Read Part 1 here.

It is important to note that the effect of the carbon tax is not double counted. The carbon tax raises tax rates generally, and so market actors react to the $100 billion income tax with the knowledge that taxes are actually higher than $100 billion. This insight is obvious when thinking about taxes that are “closer to home,” but is often forgotten when talking about complicated public policy. For example, if a state has both an income tax and a sales tax, each of the two taxes will exacerbate the deadweight loss caused by the other. When estimating the total deadweight loss of the two taxes, you cannot simply subtract their accounting cost from total output, plus whatever deadweight loss they would cause in isolation; you must assess the degree to which the existence of one tax exacerbates the costs of the other, and vice versa. The real deadweight loss generated by the two taxes is the sum of the two new, higher figures. The real world is not static, and economic analysis of tax issues, or any issue for that matter, should not work out proposals as if two policies can be assessed in isolation and then added together to produce a sum that will be of any real value.

What is left out of this analysis is that the $10 per ton carbon tax is no longer the optimal tax commensurate with the social cost of carbon. On the margin, the carbon tax is causing more damage in the presence of the income tax than it would under an isolated analysis. Therefore, under a dynamic analysis, if we reduce the carbon tax down to $5 per ton, we may come closer to maximizing total social gains. The precise numbers of course would depend on the circumstances in the economy at the time. However, this result makes intuitive sense, because we now have to worry about the carbon tax exacerbating the distortions of the income tax alone and the static social cost of carbon is no longer as useful a tool to measure to total social gains achievable.

One additional point warrants attention. The receipts of the new carbon tax could be used to reduce the income tax, making the carbon tax “revenue neutral.” In the absence of the carbon tax, reducing rates reduces deadweight loss. However, given a carbon tax, there are two effects—in addition to the benefits of reduced emissions—acting at cross-purposes. The tax interaction effect makes the marginal distortion of the income tax worse, incentivizing us to set a carbon tax lower than the social cost of carbon. At the same time, the possibility of using the proceeds of the carbon tax to lower the income tax (“revenue recycling” effect) incentivizes us to levy a Pigouvian tax that is greater than the social cost of carbon. Which effect dominates is an empirical question, but my intuition points to the magnitude of the tax interaction effect outweighing that of revenue recycling. The deadweight loss generated by the interaction of a carbon tax and an income tax is almost certainly greater than whatever reduction in the income tax may be afforded by increasing the carbon tax, especially given political incentives to make the reduction in other rates as small as possible, and the narrow tax base and corresponding decreasing marginal returns to revenue that would be generated per unit of deadweight loss inflicted as the Pigouvian tax grew. Additionally, the carbon tax base is smaller than that of the generic income tax. Therefore, theoretically, raising $1 billion from the carbon tax should be, ceteris paribus, more distortionary than raising $1 billion from the income tax.

Hence, we have the (at first counter intuitive) result that the presence of general distortionary taxes may actually weaken the case for imposing a revenue neutral carbon tax, even if the proceeds of the tax are distributed back to taxpayers in lump-sum payments. That the case for the carbon tax is so weakened does not automatically show that it is vitiated altogether; however, that result is heavily implied, given that the “optimal” rate of a carbon tax — in our economy with both an income tax and a payroll tax to interact with — will necessarily be less than the social cost of carbon. Furthermore, depending on the magnitude of the tax interaction effect, and the degree to which it outweighs the revenue recycling effect, the optimal tax may be much less than the social cost of carbon.

This last bit of intuition is important because if the optimal carbon taxed is reduced so far as to reduce much of the positive social benefit stemming from reduction of emissions, then the tax is simply another mode of revenue collection, and there are far more efficient means of accomplishing that objective.

Overall, a carbon tax may generate extremely short-term environmental benefits, but in terms of a social benefits calculus, in the longer-run their interaction with other taxes makes assessing such a Pigouvian tax both counterproductive and distortionary. The renowned economist and Nobel Prize winner Friedrich Hayek once wrote that he preferred “true but imperfect knowledge, even if it leaves much undetermined and unpredictable, to a pretense of exact knowledge that is likely to be false.”  Pigouvian taxes provide us with exactly that: a pretense of exact knowledge of the so-called “social costs” created by various market exchanges. However, it is a fatal conceit — to borrow another term from Hayek — to assume the accuracy of such cost projections when the models employed for discovering them necessarily leave out important interactive effects. Let us instead rely on the true but imperfect knowledge that is generated by the millions of individual exchanges that make up our market system for guidance as to the best way forward.