Bryan Caplan  

A Closer Look at Adverse Selection and Mandatory Insurance

Casnocha on Tyler's Next Book... Virtual Secession...
If an economist wants to ward off the spirit of laissez-faire insurance policy, all he has to do is repeatedly chant "moral hazard and adverse selection."  The funny thing about this two-part mantra, though, is that the "moral hazard" part doesn't do any of the work.  Almost no one even pretends that governments do anything to mitigate it.

When we get to the "adverse selection" part, the plot thickens.  There is, in theory, a regulation capable of solving the problem: mandatory insurance.  To see how mandates can help, consider a simple example.  Suppose there are two equally common types of people who buy insurance:

High-Risk Consumers: They have a 20% chance of losing $2000.  Since they're risk-averse, they value full insurance at $1000 ($600 more than the actuarially fair premium of $400).

Low-Risk Consumers: They have a 1% chance of losing $2000.  Since they're risk-averse, they value full insurance at $50 ($30 more than the actuarially fair premium of $20).

If insurance companies can't distinguish High- from Low-Risk consumers, an actuarially fair premium for an average consumer would cost .5*$400+.5*$20=$210. 

If consumers purchase insurance voluntarily, though, the Low-Risk will drop out of the market - they won't pay $210 to get a policy worth $50 to them.  With only High-Risk consumers in the market, the competitive price of a policy is $400.  The market fails to realize $30 worth of consumer surplus per Low-Risk consumer.

In a mandatory insurance regime, however, the Low-Risk have to buy the policy.  The result: The regulation is efficiency-enhancing, because it takes $160 from every Low-Risk person in order to give $190 to every High-Risk person.

So far, so good.  It's conceivable for mandatory insurance regs to improve market performance.  But their argument jumps the shark when defenders of government insurance regulation notice the existence of mandatory insurance regulations, and infer that these regs are doing something about adverse selection.  When you actually look at these regs, you'll notice some peculiarities:

1. Mandatory insurance is most prominent in the auto insurance industry.  But these regulations don't target low-risk drivers.  Their main purpose, contrary to the adverse selection model, is to make sure high-risk drivers get insurance. 

2. Even more shocking: The regulations usually go on to somehow subsidize the rates that high-risk drivers pay.  This is necessary because, contrary to the adverse selection model, insurance companies are able to detect high-risk drivers, and do not want to cover them at a loss.

3. Economists usually mention adverse selection in the context of health insurance.  But in the market for individual health insurance - precisely where you'd expect adverse selection problems to be most severe - governments very rarely mandate insurance coverage.  Instead, they focus on mandatory employer-provided health insurance, where the adverse selection problem is likely to be milder.

4. When governments do mandate health insurance, they almost always subsidize the rates that high-risk buyers pay.  This is once again necessary because, contrary to the adverse selection model, insurance companies are able to detect high-risk customers, and do not want to cover them at a loss.

Bottom line: Real-world insurance regulation has little or nothing to do with economists' "moral hazard and adverse selection" mantra.  The "intellectual" bases of real-world regulation of insurance are rather populism and paternalism: Big bad insurers won't cover people unless it's profitable, and simple-minded consumers don't care enough about their own health to pay for it themselves. 

Contrary to e.g. Krugman, insurance isn't a "special" market where laissez-faire doesn't work.  Instead, it's a normal market where democratic politics doesn't work, because both the public and economists remain wedded to populism and paternalism.

Comments and Sharing

COMMENTS (26 to date)
Willem writes:

In the Netherlands, health insurance is mandated on an individual level (theoretical fine is 5 years of insurance fee).

An extensive risk model levels the costbase of insurers ex ante for the expected cost of their sample. Not all costs are compensated (so some prevention should pay), and if a insurer pays higher-than-average prices for care, they pocket their higher costs. Most of the competition then lands between the insurer and care deliverer.

We still have abundanted (growth of) health costs (absence of serious copayments), but insurers operate privately and we still have mandatory insurance. I quite like our model. It puts responsibility for insurance at the consumer level; uses private insurance and health institutions and still makes sure poorer people get (or have to buy) care.

I understand that being able to privately decide if i want to insure myself is to be prefered as a libertarian, but perhaps this is a loss of liberty I would accept if it gives universal health care access.

El Presidente writes:

There is an important difference between mandatory car insurance and mandatory health insurance. Under mandatory car insurance policies, you may still elect not to drive, and thus not to purchase insurance, and you can only drive if an insurer is willing to insure you at a price you are willing to pay. There is still a market in which price-elasticities and risk matter, just less so.

There is also an important lesson I think we can apply to mandatory health insurance from our experience with mandatory car insurance. Mandating car insurance in a consolidated insurance market has led to greater coverage, but I don't know that it has led to a proportional decrease in premiums (in California). It appears that insurers may be capturing a disproportionate amount of the surplus. One insurer covers roughly half of the market in California. That means they are likely to be representing both parties in any accident, giving them a great deal of cost savings by forgoing litigation and reducing payouts without the threat of adversarial challenges. After all, most people would be outgunned by their own insurance carrier. That's why they would rely on their carrier to fight the other person's in court, if need be, to get a fair settlement. The increased surplus doesn't necessarily translate into proportional consumer surplus. This is what troubles me about the notion of universal coverage without any guarantee to direct some of the surplus back to consumers. Sprinkling fairy dust on the magical market might not work. It seems the pressure provided by a public option tends to force the issue a little more.

If anybody out there has data or studies that would disabuse me of my sense that mandatory insurance has resulted mostly in producer surplus, I am eager to read. I hope it isn't true, but I need to be persuaded.

B.B. writes:

I think your points are right on. But I think you may be missing the major issues in your eagerness to refute Krugman.

Issue #1: the free rider problem. In a world in which the majority are mildly benevolent and altruistic, the majority doesn't want to see people suffer unnecessarily. Selfish individuals can refuse to buy insurance against severe low-probability events. They can selfishly consume what would have been spent on insurance premiums. If the bad event happens (severe injury or illness), they appeal to the majority for assistance. The majority care too much to do nothing; it provides "welfare". Implicitly, the majority are giving insurance premiums to the free riders.

The rational democratic response is: (a) the majority cease to be altruistic, which is the Randian solution; (b) the majority votes for mandatory, universal insurance, thereby forcing the selfish to pay for the insurance they used to get free.

One might suppose that the selfish free riders tend to be those with less income and education, but that need not always be so.

Consistent with this view, I have always thought that the point of the European social welfare state is to tax and control the poor. The poor are going to get medical care anyway, so tax them. To just call that paternalism is to miss the point: the model is trying to solve the free rider problem.

Enforcing mandatory insurance is hard. That is why the approach tends to be an employer-based system or a payroll tax. I hope that we can find a better approach, but I appreciate the difficulties.

Issue #2: the poverty problem. The single-payer advocates want a payroll tax or VAT to pay for health insurance premiums. But why? If per capital health spending is "X", then why not just charge every person "X"? A lump-sum charge would avoid the distortions from taxation and would truly diversify health risk. There is an element of dishonesty in single-payer advocates in that claim they want to reduce inefficiency but are using they program for a huge tax-based redistribution program.

But there is also a element of dishonesty in advocates of purely private insurance: those who are unemployed or have low wages cannot afford to buy health insurance, just as many cannot afford to eat or pay rent. An employer-mandate would be so expensive that the jobs would be destroyed; the cost of health insurance could exceed their wages. Such people must get public health insurance or must do without health care.

A final issue. We know that many people have health issues from birth, set by genetics or fetal environment. Those unfortunates are not able to buy insurance against predetermined risk. It is like trying to buy insurance after the auto accident. Mandatory social pooling of the risk of genetic flaws or birth defects would raise efficiency and fairness. That is neither populism nor paternalism.

ryan yin writes:

B.B., the problem with making arguments about poverty and altruism is that you can do more good with the same budget by giving money rather than health care (or any specific good). At this point, some argue that if we give the poor money they won't buy health care (or whatever it is we want them to buy); however, if we're actually altruistic, choice is a feature, not a bug.

As an aside, I would further argue that Caplan is too mild about "moral hazard doesn't do any of the work." Moral hazard isn't just a lazy deadbeat; it's a saboteur. If you're worried about moral hazard, you should be more skeptical of insurance, not less.

BDM writes:

I used to completely agree with the standard libertarian argument that insurance companies can solve the problem of adverse selection by raising the premiums of high-risk individuals, or by preexisting condition exclusions, and that most people use the adverse selection argument as a cover for their desire to redistribute wealth to the sick.

What changed my mind was reading John Cochrane's Cato report on health-status insurance. He points out that people don't merely want annual insurance. Individuals want long-term protection, as they may develop chronic illnesses - and so you would expect that people would be buying and selling "health-status insurance", which guarantees you the right to purchase health insurance at your original premium.

Unfortunately, it does not exist. Cochrane argues that this is because regulation prevents it from emerging. The more likely story, however, is that adverse selection makes health-status insurance a dud. Discussing a similar idea of "guaranteed renewability", David Dranove writes in "Code Red": "Healthy, older individuals may drop out in favor of an insurance plan that charges lower premiums more in keeping with their better-than-expected health status. This destroys the risk pool." (187-188). And so we finally have a case where adverse selection destroys a beneficial market.

My question for Bryan is this: do you believe that health-status insurance (or guaranteed renewability) are "illegitimate" products (in other words, it doesn't matter that they don't exist), or that they would exist if only we had fewer regulations? These are the only two libertarian responses I can think of - but I am open to the possibility that I have created a false dichotomy.

Scott writes:

I propose we track any user of government health care. Garnish the wages of anyone who uses government money until it is paid back. This, combined with what is undoubtedly crappy government-provided health care, would encourage most people to buy insurance.


Josh writes:

It's amazing given all the success of insurance over the past several centuries that health would be the one area where asymmetrical information would be the most problematic. It seems like health should be one of the easier things for an insurer to measure on par with the insured since nearly everyone gets virtually all of their health information from a 3rd party (perhaps there are some doctors that treat themselves?).

If some people are then more costly to insure than others, and if we object to that, then we can solve that problem independently through direct cash transfers. But I think that in the end, information asymmetry is not a serious problem in medicine.

Jim Ancona writes:


It seems to me that the same adverse selection problems you expect for health-status insurance would apply to life insurance as well. Yet guaranteed renewable term life insurance is available at very reasonable rates. Why don't healthy, older individuals drop out in favor of an insurance plan that charges lower premiums more in keeping with their lower-than-expected risk of death?


Jeremy, Alabama writes:

Existing government and state mandates prohibit the emergence of genuine innovation in health insurance.

Arnold emphasizes that the thing we buy right now is NOT insurance. I just went to the doctor for back trouble - I pay $30, BCBS pays $60. We just had a baby boy - I pay $500, BCBS pays $5000. These should NOT be insurable "risks". I should shop around for doctors and natal clinics, I ought to have less restrictions on what natal care I want to afford (e.g. midwife instead of hospital).

But no - I pay my $1000 a month and BCBS gets in the middle of every health decision. Government intervention would be incalculably WORSE.

An acquaintance of mine recently had her pancreas or something removed - she was uninsured, it was about $2500. Personally, I would like to buy cheap insurance that would NOT cover me for an urgent but inexpensive procedure like this, but would cover me for, e.g. another acquaintance who had his colon removed (multi-hundred-k procedure).

Law PROHIBITS meaningful evolution of insurance along these lines. Perhaps poor people and young people really WOULD buy cheap insurance for named, scary, specific risks such as cancer, and take their chances with back trouble.

ionides writes:

Does anyone have the same problem as I do in distinguishing moral hazard from adverse selection? I have looked them up about 100 times and I never come away with any conviction that I understand them.

Moral Hazard: Someone gets insurance against breaking his leg doing the charleston in an unlighted room. Before the insurance he was careful; after the insurance he charlestons with abandon.

Adverse Selection: Someone gets insurance by assuring the agent that he never does the charleston in an unlighted room. With policy in hand he continues his wanton dancing.

What is the real difference?

Consider this sentence from Wikipedia: "A bank that sets one price for all its checking account customers runs the risk of being adversely selected against by its low-balance, high-activity (and hence least profitable) customers."

Selected against? What does that mean?

Could someone post or link to a crystal-clear definition of these terms?

Does the consumer adversely select the insurance company, or does the insurance company adversely select the consumer?

Does moral hazard mean the risk of selling insurance to a liar? Doesn't it apply to any transaction which is distributed through time?

Thomas DeMeo writes:

"Mandatory insurance is most prominent in the auto insurance industry. But these regulations don't target low-risk drivers. Their main purpose, contrary to the adverse selection model, is to make sure high-risk drivers get insurance. "

What does this mean? Mandatory means everyone is in. In my state (Massachusetts) we have highly regulated auto insurance and low risk drivers don't escape coverage. There also isn't much in the way of cherry picking by high risk drivers either. Insurance is priced too high here and produces some bad market effects with auto repair, but it clearly solves adverse selection.

"Even more shocking: The regulations usually go on to somehow subsidize the rates that high-risk drivers pay. This is necessary because, contrary to the adverse selection model, insurance companies are able to detect high-risk drivers, and do not want to cover them at a loss."

Again, wrong in my state. Insurers can't turn any legal driver down, and we have a significant penalty system to target drivers with poor records.

El Presidente writes:


What is the real difference?

Moral hazard is about the individual's behavior once they are insured (i.e. changing opportunity costs).

People may go from being low-risk to high-risk simply because the cost of engaging in high-risk behavior is reduced as a result of the insurance. People may go from using few services to using many simply because there is low or no marginal cost (I'm skeptical of this argument).

Adverse selection is about whether or not the person becomes/remains insured (i.e. assessing and limiting risk).

High-risk individuals want underpriced insurance. Insurers wnat to exclude high-risks or charge them higher rates. This may take behavior into account, but it also includes innate risk factors (e.g. genetic, environmental). If insurers are likely to take a loss by insuring these individuals, they would prefer to charge more (i.e. price discrimination), limit maximum liability (e.g. health savings accounts), or simply not insure them. If they cannot (e.g. asymmetrical information favoring consumer), these people are insured and the losses are spread to other rate-payers.

I took a stab at it. Better?

Perhaps they would more closely mirror one another if coverage was dependent upon guarantees of behavior and indemnification for unknown or undisclosed risk, but then what would be the point of purchasing insurance instead of long-term financing?

ionides writes:

El Presidente,
You cleared up moral hazard for me; thanks.

An adverse selection occurs when an insurance company covers someone who has high risk factors unknown to the insurerer?

Vichy F. writes:

"Contrary to e.g. Krugman, insurance isn't a "special" market where laissez-faire doesn't work. Instead, it's a normal market where democratic politics doesn't work, because both the public and economists remain wedded to populism and paternalism."

Absolutely, every time I hear people ranting about how some market or service is 'special' one wonders what they mean. Every market is 'special' and 'different', that's why it's not any other market (IE, why insurance is not the same as hammer manufacture). But in none of them do the populist-paternalist government work.

Dr. T writes:

In a libertarian society, this discussion would not be needed. In our existing society, it's important. And, much as I hate government mandates, I believe that mandatory catastrophic health care insurance is better than the alternatives.

The first alternative is our current system where expensive hospitalizations of uninsured people with few assets end up being losses for the hospitals, unless government (eg: the taxpayers) covers the bills.

The second alternative is Medicare from cradle-to-grave. Taxes pay for everyone's health care; the bureaucracy will overflow; and we'll be deeper in debt each year.

The third alternative is the current Obama proposal that creates a federally-backed, full coverage health insurer to compete against the scores of private health insurers. This is like the first alternative with more federal involvement.

The three alternatives make mandatory catastrophic health insurance seem benign in comparison.

El Presidente writes:


One out of two aint bad. :)

Maybe somebody else can help us with adverse selection.

MHodak writes:

I think it's ironic that Krugman won his Nobel for upending a "standard competitive market model" for trade. His theory was explanatory and predictive.

Unfortunately, Krugman could not have predicted E-bay-- a market riddled with asymmetrical information, made efficient by a clever mechanism. Krugman's remarkable faith in government is matched by his remarkable lack of imagination of how businesses evolve mechanisms to overcome "standard competitive market" problems.

Kurbla writes:

Mandatory insurance really doesn't solve the moral risk problem, but it is minor problem. People do not smoke just because insurance covers potential chemotherapy. Non-mandatory insurance doesn't work well if people are motivated selfishly and if there is significant knowledge about risk. The omniscient insurance company rates risk with only two degrees, 0% and 100%, and only those with risk 100% are willing to pay insurance. Mandatory insurance still work (additionally, there is no need for individual risk estimation) so it solves negative selection, and yes, it is more like solidarity than insurance, but who cares.

Gary writes:


Now that El Pres has helped you with moral hazard, I'll give adverse selection a try.

In this context, adverse selection occurs when an insurer is incapable of charging (or isn't permitted to charge) different prices for different risk categories. So the insurer tries to charge a single price -- one that will balance its costs and revenues. That price is higher than the costs of the low risk group and lower than the costs of the high risk group (just the way the math works out).

The price is a bargain for high risk folks and a burden for low risk folks. So the high risk folks pile into the plan and the low risk folks abandon it. Now the pool is full of high risk folks, and the price that makes up for the costs will shoot up. In fact, the price will rise to the point where it would have been if the insurer could have distinguished between low and high risk folks in the first place.

Net result: high risk folks are no better off and low risk folks don't get insurance which they could have benefited from.

Make sense?

Jeremy, Alabama writes:

Kurbla has good points - except insurance companies are not yet omniscient.

Our existing health insurance system willfully conflates at least 3 unrelated things:

- health "maintenance", such as minor doctor visits, pregnancies

- "catastrophic" risk such as cancer or heart attack

- socialization of cost (what Kurbla calls "solidarity").

Through laws that require bundling of "maintenance" and "catastrophe", we socialize more than is necessary.

If they were unbundled, "catastrophe" could be socialized, while market forces would force down "maintenance" prices.

mark writes:

El Presidente identifies "an important difference between mandatory car insurance and mandatory health insurance," the ability to choose to live without auto insurance, simply by not driving. I agree with him but I think there is another more important difference, not a conceptual one but a quantitative one: people just put in far more claims under health insurance than auto insurance. That's why your premium is so much higher. I can put my family in my car and drive them around every day for an auto insurance premium of about $1,000, but my health insurance premium for all of us is about $15,000. That's why when I hear progressives say, if society mandates auto insurance and fire insurance, there's no logical reason to resist mandatory health insurance, I have to say, we can afford mandatory auto insurance and fire insurance because the cost is so small. We can't afford to apply that logic to health insurance; if someone could deliver a model where health claims occurred with a similar frequency as auto and fire claims, that would be a different story.

I would also note that since we don't subsidize citizens' auto and fire insurance premiums, the progressive logic should lead us not to subsidize health insurance.

Finally, I note that we have fire safety codes and driving laws that reduce claims on auto and fire insurance, so does the auto and fire insurance analogy suggest we should have health laws that limit your sugar intake, fat intake, require exercise three times a week, etc and give you tickets for noncompliance - because that is where the auto insurance analogy takes us

Dan Weber writes:
do you believe that health-status insurance (or guaranteed renewability) are "illegitimate" products (in other words, it doesn't matter that they don't exist), or that they would exist if only we had fewer regulations?

The vast vast majority of private health care is purchased through an employer. It is not in the employer's interest to give you an insurance policy you can take with you. Plus the additional cost would select against the long-term employees, who will probably have it anyway.

See also this citation from here at EconLog:
Contrary to popular claims, state laws generally prohibit raising a sick individual's premiums unless an insurer also raises the premiums of everyone else in his rating class.

ionides writes:

Thank you Gary.

Jimbino writes:

The existence of insurance depends on ignorance, irrationality and force.

In a game of risk like Roulette, where the exact odds are known, a person would have to be irrational, or compelled, to buy "insurance."

In a utopian world where everyone knew his own health risks and acted rationally, insurance without price discrimination would disappear because of adverse selection: the best risks would drop out right away, causing premiums to rise, at which point the best risks in the higher risk group would drop out, and so on, by inductive argument, until insurance would disappear entirely.

Of course, the insurance industry can count on Americans to be irrational and they do their level best to keep them ignorant.

Google "claims ratio" or "loss ratio" to see how much info on the great American insurance ripoff you can find. There are some data from Indonesia and New Zealand, but almost nothing from America. But because of government involvement, you can find out that the claims ratio for NFIP is a piss-poor 60% (for roulette it's about 96% and roulette is fun).

Apparently the American people are getting less ignorant and less irrational, because it seems the Insurance Industry now favors force.

Alfredo deLorenzo writes:

The talk of mandatory insurance, auto - health compared leads me to think where it's going next maybe to mandatory life insurance or air insurance for breathable air or food insurance for the folks who still eat. Why don't we just establish public clinics for folks who are unable to pay for medical and can be run under the rationing/control of proceedures like the government wants everyone to fall under. That could provide the emergency room type care - bare-bones care - which might be a joint private/government undertaking. We will always have folks who cannot afford insurance and they should be identified and their needs addressed.

Evan writes:

I think the comparison between health and auto insurance is a little off. Auto insurance does not cover general maintenance, there is a relatively high deductible, plus you take a big hit in premiums if you get into an accident. Of all the times that I or people I know have been in accidents, the insurance companies were involved less then 50% of the time. For fender benders, the driver at fault is better off paying $1000 to fix the other persons car than see their premiums jump $150 a month for eternity.

Earlier posters mentioned that distinction between "catastrophic" and "maintenance" care; its an important distinction that should be front and center in the debate.

Also, I think people do not take the cost of proposals like these seriously because they don't look at their own costs. It may be a stretch, but I'd reckon that most of the people who want to "help the poor get healthcare" have employee sponsored plans and are thus blissfully unaware of the true cost of these programs. In most of these programs, the employer is paying 80-90% of the premium, so the employee only sees $50-$100 a month coming out of their pay check (many people find even that to be expensive), when in reality the plan is costing $500-$1000+ a month. If they knew the real cost, they might opt for less "maintenance" coverage and just pay out of pocket for everything but the big stuff.

Comments for this entry have been closed
Return to top