I think I have now answered Steve Landsburg’s puzzle. The difference between his example (or mine) of an action that imposes only subjective costs and his example of an activity such as reading pornography, or Bork’s of using contraception, that imposes only subjective costs, is not the nature of the harm. The difference is that in the one case the cost is of a sort that can be measured, the action controlled, via a property rule. In the other, it is not.More precisely, the property rule under which I have a right to read porn and you can only stop me by offering to pay me not to do so produces its result by ignoring the cost my porn reading imposes on you, since, as with the case of risks imposed by careless driving, including that cost requires an unworkable contract between all of the prudes and all of the would-be consumers of porn. The property rule under which you have a right to forbid me, or anyone else, from reading porn, produces its result by ignoring the cost your ban imposes on me, for the same reason.Neither property rule gets the cost/benefit calculation correct, but the former rule is a great deal less expensive to enforce than the latter, which is an argument for it.What about a liability rule? That is the point at which the subjective nature of the harm comes in. It is true that, from the standpoint of economics, all harm is ultimately subjective—having my arm broken or my car dented would not be a cost under sufficiently bizarre assumptions about my preferences. But some subjective costs are a lot easier to measure externally than others. When I claim damages for my wrecked car, there are market prices out there for repairing or replacing it that provide a court with a reasonable basis for estimating the cost. When I announce that your reading of porn, or oil drilling in a wilderness I never plan to visit, inflicts large psychic harm on me, there is no such basis for checking my claim.
A good answer to Landsburg’s excellent question
April 7, 2013 by Eric Crampton
In other words, the property rule is more efficient where it is harder to measure the harms of transgression and where it is easier to arrange payment and trade if the default allocation is inefficient.