only unethical thing would be to keep spending trillions on this program without knowing whether it's even effective (much less cost-effective).
Others are making less-solid contributions. Here's a question for them.
Since the OHIE shows that Medicaid makes no difference in the diagnosis or use of medication to treat high blood pressure or high cholesterol, and has no effect on blood-sugar levels despite increasing diabetes diagnoses and medication use, would you support eliminating Medicaid coverage for these screenings and medications?
If not, why not?
Columbia Business School economist Ray Fisman has a piece at Slate.com discussing the first-year results of the Oregon Health Insurance Experiment. In brief, when Oregon transferred an average of $3,000 from taxpayers to poor people in the form of Medicaid coverage, it did those poor people some good.
Fisman's interpretation of the results is different from mine in mainly two respects. First, I describe the one-year benefits of Medicaid coverage as modest; he says they're "enormous."
A more fundamental difference concerns whether expanding Medicaid was a cost-effective use of the taxpayers' money. Fisman writes:
Given the added expense, did the Medicaid expansion prove to be cost-effective? That is, did the treatment group actually have better health outcomes?
That's not what cost-effectiveness means. For Medicaid to be cost-effective, it must (A) produce benefits and (B) do so at the same or a lower cost than the alternatives.
The OHIE establishes only that there are some (modest) benefits to expanding Medicaid (to poor people) (after one year). It tells us next to nothing about the costs of producing those benefits, which include not just the transfers from taxpayers but also any behavioral changes on the part of Medicaid enrollees, such as reductions in work effort or asset accumulation induced by this means-tested program. Nor does it tell us anything about the costs and benefits of alternative policies.
Just as some opponents of ObamaCare over-interpreted previous Medicaid studies, Fisman and other ObamaCare supporters are over-interpreting the OHIE.
The Oregon Health Insurance Experiment is the first experiment since the dawn of time that randomly assigns some households to receive health insurance (Medicaid) for purposes of comparing their medical consumption, health outcomes, and financial security to similar households that do not receive Medicaid coverage. Some of the nation's top health economists have released the first batch of results from the OHIE.
At National Review (Online), I summarize the OHIE's first-year results and offer the following analysis:
Supporters of President Obama’s health-care law may tout these benefits, but the OHIE does not provide the vindication they seek. First, despite being eligible for Medicaid, 13 percent of the control group had private health insurance — suggesting that on some dimension, Medicaid’s eligibility rules are already too broad.
Second, the OHIE extended coverage to the most vulnerable population of uninsured Americans, yet the improvements in health and financial security are so far apparently modest. At higher income levels, where individuals have greater baseline access to health insurance and medical care, the benefits of expanding coverage are likely to be smaller and the costs (to the extent that crowd-out is higher at higher income levels) will be greater.
Third, supporters must show not only that expanding coverage improves health but also that it does so at a lower cost to taxpayers than alternative policies. Health economists generally agree that discrete programs promoting highly effective treatments (for hypertension, diabetes, etc.) could produce health gains as large as expanding health insurance would, but at far less expense. Reducing taxes could plausibly reduce financial strain to a similar degree by expanding job creation.
Finally, the OHIE illuminates an unflattering feature of the push for Obamacare. For a century, the Left has advocated universal health insurance despite not knowing what benefits it might bring. In 2010, Congress and President Obama vastly expanded Medicaid without waiting for the results of the one study that might tell them what taxpayers would get in return for their half a trillion dollars. As the law’s supporters seek to cajole doctors into practicing evidence-based medicine, it is no small irony that they themselves dove head-first into evidence-free policymaking.
- If Medicare subsidizes enrollees by giving them a fixed amount of money, much like Social Security does, they would be more cost-conscious than they are under the current open-ended subsidy, because enrollees who avoid wasteful spending would themselves get to keep the savings. Put more plainly, people spend their own money more carefully than they spend other people's money.
- Health insurers and health care providers would compete to serve these cost-conscious Medicare enrollees on the basis of both cost and quality. Prices would fall while quality improves.
I'm not really sure to what extent all this would occur under the Medicare reforms the House passed a couple of weeks ago, because we don't yet know to what extent each enrollee's subsidy would resemble a fixed amount of money.
Here's what Frakt does with my claims:
[A]s I heard these words I wondered if we had any evidence on hand about the relationship between lower premium subsidies and health care cost inflation. Indeed we do! Premiums in the commercial market are subsidized by the government at a lower rate than those in Medicare. [Emphasis added.]
He then throws up the chart, shown below the jump, showing that for common benefits, the rate of growth in per-enrollee spending is "pretty similar" in Medicare and private insurance. He concludes: "With data like this, I think we need to reexamine some of our theories about what lower premium subsidies can do."
Oy, where to begin?
First, Frakt does not actually challenge my claim. My claim is that voucher-like Medicare reforms will lead to reductions in the per-unit cost of producing certain goods and services, and therefore to lower prices. Frakt responds with data on health care spending. When someone predicts that P will fall, introducing into evidence what has historically happened to P x Q is no kind of rebuttal. The confusion Frakt sows is rooted in the fact that both prices and spending can be accurately described as costs. Such confusion could be avoided were everyone to honor Cannon's First Rule of Economic Literacy: Never say costs when you mean spending.
Second, though the tax preference for employer-sponsored health insurance distorts relative prices the same way an open-ended subsidy does, it is not a subsidy. This isn't really relevant to the matter at hand, but it's worth emphasizing to avoid such silliness as this.
Third, it would be great if there existed one chart that would settle once and for all whether Medicare or a free market does a better job of containing costs. Alas, this is not that chart. Frakt uses it to make a less-ambitious point about the effects of larger versus smaller policy-induced price distortions, but its shortcomings nevertheless confound that comparison, too. In addition to measuring increases in spending (whose effect on social welfare is ambiguous) rather than increases in cost (which are always bad), this chart handicaps private insurance by leaving off three or four years of explosive growth in per-enrollee Medicare spending (1966-1969). Worse, it reeks of endogeneity problems. Amy Finkelstein finds evidence that Medicare itself increases spending among those with private insurance (mostly by increasing hospital capacity), and that this effect has grown over time. Moreover, as Medicare spending rises, so do average marginal tax rates, which increases the price distortion created by the tax exclusion, which increases spending on private insurance. For all the play it gets on the Left, this chart serves no useful purpose that I can discern. Economists should be embarrassed to use it. If it were a farm animal, and social scientists farmers, they would have to take it behind the barn and put a bullet in its head.
Fourth, we are not yet at the point where we need to reexamine the theory that demand curves slope downward. There is ample evidence to show that Medicare enrollees will respond to voucher-like reforms by choosing more economical health plans, and that health plans and providers will respond with greater efficiency. Thomas Buchmueller reports:
Two notable experiments...took place in the mid-1990s: the University of California (UC) and Harvard University both offered a menu of plans that varied in generosity, but adopted a "fixed dollar contribution" policy. The plans also varied significantly in cost, so employees had a greater incentive to consider price when selecting a health plan...
In both cases, employees were quite sensitive to price, and were willing to switch plans to save as little as $5 per month in out-of-pocket premiums...In addition to this demand response, participating insurers lowered their premiums in order to compete for enrollment.
Fifth, there is plenty of evidence that prices can and do fall in health care -- from research on the markets for laser-eye and cosmetic surgery, to the work of Clay Christensen and his colleagues, to the research of David Cutler and Mark McClellan. (If spending increases while prices are falling, it is because changes in Q dominate changes in P -- which could be due to all these open-ended government subsidies and other price distortions.)
Finally, as a response to the general theme of that NPR story: we're a long, long way from the point where we have to worry that reductions in the growth of Medicare spending are going harm enrollees' health. Relying on data from the Dartmouth Atlas, President Obama's Council of Economic Advisers reminds us that "nearly 30 percent of Medicare's costs [spending!] could be saved without adverse health consequences." And there is plenty of evidence, from the RAND Health Insurance Experiment and elsewhere, that plans with greater cost-sharing or care management reduce utilization without harming patients' health.
I cannot fathom what has opponents of Medicare vouchers so spooked. It cannot be the effects that vouchers would have on Medicare enrollees and taxpayers.