David R. Henderson  

Krugman Misstates Arrow

Genetics and the Future of Rel... Resources for Home-Schooled Ec...

My co-blogger Bryan Caplan comments today on Paul Krugman's blog post on Arrow's famous 1963 article on health insurance. There is more to be said. Krugman writes:

Both George Will and Greg Mankiw basically argue that we don't need a government role because we can trust the market to work -- hey, we do it for groceries, right?

Um, economists have known for 45 years -- ever since Kenneth Arrow's seminal paper -- that the standard competitive market model just doesn't work for health care: adverse selection and moral hazard are so central to the enterprise that nobody, nobody expects free-market principles to be enough. To act all wide-eyed and innocent about these problems at this late date is either remarkably ignorant or simply disingenuous.

In a literal sense, Krugman is right about the standard competitive model. After all, at UCLA, Ben Klein and Armen Alchian taught us that the standard competitive model, if by "standard competitive model" you mean "perfect competition," doesn't work well even with gasoline stations and repair shops. When a company can invest in reputation, what Ben Klein called "brand name capital," the perfectly competitive model goes out the window. But if you read just Krugman's short post, you might think that Arrow is arguing for a government role in health care, as Krugman is, right? And I would bet that Krugman wants you to think that. Yet, nowhere in Arrow's article can I find such an argument. Rather, Arrow is saying that there are things peculiar to health care and health insurance that mean that we have to supplement our standard models. And, as for free-market principles, although Arrow might think that they are not enough, he doesn't say that in the article.

Unfortunately, many economists of various persuasions have said that Arrow's article makes a case for why competitive markets in health insurance will fail [which is different from saying that standard competitive models will fail] and why government regulation is needed. That's what I had remembered it saying. But, as I noted in January, when I went back and read the whole thing, I found no such thing. In fact, Arrow's article is much more careful and nuanced than I had remembered.

Also, and contrary to Bryan's claim, Arrow points out that optimality in health insurance requires that higher-risk people be charged higher premiums.

I won't be surprised if someone can find Arrow saying good things about socialized medicine or other government interventions in health care. What I am saying is that you can't find him saying those things in his seminal 1963 article.

Comments and Sharing

COMMENTS (12 to date)


You (and your readers) might be interested in the October 2001 issue of the Journal of Health Politics, Policy and Law 26(5). The entire issue is devoted to Arrow's 1963 paper. In a recent post on my blog I reviewed the article in that JHPPL issue by Uwe Reinhardt, which I recommend to anybody wishing to understand welfare economics and its limitations.

RL writes:

Well, I haven't read Arrow's classic piece, but the claims are confusing. I can understand how adverse selection and moral hazard create a problem if the insurance policy is designed to pay for whatever the doctor orders. But insurance of other things--home, auto, etc.--aren't set up in this manner. You get a lump sum if your home or car is damaged and you had the appropriate coverage. You don't HAVE to use them to fix your car or house.

Assuming proper pre-acceptance testing, why would lump sum payments for covered health problems lead to any of these problems? Isn't Arrow simply pointing to a problem in how health insurance is structured rather than to a problem with health insurance qua health insurance?

gnat writes:

I read Arrow's article to say that healthcare has a litany of market failures (moral hazard, adverse selection, asymetric information)that violates the First Fundamental Theorem of Welfare Economics. http://en.wikipedia.org/wiki/Fundamental_theorems_of_welfare_economics
The fact that the private market has made such a mess of healthcare suggests Arrow was right.

David Hendersen points to Arrow's statement that
optimality requires that higher-risk people be charged higher premiums, but I think David misstates his meaning. Private healthcare insurance has developed such that it covers me for a one year term. If my risk changes this year as a result of an illness I may not be able to get coverage next year. The problem is that the private insurance market has not responded appropriately. In part, this may be due to consumer's dynamic or time inconsistency.

mark writes:

You've encountered a specific example of the pundit method:

1) State a really important problem
2) State two possible solutions to it
3) Show problems with one solution
4) Leave the other unexamined but pronounce it the winner.

Obviously this a) ignores the possibility that there are more than two possible solutions and b) ignores the defects in the unexamined solution.

Yancey Ward writes:

Gnat gets at one of the problems with health insurance- it's renewal problems. Catastrophic coverage should be like level term life insurance with the rates set at the time of initial purchase. You can cost the policies based on risk, and people have an incentive to continue coverage. Of course, such policies will have maximum payouts, above which you can no longer pay for your healthcare, but that is only recognizing the reality that consumption of healthcare at the individual level has to be capped somewhere.

There really aren't market failures here- the failures are in our silly belief that everyone is entitled to unlimited resources in the case of medical care.

GabbyD writes:

i have a question:

you write : "Arrow's article makes a case for why competitive markets in health insurance will fail [which is different from saying that standard competitive models will fail]"

in what sense is it different? from a practical perspective, whats the difference between competitive markets and competitive models?

Dan Weber writes:

We had a cite here some weeks ago that said that most states stop insurance companies from denying renewal or jacking up your rates if you get sick. It was an academic-like paper but I don't know that it was peer-reviewed or fact-checked, so it could be wrong.

(Insurance companies can increase your rates, but they need to increase them on everyone in the pool. This still allows them to play games, but it's nowhere near as straightforward.)

Artturi writes:

GaddyD:"in what sense is it different?"

It is different in the sense that if the competitive model fails it means that the market does not behave in a way that the model predicts. Usually when people talk about markets failing they mean that it produces worse outcomes than its alternatives. (Real alternatives, not imaginable.)

Hence, the competitive model can fail (e.g. market does not work in the way the model predicts), when market does not. (e.g. it produces better outcome for participants than its alternatives.)

IF we assume that by markets failing it is ment that the market produces an outcome that is not efficient, then the model and the market failling are the same thing. However, no real market is completely efficient.

kingstu writes:

"The fact that the private market has made such a mess of healthcare suggests Arrow was right."

I'm not sure where this private healthcare market you are referring to exists.

gnat writes:

Here is a good summary
My reading of this and other histories indicates that the "insurance" institutions--really prepayments to ensure payment-- were developed by hospitals, the AMA and corporations: the private market. The result was lack of market coordination, and of risk-based insurance, employer-based compensation that that accentuated adverse selection, moral hazard and asymetric information, and market power.

GabbyD writes:


but the competitive model is a model where the market maximizes total surplus.

i don't get the distinction Prof. Henderson is trying to make.

Jeff writes:

I haven't read Arrow as yet, and am relatively speaking an amateur as far as economic principles are concerned, *but* am a reasonably adept systems analyst.

It strikes me that a fundamental flaw with a market approach to health care is that it ignores the relationship between the health of individuals and that of the society as a whole.

Unlike my economic ability or inability to buy something like a stereo, car or other service, such as education or legal services, if I am unable to obtain healthcare it has a non-trivial and quantifiable effect on many other people around me who are not directly envolved in a medical "transaction".

Let us for a moment consider an extreme model, where we reduce economic transactions for medical services to the same status as buying a shirt.

Unlike the purchase of a shirt, if I have a medical problem, it can have a distinct and disproportionate effect on the society around me.

That can be indirect, such as an inability for me to get to my employer and produce economic output, or my being unable to take care of family or other specific responsibilities, or it can be very direct, as in the case of infections disease.

There, if my medical problem is not managed, it can spread to other individuals, impacting society as a whole. If it is a disease, it could potentially spread to a significant number of people, causing considerable cost to the rest of society in the form of energy expended to treat those other people, or, in lost productivity as a result of the illness. The problem could be mental or emotional, in which case, a problem could evolve into anti-social or destructive behavior that could cause harm to individuals around me.

In short, my assertion is that my health, unlike a pure economic transaction, has indirect effects which the market *cannot* and *will* not accomodate as part of its calculation of cost to those who provide the service (insurance). An insurance company has no stake in the welfare of my employer, nor my children. If the needs of those individuals are not dealt with, the insurance company incurs no cost.

My society as a whole however, can suffer significantly if I am not healthy.

My assertion then is that health care, at least basic health that affects my ability to perform useful work, *MUST* be considered from the perspective of society as a whole, rather than as an individual. I also assert, that that is where a purely capitalist transactional model breaks down, as it cannot accomodate or calculate those broader indirect costs.

As such, I do not see how we can approach public health as an economic enterprise *unless* that competition takes place at the level of nations or communities rather than individuals.

My 2 bits ;)


Comments for this entry have been closed
Return to top