• Why not give cash instead of Medicaid?

    This question is kicking around the blogosphere and Twitter. The current interest is inspired by the incorrect view that the latest Oregon Health Study paper shows that Medicaid does not improve physical health outcomes but does confer financial benefits. In fact, though the study does show it confers financial benefits, it does not show that Medicaid confers no physical health benefit. It shows that the study had insufficient sample to detect reasonably sized, clinically significant physical health improvements over two years. The authors explain this in their concluding discussion, though I acknowledge it is not in written in a way that is as accessible as it could be. (See Kevin Drum.)

    Nevertheless, a great deal of the benefit of health insurance is financial security. And, even though Medicaid does more than that, we can still entertain the idea of giving cash instead. Here are a few points and questions of relevance:

    • The Oregon Health Study paper based on one year of follow-up, published in the Quarterly Journal of Economics in August 2012 (ungated PDF here), includes the result that Medicaid led to a statistically significant 32% increase in overall happiness. The authors point to some other work that shows this is equivalent to a doubling of income. (See page 29 and footnote 31.) This begins to suggest how much cash would be required to do the job of Medicaid. But it is only a start.
    • Presumably an income transfer would be paired with a catastrophic-only public insurance benefit with some deductible considered high for poor Americans. What would we do with people who blew through their cash, leaving insufficient funds for the deductible? Would we deny them care? Even life saving care? Hard to imagine.
    • What would we do for people with consistent, non-catastrophic range annual health care needs, e.g., due to a chronic or congenital condition? Would cash payments be risk-adjusted?
    • Another feature of the in-kind benefit of health insurance is that it supports the health care infrastructure to an extent cash payments may not. This could be good or bad depending on your view of that infrastructure and public support for it.
    • If we were to study cash payments vs. traditional Medicaid, what outcomes would you select?
    • How likely do we think it is that Congress would consider allowing a state to swap Medicaid for a cash benefit? I consider the chance of this extremely low in the foreseeable future. Meanwhile, this year, states need to choose an allowable Medicaid expansion variant or no expansion. Among the feasible set of options, which do cash benefit proponents dislike least?

    What are some other considerations? Any relevant research come to mind?


    • Why not just give people access to free health care without going through insurance?
      No need to worry about sign up, cash, payment, perverse incentives.
      Just give people access to health care.

    • The Cash and Counseling demonstration and subsequent waivers provide one example of generally positive results, I think.

      • Excellent post Austin. In addition to these points, the fundamental point that seems to be lost here is the lack of understanding of the theory of uncertainty. That is when we buy a non-cash transfer program for a person, such as Medicaid, it is not equivalent to a cash transfer for the fundamental reason that the Medicaid program buys the person on the program freedom from risk, which as we know from the theory should be worth a lot to a person with risk aversion. This would be worth a lot more than cash. Yes, with an equivalent amount of cash they could theoretically buy the same amount of “access” to medical care, which is also what Medicaid buys, but not the pooling and transfer of risk.

        As you also point out, since we have EMTALA, if someone argues we could buy them a catastrophic plan (which also would deal to some extent with the risk problem) and then give them the residual cash, this would theoretically work except that if a person blows through the cash then we would have to deny them care. EMTALA does not allow that, at least for life-threatening issues.

      • With all due respect, Cash and Counseling is not comparable to giving people cash in lieu of Medicaid.

        Cash and Counseling was a successful demonstration in long-term services and supports. Summarizing from memory, demo participants were given hiring and firing authority over caregivers and an actual budget to manage. They were allowed to hire and fire family members under circumstances. Given that a number of Medicaid beneficiaries who are approved for LTSS don’t actually get it because of no workers available, this is good. It also increases people’s sense of autonomy, etc.

        However, LTSS is entirely different from medical care. LTSS is a relatively simple service. You need help with daily tasks like bathing, dressing and eating, you hire someone to help you with that. There is definitely some diversity to the types of services you might need (e.g. adult day care, specialized services for people with developmental disabilities, home modifications, traditional personal care, assisted living). However, it’s nowhere as diverse as the services available in medical care. In addition, in LTSS, it is traditional for providers to deal on a cash basis with consumers, whereas it is not in medical care. It is not traditional for LTSS providers to offer large volume discounts to insurers, whereas this is the status quo in medical care.

        Furthermore, everyone in Cash and Counseling needed long-term care in the community. I am not sure if the study excluded people needing 24-hour care, but those folks usually go to nursing facilities anyway. In other words, the range of possible need was not that much – LTSS services don’t vary in cost as much as medical services. In an experiment involving cash for Medicaid, you would need to deal with the fact that medical needs can vary by a pretty large factor. You would be overcompensating some people and undercompensating others, even if you had a great risk adjustment system. For example, someone diagnosed with asthma with complications might be able to control it with inhalers and immunotherapy, or they might need to visit the ED for, say $10,000, or they might need to an inpatient stay for upwards of $80,000. You would have to have a catastrophic plan, for example, and you’d need to choose an appropriate attachment point.

        I realize this might sound doctrinaire. But if you want to convince me that a cash-based experiment for Medicaid is feasible, I would argue that Cash and Counseling doesn’t prove the point as it isn’t sufficiently comparable. I would need to see the results of some sort of large scale HSA experiment. Have those been done successfully with Medicaid?

        By the way, Cash and Counseling has gone on to be a regular part of many Medicaid programs. It’s usually referred to as participant-directed or self-directed care. Imo it’s going to continue to expand, especially among individuals with developmental disabilities, younger individuals with physical disabilities, and maybe people with mental illness.

        • I was not suggesting that Cash and Counseling was completely comparable to substituting cash for Medicaid. I was simply responding to Austin’s question seeking any relevant research. Certainly, a Medicaid waiver program that substitutes participant directed cash payments for Medicaid provided LTSS that has not only been demonstrated but expanded into other areas of care is not irrelevant to the question.

          • Dennis, let me clarify what I said. Austin’s question as I understand it is: how should we think about giving people unrestricted cash in lieu of traditional Medicaid benefits?

            I take your point that C&C was a successful experiment that improved beneficiaries’ independence and outcomes.

            However, I still don’t think it’s useful in answering the question.

            *The funds in C&C were restricted to LTSS services only, and you couldn’t use them on nonrelated stuff (btw, the people who are concerned about expanding participant-directed services often state fraud as a concern).

            *C&C doesn’t demonstrate how we might compensate for significant disparities in medical expenditures in a world where we are cashing people out. In such a world, we’d be very concerned about the person with HIV/AIDS or cancer, for example, if they were just cashed out at the average actuarial value of a Medicaid plan (because these conditions are very very expensive iirc).

            *And, as I explained, LTSS is different enough from medical care that while people may be willing and able to manage an LTSS budget for themselves, I would need more convincing that they could do the same thing in health care and produce similar or better outcomes than with a traditional insurance policy. We would be very concerned that they’d be unable to buy equivalent coverage.

            If you disagree, can you state what you think is transferable?

    • Medicaid gets lower prices because it has a great deal of bargaining power. Give people cash and their medical care becomes more expensive and Medicaid becomes smaller, lowering its bargaining power.

    • Maybe the study had insufficient sample to detect reasonably sized, clinically significant physical health DECLINES over two years? There were mine improvements, I know. But it didn’t not say there could’ve been declines, so, you know, it still could’ve.

      There was a better way to your point, I think. “It didn’t not say it” sounds weaslely whereas, “health improved, but not, especially given the sample size, in way that was significant.” concedes that there were, per the study, none but with a caveat.

      • Eric,

        It is inaccurate to say that the study found that health improved. A point estimate that happens to be positive but is wildly insignificant is not an improvement; it’s noise. Being “weasily” is a necessary part of scientific honesty when your results are non-conclusive. Perhaps a more direct statement would have been, “For the individual health measures, the data could not differentiate between no effect and a clinically meaningful effect, so it would be better to focus on the measures for which the study had sufficient powerto say something.”

        (That’s assuming you’ve concluded that a clinically meaningful effect size is contained within the study’s confidence interval, a point on which I’m withholding judgment until this study has been digested a bit more.)

    • Oh yeah, like the people kicking Medicaid in the teeth want to give cash to Medicaid enrollees…..

    • Considering the discount insurers and especially Medicaid gets from all types of Healthcare providers, I don’t think an individual patient would get the same care for their ~$1100 if they tried to purchase it for cash.

      • Why doesn’t Medicaid have a buy-in option (I know, self-selection)? Something along the lines of $1,500/year for the first ineligible tranche to income X, $2,000/year to income X+10,000, etc. If you can’t get Medicare for All, why not Medicaid for All as a start? (There’ll still be insurance companies.)

        As I mentioned earlier, the province I live in spends nearly $3,700 per person per year on Medicaid for All (including seniors) which represents about 65% of total medical costs. The balance is paid by individuals, employers, private insurance for dental, vision, pharma, psych, “not medically necessary”. http://www.cihi.ca/cihi-ext-portal/internet/en/document/spending+and+health+workforce/spending/release_30oct12

    • It seems to me that studying the difference between cash payments and traditional Medicaid would founder the same way the Oregon study did, unless all the participants chose to buy insurance, which is quite unlikely.

      One presumes sickly people will buy insurance on the exchanges. Healthy people might not. Unless they get sick, they will be better off with cash. When you look retrospectively, everyone who didn’t get insurance and didn’t get sick is better off with cash. Just like if you gamble your life savings on a horse race, if the horse wins, you are much better off than if you hadn’t made that stupid bet.

      So, the issue is to estimate how many people are currently healthy and might choose to forego insurance are likely to have an expensive medical problem and how big does your study have to be to draw statistically significant conclusions about them.

      • The Oregon study did not founder due to biased selection. It explicitly avoided that problem through randomization. Increasing sample size is not a substitute for randomization. You need to avoid selection bias and have sufficient sample. You need both.

    • Free health care should be considered instead. Everyone deserves it. Let health professionals provide a valuable service, and not be involved with charging fees.

    • There is a review of in-kind versus cash transfers by Currie and Gahvari in the Journal of Economic Literature (2008). Unfortunately, they seem to miss some of the obvious reasons for in-kind transfers: 1) if people are risk-averse and public insurance has the lowest loading costs, then in-kind public insurance benefits are welfare-improving relative to cash benefits. 2)If we assume that governments cannot directly observe citizens’ ex-ante health risk, then it has been shown in a number of studies that public provision of health insurance can enable some amount of redistribution of wealth not available to the government through normal taxation, I’m currently working on a structural model along these lines about optimal regulation of health insurance premiums.

    • Austin,
      You would put the “cash” into an HSA-like account, similar to the ones being used in the Medicaid pilot program in Florida and the expansion population in the Healthy Indiana Plan.

      Josh (@josharchambault)

      • Can that cash only be used for medical expenses? Is it always sufficient to cover the deductible? Feel free to provide a link to more information.

        Is this what every pro-cash advocate means?

        • I won’t speak for all pro-cash advocates as you probably would not want to speak for all economists. But for those that want to earmark the money for healthcare, they are talking about HSA-like accounts.

          Correct the cash could only be used for medical expenses. However you can imagine a broad definition for this in some states: gym memberships, some alternative medicine, long-term care– as we see in the HSA marketplace now. However, I would expect most of the money would be used for more traditional medical expenses.

          I could also see a role for HRAs, but I am not aware of any state exploring such an arrangement.

          In a place like Florida, the health credits are about $125 a year from the state. The amount of the deductible will vary because patients have a choice in which plan they pick. A paper was written for Heritage on the topic a few years ago before a statewide roll out was approved by CMS more recently.

          In Indiana participants fund the $1,100 deductible, either in full or in part, through monthly contributions they make to their POWER accounts. They do this instead of small copays in most cases.

          Monthly contribution amounts are assessed on a sliding scale based on a percentage of household income, with
          no participant contributing more than $1,100 or 5 percent
          of his or her income net of any contributions paid
          to other programs, such as Indiana’s Children’s Health
          Insurance Program (CHIP).

          Members who fully fund their POWER accounts contribute approximately $92 a month to them. If a member pays less than $1,100 into his or her POWER account during the year, the state
          pays the balance to ensure the account is fully funded.
          The full $1,100 is available to the member after he
          or she makes the first monthly contribution. Ongoing
          monthly contributions allow the member to spread
          the cost of the deductible over the year, regardless of
          when the costs of services are incurred—an important
          feature for many low-income adults.

          More info here:

    • There is zero case for cash assistance when our entire health care system uses insurance not only for financial insurance but also for bargaining power and purchasing expertise. Medical pricing is where most of the actual action is in managed care; see, for example, Altman, Cutler, and Zeckhauser (J Health Econ 2003) or Gruber and Rodriguez (J Health Econ 2007).

    • One good reasonable change that could reasonably come out of the Oregon stats would be more informed voters wanting to to force the Federal Government to cover everyone (all citizens) without spending any more money that they spend now. Make them prioritize and do the things that work and make the squeeze providers. If people want more care they could still buy it on their own.

    • -One of the reasons that critics of the study suggested comparing the effects of winning the medicaid lottery to cash is that the results of the Oregon Medicaid study cannot distinguish between the health effects mediated specifically by health insurance in general, and Medicaid in particular, and the effects that would have been produced by giving them access to alternative assets with an equal value, such as cash, healthcare vouchers, or a plan that couples a pre-funded HSA with a catastrophic policy. In that absence of those controls, you can’t determine how much of the effect is due to Medicaid enrollment relative to the alternates mentioned above.

      Most of the above discussion about the particular mechanics of cash transfer schemes seems to be missing this point.

      -Another point that seems to be lost in the above discussion is that obtaining health care and/or health insurance may not be the highest priority for a given person in the Medicaid eligible pool, and that allowing them to do so would generate subjective improvements in their sense of well-being and/or health that equal or exceed those reported for the lottery winners.

      – The major unstated premise in a great many of these conversations is that people with low incomes can’t possibly handle health insurance arrangements with the same incentives and constraints in high-deductible/catastrophic plans like those with higher incomes do.

      The idea that vast majority of people on Medicaid would blow through their deductibles every single year, never accumulate enough assets in their government funded HSA to see them through a rough year, and that we couldn’t possibly design policies that would enable those who do exceed their deductible to cope simply isn’t credible. You could address the vast majority of shortfalls by making HSA balances transferrable between family members, and the remainder by making low interest, publicly financed loans repayable in cash or deductions from future transfers.

      For the cohort of people who are so chronically ill their annual out-of-pocket costs will always exceed the annual value of the their transfers, just give them transfers equal to 100% of their deductible every year and be done with it. Whatever percentage of the Medicaid eligible are under the ACA, it’s certainly far from a majority.

      • In any case, if such an arrangement were permitted for the Medicaid expansion, would we any states still opt for no expansion? Is it likely such an arrangement will be made available soon?

        • 1) Based on the nature of the ideological opposition to the Medicare expansion, I think that plans that allow variable, income indexed government contributions to personal, transferable HSA balance plus a high-deductible plan would face dramatically less opposition on the right.

          Most on the right oppose Medicaid expansion for two central reasons. The first is that it uses highly centralized bureaucratic mechanisms and price controls in an effort to control costs and drive efficiency, as opposed to market mechanisms that harness price transparency and competition. Whatever you think of the merits of that stance, plans that have more market and consumer oriented incentives and constraints, and grant state governments more control are far less likely to encounter ideological opposition from the right, even if they redistribute as much (or more) tax revenue as Medicaid does on a per-capita basis. Anyone who says the right is always and everywhere opposed to transfers to the less fortunate should consider the case of the EIC.

          Ironically – even if such plans were on the table, vehement opposition from the left would likely block the adoption of any plan like the one outlined above *because* it involves market mechanisms like choice and competition, even if the total amount of money transferred to the lowest income quintiles were higher.

          One way to break the ideological deadlock would be to let individuals who qualify for Medicaid choose one option or the other.

          -The answer to the second question is clearly, no, which is a shame but the left bears far more responsibility for that than the right, IMO. Anything that diverts significant institutional momentum away from single payer is going to attract vehement opposition from every left of center constituency, irrespective of the actual merits of the plan. Even letting the Medicaid eligible decide from themselves, and letting them switch form one to the other is going to be a non-starter.

    • The year 1 results show a 32% increase in happiness. But the latest 2-year results show no significant difference in happiness levels. In that context, how is the year 1 result relevant? It strikes me that it could as easily be explained by “winning the lottery” as Medicaid coverage itself. The fact that it didn’t persist over a 2-year follow-up gives some credence to that explanation.