Oops, I forgot...

So I just made a comment over on Delong’s blog in which I discussed the positional aspect of the human utility function and referenced this paper. I closed with this thought:

The important liberal implication of positional utility is that we can redistribute wealth to a point of equality and still get large and significant motivation of effort by allowing tiny increases in inequality, gaining huge motivation by offering a change in social position without harming the least well off very much (and indeed not at all after making a Rawlsian accounting of incentives).

Unfortunately I did not make clear that this only applies in the case that the positional component of utility is a function of the relative ordinal wealth of the individual, not if it is a function of the magnitude of inequality; in other words, it only works if people care only about their order of wealth (as would be the case where they are concerned about which mate they are paired with), not in the case where they care more if differences are larger (as would be the case in the hypothesis suggested by Delong that rich people attempt to generate envy in others, more disparity could make more envy).