Paul Krugman on health care

…Basically, everyone agrees that health care is a messed-up sector. But
there are two opposing doctrines about what the problem is.

I believe — and the evidence, I think, supports this belief — that the
big problem is "adverse selection." An insurance plan offered to
everyone at the same rate would be a great deal for relatively sick
people, a poor deal for the healthy. So one of two things happens to
private insurance. Either plans go into the "adverse selection death
spiral," as sick people flock in, driving up rates, driving out more
healthy people, and so on. Or insurance companies spend a lot of the
money they receive in premiums screening out "high-risk" clients, so
that the system has huge overhead and the neediest cases are excluded.

The clean solution to this problem is for the government to provide
insurance to everyone. Other rich countries do that. So do we, for
older Americans, veterans, and others. Actually, government health
insurance is already bigger in America, in dollar terms, than private
insurance — it covers fewer people, but that’s because the elderly, who
cost more, are handled by the government.

Employment-based insurance is a distant second-best, but better than
nothing. Large employers, in particular, can spread risk widely,
creating the kind of risk pool that dies from adverse selection in the
individual market. And the tax preference for employer-based care, more
or less by accident, has helped sustain this imperfect fix — which is
why I’m highly skeptical of anything that might erode that preference.

What conservatives in the "consumer-directed" health movement believe,
however, is that the big problem is "moral hazard" — people consume too
much medical care, because someone else pays for it.

Now, this isn’t entirely wrong. People probably do undergo expensive
surgery with questionable effectiveness, and so on, because it’s not
out of pocket. Curbing that was supposed to be the point of managed
care. But managed care didn’t deliver, because people — rightly — don’t
trust private H.M.O.’s to make life and death decisions on their
behalf. Successful managed care only takes place in institutions like
the V.A. where there’s more trust in the institution’s motives.

The whole consumer-directed thing is, in my view, just at attempt to
avoid facing up to that failure. Rather than admit that private-sector
institutions aren’t any good at rationing, conservatives now say that
patients should be induced to ration their own care by being forced to
pay more out of pocket. And that’s where Bush’s attack on gold-plating
comes from: reduce the tax advantage of employer-based care, and
deductibles and co-pays might go up.

The trouble is that the big money is in stuff like heart operations or
other areas where (a) people can’t pay out of pocket in any case — they
must have insurance or go untreated — and (b) people really aren’t
sufficiently well informed to make the decisions. Yet the whole focus
of consumer-directed doctrine is on things like routine visits to
doctors’ offices and annual dental checkups. It’s going where the money
isn’t — because the advocates just can’t believe that markets aren’t
always the answer.

Now here’s the thing: in the name of consumer-directed health care
theory, Bush is proposing changes that would essentially encourage
people to move into the individual market — which wastes a lot of
money, and doesn’t and can’t work for those most in need — while
undermining the employer-based system, which isn’t wonderful but is
still essential. In particular, healthy high-income people would be
encouraged to drop out of employment-based plans, leaving behind a
sicker risk pool, driving up rates, and pushing employer-based care in
the direction of an adverse selection death spiral. The plan we’re
supposed to learn about tomorrow doesn’t sound big enough to have
catastrophic effects, but it’s a step in the wrong direction.

I might add I don’t think adverse selection is the major problem, nor for that matter would I cite moral hazard.  I would cite the (temporary?) difficulty in evaluating procedures and outcomes, plus consumer irrationality, noting those same consumers still could be rational in choosing insurance plans, even if they screw up their on-the-spot medical decisions.  In my preferred but still imaginary model, consumers rationally choose insurance companies ex ante, those insurance companies make and fund key medical decisions ex post, and other third party intermediaries keep those insurance companies honest.


Adverse selection is a big problem in how the insurance industry works. There is substantial competition for the healthy and incentive to discourage enrollemnt or reenrollment of the sick, rather than assure their optimal care. Risk-related premiums would overcome this but make insuring the sick expensive and, in an employment-based insurance world, create incentives for employers to avoid employing those who are sick or whose families include the chronically ill.

I've written a long comment in response to the earlier post "The new Bush health care plan" expanding on this and its implications.

"In particular, healthy high-income people would be encouraged to drop out of employment-based plans..."

But then they take private insurance. And if the adverse selection problem is worse in private insurance markets, then that's where we want healthy people going, yes?

It looks like the employer-based system worsens your adverse selection problems, because less healthy people are less likely to be able to work, so incentivizing people towards employer-based health-case would make your problem worse. So, conversely, Bush's undoing of that bias should alleviate the problem.

From the linked-to post on adverse selection:

"Everyone talks about adverse selection in the market for health insurance but in fact non-group policies in these markets are not relatively expensive and not hard to get. The national average annual premium for reasonably generous coverage for a single person is just $2,268."

Either I'm a liar, or you're misinformed. Since I had to change my entreprenurial plans in order to get health insurance (the spouse went to work for a "big enough" company), I think you're misinformed. If a healthy person with about as mild a history of cancer as it's possible to have can't get insurance for *any* price, it doesn't matter at all what the average price is.

This is all very disappointing. Extremely poor information dissemination among people who should know better bodes poorly for sanity improvements in the future. Have any of you policy wonks ever worked for a company with 10 employees or fewer? Ever talked to a lawyer who just made partner?

Jack Needham and Russel L Carter:
You don't know what adverse selection is. Read Alex's post on adverse selection again, particularly what he says in the comment section.
If Krugman is right that insurance companies spend a lot of money screening applicants, that would be adverse selection. As far as I can tell, they don't spend a lot of money, they just reject a lot of people. That's not adverse selection.

If they're overzealous in screening people out, maybe that's adverse selection, but it's not an "adverse selection death spiral." It is mysterious to me why they don't offer a plan to Carter that simply excludes cancer treatment.

"As far as I can tell, they don't spend a lot of money, they just reject a lot of people."

"It is mysterious to me why they don't offer a plan to Carter that simply excludes cancer treatment."

Doug, it is essentially free for a computer to scan for a fixed set of rejectable premises on the very detailed individual health insurance applications. Have you ever filled one out? This is done with all small companies by the way, not just individual applications. I was the vice president of a 10 person company that was rejected by several insurers because *as a group, including children and spouses* our collective pool was too risky[1].

Why don't all you smart people know this? I find this very mysterious.

Now where is all this going? I don't think any of you that don't think adverse selection, or whatever it is, that keeps people who are willing to pay even exorbitant costs in order to obtain at least catastrophic insurance off the insured rolls have made your case. I can appreciate your position if you've always had a large company or university health plan; I had the best in the world when my doctor in Healthnet was a tremendously cool professor in Stanford medical school. But there's a whole nother world out there of small business and garage entreprenuers, and we see a vastly different world than you do.

[1] At the time, I didn't think much of it, you know, we'll just find somebody else to insure us. It didn't get rammed home until I went out on my own, and couldn't get insurance for *me, and my family*.


Adverse selection means that individuals who are more likely to use insurance seek it. If premiums are based on the average risk in the population, the premiums will not cover the costs of the insured group. If an insurer prices the policy higher to reflect the actual risk profile of the insured group, healthier individuals are more likely to drop the coverage, since the value of the insurance relative to expected risk is lower. As they drop out, the premium to cover the less healthy risk pool needs to keep going up, further encouraging dropping of coverage by the lower risk. That's the adverse selection death spiral.

Health insurance companies would prefer to insure the healthy, and the process for estimating the risk of individuals via underwriting is imperfect at best. They respond by using very aggressive underwriting where they can, and may also respond by making the coverage they provide less attractive to the sick (more restricted lists of specialists, more hoops to jump through to have care approved, etc.). Perhaps not adverse selection per se, but a response to it, and one that contributes to health insurance serving the needs of the sick poorly.

Seconding buzzcut, I would have to suggest that Paul Krugman doesn't know many veterans (not too surprising, perhaps). In my experience, veterans complain about the VA at least as much as the average American complains about his or her HMO. Alternatively, maybe Krugman walks by faith, not by sight, and has the will to believe that patients like the VA, so he simply can't hear the veterans he knows.

Russel Carter and Jack Needleman:
I'm sorry for my tone. Adverse selection is a problem, but I agree with Tyler that it's not the problem. I object to Krugman's claim that private insurance is in a death-spiral, and everything he says follows from that; everything you say seems compatible with my belief. I think a more central problem is that everything is absurdly expensive. If the actual cost of insurance for RC did not look "exorbitant," perhaps a company would have made him an offer.

Mainly I object to Krugman's claim that private insurance is in a death-spiral. JN: your description of a death-spiral is correct, but I don't see it occurring in private insurance. Krugman claims that these companies dissipate money trying to find the healthy, while I claim it's cheap. RC seems to agree with me, but seems to claim he doesn't.

Since I don't believe Krugman's claims about how much worse private insurance is, I don't see moving in that direction as a bad thing. It's not obviously a good thing, either, but it is plausible that if the private insurance industry grew, it might become more flexible, and be willing to insure RC. Of course, if it doesn't, and the employer-sponsored system is destroyed, then he's worse off.

Anything that gets us closer to Singapore, like moving to catastrophic insurance, seems like a pretty good thing to me. The complaint about risk pools is absurd. If you want the healthy to subsidize the sick, do it through taxes.

I said I think the problem is that everything is expensive. Perhaps it's worth the cost, but we can't decide without knowing what the costs really are, and we don't know that because all the prices are fake numbers produced by hospitals and insurance companies trying to trick each other. Paying for medicaid, medicare, and charity cases are other sources of deception, but probably not as bad.

It's not clear that moving to catastrophic insurance would actually solve this problem, but it's hard to see how it would hurt. It's not clear that solving this problem would solve the adverse selection, but it's hard to see how it would hurt. Singapore shows that these problems are solvable, though I admit that doesn't mean it can be replicated.

Moving to some kind of socialized system would solve the adverse selection problem. It might even make price information better by getting rid of the insurance companies, but I suspect it would get worse in the long run. I suspect that European systems, which are all rising exponentially and faster than GDP, get some price information from the US. I must admit that Singapore's health care is also rising faster than GDP, but I think slower than elsewhere.

Mauro Kochen: Maybe this what you're saying is not true, but I'd guess they have some kind of agreement. My last insurance plan involved 20% being taken off the list price immediately. This is the kind of distortion I was talking about above. The other thing that springs to mind is that maybe the rich don't have insurance.

I think Krugman is largely convincing, but why not a combination of government-supplied insurance *and* 'consumer directed' policies?

That is, the government policy would not cover 100% from dollar one but would require significant deductibles and co-pays (the amount could vary as a function of income). This would make government-provided insurance substantially less costly and create incentives for individuals not to over-consume routine care. HSAs could even be included to help with the deductibles and co-pays. Patients no longer run into the doctor's office to demand a useless antibiotic for every sniffle (or if they continue to do it, they pay themselves), but everybody's covered if they need a heart bypass operation or cancer treatment.

Also, anyone discussing individual health insurance should mention which state of the Union that they live in. There's a law banning buying health insurance coverage from another state, and state regulations about health insurance vary considerably. Not just in states mandating that plans cover certain items, but also New York and New Jersey attempting to ban insurance companies from taking prior conditions into account. The latter results in a real problem of adverse selection, since the insurance companies are not allowed to price the customers differently, and hence it doesn't make sense for the healthy to get health insurance. This does result in a death spiral. NY and NJ individual health insurance rates are MUCH, MUCH higher than the rest of the USA.

Since Professor Krugman is at Princeton, his perspective about the adverse selection death spiral might well be skewed.

The debate here sounds so *normative* with almost complete absense of counterfactual examples or cohorts (read: every other industrialized nation with the exception of a city state). You are well trained.

I'm surprised nobody mentioned the reinsurance option. Mr Krugman, shame on you.

I think people are completely missing the point by focusing on the now and not the soon. More precisely you need to think about the future of dna profiling relative to adverse selection. It is relatively easy to scan ones DNA for markers even now and the number of known markers that change the probability of a given disease is increasing rapidly. (For example I am homozygous for a mild genetic disease called Hemochromotosis that increases the probability I will develop the disease 9X). It is clear that, within a few years an insurance company will be able to very accurately calculate "the actuarial value of a persons genes", and at this point health insurance that is based on individual coverage must end.

More precisely you need to think about the future of dna profiling relative to adverse selection.

You don't mean adverse selection, you mean something else. Adverse selection is when there is an asymmetry of information-- when certain customers know that they are a worse risk but insurance companies don't. In that situation, the customers who are a worse risk buy lots of extra insurance, so the companies pay out more to those people and have to raise premiums. This causes premiums to increase past the level where the people who know that they are a better risk feel that it's worth it, so they drop out. This causes the insurance companies to have to raise premiums again, since the risk pool just got worse. That is the death spiral of adverse selection that Professor Krugman and others warn about.

The problem you point to is not the economic problem of adverse selection, since both individuals and insurance companies would have access to the information. It is more of a social justice problem, where someone has drawn a bad result in the genetic lottery of life. The more certain the knowledge of congenital diseases become, the more it's about redressing the bad genetic hand someone drew rather than guarding them against an unlikely occurance that could happen to anyone. There are, of course, some genetic outcomes that we don't compensate for; we don't pay men who were born short or women genetically predisposed to be fat for the loss of income and mating possibilities. However, diseases are generally a bit more serious, and I certainly am not arguing that mandatory redress for rare genetic illnesses will lead to Harrison Bergeron. Certainly states have considered and in some cases implemented funds for children born with cerebal palsy, paying them and their parents for their poor luck. (Not in NC, of course, where John Edwards successfully led the opposition, since it would keep him from being able to sue on their behalf.)

Of course, one can hope that earlier screening and better knowledge will also come with better abilities to treat and mitigate congenital diseases. There's also the possibility that nearly everyone will have some congenital disease, it will just differ what from person to person and risk pools will still be possible. Also, if the chance of getting a disease is rare and it happens late in life, then it may have no more effect than finding out what is most likely to be a problem, but not necessarily raise costs.

(It's also possible that companies will offer insurance that will pay for anything other than the disease one is likely to get, or pay a reduced amount for that, depending on the propensity to get it.)

John Thacker nails it. We can have extremely cost-effective and beneficial national health care the moment Americans are willing to submit to technocratic regulation of their health care. It would be easy to make a list of cost-effective medical procedures and interventions that would be universally covered in a system with relatively small tax costs, particularly if the system replaces Medicare and Medicaid.

Ain't gonna happen. Hell, look at the unnecessary and cost ineffective treatments that insurance companies cover for fear of being sued or out of fear of bad press. Our political system will inevitably lard up a national health care system that will cover everything for everybody, particularly the politically powerful senior lobby.

Our real hope is continued cost savings due to pharmaceutical treatments, which save a lot of money by enabling non-invasive surgical treatment, outsourcing of health care to cheaper but still high-quality foreign health care providers, and technological innovations (and openness to those innovations) that give more people access to good treatment. For example, diagnostic software outperforms most doctors, so incorporating that more widely could be very effective.

That doesn't leave any insurance for the sick. Therefore, I think it is reasonable to conclude that people who hold Tyler's views don't actually intend for a functioning private health insurance market to exist. This is all about keeping healthy person's expenses as low as possible.


You are misrepresenting my position horribly. I am not claiming that the problem doesn't exist, I am simply claiming that it's not adverse selection, and it's not insurance. Insurance is about defending against rare outcomes.

Once you know *for certain* that you're going to get a disease, it ceases to be insurance at all. Even if you get totally reimbursed for it, that's not insurance, that's being reimbursed for being unlucky enough to be born with bad genes.

I agree that it's a real problem, because it transforms the entire problem from insurance to transfers from people lucky enough to have good genes to the unlucky ones. I didn't say that I was against that sort of compensation at all, either. Merely that it's not adverse selection.

There is no "intending for a functioning private health insurance market to exist." The problem is that such a thing *can't* exist as regards diseases that can be totally predicted by either just the patients, or the patients and the insurance company. It ceases to be insurance. It's still a huge problem for those people and for society, but it is not insurance and it would be impossible to solve within a private health insurance market. Such redress for losing the genetic lottery could only be sought at a social justice level, only from government. But it's not insurance. It's healthy people taking care of sick people because it's the right thing to do.

Insurance could possibly be saved for most people if the genetic markers simply show a slight propensity to various diseases, not a certainty, and nearly everyone has some sort of propensity towards something, and the genetic screening actually helps save costs by helping us target treatment and know what to avoid. But there is no way to have a functioning private insurance market for people for whom congenital diseases are a surety-- most attempts to force it to will suffer from adverse selection or other problems.

That's what I'm saying.

John Thacker has created a false dichotomy, between knowing for sure that you have a disease and what it will cost and being completely ignorant. He also ignores the temporal aspect of our knowledge and the reality that health insurance is purchased year by year.

The bulk of health spending occurs among the top 10 to 20% of health spenders. Some of that spending is unpredictable before it occurs, e.g., the costs of being hit by a bus, or having a child with a birth defect. Some is partially predictable. An overweight individual with high blood pressure is at greater risk of stroke, but it is not determined. An individual with diabetes will generally have higher routine health expenses and may have higher extraordinary expenses if their condition worsens and they go blind or need to have a limb amputated. A person with diagnosed colorectal cancer will have high expenses. The risk pool for insurance is a combination of all these groups, with the largest group at low risk in any given year.

Once an individual has a chronic condition, their costs are likely to be higher in future years.

Given this, I don't see how we can discuss insurance reform and say the support for the sick should be separate. You either need to endorse risk related premiums that will price many of those with chronic illness out of the market, or you need to accept the social insurance aspects of health insurance.

"Personally, I'd like to see a mandated national health system that limited treatments to the most well-known and cost-effective, or was just catastrophic high deductible coverage."

Exactly. But is that what we'd end up with? I really don't think so.

A couple of comments. The first is that Michael Sullivan's proposed solution seems to be a bit blind about what happens to people who don't fit into his plan. Care to elaborate?

The second is that I specifically addressed my assessment of 'facile narcissism' to the esteemed proprietor of this blog, not John Thacker. However, I completely concur with Jack Needleman's much more professional assessment of the problem. Thus I too think John's dichotomy is fundamentally flawed, for Jack's exact reasons. I think I covered the implications of a system built on the already observed flaws adequately. Certainly there's been no speculation otherwise, neglecting martian moon bases and otherwise delicious boing boing nano nonscience.

Surely the rest of Tyler's lunch table can provide productive suggestions for the resolution of this impasse.

It's a blog, after all.

First, Tyler, what DO you agree with Krugman about, if adverse selection nor moral hazard?

Second, does no one else find Krugman's strategy to be curious?

"Employment-based insurance is a distant second-best, but better than nothing. Large employers, in particular, can spread risk widely, creating the kind of risk pool that dies from adverse selection in the individual market. And the tax preference for employer-based care, more or less by accident, has helped sustain this imperfect fix — which is why I'm highly skeptical of anything that might erode that preference."

Do we really want to *encourage* people to be at the mercy of large employers? Do we really want to *discourage* people from being financially independent from large employers? Do we really want to encourage large employers while discouraging small ones?

With 7% of the population 65+ (compared to the US 12%), the government of Singapore provides 1/3 of the funds for health care. Looks like a large government involvement. Singapore needs to regulate insurance for 4 million people over an area of 253 square miles, less than the size of New York City, and has an authoritarian government that has been strict in imposing regulation. Furthermore, the government imposes controls on the supply side of the health market that impact cost and use. I'm not convinced Singapore's experience provides a particularly useful example for the US.

Taking the chronically ill out of the insurance market, making them the government's responsibility, takes much of the health spending out of the insurance market as well. (80% of the spending by 20% or less of the population, much for chronic care.)

Even for those remaining in the insurance market, the bulk of the spending is among those with high expenses -- the heart attack, cancer, bus accident, birth accident or congenital problem -- and these folks will blow through even high deductibles. You will still be relying on the insurers' models for cost control for the bulk of cost savings for this population.

I'll avoid getting into a discussion of the net impacts of health savings accounts in this post.

I would like to know if the new health care plan will pay for my acupuncture visits? I stay healthy by using preventative care and that is what I believe will reduce costs. This is what Dr. Oz and Dr. Weil have been trying to say to everyone. Learn how to breathe, move the body, make wise eating choices and exercise the brain.

Comments for this post are closed