Tuesday, April 29, 2003

“Isn’t there an existence theorem proving that there is a second best argument for any policy ?”
Robert Barro 1988.

Now there is. It is a simple generalization of the Pareto liberal paradox also known as Sen’s paradox. This note is the fruit of the meeting of the minds of Barro and Sen.

Sen’s paradox is well known, even if the example is painfully dated. Once upon a time Lady Chatterley’s Lover was considered obscene. Amazingly recently, the courts came to their senses. Sen considered what Pareto might say thinking of two agents lewd and prude. Lewd wants to read porn but finds even more exciting the thought of Prude being forced to read porn. Prude finds porn appalling but is even more disturbed by the thought of Lewd reading and enjoying it. Freedom is not Pareto optimal in this case. It is Pareto better to block Lewd from reading porn and force Prude to read it. This is crazy. In any case it shows that liberalism and the Pareto principal might hypothetically be in conflict.

Here an important point is that ethical principals the Pareto principal and liberalism are confounded by assumptions about the way things work which reconcile them. Ethics should apply to all conceivable worlds so such a reconciliation is hiding rather than resolving a conflict. I suppose one might argue that we have enough principled disagreement about right and wrong in this universe to have any need to dream of more problematic ones.

Another important point made once or twice by Sen is that alleged implications of rational utility maximization are, in fact implications of assumptions about the form of the utility function. If one is allowed to consider agents like Lewd and Prude who care about each other’s consumption, even the obvious point that people should be free to do anything which has no physical effect on anyone else, can conflict with the desire to Pareto improve the world.

Now consider some aweful policy X. Just what X is does not matter (this aims to be a general proof). What if everyone really really loves the idea of X being imposed for its own sake. That they want desperately to know that it is written in the law and being imposed on themselves and other people. This is a second best argument of a sort because “X is the law” is a public good.

X might make them poorer, less free, dead … whatever. However if we consider all utility functions we can consider one in which the joy of knowing that X is the law and is being imposed outweighs all costs. Now one might imagine that the lawmaker could just trick people into thinking X was the law. This possibility can be eliminated by another assumption about utility functions. Assume that lying about whether X is the law is so horribly painful that no one will do it for any reason.


QED

2 comments: