I agree with Diamond and Saez that the top marginal income tax rate should be slightly over 70%, but I don’t agree with what I take their reasoning to be. My argument after the jump.
The assumptions aren’t even crazy. Assume there are three types of people, the super-rich, the rich and the middle-class. Labor of the super-rich and of the rich are perfect substitutes with the super-rich 10 times as good at everything. Labor of the middle-class and labor of the richandsuperrich enter in production cobb douglass. Reduction of labor supply by the super-rich reduces GDP by their wage. It also causes the wage of the rich to go up and the wage of the middle class to go down. This second effect reduces welfare. At the top of the supper-rich Laffer curve, a tiny reduction in the tax on the super-rich causes the same income of the rest of society and a more equal distribution of that income. So it causes increased welfare. This is true for any fixed tax rates on the rich and the middle class and for tax rates which are adujsted optimally (the second by the envelope theorem).
I haven’t read the paper, but I think that Diamond and Saez must be doing something odd. They consider utils when discussing the income of the rich and equivalent variation when considering the effect of the labor supply of the rich on the rest of us.
There is an argument which seems to suggest that the effect of the labor supply of Mr Jones (“Smith” is already in use in this context) on the rest of us is zero. Standard (false) positive assumptions imply that Mr Jones is paid his marginal product, so the effect of his labor supply on the total income of the rest of us is zero. This means that the effect on money metric not-welfare-at-all is zero. So if we don’t care about the distribution of income and wealth among the rest of us, we don’t care about Mr Jones’ labor supply. So if money can’t help Mr Jones more, we tax him to the peak of his personal Jones/Laffer curve.
But wait the whole point is that we should care a whole lot about the distribution of income.