Of Genomes and Lemons
How Well-Intentioned Laws Can Harm the People They Mean to Protect
SEPTEMBER 01, 2001 by E. FRANK STEPHENSON, MICHAEL E. RUPERT
Michael Rupert is a senior majoring in economics at Berry College in Rome, Georgia. Frank Stephenson is an assistant professor of economics in Berry College’s Campbell School of Business.
While the recent announcement of the mapping of the human genome was greeted with optimism about cures for dread diseases, it also led to predictable teeth-gnashing about possible genetic discrimination.
Genetic discrimination ostensibly occurs when economic decisions are based on genetic information about people’s susceptibility to disease. For example, medical or life insurers might use genetic information in deciding whether to cover an individual or what premium to charge. Likewise, employers seeking to minimize employee-benefit costs might use such information in deciding whom to hire.
Predictably, bills have been introduced in Congress to ban genetic discrimination by insurers and employers. (The 1996 Health Insurance Portability and Accountability Act already prohibits the use of genetic testing by group health insurers; this barrier has probably had little impact thus far since the genome has only recently been mapped and because genetic testing is still in its infancy.) Some 18 states and at least one locality (the well-meaning Montgomery County, Maryland) have enacted genetic-discrimination bans of one form or another, and the Equal Employment Opportunity Commission (EEOC), not waiting for additional federal legislation, claims that genetic discrimination is already illegal under laws prohibiting discrimination against the disabled. It is under this legal theory that the EEOC recently sued Burlington Northern Santa Fe Railroad seeking to prohibit its testing workers who submit carpal-tunnel-syndrome complaints for “a predisposition [for the syndrome] within the body chemistry of the individual” that “has nothing to do with work.”
Legal bans on the use of genetic information appeal to a perceived, though somewhat perverse, right of privacy. After all, people can keep their genetic information to themselves by not purchasing insurance. And people’s privacy could be protected by prohibiting insurance companies from disseminating information gleaned from genetic testing without the consent of the subject.
If, however, individuals wish to purchase insurance, they might be required to submit to testing as a condition of coverage. (Use of such background information is analogous to banks performing credit checks before making loans.) Such use of genetic testing would not be fundamentally different from the now-common use of blood and urine tests and the required submission of information on lifestyle factors such as diet and smoking habits.
Legislation prohibiting genetic testing as a condition of purchasing insurance could ultimately undermine the insurance market and make it difficult for people to be able to purchase insurance. Consider the life insurance market. Currently individuals and insurance companies enter into policies under a large cloud of uncertainty. While people have some inkling about their expected life spans from family history and lifestyle, they nonetheless know little because family history is not definitive and their parents often come from families with different histories. Of course, uncertainty about life span also exists because accidental death is possible.
Similarly, insurance companies can obtain some information about potential policyholders from blood and urine testing and lifestyle questionnaires, but they remain largely uncertain about the expected life spans of particular individuals. Insurers can, however, make reasonably accurate predictions of the mortality rate and life expectancy for the population as a whole, and they use that information to set premiums. Insurance is essentially risk-pooling, which works well when both insurer and insured have similar levels of uncertainty.
Delicate Balance Upset
Bans on genetic testing threaten to upset the delicate balance of mutual uncertainty in the life-insurance market by creating the possibility of asymmetric information. Asymmetric information exists when individuals and insurers have different information about life span. Outlawing genetic discrimination could create informational mismatches by prohibiting companies, but not individuals, from engaging in predictive genetic testing. Individuals could legally undergo tests to determine their life expectancies and could use the resulting information in buying insurance.
To consider how asymmetric information might undermine the life-insurance market, consider the following example. Imagine three people, Anne, Becky, and Cara, who because of genetic differences have life expectancies of 60, 70, and 80 years, respectively. Their life-insurance policies are with Big Global Insurance Company, Inc. Big Global knows that the average life expectancy of its policyholders is 70 years, and it sets its premiums accordingly. Of course, Big Global also knows that some of its policyholders will die before 70 and some will die after 70, but it cannot, based on blood and other tests, predict which customers will die young and which will die old.
Now suppose that a predictive genetic test is introduced. Individuals can get themselves tested, but insurers are legally prohibited from using the test in setting premiums or making coverage decisions. Anne, Becky, and Cara all avail themselves of the test and each learns her life expectancy based on genetic factors. Since Cara learns that she has a life expectancy of 80 years, barring accidents, and therefore has a very low probability of premature death, she judges that life insurance is not a good deal for her. Put differently, Cara is paying a premium that is too high because it is based on her dying at 70 rather than 80. The information gleaned from the test enables her to enhance her well-being by spending her money on things she values more than life insurance.
When Cara cancels her policy, Big Global has only two policyholders remaining. Their average life expectancy is 65 years, and Big Global now raises its premium accordingly. Becky, who was happy when her premium corresponded to her 70-year life expectancy, now finds life insurance to be too expensive and cancels her policy. As a result, Big Global has only one policyholder, Anne, left, and it adjusts its premium to match her 60-year life expectancy. (In practice, since Big Global is proscribed from performing genetic testing, it only learns of the change in the life expectancy of its policyholders over time as it notices its customers dying younger than before. Hence, it is probably more accurate to say Big Global raises its premiums over time and scares off future Beckys. For simplicity, we ignore this complication of timing, but it does not alter our conclusions.)
Perversely, the result of the ban on genetic testing by insurance companies is harmful to two consumers, Becky and Cara, who would still like to purchase actuarially fair life insurance because of the uncertainty arising from accidental death. They are unable to do so, however, because the insurance company is barred from using technology to learn, as Becky and Cara have, that they are genetically low-risk (they have high life expectancies) and then setting a correspondingly low premium.
Moreover, while genetic antidiscrimination laws are supposed to help people like Anne, who have short life expectancies because of “bad” genes, future Annes will nonetheless pay premiums that match their riskiness, because future Beckys and Caras will forgo insurance. So antidiscrimination laws harm the Beckys and the Caras of the world, while providing only temporary relief to the Annes.
In other words, the imbalance of information created by the ban on genetic discrimination results in an adverse selection process in which the genetically healthy choose to forgo life insurance because companies are unable to identify them as such. People who are genetically risky continue to purchase insurance because companies are not able to immediately adjust their premiums to fully account for these people’s higher risk. Over time the average level of riskiness in the pool of policyholders rises, causing firms to further increase their premiums and scaring off more potential customers. As a result, many people are unable to purchase insurance against the risks they do face and a lot of insurance companies go bankrupt.
Recognizing the possible harmful effects of asymmetric information on its insurance markets, the British government has explicitly allowed life-insurance companies to ask for genetic screening for Huntington’s disease and is considering granting approval to test for other diseases.
Though well-intentioned, laws banning predictive genetic testing by insurers are lemons that will, oddly enough, harm the very people they are intended to protect.