Kenneth Arrow on the welfare economics of medical care, a critical assessment

Kenneth Arrow wrote a paper in 1963, Uncertainty and the Welfare Economics of Medical Care. This paper tends to appear in debates regarding whether healthcare can be left to the market (like bread), or if it should feature heavy state involvement. Here I explain what the paper says, and to what extent it is true.

So lets see what the paper says.

Well, first of all, Arrow doesn’t intent for the paper to be anything more than “an exploratory and tentative study of the specific differentia of medical care”. If he says that, one wouldn’t be surprised to find to find mistakes in the paper, as it is also quite a decades old. But on the other hand, Arrow won a Nobel Price, perhaps we was truly prescient and the whole thing has stood the rest of time.

He than claims that the “special features” of the industry essentially stem from the fact that healthare demand is strongly uncertain from an individual perspective, and that when markets fail, societies will, to some extent, come up with non-market social institutions to fix those market failures. But, he says, these institutions induce non-competitive practices in the healthcare sector, which prevents it from moving closer to optimality.

Special characteristics of the Medical-care market


(Unpredictability) First, unlike food or clothing, demand for healthcare is not as predictable. They are only purchased when one is ill, and illness is be hard to predict. That is, demand is driven by exogenous shocks, and not by one’s own planning. Arrow finds it hard to think of goods and services for which the same thing holds. He can think of legal services (and he points out that there are some similarities with the healthcare sector).

(Criticality) Second, illness can cause death, permanent injuries, and loss of the ability to earn. Given that our life goals need us to be alive, and relatively healthy, it is of supreme importance to us to be healthy. The same holds for food, but food deprivation can be avoided if one has sufficient income.

Here Arrow is largely correct. The list of examples for the first special characteristic would be: healthcare, legal services, gadget repairing, home maintenance, or car maintenance. For the second characteristic, we can cite healthcare, food, water, and housing. Thus only healthcare meets both condition: It is the only possible critical good that has an unpredictable demand.

Or, to be precise, part of healthcare is. Like with some goods usually bunched in a single category, healthcare can be unbundled: cancer and heart diseases can be quite unpredictable and fatal, flu, headaches and cold  can be mild and fairly predictable (usually around winter). There are few predictable and critical malaises, among them aging is probably the chief one. For highly unpredictable and low criticality, mosquito bites is one example.

Expected behaviour of the physician

(Moral provision) Healthcare is a service, not a physical good. Producing healthcare is synonymous with providing it, says Arrow. The customer cannot test the product before consuming it, and there is an element of trust in the relationship. This also holds for a barber. But the ethical standards required of the physician are higher. The physician must honestly take the patient’s interests into consideration and act to further them instead of acting to pursue his own private gain. As an example of the special moral status of physicians, he mentions these special characteristics:

  • Lack of advertising and overt price competition between physicians
  • Advice given by physicians for further treatment is supposed to be divorced from self-interest
  • It is claimed that treatment must be dictated by “objective needs” and not limited by financial considerations
  • The physician is relied on as an expert in certifying the existence of illnesses truthfully, outweighing his desire to please his customers.
  • Nonprofit hospitals predominate over for-profit hospitals (3 per cent in for profit, 30 per cent in voluntary non-profits. The remainder is in state owned hospitals). Arrow posits two possible explanations

Two of this items are at least quite false, the others are argueable. Starting from the bottom, he might have been true about the US in 1963. But he didn’t look further beyond (Perhaps the information wasn’t available then). Had he done so, he would have found this (from OECD data, 2015)


For the countries that we have data for, most private hospitals are for-profit, except Korea, Luxemburg, and the US. And even in the US, while there were 10x more non-profit hospitals back in 1963 as Arrow says, but 1975 there were only 4.3x more. By 2013, only 2.7x more.

Next, lack of advertisement and overt price competition between physicians. This also has to be unpacked. We don’t see either advertisements from particular individuals for commercial purposes. Smaller entities usually rely on word of mouth and small scale networks to gain reputation and customers. This holds for bakeries and small medical practices. For larger entities, advertising does make more sense. We can find advertising for health insurance (here  or here). We can even find advertising for hospitals (herehere, here, here) I will grant that there is’t much advertisement for hospitals, it is true.  This may be because people don’t care much for a particular brand (like, in many cases, with supermarkets). The services you can get at one, you can get at all of them, and many people will go to the nearest one. The rare exceptions would be elite hospitals in one’s own city or country where they may have state of the art facilities and highly skilled specialists. These hospitals would also fall under the local network advertisement effects. In a given city, there may be two or three of these, and given that hospitals tend to stay in the same place for a long time (decades), they will get to become well known. This is to some extent like hotels. Hotels don’t advertise that much for the same reasons.

Next, price competition is indeed an issue in the US. It is difficult to find prices for treatments online in the US. In the UK it doesn’t seem that difficult. It is more difficult for Spain. My overall assessment is that indeed there is little price competition -as Arrow says. Why the situation in the UK is like that and not in the US? This deserves further investigation, but again shows that lack of price competition is not an intrinsic characteristic of this market.

Next we have the extra bundle of ethical obligations that physicians are said to have. But in this regard, they are not prima facie dissimilar from a butcher counseling you on what the best cuts of meats are for a given purpose, or a data science consultant (like me!) you hire to help you improve your business, or a car mechanic, or even a CEO. In those cases, you know less than the professional about what you really need, so some intrinsic moral motivation is needed for a fully fruitful and honest transaction -at least in the short term. In the long term, reputational effects may kick in-. So this is not a special feature of the moralisation of healthcare, but of the fact that doctors as professionals know more than you do. Fortunately, not only in these cases but in most, people want to be good at doing their jobs. Having a self-image of “a good doctor” can be very important for one. (See section 2 here for example).

Product uncertainty

(Asymmetric information) With many commodities, you can learn from your or other’s experience about the nature of the good purchased. With illness, it is more difficult. You sort of know what a good mango is, and how pricey it should be. If not, you can easily find a mango-eating friend. But it is far harder to do for cancer, or hip surgery.  You don’t know beforehand what you will be getting. The physician is aware of this, and knows more than the patient. Arrow says this is uncommon.

While in general it is true, it is not true for the services mentioned in the previous section. You don’t know what is needed to repair a car beyond broad details, for example…. unless you want to. What does a hip replacement surgery involve? No idea, but then I can google “hip surgery” on google and get an idea of what it takes. You can see that it will involve surgery with anesthesia, that it will cost between 8 and 16 k£, that it involves removing the damaged hip joint and replaced with a metallic or ceramic artificial joint. You can also see the side effects, and expected recovery time. This greatly closes the gap in knowledge. Perhaps this information wasn’t easily available in  1963, but now it is.

Supply conditions

(Licensing) The supply of is commodity is tied to the expected return from its production. For healthcare some elements insert a wedge between returns and supply. One is licensing, which restricts supply. This policy tends to be justified by the desire to provide a minimal level of quality. I personally doubt that to get decent healthcare (or services in many other regulated professions) the current standards are needed, but that’s another matter. In the US, education of physicians is also tightly regulated, resulting in expensive education, and adding to the limitations on the supply of physicians.

Pricing practices

(Price discrimination) The sector practices it. Hospitals charge the rich more and the poor less (even nothing for the truly poor). There is also opposition to pre-payment from the side of hospitals, and to closed-panel practice (binding the patient to a particular group of physicians). Arrow provides no evidence for this, altough the first part seems plausible enough to accept it without further investigation. The second part is hard to understand, as one common insurance offering is precisely to tie the user to hospitals or medical experts tied to the insurance groups.

Comparison with the competitive model under certainty

Nonmarketable commodities

(Negative externalities) The case of vaccination as fixing a negative externality.

(Other-regarding preferences) People care for other people’s health. People usually don’t care about what other people do, so this is another difference.

Well, yes.

(Increasing returns) Hospitals show them. Equipment and specialists are indivisible, so it makes sense to aggregate them in units (hospitals) that work better than if the components were separate. This can decrease competition.

But, as he admits, transportation and concentration in cities ameliorates this issue. In addition, the prevalence of small clinical practices in many parts of the world suggest that returns are not that increasing in general. They may be for certain surgeries, or certain high-end equipment.

(Licensing and educational costs) Discussed above

Arrow says that if the requirements were relaxed, the output would be of lower quality. This need not be so if education acts to some extent a signaling rather than as human capital buildup, or if there are faster ways of training doctors than currently practiced, or if some procedures can be left to less trained specialists (Who will perform them equally well).

(Price discrimination) Price discrimination, says Arrow, is inconsistent with the competitive model.  Here I wonder to what extent price discrimination exists in reality. I imagine that it will be more prevalent there were hospitals do not publish their prices, as if they did it would be difficult afterwards to charge more to some. There is also some rationale for that practice if hospitals seek to subsidise the poor at the expense of the rich. I am thus willing to concede that Arrow is right for the US.

(Moral hazard) Being insured reduces one’s incentives to avoid illness, and on the other hand insurance companies will try to pay as little claims as possible.

(Predictability and insurance) The more unpredictable an illness is, the more risk averse individuals will be willing to pay to insure against it.

(Pooling of unequal risks) In theory, insurance companies would charge more people who are less healthy, but there appears to be a tendency to equalise costs, which would not exist if markets were competitive, says Arrow.

(Gaps in coverage) Not everyone is insured.


What then, makes healthcare special? Summarising the above, and dividing healthcare into its components, plus adding some of examples that share similar “weird” features:


Externality Uncertainty Life Critical Price Information assymmetry Moralisation
Communicable diseases (vaccines) Yes High Some Low Medium Medium
Major medical treatments (cancer, heart surgery) No Very high Yes Very high High Very high
Minor medical expenses (painkillers, cold medicine, medical checkups) No High Not usually Low Low Low
Car breakdown / Legal services No Very high No Medium High Low
Food No None Yes Low Low Low
Consulting services No Medium No High High Low

Most of the debate around healthcare tends to focus on a piece of it: major medical treatments (and secondly on drug prices). One rarely knows when one will need them, they are very expensive, and you don’t know exactly what the treatment will involve (initially). Because most people cannot afford them, insurance is typically needed. Insurance need not (but can) be expensive.  In the US, monthly premia can range from 150 to 470$ in the individual insurance market, while in Spain, you can get the best insurance available in the market for around 115€, and a cheap one for 50€.

Those treatments are also moralised in the minds of many people: they have to be available to everyone, without regards for the cost (!). Some people think some features of this market are proof that the market for healthcare doesn’t work. One example is pre-existing condition coverage. Someone with cancer who purchases a new policy (even if he manages to!), won’t be covered for expenses due to cancer. But this is a feature, not a bug: The premium is really the expected cost to you plus a markup. If the cost is known with certainty, then the premium equals the total amount. Trying to “fix” this feature of the market by doing things like mandating that everyone pays the same, or that insurer cover everyone undermines the workings of the market. If one really wishes to cover those people, it would be far better to redistribute towards them.

The second aspect of the paper is to explain how these peculiarities incentivise the creation of institutions to overcome them. Insurance is what happens when you have budget contrained, risk averse people, and highly uncertain pricey treatments.

Other features of this market are not as clear cut as Arrow explains in his paper (the uniqueness of medical ethics, or the status of hospitals as non-profits, or medical licensing).

What special features, I add, would we really expect to see to overcome the peculiarities? Transparent pricing at hospitals, websites to do healthcare-shopping easier, and lifelong all-risk health insurance. As I’ve shown in with a few examples, these features are not similar everywhere. The US ranks very low on transparency, while the UK ranks high. It is easy to get affordable cover for life in Spain (covering even non-preexisting chronic conditions like diabetes), but not so in the UK, except for cancer. Why are these markets different? I don’t know, but it is a topic worth looking into.

These points and many others have been already addressed in this article, which I didn’t initially notice. This other article deals with the relevance of asymmetric information in modern times. Here’s John Cochrane reply to Noah Smith.

I am currently reading through Jonathan Gruber’s Public Finance, where a modern treatment of this topic is presented. I haven’t gotten to the health insurance section, so for now I can say that reading Arrow’s paper fails to convince me that healthcare cannot be mostly left to the market. I don’t expect that this post will win you over to my side of the question. For that I would need something focused specifically on that, not just a critical assessment of an old interesting paper. I should probably also have a look at chapters 20-25 of the Handbook of Health Economics.

A fuller assessment of the question would involve a presentation of the most up to date theoretical treatment of the issue, empirical evidence on how modern day healthcare markets work around the world (beyond the US or Europe even), and historical evidence about how healthcare markets became what they are now.

Finally, I want to mention that some people grossly misunderstand what Arrow says in his paper. There he is not arguing or defending a single payer system. He is not even saying that “the laissez-faire solution for medicine is intolerable” as some people have said. That last line he does say, but referring to medical licensing, and that is not something he believes, but something that he says is the general social consensus.

If you have sources that deal with this market that I should read, please do tell me in the comments. The landscape of the debate seems one where there is consensus among economists that a free market in healthcare is too troublesome, yet then there are people who easily poke holes in those arguments and then they go unanswered. I stick to my principle that one should defer to a scientific consensus where it exists unless there are reasons to doubt it, and this case seems like one of those exceptions.


This entry was posted in Blog. Bookmark the permalink.

2 Responses to Kenneth Arrow on the welfare economics of medical care, a critical assessment

  1. Pingback: Los sistemas sanitarios: algunos enlaces. | PHILONOMICS

  2. Pingback: Rational Feed – deluks917

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s