« WaPo Hearts Haley Barbour | Main | Thanks Pops! »

October 10, 2005

Adverse Selection: A Big Problem with Private Insurance

By Neil the Ethical Werewolf

One reason why it costs so much to buy health insurance as a individual is because of a problem that economists call “adverse selection.” Adverse selection occurs when people on one side of a market have better information than people on the other side. Then the people with less information start doing things that keep almost anyone from getting a good deal. It's an interesting problem, and the rest of this post is devoted to explaining how it works and how to beat it.

Suppose we’re all 25-year-olds who want to buy some health insurance. Then suppose the insurance company sees that average health care costs in America for people at our age come to a total of $1000 per year*, and decides to charge each of us $1000 for a year’s insurance. What do you think is going to happen?

Personally, I’m not paying the $1000. I’d like the security that comes with insurance, but my annual health care costs are pretty sure to be a whole lot lower than that. I hope you’re healthy like me so you’d do the same thing too. We do this because we know we’re healthy people, and it’s not financially smart for healthy people to pay so much for insurance.

But this has an unfortunate consequence. We healthy people are the ones that the insurance company is counting on to stay afloat. (This is why the problem is called “adverse selection” – just because of how the market works, the insurance company ends up with the worst selection of clients.) It’s only from the profits on healthy people like us that the insurance company can pay for the $30,000/year cancer patients and still stay in business. So the only thing that the insurance company can do to stay in business is raise its prices further.

So suppose the insurance company starts charging $1500/yr. But then some of the slightly sickly people who estimated that their health care costs would be in the $500-$1000 range will jump ship, just as you and I did with the $1000 price. They were willing to pay a little extra for the security of insurance, but not that much! So the company still isn’t making enough profit. The price goes up further. This is the “adverse selection death spiral” (it’s actually been called that in academic papers) and when it ends, hardly anyone can afford insurance.

So how do we stop an adverse selection death spiral? Insurers try to ask you some questions before you sign up, so they can price people appropriately, but this usually isn’t enough to overcome the differences in information. Because you actually lived through your life, you’ll still know more about your own health situation than a company that made you fill out a form. So you’ll use your extra knowledge against the insurer. This leads to smaller spirals, but they’re bad news nonetheless, and you have to be pretty risk-averse before you'll even buy this kind of insurance.

One thing we can do is push more healthy people into the system. This is why it’s cheaper for employers to provide health insurance than for the private market to do so. Insurance companies know that when they deal with an employer, they’ll be getting a reasonably good selection of healthy people, so they’ll give the employer a better price. Your employer probably doesn’t offer you the option of taking a cash payment instead of health coverage. If you could get that kind of deal, the per person cost of health insurance would go up a lot for your employer, because they'd be buying only for an adverse set of clients.

Pushing more people into the system is basically what we universal coverage fans want to do. If the government covers everyone, everyone can have the security of insurance without having to pay the insane prices at the top of the adverse selection death spiral. Sure, this means that healthy people might not be able to take on some risk for a good deal. And if a large percentage of the American population took wild gambling pleasure in the idea of going without health insurance to save money, maybe that’d give us an argument against universal coverage. But nobody actually thinks about health insurance that way. So even if some healthy people are stuck with a mildly bad deal, the aggregate population is better off due to everyone having security.

For those of you who liked this, and want to see a badass young wonk doing something cool, watch Matt Yglesias point out how adverse selection afflicts the annuity market and bedevils Social Security Privatization schemes. (That’s the first place I learned about adverse selection. Thanks, Matt!)

*If anyone can find actual numbers on this, I’ll redo the entire post with adjusted numbers. I’ve been looking but I don’t know where to find them. I'm pretty sure that $1000 is a lowball estimate, given that the average for children in 1999 was less than that.

October 10, 2005 in Health Care | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Adverse Selection: A Big Problem with Private Insurance:


Does Matt really believe that life annuity prices are materially affected by the presence of social security?

How exactly does he think the Life Insurance market functions? And how does he account for the growth of the annuity market over the last decade? Look, its one thing to argue that adverse selection exists in the annuity and life insurance markets. Its quite another to present it as having such a huge influence in these markets.

Conversly, your point about adverse selection being an important determinent in Health Insurance is very valid.

Posted by: Mihir | Oct 10, 2005 8:20:17 PM

Mihir, the point isn't that Social Security affects annuity prices. It's that adverse selection affects annuity prices, and that the universality of Social Security gets us around that. So if you try to take the universality out of Social Security, as some privatizers want to, you'll be stuck with all the problems of the annuity market.

I'd account for the growth of the annuity market in terms of the aging population. Increase the demand for a bad deal, and more people will take it.

If your position is that there's less adverse selection in annuities than in health insurance, that seems reasonable to me.

Posted by: Neil the Ethical Werewolf | Oct 10, 2005 8:32:34 PM

Because you actually lived through your life, you’ll still know more about your own health situation than a company that made you fill out a form.

Neil, do you really believe that people know more about their own health risk than the actuarial profession? This seems highly, highly dubious to me.

Also, a death spiral won't necessarily mean "hardly anyone can afford insurance." Mark Pauly et. al. recently found that "high premium rates for high risks are not a significant contributor to the uninsured population in the United States." Even among low-income families. It's a moderately unexpected result, and the paper wasn't evidence one way or the other for the existence of adverse selection (although it sounds that way, since they found that rates of insurance are positively correlated with risk). I'm all for universal health insurance, so these findings don't matter much, but interesting still.

Another interesting problem is that standard models of adverse selection don't predict a large continuously uninsured portion of the population. Either you'd reach a) an equilibrium point of low, sub-optimal coverage for low risks, and complete coverage for high risks, b) an equilibrium with pooling into a single level of coverage, or c) cycling behavior. (I think, I hope that's right.) But we obviously have a continuously uninsured portion of the population in our dear country. Why? Adverse selection can't seem to explain all of it. Immigration? Or else the models are fucked.

To further complicate things, here are two papers (1, 2) in which adverse selection doesn't lead to the death spiral, but rather advantageous selection. (The examples are in auto and life insurance, though.) We're having fun now...

Posted by: Brad Plumer | Oct 10, 2005 9:11:54 PM

I think you are wrong, adverse selection is only a problem if you want to charge everyone the same rate. You say people can estimate their expected medical expenses better than insurance companies. I am not convinced. Do you think the average person can tell you how expected medical expenses vary with age and sex for example?

Another problem with the current medical insurance market is that the plans are generally required to offer all treatments whereas if you were paying for your own care you might choose the treatment that is 90% as good and 10% as expensive.

Posted by: James B. Shearer | Oct 10, 2005 9:24:00 PM

I'm pretty confident that individuals are better informed about their own health situations than the actuarial profession. In preparing this post, I went online and filled out a questionnaire that would allow me to buy insurance. I gave them some basic information -- age, height/weight, etc., and told them I didn't have any of ten diseases. Oh, and I had to put in that I didn't use tobacco. For all they know, I could have a rare disease in my family which sets in at age 30, or do crazy things with motorcycles, or use meth. I could have just come down with my first migraine, or felt some intense chest pains for the first time, or started thinking that I'm bipolar. In less than 5 minutes they just offered me a list of insurance quotes, which you can see here.

There's plenty of other explanations for why we have so many uninsured people. The fact that a lot of people have very little money is a huge one. Then you get the huge administrative costs from having many private providers, each with their own bureaucracy. Then there's the expenses involved in developing better ways of sifting through clients so you can cherrypick the healthy ones. (Feel free to name the others you know.) Adverse selection is only one reason among many.

Posted by: Neil the Ethical Werewolf | Oct 10, 2005 10:14:16 PM

Neil --

In preparing this post, I went online and filled out a questionnaire that would allow me to buy insurance.

Touche. But you better learn to read the fine print:

"Your premium is subject to change based on your medical history, the underwriting practices of the insurance company, the optional benefits you selected, if any, and other relevant factors. The insurance company always determines your actual premium. Insurance companies reserve the right to change the terms of a policy upon proper notification."

At any rate, sure, individuals can certainly beat the actuaries from time to time -- that's the whole idea here -- but on average, I think it's safe to say that the number-crunchers know more about health risks than the people themselves. As James says above, "Do you think the average person can tell you how expected medical expenses vary with age and sex for example?"

As for your second point, totally agreed, there are many possible reasons as to why 45 million people are uninsured. The point is to try to pin down exactly what's going on, though. And I think it's tough to pin down how adverse selection works in practice, if relatively straightforward in theory.

Posted by: Brad Plumer | Oct 10, 2005 10:35:27 PM

Neil the Ethical Werewolf, ok so why don't you tell us what each of those insurance companies should be charging you? Please explain the cost effect of any plan differences.

Most people don't in fact know they have a rare expensive disease which is about to manifest itself. A few people paying underpaying because of private knowledge won't break the system, all insurance has some frictional costs.

Posted by: James B. Shearer | Oct 10, 2005 10:44:19 PM

Do you think the average person can tell you how expected medical expenses vary with age and sex for example?
No, but I don't see the relevance of this. The average person doesn't need to know expected medical expenses for people of any age or sex, including her own, to beat the insurance companies. All a person needs to do is estimate her personal medical expenses in the near future, and see how they relate to the number that shows up on the screen. (The insurance company, btw, didn't even do anything to screen out any chronic diseases I might have which I could easily estimate the costs of.) So James, I have no idea what any of those insurance companies should be charging me. But I've met a couple fairly unfortunate students who would give answers on those questions roughly similar to mine, and who I'm sure the insurance companies would need to charge more than that.

I really don't know how to analyze the fine print there, Brad, as it could be interpreted in a wide variety of ways. Certainly they could use that in an abusive fashion. But even then, if I had a big one-time expense looming in the future, or the risk of one, they'd be on the hook for a big loss, after which I could just cancel the plan.

Posted by: Neil the Ethical Werewolf | Oct 10, 2005 11:25:45 PM

And Brad, perhaps I don't understand the abstract of the Pauly paper. "High premiums for high risks" isn't the adverse selection problem, is it? Adverse selection is about how even low-risk people get high premiums because insurers can't tell the difference between them and the higher-risk folks.

Posted by: Neil the Ethical Werewolf | Oct 10, 2005 11:31:42 PM

Pushing more people into the system is basically what we universal coverage fans want to do. If the government covers everyone, everyone can have the security of insurance without having to pay the insane prices at the top of the adverse selection death spiral.

Don't neglect the other huge cost being elided, which is the cost of determining who should be offered insurance.

Posted by: Allen K. | Oct 11, 2005 1:54:16 AM

No way I would elide that one, Allen. In fact, I was thinking about putting something about that in here, but the post is long enough as it is.

Posted by: Neil the Ethical Werewolf | Oct 11, 2005 1:59:43 AM

Another way of thinking of the problem of adverse selection is to remind ourselves of the concept of risk pooling. Risk pooling is the underlying base for ALL insurance - for both those covered as beneficiaries and those providing the coverage.

No one knows when they will be hit by a car and die, so accidental death insurance makes sense because it is cheap, and it is cheap because lots of people buy it, increasing the risk pool - thus making payouts low relative to premiums, because it IS cheap and offers protection for someone in the event of being steam-rollered by a SUV.

For any given population, the cost of the insurance will be lowest when the risk pool is the largest possible, preventing adverse selection as a precaution. Elementary, Dr. Watson.

If offered a CHOICE, the wealthy will always choose to opt-out of the risk pool since they have the means to buffer the effects of an unforseen event.

The poor opt-out because even at low rates, it is a hit on their precious spendable dollars, and they likely have a lower expectation of 'protecting' themselves or others in their family from unforeseen events.

The case for universal health insurance is, IMO, based on making the risk pool equal the population - thereby making the cost per person as low as possible.

The case for single-payer universal health insurance is based on eliminating administrative overhead, and eliminating the possibility of people opting out because they think they can beat the odds and spend less overall.

Univeral risk pooling is social coercion, but the good kind - like a requiredment for auto insurance to get a vehicle registered. I don't want to be injured by an uninsured driver - at no fault of mine, but at great risk if involved - so I support an auto insurance requirment for driving. Similarly with health care: do I want people without medical care wandering around my supermarket during an avian flu pandemic? Hell no! I want them to obtain medical care without regard to the ability to pay immediately.

Whatever the effects of adverse selection, it is clear that the problem can be reduced or eliminated by making the risk pool as large as possible - using the power of the state to ensure the greatest benefit for the largest number of people in an area where individual choice can only weaken and kill whatever risk-sharing system is involved.

All opt-out plans, or plans that require opting in (for a risk-adjusted premium) for a limit on risk from universal problem like health care or death, will increase the average cost of those covered.

Sickness and death largely cannot be avoided - it will be someday for most or all people, but are unpredictable as to occurance even with a great deal of information at hand.

Market forces do not cure this problem and the administrative duplication involved in private insurance solutions plus the constant danger of adverse selection make a joke of claims of less overall cost for plans with limited risk pools (or no pool at all - for the wealthy).

And, while I am ranting, exactly WHY is the insurance industry exempt by federal law from ALL federal regulatory oversight? Does disease or death not cross state lines?

Posted by: JimPortandOR | Oct 11, 2005 4:06:33 AM

Your employer probably doesn’t offer you the option of taking a cash payment instead of health coverage. If you could get that kind of deal, the per person cost of health insurance would go up a lot for your employer, because they'd be buying only for an adverse set of clients.

But when you opt out of health insurance at your workplace, the employer pays directly to you the portion of your premium that otherwise would have come out of your paycheck. So they are exactly giving you a cash payment in exchange for not buying insurance. This is a false choice because it is not a choice between employer-sponsored health insurance and privately purchased health insurance. It's a choice between 15-30% of the cost of employer-sponsored health insurance and 100% of the cost of privately purchased health insurance. The premiums, after a group adjustment, are usually pretty close to the same. It's the subsidy that makes it work.

Posted by: diddy | Oct 11, 2005 9:50:51 AM

It's not really a question of "security." One of the surest ways to ruin your life is to not have health insurance. Even if (you think) you're healthy, you never know what's going to happen to you.

I hope this was just a for-the-sake-of-argument post and not the decision you've actually made.

Posted by: hmm | Oct 11, 2005 10:25:27 AM

People questioning whether "adverse selection" exists in the health insurance market are weird. Neil isn't just making this up randomly, but rather it's pretty clearly what the health insurance market is based around. In particular, I assume Brad has taken some Ec 101 and is aware this is basic stuff. This stuff is absolutely true in regards to health insurance, auto insurance, even homeowner insurance, and anything affected by our personal choices. The more you get into random occurances that are universally the same over an area (fire, flood, terrorist insurance) getting coverage becomes radically less complicated.

In the modern era, simply by being aware and thinking about the problem many insurance companies probably have more information than most of us about longevity. But here's the thing: the "most of us" I reference are the people who are healthy and normal. The people who know more than the insurance companies are the exact people who the insurance companies are afraid of, because they have specific health issues.

Neil has correctly identified market failure... however, that doesn't mean the solution is necessarily government takeover. I believe in universal healthcare of course, but one of the traps liberals must avoid is "the market isn't operating perfectly here - so let's nationalize the sector".

Posted by: Tony Vila | Oct 11, 2005 11:56:43 AM

Neil and Brad,

I don't think that the actuaries have figured out how to estimate a projected insured's costs on an individual basis. I think they're quite good at doing it for a large population--yet another reason why we should have universal coverage and probably single-payer.

I also think that there's a lot of CYA stuff where they jack your rates up for things that will not affect their costs much at all. I've heard of people with a past history of taking Flonase seeing their quoted rates double. Do you really think that Flonase usage will make you that much more of a health risk.

Plumer, you should talk to Matthew Holt about Pauly to get another perspective. So, people, how do we reach the unconvinced?

Posted by: Bostoniangirl | Oct 11, 2005 12:09:26 PM

I don't think that the actuaries have figured out how to estimate a projected insured's costs on an individual basi

I work as an actuarial consultant - i.e. an actuary who provides consulting services to various insurance companies.

I can tell you that even if actuaries could estimate projected insured costs on an individual basis it would probably be illegal to price such policies on an individual basis without accounting for the group.

Adverse selection is certainly a problem for the insurance industry. We struggle with it on a regular basis. Actuaries have got pretty good at "weeding out the adverse selector" or right pricing. In life insurance industry (annuities are a part of this) this is accounted for in various ways. I certainly dont buy is argument that annuity prices are materially affected by the presence of social security.

Neil is completely correct that adverse selection is a huge problem in the health insurance market. While insurance companies can account for some adverse selection they will never know as much as an individual does about his own health. Adverse selection only occurs when a person knows his health is worse than average and does not say anything about it.

As an example - say Sam and everyone in his office have a health insurance policy through their employer. Sam finds out that he suffers from a disease that will require a lot of expensive medicines but does not stop him from working. Now say the employer lays off everyone in Sam's office. Sam will avail of the regulations in place and pay his premiums to keep the same policy inforce. Others in his office wont do the same and will let their policies lapse.

Adverse selection just occured. Keep in mind most adverse selection is not illegal. If you knowingly hide information (or lie) then you are commiting fraud. Sam is doing no such fraud. The insurance company typically prices for some adverse selection. Before the concept was popular, the insurance company would have not accounted for adverse selection and assumed everyone would lapse since it is quite a bit more expensive to keep that policy inforce.

Posted by: Mihir | Oct 11, 2005 12:54:41 PM

Bostoniangirl -- Thanks for the Mark Pauly reference. Not sure that makes his research wrong, but always good to know.

And yikes, I certainly wasn't trying to argue that adverse selection doesn't exist...

Posted by: Brad Plumer | Oct 11, 2005 1:20:43 PM

Brad-- I haven't read any of Pauly's stuff, and I can't really comment on it myself. I just wanted to point out that he's really rather controversial.

Posted by: Bostoniangirl | Oct 11, 2005 1:28:03 PM

As my name's been taken in vain, let me tell you what I think about Pauly. Note that he is basically the only commentator on health policy who believes that our current system is essentially the way "things ought to be" other than needing a little more "free market" who is NOT from an ideological right wing think tank. Almost every other major health economist academic (Rheindhart, Enthvoven et al) whether writing from a market or more liberal perspective agrees that the individual insurance market is a complete zoo.

Pauly believes that because, in his terms, the indivdual insurance market works OK for 80% of the people in it, then it's OK. Yup he really said that in Health Affairs a few years back. The problem of course (as the actuary Mihir can tell us) is that only 20% of the people in a risk group actually need health insurance, and the current market is set up to give insurers every incentive possible to get out of offerring them insurance. This starts with medical underwriting (which will increase premiums 5 to 10 fold if they discover that you're sick) to games insurers play re-rating groups by claiming that the "group" has changed if they are forced by state laws to keep insuring them. So the 80% for whom it works well don't need it and the 20% who do need it are screwed.

The problem is that it is absolutely in the insurers interest for this situation to play out as it does because caring for one sick person negates the profit they make by insuring four (and sometimes many more)healthy ones. Until everyone is in the same insurance group with true community (i.e. national) rating, or we invent a risk-selection payment system that can't be gamed (and I don;t think we'll ever get there), the problem of adverse selection and its inverse adverse dis-selection will continue.

I've decided that the best way to get this solved is for an act of Congress to force Mark Pauly, and everyone in the Congress and the Administration to buy their insurance in the individual market. I think we'd get universal health insurance about 20 minutes later.

Posted by: Matthew Holt | Oct 11, 2005 5:08:12 PM

Adverse selection and its bedfellow, cherry-picking, are why private insurance will never get us decent health care, regardless of what the "markets are magic" crowds argue.

Fundamentally, it's a much better business model for insurance companies to avoid selling insurance to sick people than it is to try and turn sick people into healthy people by spending money on them. It's a lot easier to avoid the sick than it is to cure them.

So I don't really understand why we're the least bit surprised that our health insurance system doesn't work well for sick people. There's no reason to expect it would.

Of course, given that the whole reason healthy people buy insurance is so that they will have money to spend on treatments when they get sick, I can see where people would be a little upset by this fact. But surprised? No. Shouldn't surprise them at all.

Posted by: theorajones | Oct 11, 2005 6:27:43 PM

So Neil, is adverse selection really your gripe with the current market? AS is just a market failure problem having to do with how asymmetric the information about health-conditions is. It does not specifically resolve the fact that some people simply do not have the resources to guarantee a healthy life, which many left-wingers feel should be a basic right.

Let’s say tomorrow medical-diagnosis improved ten-fold and insurers knew everyone’s exact chance for getting a catastrophic health loss. There still would be some people who are not only poor/middle class, but have a 40% of disabling loss, and would either have to pay very high insurance that ruins their life, or roll the dice and hope nothing happens. In the game of life, they’ve just been handed a bad scenario (bad genes, whatever), and the liberal instinct is to compensate them for it – redistribution of some form is the only solution.

I’m all for universal care, but let’s be honest. The 45 million people who aren’t insured are largely a concern only because many feel your medical condition should not be left to capitalistic fate. Not because the market is inordinately broken in terms of adverse selection.

Posted by: Tony Vila | Oct 12, 2005 12:59:06 PM

I'll say for the third time, Tony, that adverse selection is only one of my many gripes with the current state of American health care, and far from the most significant one. The thing is that it's a theoretically interesting gripe that non-wonky people might not be aware of, and that's why I wrote this post.

Posted by: Neil the Ethical Werewolf | Oct 12, 2005 1:42:39 PM

But the point is, your solution to adverse selection involves inviting an agent in who doesn't want to/can't save money. Either force healthy people to be in the same risk-pool as risky people, or have the government cover the risky people in their own special high-cost risk pool. Either way, you aren't fixing the problem of adverse selection, you're just moving the costs around - with governmental force. And either goal could be accomplished somewhat with new insurance rules as much as actually establishing a large bureaucracy.

Which is fine to me of course, but for any intellectual conservatives/libertarians you're trying to convince, doesn't really fly.

Posted by: Tony Vila | Oct 12, 2005 4:48:26 PM

Maybe it looks like you're playing a zero-sum game in moving the costs around. But if you think there's an automatic plus in having the peace of mind that health insurance gives you, a system that insures everyone is better overall. (This is the point of my comment above that nobody takes pleasure in gambling with their health situation.)

Posted by: Neil the Ethical Werewolf | Oct 12, 2005 6:09:30 PM

The comments to this entry are closed.