Genetic Testing: Further Debate with Richard Epstein

dna7.jpgRichard Epstein has posted a reply continuing our debate over whether employers should be able to use genetic testing information to make employment decisions regarding employees. Here are the posts in our debate so far:

1. Solove, IBM vs. NBA: Using Employee Genetic Information

2. Epstein, Two Cheers for Genetic Testing

3. Solove, A Reply to Richard Epstein on Genetic Testing

4. Epstein, A Third Cheer for Genetic Testing

Epstein’s latest reply, A Third Cheer for Genetic Testing, slips in another cheer for genetic testing. He asserts that my argument that genetic information only reveals propensities, not the presence of certain conditions, actually cuts in favor of employers using genetic information:

That information should not make the employer instantly hand out a pink slip. It is one factor among many to be taken into the overall assessment. The insurance could be supplied, but in exchange for an additional premium that reflects that additional risk. Or the health insurance could be supplied subject to an exclusion for the risky condition. Judgments like that are made all the time in the insurance business, and there is no reason why they cannot be made with the processing of genetic information.

Epstein is certainly correct that genetic information is helpful in assessing risk, and he is right that employers need not just fire or refuse to hire people with genetic predispositions. But there are larger normative issue at stake: What risks ought people to bear? Who ought to bear these risks? How ought these risk to be distributed throughout society?


With insurance, there’s the issue of how equal a risk distribution one wants to achieve. We could have health insurance, for example, that is prohibited from charging people extra based on pre-existing conditions. In other words, we all bear the cost of health insurance equally. Another model is for those who are more at risk to pay higher premiums and those who are less at risk to pay lower premiums. I generally support insurance that is more equal – even if that means I pay more. The primary benefit of insurance, I believe, is to spread risk throughout society so as to eliminate certain contingencies in life. Society as a whole absorbs these losses, rather than particular individuals. Unfortunately, insurance only partially functions in this way today.

Epstein also takes issue with my arguments about countervailing values such as freedom and autonomy as well as privacy. He argues that employers have freedom too:

There is nothing in the ideal of personal freedom or autonomy that requires one individual to offer a subsidy for the health care of a second. . . . Solove’s revised notion of autonomy and freedom, but only for some, means additional duties imposed others. Yet no justification for that imposition on employers is given.

Everything in life involves a tug-of-war between freedom and duties. Tort law, for example, imposes duties on everybody’s conduct, and thus limits everybody’s freedom. The law routinely limits the freedom of some to promote the freedom of others. That said, Epstein makes a fair point in demanding a justification for why employee freedom should trump employer freedom with regard to genetic testing. I believe that employers already stand in an incredibly powerful position vis-à-vis employees. Employers have a vastly greater ability to limit employee freedom than vice versa. The law recognizes the profound power of employers and will often try to equalize the employment relationship, such as through laws regulating sexual harassment and workplace safety.

Epstein challenges my argument about privacy as a countervailing value:

The appeal to privacy is subject to similar limitations. No doubt privacy covers the case where individuals want to snoop into the private lives of others. But here the deal is quite simple: either you waive your right to privacy or you will not receive a job from me. People waive their liberties when they agree to work for another individual. They can, if they choose, waive their privacy rights as well.

People indeed are forced by employers to waive many rights, but to say this still leaves open the question of whether the law should allow employers to force people to waive particular rights. We could allow people to sell themselves into slavery; we could allow employers to force people to consent to sexual harassment or to ultra-dangerous workplace conditions. But we impose limits on what employers may reasonably require an employee to consent to.

Genetic testing should be one of those limits. Employers should not be privy to everything in a person’s life, even things that are relevant for the job. For genetic information, there’s a social value in preventing people from being disadvantaged for reasons beyond their free will.

One of the greatest advances of civilization, in my opinion, is the attempt to help ameliorate the horrible contingencies nature and chance create: hurricanes, earthquakes, diseases, accidents, pestilence, fire, and countless other calamities. Without protection against these catastrophes, life becomes all the more precarious. A person’s entire life’s labor and toil can suddenly be wiped out in a stroke of bad luck. Fortunately, we’ve found ways to eliminate these contingencies. We spread and redistribute risk. We protect those who have suffered bad luck. And in return, we can live life with at least some measure of protection from the vagaries of chance and the harshness of nature.

I believe that such a system is much more preferable than one that places the risk on the lone individual, who, through no fault of her own, finds herself dealt a bunch of “bad” genes. So we could say “tough luck” to the person who has “bad” genes or “tough luck” to the employer who hires that person. I’d rather have the employer bear that burden. Employers generally have more power and options to absorb the loss and to figure out how to insure against and ameliorate the risk.

You may also like...

5 Responses

  1. meep says:

    Perhaps a little off-topic, but there’s a problem with genetic testing and insurance:

    http://www.livejournal.com/users/meep/336896.html

    I go through a hypothetical example, with numbers, about what the impact of anti-selection is (that is, those who know they’re at a high predisposition for certain conditions take advantage of policies that work in their favor… this is what happens when one doesn’t price for known risks.) I don’t see this is necessarily a problem for employers, other than anti-selection of this kind might make certain generous benefit programs go away, as employers are under no obligation to keep a particular benefit plan — of course, the employers (or the insurance company providing group benefits) won’t find out about their extra risk until the condition actually occurs (though, of course, it may never occur.)

    As for employers being more able to absorb the risk – well, even small businesses do a better job than the individual (as there’s more people to spread the risk over) – but why should it be a function of the employer? Why not a function of the government or private market at large? I don’t get car or home insurance through my employer, for example — I go and buy my own in the private market. These risks are something that could also be borne by my employer, as the company I work for is larger than just me and could spread risks. But they don’t bear those particular risks, and nobody complains about that.

  2. geoff manne says:

    This seems crazy:

    With insurance, there’s the issue of how equal a risk distribution one wants to achieve. We could have health insurance, for example, that is prohibited from charging people extra based on pre-existing conditions. In other words, we all bear the cost of health insurance equally. Another model is for those who are more at risk to pay higher premiums and those who are less at risk to pay lower premiums. I generally support insurance that is more equal – even if that means I pay more. The primary benefit of insurance, I believe, is to spread risk throughout society so as to eliminate certain contingencies in life. Society as a whole absorbs these losses, rather than particular individuals. Unfortunately, insurance only partially functions in this way today

    The errors of economic logic in this one paragraph (to say nothing of the others) are pretty serious.

    For starters: We could have insurance that is prohibited from charging a market-clearing price (it’s called price controls). The effect is not, however, simply a re-spreading of risk. Insurance will become a losing proposition for the insurers (and, of course, the low-risk insureds) and they’ll go out of business. Insurance won’t function as insurance — it’ll just be cumbersome and limited redistribution.

    What most amazes me is the statement that the purpose of insurance is to spread risk through society. In what world? Insurance is a private decision (except where it is offered or mandated by the government) and its “purpose” is to make money for the insurers and to smooth income and consumption for insureds. It’s true that insurance companies (or other effective insurers like large corporations, etc.) are better risk bearers than individuals in part because they can aggregate risk across a larger population. But this isn’t even close to what you said. Insurance is not a giant redistribution scheme. It’s not about “equalizing” risks or protecting the uninsurable. That’s just wishful thinking on your part.

    In the end, any of your suggestions for distorting insurance markets would be just that — distortions. Each intervention will beget further interventions to correct the prior distortions until you’re left with a hollow shell of an effective insurance market. If you’re bothered by genetic inequality, the way to correct it is not by crushing the insurance markets. Try basic income redistribution or insurance vouchers or some other mechanism for subsidizing the unluckys’ participation in a functioning market rather than eviscerating the market itself.

  3. Bruce says:

    “It’s true that insurance companies (or other effective insurers like large corporations, etc.) are better risk bearers than individuals in part because they can aggregate risk across a larger population.”

    OK Geoff, engage in a little charitable interpretation here and assume Dan meant just that. The issue is how genetic testing alters the existing dynamic which I assume most people are comfortable with. You argue a slippery slope will occur if the insurers are prevented from genetic testing by law, but I’m not sure that the slope is all that slippery. One can easily conceptually distinguish between disclosure of existing medical conditions and habits and undergoing genetic testing (or disclosing the results of such tests) for predispositions that have not manifested themselves. A genetic predisposition is something that, at least at present, is entirely beyond individual control (I suppose there’s a middle ground for genetic predispositions that only manifest given certain behaviors, but let’s just admit for now that that’s going to be the hard case). At present, given the limited information available through genetic testing, both individual and insurer are largely ignorant of the risks. As a result, those with higher risks are essentially “subsidized” by the insurance premiums paid by others (which I think is what Dan was getting at). Why not preserve this status quo, which is not currently “eviscerating the market”?

  4. Paul Gowder says:

    Stuart Buck just posted a link to this paper, which apparently (I haven’t read it yet) argues that statistically-based discrimination is likely to impair market absorption of more individualized information, at least when the statistical information applies to skill. To the extent that some of this genetic discrimination goes to skill-relevant factors, this might be an interesting contribution to the discussion. (I’m not sure that it would be generalizable to likelihood of death.)

  5. Geoff, pardon the fact that I’m not an economist or an expert in insurance, so I don’t speak the lingo or have an extensive background in the functioning of insurance markets.

    On the insurance front, when I was speaking about “insurance” I wasn’t just referring to private insurers, but of the larger concept of distributing the risk of loss. This isn’t the goal of private insurers, to be sure, but I was speaking in terms of larger societal goals. Perhaps national health insurance will achieve these goals better than the current system. Perhaps these goals are the wrong normative aims. I didn’t want to have a debate over the nature of insurance, and I regret using the term. I wanted to have a debate over which risks were appropriate to be distributed and which risks were not.

    For every risk that can be knowable, the question arises whether or not insurers should be permitted to use information about that risk in calculating insurance premiums. Some risks are just unknowable as of now, and so they are not taken into consideration in calculating premiums. Other risks are knowable from certain information or factors, but the law may not allow insurers to use the information and factors.

    Suppose that for Disease X, there’s no good factor or test that will help assess the risk for developing the disease. As a result, insurers treat everybody as having an equal risk of developing Disease X. Now suppose that suddenly a test is developed that can predict with a great degree of accuracy whether a person will develop Disease X. With this test, those at risk for Disease X now are suddenly disadvantaged. Before, there wasn’t information to reveal their risk factor; now there is, and if insurance companies can use it to calculate premiums, then the losses for this disease can be shifted from everybody to those at risk for Disease X.

    Is such a development a good thing? What if nearly all diseases followed this trajectory? What would become, then, of health insurance?

    I’m not an insurance expert, so I don’t know the answers to these questions. But there is a larger normative question at stake that is raised by this hypo and by genetic data — to what extent should we allow better information about assessing risk (as it becomes available via genetic testing or other developments) to be used by insurers?