I found the Caplan/Boettke debate on Austrian economics very interesting – if only because this is possibly the only issue where I would side with the Austrians against Caplan. The debate is over probability theory, and while Caplan forcefully mocks the vacuous formulations of “radical ignorance”, Boettke discusses some interesting tangents but never makes the most powerful argument for his position.
Critics of probability theory make at least one incontrovertible argument: probability statements, understood as statements about observable frequencies of some event occurring over a large set of cases, are not useful (or even coherent) when applied to unique, unrepeatable historical events. It is meaningless to say, for instance, “there was a 60% chance that Nazi Germany would lose WWII”, because we cannot reset the initial conditions and test the scenario a few hundred times until we arrive at data with statistical regularities – and we certainly can’t do that for complex events that haven’t happened yet. Caplan (following the lead of many others) tries to resuscitate probability statements in those cases by interpreting them as statements about subjective degrees of belief. I would agree with Caplan that subjective theories of probability (in particular, Bayesianism) are both coherent and very useful (I would also agree with Caplan against the Austrians that mathematical analysis of probability is useful). The key problem with Caplan’s argument is that subjectivist theories and frequency theories are useful in different ways. Bayesianism is very useful for establishing standards for how a rational agent should respond to new evidence. But frequency theories are useful because they allow us to reduce or eliminate risk by distributing costs over a large group – i.e., insurance. Insurance companies are able to reduce the risk of individual policy-holders without merely transferring that risk to themselves, because they are able to benefit from the law of large numbers: they collect premiums from all of the policy-holders, while only a small fraction of the policy-holders will need to be compensated. Everybody benefits. Unfortunately, much (if not most) of human life is not reducible to observable frequencies, so not all risks can be insured. Different economists come to different conclusions from this argument: Austrians conclude that entrepreneurship is an indispensable component of capitalism, while Post-Keynesians conclude that financial markets are inherently unstable. Whatever conclusions are to be drawn from it, the argument itself is significant and undervalued by mainstream economics.