David R. Henderson  

Robin Cook's Mistaken Idea about Pooling Risk

PRINT
How Much Liberty Will We Sacri... Tyler's Breakdown...

I'm on vacation in Hawaii and just finished the first novel I've read in months. It's Marker by Robin Cook. While it's a real page turner and I enjoyed it, I found the ending, which I began to anticipate halfway through, highly implausible. Still, there's a grain of truth in it, but Cook totally misses that grain.

SPOILER ALERT for what follows.

The bad guy in the novel is a managed health care organization called AmeriCare that is out--gasp--to make money. AmeriCare pays people to go into hospitals and kill some of their clients once specific genetic markers have been identified, markers that indicate expensive health problems for those clients down the road. I trust you see why I found this implausible.

Still, as I said, it's understandable that the fictional AmeriCare would not want such customers because those customers could cost them lots of money. But it's understandable only if the insurance company is not allowed to price for the clients' risk. Cook even gets that. Towards the end, he has his hero state:

Using single, easily performed tests, people who are destined to cost them significant money can be recognized. The problem is that the large healthcare companies cannot show discrimination, so they have to take them. At that juncture, from a purely business perspective, they should be eliminated.

In other words, Cook understands that the problem is that insurance companies can't price for risk.

But then he doesn't advocate allowing health insurance companies to price for risk, the way auto insurance, home insurance, and life insurance companies are allowed to do. Instead, he has his hero state:

Unfortunately, that's the way the business works and why some level of government oversight is necessary as a general rule in a free-market economy.

The hero makes this statement only two paragraphs after pointing that it is precisely because of government oversight that the insurance company can't price for risk and that this inability to price for risk is what leads to the problem.

In case anyone thinks this is just a fictional character spouting off, Cook then adds an Author's Note in which he writes:

The end result will be that the concept of health insurance, which is based on pooling risk within specified groups, will become obsolete. In other words, risk cannot be pooled if it can be determined.

Cook is mistaken. The whole idea of insurance is to pool like risks. And the only way to pool like risks is to determine them.

Cook then jumps to his solution, writing:

I now feel that there is only one solution to the problem of paying for healthcare in the United States, indeed for all developed countries in this economy: to pool risk for the entire nation.

That way, he writes, "we" can "decide rationally how much we should spend on healthcare i general." Of course, he doesn't say how "we" should decide rationally--or who "we" is.

For more on insurance, see Richard Zeckhauser, "Insurance," in David R. Henderson, ed., The Concise Encyclopedia of Economics.


Comments and Sharing





COMMENTS (15 to date)
JohnC writes:

Still not as bad as Sean Bean's implausibly stupid plan in
Goldeneye to steal £100 million and then immediately destroy the Bank of England (or pretty much the whole of Prometheus).

Daublin writes:

Don't we get exactly this from Obamacare? Insurance providers have much less ability to price discriminate. Young, old, sick, you name it, they have to take you, and for the same rate as everyone else.

Andrew writes:

Alas, I tried to find a Twitter handle for Robin Cook but none existed. I would love to read his response.

Andrew writes:

Another point. I wonder if Mr. Cook is aware that having nationwide pooling by "we" doesn't reduce the probability of killing people "we" have identified as higher risk/more expensive.

Max writes:

Another point. I wonder if Mr. Cook is aware that having nationwide pooling by "we" doesn't reduce the probability of killing people "we" have identified as higher risk/more expensive.

I realize it's considered a bit low-class in certain circles to utter the phrase "death panels," but yeah, this.

Mike Davis writes:

I don’t think you can simply dismiss the adverse selection problems created by genetic testing, a point revealed by your comment that insurance is intended to pool “like risks”.

Suppose, for example, scientists discovered a genetic marker that is highly indicative of the risk of getting, say, prostate cancer--if you have the marker, there is 75% chance that you’ll have prostate cancer before the age 60, without the marker you’re good. Suppose further that a cheap home testing kit for the marker is available. To make the hypothetical even more challenging, suppose that prostate cancer is really expensive to treat and so guys with the marker are likely to have significantly higher health care costs than guys without. (I know there’s some relatively implausible assumptions here but this whole thread is based on David’s beach book.)

An unregulated health insurance market (at least one that covered prostate cancer) would then pool guys with the marker, charging them higher rates. Adverse selection would prevent the at-risk guys from forcing the no-risk guys to share.

But that stinks. What we really want is insurance against having the genetic marker. Sure, the guys who took the test and found out they were not at risk feel lucky and don’t want to buy insurance that covers prostate cancer. That doesn’t mean there is efficient risk sharing. If you had asked them before they took the test whether they wanted to pool with everyone, the risk averters among them would have said yes. Again, the real risk is having the bad genes. With cheap genetic testing that risk will be hard to insure.

I can think of market solutions much less extreme than having insurers murder patients or single payer health care (yes, I know some of you think the two are equivalent). It’s just not as simple as it seems.

Russ Roberts needs to write the next health insurance novel.

Steven Flaeck writes:

Risk pricing wouldn't change this; you profit from assassinations by reducing the payments paid vis-a-vis the total income of the company.

Glen Smith writes:

Mike,

You seemed to miss how risk is pooled. The guys at no risk don't pay for the guys at risk. The guys at risk who don't get cancer pay for the guys at risk who do. An issue would only arise if having the genetic markers was 100% predictive of cancer but that also would eliminate the risk.

Aaron McNay writes:

I think part of the problem with discussions on health insurance stems from different understandings of what health "insurance" should be and what it should do.
For some, health insurance should be used to financially protect a specific individual against particular events from occurring to them. In this instance, the cost of insurance is simply the probability of the event occurring to the specific individual * the expected cost of the event + the administrative costs of the insurance. Because people are risk adverse, they are willing to pay the higher costs that insurance impose to protect them from risk. As they are protecting themselves from their associated risk, they have every incentive to stay in the market. The problem with this system for some people is that sick people will also face very high insurance costs, as the probability of an insured event occurring is very high. (As Mike Davis said, some people think this "stinks")
For others, health insurance is a system where healthy people subsidize the health care of sick people. In this system people are not charged premiums that are associated with their risk levels, but with that of the group as a whole. The problem with this system, however, is that healthy people have a strong incentive to either drop out of the insurance pool completely, or form their own pools. I think for people who support this second type of insurance, preventing the pricing of risk is ultimately one of the reasons why they think insurance exists.
I think the problem is David Henderson sees health insurance as the first type, while Robin Cook sees health insurance's purpose as they second type. In addition, I think that these differing views on what insurance is, and what it's purpose is, can explain a lot of the disagreements that exists on what health care policy should be.

Hazel Meade writes:

The whole notion that insurance is about "pooling risk" is wrongheaded and leads to all sorts of errors.

From the consumer's perspective, the point of insurance is to trade financial risk for a fixed predictable cost. You pay a monthly premium, instead of taking the chance on a unexpected $100,000 cost. There is no pooling involved. You don't care about anyone else's risk, or who else is in the pool. All you care about is the premium being charged, the probability of the risk being realized and the expected costs.

Pooling only comes into effect from the insurance company's perspective. By putting many individuals in the same "pool" you average over the statistical probabilities and your costs for the "pool" become predictable. You can thus know what your expected monthly expenses are and set a premium for everyone in the same pool based on those expenses and thus ... Profit!

Now, these days, we have lots of knowledge about genetic variation and risk, so the pools of "like" risks are getting smaller. However, because we have all this knowledge about risk factors we can also calculate the expected costs, and the insurer still has many different customers, so he can still average out the risk. It's more complicated, but that's what computers are for.

The problem is that most of the public doesn't really understand "risk". Some people assume that all risks are uniform-distribution. And some people think that all insurance should operate as if it were. There's a curious transferance of the idea that it's "unfair" for some people to be born with a bad gene, and the idea that it's "unfair" for the insurer to charge them more. The insurer is just pricing their risk, not *creating* it, but because people did not innitally have knowledge of their own risk, they think insurers should behave as if all risk was unknowable and everything was a perfect uniform distribution.


Hazel Meade writes:

Mike and Aaron both bring up some interesting points.

I think health insurance is Aaron's first system. Pooling is just a tool for the insuerer to average out risk. It has nothing to do with the consumer.

But as Mike points out, increased knowledge makes pooling more difficult. But, as I suggest, the insurer can still average out risk, it's just more complex. Instead of a single pool with the same risk and everyone paying the same premium, you have a vast group with varying premiums according to varying risks. What's wrong with that?

Mike brings up the fact that people don't know what their genetic markers are. There could be some market based solution where you insure against having certain markers, maybe even before birth. But that doesn't exist right now, so people are absorbing the risk of having a risk-factor, and will end up paying higher premiums when they find out.

Still I think that is a pretty small problem. There are lots of risk factors and very few people are going to have a lot of bad ones. The notion behind Aarons second system is really to correct for the fact that it's "unfair" that nature endowed some people with bad genes. This is a kind of Rawlsian idea about having society correct for the effects of bad luck. Everyone pays a flat rate. Nobody ends up with unequal healthcare expenses.

Of course, even with system 2, I don't see why you have to make private insurers (and their stockholders) the victims of this.

mike davis writes:

I’m not a Rawlsian but I do see some value to “veil of ignorance” thought experiments. Consider this one (along the lines of my earlier post).

Suppose you and a large group of otherwise similar individuals have a 1% chance of having a genetic defect that raises your chances of getting an expensive disease from 0% to 75%. That is, you have a 0.75% chance of the disease. Now suppose you’re about to get a test that will reveal whether you have the genetic defect and I come up to you and propose the following:

“ My mutual insurance company offers two policies that cover the cost of the disease. The first policy costs 0.75% of the cost of the disease but to qualify you have to pay us now. The second policy costs 75% of the cost of the disease but you can wait and find out what your test says before you pay us. Of course none of our products let you buy the insurance if you already have the disease. Remember, we’re a mutual insurance company and so all of our administrative expenses are covered by the profits from bake sales.”

I’m pretty sure all risk averters would prefer the first policy. (I think concave utility is enough but I’d want to want to do the math to make sure.) I’m also pretty sure, there’s no way to offer that type of policy.

Hazel’s point—which I think is a good one and I hope is true—is that this kind of problem isn’t such a big deal since all the genetic risk factors that are being discovered are largely uncorrelated with each other. If I’m at high risk for disease X and your at high risk for disease Y, the lack of the ex ante insurance market doesn’t matter—we get close to the same result.

MingoV writes:
The whole idea of insurance is to pool like risks.
That is incorrect. The purpose of insurance is to pool money from covered entities to prevent large monetary losses after adverse events. There is no requirement that insurance premiums have to be stratified based on relative risks. Thus, it is plausible to have health insurance with the same premium for every customer. That's essentially what Robin Cook advocates. I think its a poor decision for a number of reasons. One reason is that there is no reward for reducing risk (e.g.: by exercising) and no penalty for increasing risk (e.g.: being an overeating couch potato). A more important reason is that high risk customers will switch to the "one pool" insurer and force the premiums up. Similarly, the low risk customers of the "one pool" insurer will switch to another insurer with lower premiums. Thus the "one pool" insurance idea will work only if there is a monopoly. Like ObamaCare.
Hazel Meade writes:

mike, I think your numbers are reversed in your hypothetical.

You would offer the policy that pays 75% of the cost before you take the test, and the policy that pays 0.75% after you take the test (if you don't have the risk factor you won't buy, and if you do you're effectively getting nothing.)

mike davis writes:

Hazel,

I was imagining a zero-deductible, full coverage policy. If the treatment cost $100 K, your premium would be either $750 or $75,000 depending on whether you committed before or after you got the results of your genetic test.

MingoV,

In disputing the claim that insurance pools "like" risks, you're making a semantic argument (which doesn't mean you're wrong or irrelevant). If you think about the math, insurance is just a mechanism that creates a portfolio of (mostly) uncorrelated random variables. There are lots of ways to make up such a portfolio. Obviously "similar risk" can't mean that the random variables are correlated. My neighbors and I share a very similar risk of suffering hail damage to our roofs, but there is no way we would form a mutual insurance company that pooled just our risk.

Your larger point about adverse selection and moral hazard reminds of us of a bigger challenge. Those of us who are very uncomfortable with government organized monopoly providing healthcare have to do a better job of thinking about and then explaining how private insurance can help resolve those issues. We can't simply say "deregulate insurance" without explaining how that would work.

Comments for this entry have been closed
Return to top