UPDATE: I made an important mistake. Robert Murphy corrected me in an e-mail. Correction below.
Robert P. Murphy, one of the economists who writes frequently for Econlib, has published a number of pieces on the "tax interaction effect." He has recently published another that is well worth reading. Before getting to his article, here's some background.
Many economists, when they hear about a tax on carbon to address the issue of global warming, tend to have a knee-jerk reaction in favor. This reaction is based on two things: (1) their belief that global warming is a problem, and (2) their economics training, which tells them that a Pigovian tax on a negative externality is a better solution than regulation because, if the tax is set correctly to reflect the costs of that externality, various carbon users will adjust optimally and use the efficient amount of carbon.
In a world starting with zero taxes on anything, a tax on carbon would be optimal, assuming, I remind you, that global warming is a problem.
But the academic literature on carbon taxes has moved well beyond that simple world. In the world in which we live, where there are many taxes, the case for carbon taxes is no longer so clear. And the reason? It's that there is an interaction between carbon taxes and other taxes. The leading academic economists in this literature are A. Lans Bovenberg and Lawrence H. Goulder. As I've written earlier, Bovenberg is an economist in the Netherlands who was awarded the Spinoza Prize in 2003. Goulder is Shuzo Nishihara Professor of Environmental and Resource Economics in the Economics Department of Stanford University.
About a year ago, I tried, unsuccessfully, to get good economists who favor a carbon tax to address this issue. As far as I can tell, they have not. This next statement is strong, but, as far as I can tell--and I could be wrong--the majority of economists who have embraced the carbon tax do so as a matter of faith and have simply not considered--and may not even be aware of--the Bovenberg/Goulder/Murphy critique.
Although the initial theoretical analyses tended to reject the double dividend, a second wave of models offered more scope for the double dividend by acknowledging additional potential channels for beneficial efficiency impacts from green taxes. One such channel is an improvement in the relative taxation of capital and labor. If, prior to introducing the environmental tax, capital is highly overtaxed (in efficiency terms) relative to labor, and if the revenue-neutral green tax reform shifts the burden of the overall tax system from capital to labor (a phenomenon that can be enhanced by using the green tax revenues exclusively to reduce capital income taxes), then the reform can improve (in efficiency terms) the relative taxation of these factors. If this beneficial impact is strong enough, it can overcome the inherent efficiency handicap that (narrow) environmental taxes have relative to income taxes as a source of revenue. Similarly, if the initial tax system is highly distorted in terms of consumer goods, and the green tax reform improves the system in that dimension, then the double dividend can occur after all.
The presence or absence of the double dividend thus depends on the nature of the prior tax system and on how environmental tax revenues are recycled. Empirical conditions are important. This does not mean that the double dividend is as likely to occur as not, however. The narrow base of green taxes constitutes an inherent efficiency handicap...Although results vary, the bulk of existing research tends to indicate that even when revenues are recycled in ways conducive to a double dividend, the beneficial efficiency impact is not large enough to overcome the inherent handicap, and the double dividend does not arise.
Did you notice that? The best shot we have at having the carbon tax not move us further from the optimum is to have it replace taxes on capital. Do you see that happening in the current political environment? Moreover, as Goulder points out above, even if the carbon tax did replace taxes on capital, it would still not clearly improve efficiency.
UPDATE: Robert Murphy, who has clearly mastered the literature more than I have, writes me to correct a mistake I made above. Here's his correction, as edited by me:
Strictly speaking, Goulder, in that quotation, is simply discussing whether there is a double dividend. So, if there's not, there still could be a case for a carbon tax; it just would optimally be set at lower than the social cost of carbon (i.e. the textbook Pigovian case).
The way your post reads right now, it looks as if you're saying, "If a revenue neutral carbon tax were implemented and didn't phase out capital taxes, it would move us away from efficiency." [DRH note to Bob. You can delete the "might." I was saying exactly that, due to my too-quick reading early in Pacific Standard Time. So you're right to correct me.]
But even though it would make the tax code more distortionary vis-a-vis conventional economic growth, it would also reduce climate change damages (if the models in which global warming is a problem are right, etc.), and so, on balance, might be better than the status quo.
Here's a specific hypothetical numerical example. Suppose the Pigovian externality from a ton of CO2 is $30. The government levies a carbon tax of $30/ton. Because of the tax interaction effect, this is not the optimal policy; the optimal rate should really be only $20/ton.
But implementing a $30/ton tax is better than a $0/ton tax, because the incremental deadweight loss from the tax code is lower than the marginal reduction in climate change damages.