Code Red: Two Economists Examine the U.S. Healthcare System

July 8, 2011

Another Guest Blog from Craig Garthwaite: The Oregon Medicaid Lottery

Filed under: Uncategorized — David Dranove and Craig Garthwaite (from Oct 11, 2013) @ 9:54 am

In this blog, Craig talks about a remarkable new study that was just released. The study is nominally about Medicaid, but the reaction to it says a lot about election politics. I will let Craig explain. The post is very long but the study and its implications are important. I urge you to take the time to read it all.

Considering the strength and passion of the statements about the PPACA during its debate in Congress last year, one would think that its effects would be clear and uncontested. For example, opponents of the bill were portrayed as heartless thugs who would literally kill Americans by denying them the benefits of health insurance. Supporters of the bill characterized PPACA as an entitlement on equal footing with free speech and free exercise of religion. But, in truth, even health “experts” don’t really know exactly how much people benefit from health insurance coverage. I have little doubt that health insurance provides real benefits, but it is important to understand what we know from actual research, and what we assume must be true.

A study released yesterday by some of the most respected health economists in the country provides valuable information about the benefits of health insurance. It is a compelling study that exploits a lottery for access to Medicaid that was held in Oregon in 2008 to estimate the benefits of Medicaid compared to being uninsured. This is exactly the type of random assignment to Medicaid coverage that health economists need to determine the benefits of health insurance—and a source of variation that has not been available since the RAND health insurance experiments in the 1970s. These experiments found that individuals with better health insurance spent more on medical services, but didn’t enjoy better health outcomes. While these results are informative, medical technology has grown dramatically since the 1970s. For example, we now have the widespread use of prescription drugs that treat everything from chronic pain, to cholesterol levels, to depression. Accordingly, it is an open question as to how valid the estimates from the RAND experiment are in a world of these advanced medical technologies.

Like anything related to healthcare these days, this study has generated both attention and controversy. Supporters of the Health Care Reform bill immediately held up this study as proof positive of the clear health benefits of insurance coverage. Opponents attempted to knock down the study, saying that the demonstrated benefits are illusory and the results are not applicable to the PPACA. Some of these arguments are value and some are dubious, at best. I thought it might be important to sort out what this study tells us, and what it doesn’t.

Having access to health insurance increased the use of medical services. At first glance this may not be that surprising, but remember that we are talking about having Medicaid and not just insurance. While Medicaid is known to provide generous coverage, it is regularly argued that enrollees are unable to find a doctor and therefore the insurance is not very useful. Some politicians have even gone as far as to say that individuals can get better care if they are uninsured than if they have Medicaid. The Oregon lottery study shows that the new Medicaid enrollees were clearly able to find health care providers who accepted their insurance. The lottery winners were substantially more likely to have a regular source of care and to see the same doctor over time. These are all believed to be good things, though in truth the literature of continuity of care is still a little spotty.

The “lucky 10,000” also were far more likely to have participated in preventative care activities such as cholesterol screening. All of this new medical activity increased spending on health services, showing that at least in the short run this preventative care did not decrease spending (though I think we need more time before we can say anything definitive about the long term savings of preventative care.) It is also interesting that having access to health insurance had no effect on the use of emergency room services. If anything, the point estimates suggest that use may have increased. This should not be surprising given that Medicaid does not require punitive copayments for use of emergency room services, but it certainly runs counter to the argument that expanding insurance will cure overcrowding in the ER.

What about health outcomes? Were the opponents of PPACA actually attempting to kill thousands of Americans? It is has only been a year since the lottery so any evidence on this dimension is limited. The initial returns, however, are encouraging for supporters of more health insurance. The authors found that the self-reported health of the lottery winners increased compared to those who didn’t have the option of enrolling in Medicaid. It is not clear, however, if this was a result of actual improved health or more of a “psychic” effect of securing insurance. Critics of the study quickly claimed that because a large percentage of this increase came immediately after receiving coverage (and before any actual use of medical services) this means that the benefit is merely a placebo effect. At best this view is shortsighted. Certainly, the decreased stress and increased financial certainty from having health coverage is beneficial to a person’s health. Furthermore, the increase in income these individuals receive from having subsidized health insurance coverage could be beneficial to health. In earlier work I completed with Bill Evans of Notre Dame University, we found that the increased income resulting from the federal Earned Income Tax Credit increased both self-reported and measured health outcomes among recipients. This included in risky biomarkers—measured health outcomes that are believed to predict later life health albumin levels, cholesterol, and blood pressure. The authors of the Oregon study have said that they are tracking additional health outcomes, such as biomarkers, and over time will be able to more carefully examine the impact of health insurance on health. Therefore, criticizing the study on this dimension seems, at best, premature.

Other critics of the study have claimed that the results may not be generalizable. Their criticisms range from Oregon not being like the rest of the country (something any viewer of Portlandia may agree with) to the fact that this was a relatively small Medicaid expansion. While based on facts which are all correct, the criticisms are hardly damning. Perhaps the most important caveat, which is addressed by the authors, is whether Oregon’s lottery experiment measures up to the recently passed health reform bill. Starting in 2014, it is estimated that PPACA will increase the size of the Medicaid population by 16 million Americans. This surely will have a number of effects beyond those of the Oregon lottery experiment, and therefore it is important to consider the size of the expansion in discussing its potential impact on the entire health care system. For example, my research shows that previous large Medicaid expansions have decreased the labor supply of physicians. In considering the potential effects of the PPACA, we must think carefully about how providers will respond and how we can organize the health care system to absorb the millions of newly insured individuals. The results of the Oregon lottery study show that these individuals will dramatically increase their use of medical services.

The Oregon lottery study has attracted lots of attention and was immediately attacked by opponents of the health care reform bill. Attacking the benefits of health insurance in general seems like an odd choice and a losing strategy. Of all the problems with PPACA, it can’t be that we think there is no benefit from people having health coverage. While we might quibble about the size of the benefit, and whether it is cost effective, this study clearly demonstrates that the self-reported health of individuals who received new insurance as a result of the Oregon lottery increased. In the follow-up study, after more time has passed, we will see how other measurable health outcomes have changed (spoiler alert: I predict they will also improve).

Rather than trying to convince people that health insurance has little benefit (an argument that is made more difficult when the speakers is often generously insured), opponents of health care reform would do themselves a service to focus on other aspects of the bill that will remain problematic. For example, one result of PPACA’s personal mandate is that millions of people who are eligible for Medicaid services are going to be enrolled in the program—dramatically increasing the cost to state governments. Given that states can barely afford their current programs, it isn’t clear how they are going to pay for all of these newly eligible individuals. It seems that some of the “budget neutral” nature of the bill came from shifting these expenses to state governments.

Another fair question about PPACA comes from a recently released study from the respected consulting firm McKinsey and Co. In a survey of 1,300 employers, McKinsey found that a large portion of employers plan to drop health insurance coverage because of the legislation. While in general economists don’t trust surveys about what people “intend” to do, it isn’t difficult to think that many employers will rethink their benefits policies in a world of highly subsidized health insurance exchanges. Will PPACA be the end of employer provided health coverage? What will this do the cost projections of this bill? Dumping millions of employees into subsidized coverage will dramatically increase the total cost of these subsidies, should we increase the debt ceiling even higher now to get ready? I am not saying that we should, but it would be good to be honest about where we are headed in terms of the full fiscal impact of PPACA.

These are the types of arguments that opponents of PPACA should making, not whether we think giving low-income Americans access to health insurance is beneficial.


  1. It was not critics of the Harvard/MIT study that said the results might not be generalizable. It was the authors of the Harvard/MIT study that said it.

    Comment by Dennis Byron — July 14, 2011 @ 5:14 am

  2. In addition, though, if employers “dump” millions of employees into subsidized coverage (or state-based exchanges which may or may not be subsidized), those employees lose the tax-protected status of their current healthcare coverage. In other words, if employers drop healthcare coverage, the tax base will rise significantly. That’s the secret bogey I’d like to see calculated.

    Comment by enr — August 4, 2011 @ 12:50 pm

RSS feed for comments on this post.

Blog at

%d bloggers like this: