On his Substack, Arnold Kling, resident book reviewer at Econlib, wrote:

If people pay for their own health insurance, the market is subject to selection games. The individuals with the most incentive to buy health insurance are those that will cost the insurance company the most in claims. (Although it turns out that there is a selection effect that goes in the other direction. People who are high in conscientiousness are more likely both to obtain health insurance and to take better care of themselves.) Insurance companies, by the same token, have an incentive to try to avoid writing policies for people who most need health insurance. Nobel Laureate Joseph Stiglitz was known for pointing out that this selection game might have no viable solution: the health insurance market could collapse entirely. (italics added)

I think Arnold was engaged here in what Ronald Coase called “blackboard economics.” The idea is to think through the incentives that various players face and, on that basis, make conclusions about the way the world is. It is appropriate that he cited Joe Stiglitz because Stiglitz has been one of the masters of blackboard economics.

The problem is twofold. First, under this approach, you can sometimes be tempted not to think through the incentives all the way. A clear incentive is for the insurance company not to deny insurance to a high-risk person in the individual market but to underwrite insurance. That means assessing risk and charging a premium that reflects that risk.

Second, the blackboard approach ignores the evidence. In a blog post in 2010, I discussed some interesting economics in 2010 Economic Report of the President and the way the author of the health care chapter, whom I assume was the CEA’s health economist Mark Duggan, now head of SIEPR at Stanford, twisted himself into a pretzel to justify Obamacare. The whole post is worth reading. Here’s the particular part I want to emphasize here:

A House investigation committee found that three large insurers accepted nearly 20,000 policies over a five-year period, saving these companies $300 million that would otherwise have been paid out as claims (Waxman and Barton 2009). (p. 188)

20,000 policies over 5 years is 4,000 policies per year. So the average number of policies adopted by the 3 companies individually was about 1,333. Is that a large number? Given that the author says these were large companies, I don’t think so. As one commenter on my post noted at the time:

The five largest insurers nationwide each have between 10 and 40 million members, which is ~4-15M policies. The next half-dozen or dozen are all million+.

The bottom line is that it appears that very few people wanted health insurance and were willing to pay a premium that reflected their risk of going without.

By badas

Related Post