Thursday, September 11, 2014

Actuarial Euvoluntarity

Alexandra Farrington of Fairfax, VA has posted a petition on to have George Mason University President Angel Cabrera call for the resignation of Professor Walter Williams. Her complaint is that Williams (disclosure: Williams was my Microeconomics I professor) made bigoted remarks (link 1 link 2) "against the gay community." viz:
In his column, he says that gay people are worse than smokers and those that are obese. His argument is, homosexuals should have higher health insurance than heterosexuals solely because they identify as homosexual. He relates this to non-smokers, who have lower health insurance than smokers. Mr. Williams believes that because gay individuals have a lower life expectancy, they deserve to pay more for their health insurance. Not only is his conclusion offensive and discriminatory, but his information is out of date: the basis of his evidence comes from a report issued in the late 1990's when HIV/AIDS was more fatal than it is now.
I believe this is the offending passage from the first link:
According to the International Journal of Epidemiology, life expectancy at age 20 for homosexual and bisexual men is eight to 20 years less than for all men. That's a lifestyle shortening of life expectancy greater than obesity and tobacco use. Yet one never hears of insurance companies advertising lower premiums for heterosexual men. You say, "That would be discrimination." You're right, but why is it acceptable for insurance companies to discriminate against smokers and the obese but not homosexuals? After all, they are all Americans and protected by the Constitution. It's really a matter of politics, as seen by the journal's publication of an article titled "Gay life expectancy revisited" ( The publication had to soft-pedal its study results because of complaints that pointing out life expectancy differences between heterosexuals and homosexuals had become fuel for homophobia. The bottom line is that homosexuals have far greater political power and sympathy than smokers and the obese.
Emphasis added. Williams is making a public choice claim. Actuarial science is discriminatory by its very nature. Indeed, that's precisely the point: bowtied number crunchers crawl mortality data to rigorously price mortality risk based on observable characteristics. Because the insurance industry is (relatively) competitive, firms would lose customers in a big hurry if they attempted to charge customers more for matters of mere taste—if there is no actuarially sound justification for charging gay customers more, there's a very lucrative commercial opportunity for a firm willing to serve these customers at lower (read: actuarially fair) rates. This is the basic logic of the marketplace. Econ 101, if you will.

But it's obviously incomplete logic. Those tweed-wearing, bespectacled, bowtied number crunchers have sense enough to indulge a little backward induction. They're in the business of pricing mortality risk, a skill that lends itself naturally to the pricing of political risk as well. In a competitive market, a single participant is a price-taker. Here, the marketplace is in political sympathy, and insurance firms are idea takers: unless the premium differential is of sufficient magnitude, it's simply not worth it to the firm to buck popular opinion and discriminate along sensitive margins. A firm that stands to earn an extra $50,000 from increased premiums won't spend $100,000 in legal fees and fines to obtain those premiums. This is the basic logic of the regulated marketplace. Econ 312, if you will.

Prices are information. They tell buyers to modify their consumption and suppliers to alter the goods under delivery. In the case of actuarial analysis of mortality, this signal is sometimes useful. You can send a price signal to a smoker that her behavior has costs not immediately borne by her. It's an incentive to quit. Price discrimination in this case takes a non-euvoluntary habit and makes it more so by making the regret more immediate: it (admittedly, imperfectly) shifts the future costs to the present. Ditto overeating or being sedentary. But there's no such useful signal to the consumer for homosexuality, or left-handedness, or being ginger or whatever else a clever actuary might stumble upon as being a salient variable in the tables. The normative question about the role of insurance in society is whether or not it's socially useful to force insurance firms to pool these sorts of non-modifiable risks. Should relatively long-lived women subsidize men? Should relatively healthy Asians subsidize Whites?

Like it or not, from the point of view of the practical firm, these are political questions, and in order to maintain a hard-fought reputation of being utterly and ponderously boring and predictable, the insurance industry will not attempt to fight public opinion on questions like these.

As far as the contents of the petition, is it euvoluntary for the university to continue to permit Williams to teach PhD Micro I? I'm afraid the limits of my Smithian sympathy are tested here. Williams is certainly thought-provoking in class. And he certainly communicates on the barbarism-civilization axis of Arnold Kling's tripartite rhetorical model. But in my (admittedly limited) experience, he tends to draw libertarian conclusions from his analysis. I cannot recall an instance where he advocated the sovereign exerting unjust dominion over constituents. But my experience is limited to the single class I took from him and the few hallway discussions we've had over the years. The natural counterargument is that merely by mentioning moral opprobrium (as in the second link above), he raises in an unsophisticated audience additional ire against his disfavored groups; ire that might result in state-sponsored violence (#Ferguson). If that's the case, there's a good opportunity for men and women of good conscience to respond using careful, rigorous economic analysis, the kind you might learn in Professor Williams's class.

No comments:

Post a Comment

Do you have suggestions on where we could find more examples of this phenomenon?