Economics and genomics

Yesterday, I spat in a tube, added a preservative solution, and mailed my saliva to 23andMe, a company that provides its customers with some of their genetic data. 23andMe does not sequence whole genomes—that is still extremely expensive. Instead, they examine about 600,000 locations in human DNA, selected on the basis of actual or potential informativeness. This represents only a small fraction of the genetic variation that occurs in humans, which is itself only a small fraction of the human genome. While I wish the product were more complete (perhaps in 15 or 20 years getting one’s complete genome sequenced will be commonplace), I am eager to get the results, which will be available in 2-4 weeks.

Not everyone is as enthusiastic about new genetic technologies. I was warned by my friends to keep my genetic results private. You cannot peruse 23andMe’s website for long without coming across references to GINA, the Genetic Information Nondiscrimination Act, which passed in Congress with only one dissenting vote (Ron Paul’s) and was signed by George Bush in 2008. Genetic privacy is also protected by a number of state laws.

GINA prohibits health insurers from requesting or requiring genetic tests or using genetic information about you to determine premiums. I find this prohibition to be highly unethical—what consensual arrangements I or other consumers make with private companies are none of the government’s business. Unfortunately (and as usual), I was not consulted. In any case, I suspect that GINA is not really a binding constraint yet. Genomics is not yet so advanced that health insurers would be likely to insist on genetic testing. However, as the science progresses, GINA will bind and have some potentially significant effects on the insurance market.

In a health insurance market in which all consumers have access to statistical information about their genetic predispositions to disease and insurance providers do not, economic theory predicts a classic Akerlof lemons/adverse selection situation. Those consumers with bad health genes would be eager to buy insurance and those with good health genes would not. Insurers, faced with an influx of bad gene customers and an outflux of good gene customers, would have to raise their rates, further driving away the desirable customers. It is the genetic version of Gresham’s Law: bad genes drive out the good.

Empirically, evidence for the lemons model is not clear cut. For instance, people with objectively higher risk of death do not seem to buy more life insurance. There is even evidence for propitious selection—accountants buy more life insurance than motorcyclists. However, this seems to be due to an accident of how humans view risk: those who are cautious types with respect to dangerous activities are also cautious with respect to providing for their families. If the lemons model applies anywhere, it would seem to apply to genetic information, because there is no reason why good health genes should be correlated with risk attitudes. All of the preceding assumes, of course, that insurers cannot obtain basically the same genetic information in other ways, such as detailed questionnaires about family medical history.

In an environment in which insurers could require genetic testing and set premiums accordingly, the market would not break down, but it would change. People with good health genes would get a discount on their insurance, and people with bad health genes would have to pay more. In effect, just as with the adverse selection scenario, it would become difficult or impossible to buy insurance against being born with bad health genes.

People will be eager to propose solutions to this problem. A standard proposal would be to maintain genetic privacy but mandate the purchase of health insurance. This seems to be what is happening in any case, without thought to the adverse selection imposed by personal genomics. If the good genes are coerced into staying in the market, insurers will charge the average cost of insuring the pool, effectively transferring wealth from the good genes to the bad genes. Another likely response would be so-called “social insurance,” in which the government taxes everyone and pays for medical care for everyone. It does not matter who knows what about whose genes if the government is paying for everyone’s medical care indiscriminately.

A hard-bitten cynic will be skeptical of either of the standard suggestions. It is hard to see the first as anything more than a boon to the medical-industrial complex. Mandating health insurance coverage is an extreme solution when people may have other good reasons for forgoing coverage, but it will result in higher profits for insurers and providers when there are barriers to entry. The second, even more than the first, will result in many unintended consequences, including the deterioration of quality of care as market incentives are eroded.

Maybe the appropriate response is to ask, who cares? If it is impossible to insure against bad health genes, is that such a big deal? After all, it is currently impossible to insure against low IQ, or a bad personality, or ugliness, all of which are at least partially genetic and can have profound effects on one’s life. If we are not willing to take a stand for ugly people, why should we be willing to do so for the congenitally unhealthy?

Furthermore, the real solution to the issues raised by genetic technology may be…more genetic technology. Therapeutic retroviruses hold forth the promise of being able to fix our DNA. If this or other research pans out, the social benefits will be enormous: potentially billions of additional life-years and higher quality of life. Probably the best thing the government can do to encourage this scenario is to get out of the way. And unless you are a geneticist, the best thing you can do to support the fledgling genomics industry is to become a customer.