With no revenue requirement, or where government can use lump-sum taxes, Arthur C. Pigou (1947) shows that the first-best tax on pollution is equal to the marginal environmental damage. Consumers then pay the social marginal cost of each item, the direct cost of resources, plus the indirect cost of pollution. Suppose government needs more revenue, however, and cannot use lump-sum taxes. In this second-best world, our intuition might tell us to raise all tax rates: the tax on any "clean" commodity should be raised above its first-- best level of zero, and the tax on a "dirty" good should be raised above its first-best Pigovian level (the marginal environmental damage) . Despite this intuition, a recent paper by A. Lans Bovenberg and Ruud A. de Mooij ( 1994 p. 1085 ) claims to "... demonstrate that, in the presence of preexisting distortionary taxes, the optimal pollution tax typically lies below the Pigovian tax...." This note argues that nothing is necessarily wrong with the intuition that all taxes should be raised. Nothing is wrong with the Bovenberg and de Mooij model either, but the above quote could be misinterpreted. I generalize their model to reconcile these opposing views. Earlier writers have expressed several versions of the "double-dividend hypothesis."1 These views are discussed more below, but a strong version of this hypothesis might claim that a revenue-neutral switch toward a tax on the dirty good and away from taxation of clean goods can improve environmental quality and reduce the overall cost of tax distortions. By implication, this view might suggest that any additional revenue requirements should be met by raising the tax on the dirty good by more than taxes on clean goods. The important and correct result of Bovenberg and de Mooij is that this strong view is flawed.2 Even if the pollution tax helps solve an environmental problem, it likely worsens other tax distortions. Thus, the tax on the dirty good should rise by less than the tax on the clean good. Bovenberg and de Mooij focus on the differential between the tax rates on the clean and dirty goods, but they never quite say so. They assume the tax on the clean good is always zero, so their dirt tax is the differential. With this choice of normalization, starting with the dirt tax at the Pigovian rate, additional revenue would be raised by the labor tax while the dirt tax (differential) would fall. However, other normalizations are equally valid and sometimes preferable. In their model, the extra labor tax is equivalent to a uniform tax on both goods. Thus, from the same starting point with the dirt tax at the Pigovian level, an equivalent policy would raise both the commodity tax rates. The total tax on the dirty good would then exceed the Pigovian level. Bovenberg and de Mooij clearly understand this point, but their readers might not. Therefore, the first purpose of this note is just to clarify the interpretation of their results. The second purpose is to explore the role of "normalization" in a model with tax rates on both goods and on labor. Any one tax rate can be set to zero, as a conceptual matter, but implementation of some taxes might be easier than others as a practical matter.

Additional Metadata
Keywords Economic models, Economic theory, Economics, Environmental, Environmental regulations, Pollution, Protection, Studies, Tax rates, Taxation, Taxation economics
Persistent URL hdl.handle.net/1765/11818
de Mooij, R.A., & Bovenberg, A.L.. (1997). Environmental levies and distortionary taxation: Comment. The American Economic Review, 252–253. Retrieved from http://hdl.handle.net/1765/11818