Insurance and Pooling Equilibria

In the bad old days, insurance was a way to smooth cash flows from improbable but large expenses: fire, health, auto mishaps. Through repetitious metonymy, 'health care insurance' and health care are now synonymous.

I was struck by Obama's mention last night that:

I will not go back to the days when health insurance companies had unchecked power to cancel your policy, deny your coverage, or charge women differently than men.

Emprically, women use more health care, they cost more, estimates are around 35%. Some of this is childbearing, but a lot of it comes from the simple fact they go to the doctor more often (notice women see their gynecologists rather regularly, whereas men have no comparable service). So now charging women more for something they use more of is illegal because it discriminates.

Interestingly, in the 1970's there was a law passed so that upon retirement, the annual payments to female retirees had to be the same as for male retirees even though women live longer, statistically. That is, the present value of their retirement packages, by law, are larger for women than men.

Government seems to be doing more and more to make it difficult to prevent 'pooling equilibria', cases where different types of applicants get into a pool, eventually pushing out the 'better' or 'lower cost' people who don't want to subsidize the other group. For example, due to legal rulings, it is now very difficult to give job applicants explicit aptitude tests, even though this would be very useful, and avoid the charade from those Microsoft/Google IQ tests given verbally. Interestingly, Nobel Laureate and prominent Big Government advocate Joe Stiglitz's most famous paper relates to an inefficiency from a pooling equilibrium, and his take-away was that markets were inefficient because of this problem. In practice, government encourages pooling equilibrium where it was never a problem before by preventing rational discrimination based on projected costs/benefits based on observable characteristics.

While the equilibrium efficiency loss in Stiglitz-Weiss is abstract, it usually creates something pretty simple, as if you can imagine what would happen to insurance if it could not price based on risk and allowed people to opt out: healthy people would leave in droves, which is why Obama-care made insurance mandatory. Think about the lawsuits on disparate impact for mortgage lending in the 1990s, where whites were rejected less often than blacks, and this was presumed discriminatory (in an evil way), and so the only way to make unequal groups equal is to stop making distinctions that differentiate them, which led to simply the idea that down payments and having a job were unnecessary underwriting criteria.

It's rather funny that Stiglitz's main theoretical contribution to the academic literature is so starkly in contrast to not just his politics but his obsession, which is increasing the size and scope of government which prioritizes preventing firms from rationally discriminating. Remember that in Stiglitz's model, like everything else in this literature (he didn't invent it), failure to discriminate types somewhat known by participants is what causes all the problems, the 'bad equilibria.' I guess that highlights no one takes these models very seriously--change one assumption here or there, different result.