The objective of Wildasin's paper is to estimate the excess burden (from the state's perspective) created by the state income taxation of high-income households, due to the resulting distortions to these households' location decisions. The supply of high-income households to each state is assumed to be infinitely elastic; in contrast, lower-income workers and capital are assumed to be completely immobile. The estimates assume a Cobb-Douglas production function for each state in which each income group is a separate factor of production.' The resulting estimates of the marginal excess burden from extra taxes on high-income workers range as high as 16 percent of the tax revenue raised; they are worse for taxes on higher-income households, where average tax rates are higher. While this excess burden may not seem particularly large relative to other estimates of the marginal excess burden created by existing taxes, the Diamond-Mirrlees [1971] results on optimal taxes still imply, under certain assumptions, that the optimal tax rate on mobile households should be zero. The argument can be summarized as follows: Given that the welfare of mobile workers will not be affected by any tax, since their utility will always equal the level available elsewhere, the burden of any tax must fall on other households in the jurisdiction. Given that other households are paying the tax anyway, it should be better to tax them directly, thereby avoiding the excess burden created by driving away some of the mobile households. This is strictly true as long as alternative direct taxes on the immobile factors exist which have the same incidence as the taxes on mobile households. If these alternative taxes do exist, then taxes on mobile households are dominated and should not be used. Wildasin assumes that only high-income households are mobile, but the same logic applies to any mobile factor. What matters here is the tax paid by each household net of the value of any public expenditures they receive. The result here is that net taxes on these households should be zero. Under a zero net tax, each household simply pays for the services they receive. Wildasin reached the narrower conclusion that states would be better off equalizing the average tax rate on all higher-income households. The implications of the model are in fact much stronger. I am very sympathetic to the basic tenor of these points and have argued along similar lines in Gordon [1992], though my paper assumed that all households were mobile. In spite of my basic agreement with the direction of the paper, however, I still should point out a variety of reasons for doubting the conclusions. To begin with, the immobile factors that end up bearing the tax can include not just other households but also land and long-lived capital,