The Hollowing Out Fallacy

Robin Hanson has an interesting post in which he argues that people who favor laws against employer discrimination are inconsistent. The gist of the argument is that we grant whites the freedom not work for blacks (because whites do not have to apply to firms owned by blacks) but we don't grant whites the ability to not hire blacks.

I think the distinction is clear. Employers have bargaining power and employees do not. As long as the market is monopolistically competitive then employers will have at least some bargaining power. A racist with bargaining power has the ability to harm the welfare of others; a racist without bargaining power is merely entertaining his own private thoughts.

I think Hanson, like many libertarians, is prone to what Paul Krugman might call the Hollowing Out fallacy. In his book on development and geography Krugman points out that knowledge of Africa's geography did not grow linearly from the 15th through the 19th century. It took some steps backwards before it got better. 15th century maps were inaccurate about distances and features of the coastlines but they had a lot of information about Africa. As better and more accurate techniques were developed much of this information was lost. We replaced a large amount of medium quality information with a small amount of high quality information.

Krugman's point is that this happened in economics. When economics got more mathematical economists latched onto perfect competition. The math was easier. They basically ignored imperfect markets or assumed they didn't exist - just like the African map makers. This "hollowing out" of knowledge led to a free market bias. I think many libertarians are systematically guilty of this fallacy. At its worst you see it in the "perfect competition and Ayn Rand" types who think "incentives matter" is the One Ring of economics. But you also see it in ferociously intelligent libertarians like Robin Hanson. They place beautiful and parsimonious theories of reality ahead of reality itself. "If it ain't in the theory then it ain't true."

I side with Bryan Caplan over Robin Hanson (see the debate). Start with "secure cases" and work up a theory that explains them. If your theory doesn't explain them then it isn't because of cognitive dissonance over the secure cases; it is that your theory still needs refining. Here is a secure case: buying a baby from a poor woman and torturing it to death is wrong. It may be allowed by your theory because your theory only counts self-aware preferences, or the ability to participate in market exchanges. But in that case your theory needs refining, not the moral intuitions of people who think it is wrong to torture babies.

The ironic twist at the end of this post is that Hanson relies on his moral intuitions too. How did he decide that maximizing efficiency is good instead of maximizing human suffering? Because he accepted other principles like complete, transitive orderings over preferences and pareto efficiency. But how did he choose those as his first principles? Because of his intuitions. Why does Hanson give privileged position to those moral intuitions instead of others, such as the intuition against selling babies to be tortured?