I am pleased that Jared has dropped the ambiguity and declared that he is indeed using "merit" to mean marginal product in Principle 1, as I have assumed since my original post. I still believe that Principle 1 does not accurately describe Jared's real concerns, which have little or nothing to do with marginal product. I'm certain that he would remain concerned about the increase in inequality, even if it were shown to result from widening inequality of marginal product rather than the exercise of power that has driven income away from marginal product.

Indeed, much (not necessarily all) of the rise in inequality almost certainly is due to widening inequality of marginal product. Jared acknowledges that globalization has caused part of the rise in inequality. Globalization reduces the marginal product of low-skilled workers by lowering the value of the goods they produce through competition from low-cost production abroad. If Jared really meant what he said in Principle 1, he would be unconcerned about that phenomenon because the low-skilled workers have less marginal product and therefore less "merit." In actuality, though, he still is (and should be) concerned because none of us really consider marginal product to be merit.

On a related note, I advise caution in using the gap between cash wages and labor productivity. The appropriate comparison is between total labor compensation (including fringe benefits) and labor productivity. That comparison would show less, but still some, divergence in growth rates.

On the elasticity of taxable income, the 3 to 4 percent income drop obviously applies only to the affected taxpayers. It probably consists of some combination of labor supply and shifting of income categories.

The factual finding by Gruber and Saez, while uncertain like any other econometric estimate, is not dependent upon their views of how progressive the tax system should be, particularly since other authors have made similar findings. In any case, I reject the claim that Gruber and Saez's paper (Journal of Public Economics, April 2002, pp. 1-32) "argues that the tax code should be much more progressive than it is today" and deny any "conspicuous omission" from my post. From Gruber and Saez' abstract, "Our estimates suggest that optimal tax structures may feature tightly targeted transfers to lower income taxpayers and a flat or even declining marginal rate structure for middle and high income taxpayers." From page 3, "the optimal system for most redistributive preferences consists of a large demogrant that is rapidly taxed away for
low income taxpayers, with lower marginal rates at higher income levels." A similar statement appears on page 27.

To be sure, their results allow for high marginal tax rates, depending upon redistributive preferences. At the extreme, they find that the marginal tax rate at the very top should be 49 percent for redistributive preferences that put zero weight on each dollar of wellbeing for those at the very top. (Between federal and state income taxes, Medicare tax, and excise taxes, we are already close to that value.) Of course, the optimal tax rate at the top is lower for less redistributive preferences. Gruber and Saez discuss this point in detail in the 2000 working-paper version of the article (NBER Working Paper 7512). On page 31 and Table 10 of the working paper, they report that the optimal tax rate at the very top is 18 percent under one set of not-so-egalitarian preferences and 5 percent under another set of preferences. Clearly, the optimal tax rate depends on value judgments.

Moreover, no definitive assessment of marginal tax rates can be made without incorporating saving and investment, which are absent from the optimal-tax theory that Gruber and Saez use and which are not captured by their empirical estimates of the short-run responsiveness of taxable income. Including saving and investment is likely to dramatically lower the optimal rate at the top, so long as we continue to use an income tax system. Redistribution would be less economically costly, of course, if we were using a progressive consumption-tax system, such as the Bradford X-tax, rather than the income tax.