• Theories of health insurance

    In an ungated 2004 article in Health Affairs, John Nyman explains his theory of health insurance in intuitive terms. It’s worth a full read, but I’ll summarize it anyway.

    It begins with the observation that health care spending is encouraged by health insurance. Is this problematic? Nyman wrote,

    Conventional health insurance theory provided a ready evaluation of this increased spending: It represents a welfare loss [*] and should be reduced.

    Conventional insurance theory also provided the policy solution: Impose coinsurance payments and deductibles to increase the price of medical care to insured consumers and reduce these inefficient expenditures. In the 1970s many insurers adopted copayments to reduce health care spending. In the 1980s and 1990s economists also promoted utilization review and capitated payments to providers as further ways to reduce moral hazard. The managed health care system we have now is largely a product of this theory.

    Renewed calls for increased cost sharing (more “skin in the game”) reflect the belief that insurance promotes wasteful health spending. However, it has been recognized for almost thirty years that the conventional insurance theory that supports this belief and has motivated insurance design for decades does not apply to all types of health care. Nyman quotes Mark Pauly as having pointed out that it was only intended to apply to “routine physician’s visits, prescriptions, dental care, and the like” and that “the relevant theory, empirical evidence and policy analysis for moral hazard in the case of serious illness has not been developed.”

    Then Nyman developed it. In his Health Affairs article he sidesteps the math (for that, see his book) and illustrates the crucial element of his theory with an example.

    [C]onsider Elizabeth, who has just been diagnosed with breast cancer. Without insurance, she would purchase only the $20,000 mastectomy required to rid her body of the cancer. If she had purchased an insurance policy for $4,000 that paid off with a $40,000 cashier’s check upon diagnosis of breast cancer, she might purchase the $20,000 mastectomy and also a $20,000 breast reconstruction procedure. For economists, this behavior implies that the additional $40,000 in income from the insurance pool had increased her willingness to pay for the breast reconstruction so much that it is now greater than the $20,000 market price, causing her to purchase the second procedure. This moral hazard is efficient because she could have spent the additional $40,000 on anything she chose but opted to purchase the breast reconstruction. The purchase of this additional procedure represents a moral-hazard welfare gain to the extent that with the additional $40,000 in income, she would have now been willing to pay more than the $20,000 that it cost to produce the procedure.

    In this example, the additional care used, $20,000 for breast reconstruction, was unambiguously welfare improving. Elizabeth valued it at more than its cost (the economist’s definition of welfare improving). If she hadn’t, she’d have spent the $20,000 another way. However, because health insurance policies do not pay off with lump-sum payments, but rather pay directly for health care, the interpretation of the additional care used due to insurance is ambiguous.

    For example, if Elizabeth had instead paid $4,000 for insurance that simply paid for her health care when ill, she might also purchase the same two $20,000 procedures, resulting in the same payout of $40,000 from the insurance pool. But it is not clear whether she is responding to the zero price by opportunistically purchasing a breast reconstruction procedure that she barely values, or responding in the same way that she would have responded if the insurer had written her a check for $40,000. As a result, we cannot tell whether this additional moral-hazard spending represents a welfare loss or a welfare gain.

    How much additional spending due to insurance is a welfare gain? In his book, Nyman calculates that the majority of it is, perhaps as much as 70%. A number of policy implications follow that differ from those implied by an assumption that all moral hazard is a welfare loss. Nyman lists them as:

    • Cost sharing is often not appropriate, particularly for cost-effective, life-saving or health-preserving interventions,
    • Subsidizing insurance premiums to encourage coverage is beneficial, and
    • High health care prices are harmful because they discourage use of care.

    It is not incorrect to say that insurance promotes additional health spending. It does. If you believe Nyman’s theory, it is incorrect to say that all that additional spending is wasteful, a welfare loss. A little is. Most is not. More skin in the game is not more efficient even if it saves money. Some things are worth the price.

    * “Welfare loss” here is used in the neoclassical economic sense: that the amount individuals are willing to pay out of pocket is below the marginal cost of health services rendered. Individuals only demand such services because their actual out of pocket liability is reduced below marginal cost due to insurance. More here.

    Share
    Comments closed
     
    • Missing from the above is the effect of cost insulation on the supply of health care. Without some degree of price sensitivity from consumers, there is no incentive for producers to seek out lower cost methods of care. There are abundant examples of consumers facing sets of treatment options with varying costs, but ambiguous projected health outcomes. This explains why we invest so heavily in more expensive treatment options where the marginal health benefit does not exceed the marginal cost.

      • Value-based insurance design is not inconsistent with Nyman’s theory, nor is reference pricing.

      • Cost insulation does encourage spending. But the bigger problem is that patients usually have no way to distinguish between the effectiveness of various options. No one really wants unnecessary surgery, drugs, or tests, but confusion often leads people to choose higher tech, more aggressive management since it seems “logical” that it must be better. In a world in which the information about the merit of various health care options is complex and technical and in which doctors frequently advise choices that do not reflect best cost effectiveness or even best effectiveness, patients need help from some outside source to make these decisions.

        Just having “skin in the game” does not lead to good health care decisions in a system in which most patients, even well educated ones, do not understand what is happening to them.

        • Just having “skin in the game” does not lead to good health care decisions in a system in which most patients, even well educated ones, do not understand what is happening to them.

          I have found that telling doctors that I have a high deductible is enough to have them recommend a different far less costly course of action.

          • That is very interesting. As a doctor, I have to ask, have you ever asked your doctors why, if a less expensive management approach was just as good, they chose a more expensive approach in the first place? Or are you suggesting that you are receiving less effective but cheaper care?

            It always seems to me that getting doctors to choose less expensive but equally or more effective management approaches is the central problem in the issue of cost in US health care. Unfortunately, having patients on insurance programs with high out of pocket expenses does not usually seem to solve that problem, but instead either tranfers a significant share of the high costs to the patient or encourages patients to avoid health care in general. Polling certainly suggests that that is the case..

            • Pat S asked:

              That is very interesting. As a doctor, I have to ask, have you ever asked your doctors why, if a less expensive management approach was just as good, they chose a more expensive approach in the first place? Or are you suggesting that you are receiving less effective but cheaper care?

              I never asked but, once it meant that I would need to keep an eye on my wife, the patient, myself. Other times it was generics, which I assume was a case were the doctor was thinking who cares if the insurance company is paying or maybe the generic is not as good but better to just keep an eye out for complications.

    • If she had purchased an insurance policy for $4,000 that paid off with a $40,000 cashier’s check upon diagnosis of breast cancer, she might purchase the $20,000 mastectomy and also a $20,000 breast reconstruction procedure.

      Isn’t that just like say if the patient was richer? If we were all richer we would change our behavior in many ways.

      Also along the lines of what Harry said if reconstruction was never covered by insurance a priced market might arise and cause lower cost reconstruction. IMO we need to focus more on the supply side.

    • It seems to me the definitions of welfare loss and welfare gain are not only economic but that they assume a free market economy. In this context I say to Elizabeth, you go girl. If the breast reconstruction is worth $20,000 to you then buy it — with your money, not with mine.

      On the other hand, our health care debate is largely about the expenditure of public money. In this context the welfare gain must be to the society. Do we choose to spend the $20,000 on breast reconstruction or on Head Start?

      Even if it is true that Elizabeth would choose to spend $20,000 on breast reconstruction — if she had $20,000 — that is not to say that the government should spend $20,000 for the same purpose.

      I am using breast reconstruction only because that was the example cited in Austin’s post. The same principle applies to prostatectomies, heart bypass operations and bone marrow transplants.

      • A couple of points for Adam. Why is health care debate largely concerned with the expenditure of public money? Do you assume Elizabeth is unemployed and using the ER as her family practice/oncologist?

        One could also assume that Elizabeth is employed and both she and her employer have been paying large sums of money in premiums to a private sector insurance company for any number of years. In exchange for such payments, the insurance company has agreed to cover various procedures, including mastectomy and reconstructive surgery. That’s not the expenditure of public money; that’s an agreement between private parties. And in today’s private market, the health insurance business model requires paying out as little as possible in order for the company to become as profitable as possible. That’s the conundrum that we’re all struggling to solve.

        Moving to a lump sum payment would require a transparent pricing structure, which we do not have today in our insurance dominated health care market.

        Nyman’s original example is filled with odd assumptions, including the assumption that an uninsured woman just diagnosed with breast cancer would have $20,000 to put toward a mastectomy. She’d likely be priced out of the medical market unless there is a hospital willing to treat her for free. Nyman’s Elizabeth would not need to go before a “death panel.” She would simply die without any medical care at all.

        • Reply to Main Street Muse

          1. I refer to the expenditure of public money because that is one of the key implications of Nyman’s Health Affairs article (cited by Austin and a link provided). Here is a quote from that article:

          “…the subsidies that encourage consumers to purchase insurance voluntarily, or a national health insurance program for the entire U.S. population, would improve society’s welfare.”

          2. I don’t believe Nyman assumes “Elizabeth” is uninsured, but he does hypothesize alternative scenarios in which she is/is not insured and in which the insurance pays a lump sum to Elizabeth/pays charges directly to health care providers.

    • i think a big contributor to the problem is the patient’s low “health IQ”. Patients often have little knowledge of their conditions or treatments available, much less which treatments might be cost-effective, or even just effective. Patients need to have more knowledge to better help themselves. Unfortunately, in today’s gimme culture people prefer to have things handed to them, rather than work for it themselves. They will take a doctor’s recommendation at face value, instead of doing research and having a more educated conversation with their doctors.

    • @George “Doctor worship” can be as much a factor as low “health IQ”, Some people hold doctors in such high esteem that they cannot see them as fallible. All too often people are too scared to ask a simple question like “Is there a comparable generic available for this drug?” even though it could save them hundreds of dollars and their insurance company thousands.

    • The total cost of care is just the price per unit X the number of units. When Austin says that “insurance promotes additional health spending” he means, judged by his examples, that it promotes additional units of care purchased. It is only if we hold the price per unit equal that this yields higher total expenditures.

      However, insurance affects the price per unit of care as well. Insurers in some nations have either the size or the social/political support necessary to control the price of care much better than in other nations (e.g., the US).

      Don’t forget the insights discussed just a week or two ago about how health care is not a normal market, and patients as consumers often find themselves with inelastic demand curves and poor information on which to make decisions between alternatives. Under those conditions, “less insurance” can actually mean higher prices than in a system with universal coverage, but strong controls on prices.

      In such cases, whether the total cost is higher all depends on whether the impact on the units of care is larger than the impact on the price of care. If the US had only catastrophic insurance and no regulatory controls on cost and market power of providers (including no Stark laws, etc.) would we pay less for healthcare than the UK? I doubt it. In any case, it can’t be assumed.

    • Ugh Nyman is using confusing language. It’s not moral hazard if you increase care utilization after receiving a lump sum payment. If it’s efficient, it’s not moral hazard.

      Moral hazard is substitution effect, not income effect. It is the movement along compensated Hicksian demand curves, not ordinary Marshalian demand curves.

      • I don’t know why he doesn’t call it an income effect vs. a substitution effect. Instead he uses “efficient” and “inefficient” moral hazard. I’ll ask him.

        • Public economist Raj Chetty has taken to separating the price response into “moral hazard” and “liquidity constraints”. http://www.jstor.org/doi/abs/10.1086/588585

          My sense is these correspond to substitution and income effects.

        • By email, Nyman replied:

          The decomposition is different. The Hicksian decomposition takes away enough income at the new prices to place the consumer on his old indifference curve, which would represent the income effect, the remaining would be the substitution effect.

          Because with insurance, the price does not fall naturally but a lower price must be purchased, the decomposition is different. The decomposition with insurance is to take away the income transfer, the amount of money, net of his premium and his coinsurance payments, that comes out of the insurance pool to pay for his care. This essentially places the consumer on an indifference curve that is tangent to the new prices but on the original budget constraint.

          It is a different decomposition than the Hicksian one.

          The Hicksian decomposition applies to an exogenous reduction in price–something that is generated by a change in supply or demand in the market. There are no such changes with insurance and if the prospective purchaser of insurance does not pay a premium and agree to pay the coinsurance rate as part of the contract, then he does not receive a lower price. Moreover, the lower the price, the higher the premium.

          • Thanks for eliciting this clarification! I especially found the following sentences helpful:

            “The decomposition with insurance is to take away the income transfer, the amount of money, net of his premium and his coinsurance payments, that comes out of the insurance pool to pay for his care. This essentially places the consumer on an indifference curve that is tangent to the new prices but on the original budget constraint.”

            Still, I think the main idea stands. As Cutler and Zeckhauser say in their health insurance chapter, moral hazard is the substitution effect. Maybe that substitution effect should be defined differently in the health insurance case than the price theory case, as Nyman points out, but I think moral hazard should only be used to describe changes in utilization that reduce social welfare.

            • I agree with you. Even though it isn’t quite right, I find “income” and “substitution” effect more helpful and I think of moral hazard as the latter, not the former. I’ve written as much on this blog, before I knew that, strictly-speaking, I might not be using the right terminology.

      • Would you please define and distinguish or provide references to “compensated Hicksian demand curves” and “ordinary Marshalian demand curves.” Thanks.

    • I’m trying to understand this argument and perhaps I’ve failed, but here’s my best effort.

      According to “conventional insurance theory,” insurance leads people to consume health care which they would not consume if they had to pay for it with their own money, even if they had the money to pay for it.

      Nyman’s “new insurance theory” contends that most (as much as 70% of) health care costs paid by insurance would, in the absence of insurance, be willingly paid for by the consumer if he had the money to pay for it. That is, in most cases the consumer’s valuation of his medical care exceeds the cost paid by insurance, so there is actually a welfare gain, not the welfare loss assumed by conventional insurance theory.

      Assuming Nyman’s theory and computations are correct (I’m not qualified to judge), that still leaves 30% or more of insurance paid medical care which would not be consumed if the patient were paying with his own money. If that 30% were a line-item in the national health care budget, I think we might want to cut it.

      But that 30% is largely beside the point because, in my opinion, Nyman’s argument does not justify his conclusions.

      Imagine an 80-year old man with newly detected but slow-growing prostate cancer. It’s my understanding that an 80-year old man is likely to die of other causes before he is killed by slow-growing prostate cancer.

      Nevertheless, this particular man is able and willing to spend his own money for state-of-the-art prostate cancer treatment. His valuation of the treatment exceeds the cost and there is a welfare gain, therefore, according to Nyman, we (the government) should subsidize his health care insurance to pay for his treatment. This is a case of 2 + 2 = 5.

      There is no reason to conclude the government should subsidize a medical treatment simply because an individual would pay for it with his own money if he had the money. On this basis, the government would also subsidize all varieties of cosmetic procedures — breast implants, face lifts, and hair transplants.

      At this point it may appear I’m solidly opposed to government payment for or subsidization of health care. I’m not. I’m saying only that Nyman’s argument is not the proper justification.

    • Nyman’s article does not correctly reflect the economic definition of moral hazard or the character of insurance policies. The $40,000 payout of a health-insurance policy is not an “increase in consumption”; it’s a transfer in consumption from likely, good-health states (where I pay a small amount of insurance and receive no payoff) to unlikely, bad-health states (where I get an insurance payout).

      Moral hazard is defined as a hidden change in _behavior_ which makes the insured event more likely. If the insurance policy only involves a fixed cash payout, and the insuree cannot take any hidden action which increases the probability of a payout, there is no moral hazard. Where is the purported moral hazard in Nyman’s breast cancer example?

      • Not quite. There is ex ante moral hazard, a change in behavior that makes the insured event more likely. And there is ex post moral hazard, a change in behavior that increases the insurance payment. In health care, ex post moral hazard predominates and is the focus of considerable work, including Nyman’s.

      • If the insurance policy only involves a fixed cash payout, and the insuree cannot take any hidden action which increases the probability of a payout, there is no moral hazard. Where is the purported moral hazard in Nyman’s breast cancer example?

        Even if the insurance pays out with an in-kind subsidy for medical care rather than a fixed payment, describing the distortionary substitution effect engendered by the subsidy as a kind of “moral hazard” is deeply confused and has no support in economic theory. In the health-purchase case, the insurer has full information about what purchases are made, so the insurance contract can be freely altered to mitigate any distortionary effects: the insuree can choose her preferred contract ex-ante based on her projected demand for health procedures. By contrast, under true moral hazard the agent’s actions are hidden and the insurance contract can only affect them indirectly.

    • My family experienced Nyman’s example 6 years ago.

      The American health care system does not allow a consumer of any significant service to know the cost of that service, nor the amount their insurance provider paid. Once the deductible and out of pockets have been met the consumer has not a clue about the actual cost of any procedure.

      We got a bit of a peek: we were sent a $1700 bill by the surgical lab because they refused to accept the $700 our major insurance provider was willing to pay. And this was after our surgeon said she was dropping us unless we paid an extra $1000, We agreed and then were billed an extra $2000. (All this was after 6 months of chemo and before 3 months of radiation.)

      The example is all wrong. Modern insurance does not work in the manner described, and the cash price for any medical procedure is vastly higher than what an insurance provider will pay. In the example above it is very likely the self pay price of the procedures would be at least double that paid by the insurance company. And the patient would have no idea about the costs if they have very good insurance and they would have not been able to afford it if they were a self pay.

      How do the calculations take than into account?

      The analysis and discussion is interesting, but the example seems akin to Elizabeth paying for a unicorn, or a unicorn and a leprechaun. I wish it wasn’t that way.

    • “All too often people are too scared to ask a simple question like “Is there a comparable generic available for this drug?” even though it could save them hundreds of dollars and their insurance company thousands.”

      The insured consumer is only interested in how much they pay out of pocket. For example I recently had this conversation:
      ” Mr. Smith, the wholesale price of Nexium is over 100 dolllars a month but OTC prilosec is roughly $15; perhaps we can try the prilosec or generic omeprazole”
      ” My good doctor, my copay for nexium is 10 dollars for a month’s supply; do you not understand simple math?”

    • It’s anachronistic to believe there is any more likelihood that skin in the game will convert Elizabeth into an effective medical buyer than into a concert pianist. Or cardiac surgeon.

      In theory but only that the insurance carrier could play that role. But that fatally conflict with its mission of maximizing shareholder return.

      Therefore the political support for “single payer” under which that skilled buyer’s mission is instead to maximize patient’s interests..

    • BTW 3 unrelated points

      1. One keep in mind could die or have some other bad outcome due to an error in breast reconstruction surgery.

      2. Breast reconstruction surgery could be considered to produce a positive externalize.

      3. Robin Hanson commented on this May 8, 2011 10:45 am.