Alex Tabarrok at Marginal Revolution claims that Adverse selection is NOT the problem. This is not surprising given his enthusiasm for Markets Everywhere. Most of the comments address the empirical aspects of one particular case, which is the insurance industry. But while I can’t get at the papers he references (they are behind the institutional subscriber firewall) I think there is a more general issue here about asymmetric information.
For me, the key line in the post is this:
You can buy a decent used car, for example just get it inspected or certified. Only if such adjustments are illegal, or in some other way not allowed, will adverse selection become important.
But the whole thing about asymmetric information is surely that information is not non-existent, but it is costly. No one thinks that the lemons scenario is more than a first-order approximation, and signalling and screening make second-order corrections, but the problem may still remain and be important. The information problem becomes central to the whole practice of the industry, as anyone reading an insurance contract will know. Alex Tabarrok is very close to saying that costs can be neglected as long as adjustments are "illegal, or in some other way not allowed". This claim that transaction costs are unimportant seems to be one of the dividing lines between market enthusiasts (AT) and market sceptics (me).
What is more, there is often a collective action problem in obtaining that information that is particularly important when the buyers are those who are ill-informed. When the seller is ill-informed, it may be the case that they can compensate: I think that Alex Tabarrok is right that, for example, the insurance company knows as much about my life expectancy as I do, and I know considerably less about the details of my coverage than they do. I would say that in this case overcoming asymmetric information becomes a fixed
cost, leading to economies of scale in the insurance industry (which at
least in Canada is an oligopoly with fewer members by the year) and leading to market failures of a variety of types. On the other hand, when the consumer is ill-informed, the collective action problem is real.
My favourite example of an asymmetric information problem comes from beer-drinking in the UK. Small independent breweries have no mechanism for establishing a reputation (and thereby overcoming the lemons problem) unless there is a critical mass of well-informed beer drinkers who will transmit that reputation to others. In the 1970’s the beer industry in the UK was in a
Catch-22 that is typical of asymmetric information problems: there was no demand for real ale because there was no supply, and there was no supply because there was no demand. The big breweries offered low quality, but predictability. AT would have claimed, I believe, that the market was working.
The problem has been resolved to some extent by the efforts of the Campaign for Real Ale (CAMRA), who included a prominent Trotskyist in their leadership and had a strong anti-big-business slant. Their efforts have led to 300 new breweries being set up, while in the 50 years before their formation not one was founded. CAMRA has moved the equilibrium in the market, which is presumably evidence that it was failing before. And unless AT is prepared to
accept anti-business volunteer lobbying and protest groups in the requirements for a properly functioning market, it would seem that the problem was fixed by a successful organization overcoming the collective action problem.
So yes, I think asymmetric information, including adverse selection, is really a problem in many markets.
Update: on asymmetrical information in insurance, see December 18th’s Dilbert.