We’ve all heard the ritual incantation: Kenneth Arrow showed that markets fail in health care, so government must intervene.
What comes next is dealer’s choice. You may be in for a pitch on regulating nurse practitioners. Or against physicians dispensing medicines. Or for price controls on pharmaceuticals. Or for abolishing profit, private health insurance, and human nature itself on our way to a glorious future with Medicare for All. To the cantors, there is no part of health care where markets don’t fail, no corner where government would not improve efficiency.
Only…Kenneth Arrow said no such thing.
In 1972, Arrow won the Nobel Prize in Economics for “pioneering contributions to general economic equilibrium theory and welfare theory,” in part for demonstrating that democracy kinda stinks.
Nine years earlier, the American Economic Review published Arrow’s “Uncertainty and the Welfare Economics of Medical Care.” The Nobel committee did not mention that article. Still, as Berkeley health economist James C. Robinson wrote, Arrow’s 1963 essay is “a good article by a great economist, a creative application of the theory of risk and uncertainty to the thorny problems of the health sector, exactly the sort of boundary-crossing, barrier-penetrating work that opens the possibility of progress in thought and action. Would we have more of the same.”
Indeed, Arrow (1963) became the seminal work in health economics, mostly because it concludes that multiple market failures prevent health care markets from reaching the efficiency-maximizing outcome. “The central proposition of his article,” Robinson summarized, is “that health care information is imperfect and asymmetrically distributed.” Those departures from theoretical perfect competition mean that consumers and producers often can’t determine the socially optimal choice—or rationally choose not to make it. Arrow then observed that government and market actors often attempt to overcome those limitations using government or other measures (e.g., codes of professional ethics). Along with George “The Market for ‘Lemons’” Akerlof, Arrow belongs in the pantheon of premiere market-failure theorists.
Which is why Arrow (1963) has perhaps become more seminal that it should have. The health sector enjoys a surplus of interest groups who want special privileges from government. What better way to press one’s case than to cite the Nobel Prize-winning economist who showed (read: theorized) that health care markets don’t deliver socially optimal outcomes? Robinson explained that Arrow (1963) achieved fame largely because, for both the industry and ideologues, there’s gold in them thar hills:
[Arrow’s] article…has been seized upon to justify every inefficiency, idiosyncrasy, and interest-serving institution in the health care industry…It has served to lend the author’s unparalleled reputation to subsequent claims that advertising, optometry, and midwifery are threats to consumer well-being, that nonprofit ownership is natural for hospitals though not for physician practices, that price competition undermines product quality, that antitrust exemptions reduce costs, that consumers cannot compare insurance plans and must yield this function to politicians, that price regulation is effective for pharmaceutical products despite having failed in other applications, that cost-conscious choice is unethical while cost-unconscious choice is a basic human right, that what consumers want is not what they need, and, more generally, that the real is reasonable, the facts are functional, and the health care sector is constrained Pareto-efficient.
It would no doubt surprise the median health economist that Arrow (1963) also says that government intervention can make matters worse; that many problems that existed in 1963 were due to such nonmarket interventions; that government should not limit med school slots or subsidize medical education; that government makes health care less universal by increasing prices through various mechanisms; that insurance encourages higher prices; that maximizing the benefits from health insurance requires “maximum possible discrimination of risks”; and that preexisting conditions are uninsurable and insuring them is “probably pointless.” Ideologues and rent-seeking special interests cite Arrow (1963) more than they read it, read it more than they understand it, and distort it more than they embrace it.
It might further surprise them that Arrow was not a terribly attentive student of the sector his work so dramatically shaped. By 1999, the health sector had overtaken every other economic sector in terms of congressional lobbying expenditures, a distinction it has held ever since. Such expenditures enable the industry to influence the regulations, tax policies, and subsidies that Arrow’s work helped to spur. In 2016, when advocating the creation of a Canadian-style health system in the United States, Arrow shrugged, “Of course, [Nobel Prize-winning economist] George Stigler would say that there could be regulatory capture, but so far it doesn’t seem to have happened really.”
When theory and reality conflict, what’s a social scientist to do?
Click here to read Arrow in his own words. For highlights of how Arrow (1963) differs from how ideologues and special interests portray it, read Kenneth Arrow’s 1963 Article on Health Care Doesn’t Say What You Think.
Michael F. Cannon (MA, JM) is director of health policy studies at the Cato Institute.
READER COMMENTS
David Seltzer
Apr 22 2025 at 10:22am
Michael: Good stuff. “The health sector enjoys a surplus of interest groups who want special privileges from government.” IMO, there is regulatory capture combined with public choice theory. Elected politicians, serving their clients and interest groups, are motivated by self-interest to retain power and budgets. socialized medical care advocates believe healthcare and insurance are special and require gov intervention. Some ways they are special. That’s where gov officials mistakenly seize the opportunity to reject free market trade-offs. Market based principles have made auto insurance work reasonably well. It seems those principles can be applied to health insurance to make health insurance and health care better and more affordable. I suspect only when Medicare collapses, market based policies will be considered.
steve
Apr 22 2025 at 10:35am
I would mostly agree with all of this. While Arrow noted that market failures were seemingly inevitable in health care he didnt offer government intervention as a panacea. However, he wrote this in the 60s, well before Medicare was established and we had years of govt run or provided health care throughout most of the first world and some of the second. While govt can make things worse in any given area what we actually see is that with government intervention health care can be universal, cheaper, on par or better than US health care and almost universally people are happier with health care under their systems than we are with ours.
A common complaint is that takes longer to see doctors sometimes but note that in the countries where you actually get to see your doctor fastest its in countries with govt run health care. In the areas where the US shines in access to fast care its actually access to expensive, specialist care*. So yes, there really is gold in the hills but is actually more of an issue with US health care than it is elsewhere, not that they avoid the issue entirely. Markets are the best pricing system ever found. Let’s hope someone figures out how to make it work in health care someday.
Those countries could also have faster access to specialized care, but it looks like they have made the trade off of having much cheaper care as US health care is uniquely expensive.
Steve
nobody.really
Apr 22 2025 at 7:07pm
I read Arrow as saying that there is no need to subsidize med school if other factors—such as barriers to entry—already reduce competition, making a medical career sufficiently remunerative. But Arrow then states that his paper does not examine the question of whether “there is a social interest in subsidizing professional education….”
Specifically, Arrow says, “Hypothetically, insurance requires for its full social benefit a maximum possible discrimination of risks.” (Emphasis added.) I read this to mean that this kind of discrimination is necessary in an unregulated market to avoid adverse selection.
Yet as I understand it, group health policies (such as those offered by large employers) often offer uniform prices to all employees, and even to members of employee households. It appears that the risk pool of a group policy is sufficiently large, and probably not correlated with some disproportionate likelihood of expensive medical condition, as to make the burdens of maximal price discrimination exceed the benefit. ObamaCare strives to pool the risk of everyone else, thereby providing a similar benefit to insurers and insured as traditional group polices do.
That said, I still ponder what Arrow meant by “maximum possible discrimination of risks.” At the limit, we could imagine a world in which diagnostic tools become so good that each person’s future innate medical needs would be perfectly known. Under this scenario, there would be no “risk” left regarding innate needs, and buying insurance against those needs would become useless. If we then relax our assumptions, moving to a world in which the price to address someone’s innate medical needs would not be perfectly knowable, but would be knowable to the nearest dollar. Thus, some risk would remain, but the resulting insurance market would be darned close to useless.
Specifically, Arrow says, “On a lifetime insurance basis, insurance against chronic illness makes sense, since this is both highly unpredictable and highly significant in costs. Among people who already have chronic illness, or symptoms which reliably indicate it, insurance in the strict sense is probably pointless.”
Again, I wonder if Arrow makes this remark assuming an unregulated market where insurers face unmitigated risk of adverse selection—and where government would not bear the cost of caring for indigent uninsured people. If we made contrary assumptions, Arrow might draw contrary conclusions.
Ok, I’m struggling with how to model this one. Does insurance have the effect of making an insured person wealthier after the insured event occurs? Or of reducing the marginal cost of consumption? Or both?
When a broke, starving person comes to town, she does not add to the aggregate demand for food. But if government buys extra food so that the starving person can eat, THAT will shift the demand for food—and, all else being equal, will increase both price and quantity. Likewise, health insurance might make a sick person wealthier, at least regarding health care matters, and the resulting shift in demand for health care would increase price and quantity. These strike me as desirable interventions.
On the other hand, I can imagine a patient choosing between a standard surgery that will leave a scar of the patient’s hip, or an advanced surgery that would leave no scar but cost more. The patient is unwilling to pay extra to get the advanced surgery—but if insurance were going to pick up the tab, would choose the more expensive option. This strikes me as wasteful.
Does it make sense to characterize the first scenario as a wealth effect, and the second scenario as an effect of reducing marginal cost? Or am I really describing the same scenario in both cases?
Bart Ingles
Apr 26 2025 at 1:11pm
“It appears that the risk pool of a group policy is sufficiently large, and probably not correlated with some disproportionate likelihood of expensive medical condition, as to make the burdens of maximal price discrimination exceed the benefit.”
There’s also the fact that employer-sponsored group insurance is heavily subsidized by tax policy. With government effectively picking up one-third of the tab, the net cost to healthier employees is much closer to actuarial cost. Any remaining differences are probably compensated by convenience.
Obama care does a poor job of this, especially for mid- to higher-income subscribers. As inefficient as the tax subsidy mechanism is for employer-sponsored insurance, it seems to have stumbled into a better solution by accident than government has by deliberate intent.
Knut P. Heen
Apr 23 2025 at 6:07am
Insurance only makes sense for low probability events with severe consequences. It is crazy to insure oneself for being hungry tomorrow. You only introduce unnecessary administration costs for something that happens with certainty. The same can be said about low probability events with small consequences (like breaking a cup or a plate).
It thus makes sense to insure oneself for getting cancer at age 20, but the probability of getting cancer and many other health problems increases exponentially with age, and it is questionable whether it makes sense to insure oneself for these events at age 80. The combined probability of getting some health problem at age 80 is close to 100 percent. Personal savings (self-insurance) is probably the best way to deal with most health issues late in life.
David Seltzer
Apr 23 2025 at 10:11am
Knut: you make good points. Personal story. I was discharged from the military at age 22. I was in fine health. I reasoned, implicitly, the probability of getting a serious illness was far out in the left tail. Instead of purchasing health insurance, early on, I self-insured by assembling a market portfolio of seven randomly selected stocks. Later I purchased the S&P 500 ETF. Full disclosure. As a veteran, I could avail myself of VA clinics. Fortunately, I’ve been free of any serious malady. Sixty years on, the self-insured fund has compounded at about 10% annually. Part of the bet I made on myself was because I engaged in best self-care habits. Strict diet. No alcohol. No substance use and regular physical training. If at my age, I get cancer, I can make a choice as to whether I let it run its course or treat it. Thanks.
steve
Apr 23 2025 at 12:59pm
Contrary to what most people believe not much healthcare money is spent on routine care. The huge majority goes to chronic care; heart disease, lung disease, mental health and cancer. (Cancer has an acute care stage and then chronic for many cancers.) Most of the rest is for acute care. You can reduce your risk by living healthy but there are lots of otherwise healthy young people who end up with a major acute need or a chronic illness they have no control over. Just having kids is a risk. If you have a premie you can end up with a bill in the hundreds of thousands no matter how healthy you are.
Many people decide to go without insurance. I think some of them really do intend to pay out of savings if they do get sick. I can tell you from helping to run a billing service for over 20 years that few actually pay. ( A good stat to keep in mind is that 50% of people account for 3% of health care spending. So your odds if you dont have insurance are at least 50/50.)
Steve
robc
Apr 24 2025 at 9:17am
Adding on to what he said, pre-Obamacare, insurance was so cheap for a single, healthy male in their 20s, it would be far too risky to go without.
Probably even cheaper if female and you declined pregnancy coverage.
Comments are closed.