The Institute for Energy Research is a not-for-profit organization that conducts intensive research and analysis on the functions, operations, and government regulation of global energy markets.

About IER
Latest Analysis
December 4, 2013

More Destructive Than You Think: Clarifying the “Tax Interaction Effect”

December 4, 2013
Info Facebook

I recently posted the video of Ross McKitrick’s presentation at IER’s Carbon Tax Conference earlier this year. McKitrick landed a devastating blow on those who argue that a carbon tax will provide a “double dividend” or “win-win” scenario, so long as the carbon tax revenues are used to cut other taxes. McKitrick explained that the general conclusion in the peer-reviewed literature is just the opposite: Pre-existing distortions in the U.S. tax code make the case for a U.S. carbon tax weaker, not stronger, even if the carbon tax receipts are completely devoted to (say) income tax reduction.

As I spread McKitrick’s video around in various online forums, it became clear to me that even people who are very familiar with free-market economics do not fully understand this counterintuitive result. Since this is such a critical point in the carbon tax policy debate, it’s worth going over the basics in the present post.

Summarizing the Standard Results in the Literature

Before diving into the details, I want to stress that what I’m presenting here is completely standard in the economics of climate change literature. In this post, I will merely be providing a watered-down-for-the-layperson version of the more technical explanation given by Lawrence Goulder in this 2013 summary article for Energy Economics. Goulder is a professor of environmental and resource economics at Stanford University, is a University Fellow with Resources for the Future, and was one of the pioneers in the study of the interaction between tax and environmental policy.

As his resume suggests, Goulder is by no means a climate change “denier” or even an opponent of carbon taxes. Yet as he explains very methodically in his work (such as the 2013 summary article), the standard result in this literature is that carbon taxes exacerbate the distortion in the tax code, even if the revenues are used to offset other taxes. Thus, the case for a carbon tax becomes weaker once we factor in the tax code; we should apply an “optimal” carbon tax that is lower than the so-called “social cost of carbon.”

Notice that this initially counterintuitive point overturns the entire “conservative” case for a carbon tax. Specifically, there are a few economists and policy analysts who argue that conservatives should put aside their disdain for Al Gore and other environmentalists, and realize that a “carbon tax swap” deal would be a good supply-side boost to the economy. Their argument crucially rests on the premise that a dollar-for-dollar substitution of a carbon tax for current income or payroll taxes would lead to greater output, regardless of the possible environmental benefits in the form of slower climate change.

What I am stressing is that this “conservative” case for a carbon tax is totally wrong at Step 1, according to the generally accepted views of the economists publishing in this area. To show just how strong this point is, in this blog post, I’m not relying on the views of economists who outright oppose carbon taxes. Rather, I’m going to accept—for the sake of argument—that it makes sense to speak of a “negative externality” from carbon dioxide emissions, and that it would enhance efficiency to enact a “Pigovian tax” to make people take into account these social costs when making decisions in the marketplace. I want to show why the “conservative” argument for a “carbon tax swap” doesn’t work, in light of the peer-reviewed economics research on this topic.

I have elsewhere published a lengthy critique of the standard case for a carbon tax; I disagree fundamentally with every plank in the standard arguments. But that will involve too much discussion if we bring up those objections. Right now, I want to make sure readers understand exactly what the “tax interaction effect” is, and why it absolutely devastates the claims by some that a carbon tax will be “pro-growth” and constitute a “win-win” outcome for both the economy and the environment.

The Standard Pigovian Textbook Analysis: No Tax Distortions

The standard economic approach to carbon dioxide emissions treats them as a “negative externality” because they reputedly will lead to harmful climate change in the future. Thus, when people today buy electricity from coal-fired power plants, or fill their cars with conventional gasoline, they are unwittingly setting in motion damages that will strike future generations. Let us suppose for the sake of argument that a ton of CO2 emitting today will generate $20 in present-discounted value of net future damages; this is the “social cost of carbon.”

When there’s a negative externality, the typical solution as depicted in most textbooks is to impose a corrective tax, called a Pigovian tax after A.C. Pigou. Now if there were no pre-existing taxes, the “optimal” thing to do would be to impose a carbon tax of $20/ton. This would make everyone “internalize the externality.”

Notice that there would still be carbon emissions even in the presence of the optimal Pigovian tax. In fact, this is one of the reasons that some environmentalists would prefer a direct mandate of emission levels, rather than a carbon tax: They don’t like people in the private sector having the option to emit CO2, even at penalty rates.

However, even though our hypothetical $20/ton carbon tax won’t eliminate emissions, it will reduce them. This now gives us two different effects, which cut in different directions:

First, because emissions are lower, there is a reduced amount of (future) environmental harm, in the form of climate change damages. This reduction in climate change harm is the environmental benefit of the new carbon tax.

Second, because carbon-intensive activities have now become more expensive (due to the tax), people partially substitute into other activities. For example, utilities will rely more on natural gas rather than coal to produce electricity. For another example, the higher price of gasoline will cause consumers to buy more fuel-efficient cars. Since the government’s new tax is forcing businesses and consumers to change their behavior, they are made worse off. This reduction in the options available to firms and households causes the economic cost of the new carbon tax.

Now in the standard textbook analysis, the Pigovian tax was chosen “optimally” at the correct level to account exactly for the $20/ton externality. This will cause emissions to fall until the point at which the marginal economic cost of reducing by one more ton will be higher than the marginal environmental benefit. This is the optimal point to stop, and the emission reduction will have been worth it: The total gain in environmental benefits will be higher than the total loss in conventional economic output. Including the impact on the environment, the new carbon tax has made society “richer,” broadly speaking, even though conventional measures of GDP will show that it is poorer.

Adding Pre-Existing Taxes Complicates the Story

In the standard textbook analysis, there is no consideration of other taxes; we implicitly assume that the $20/ton carbon tax is imposed in a vacuum. But in reality, of course, there are other taxes—such as taxes on labor and capital—that are very economically inefficient.

Now in the standard textbook treatment, the government is usually assumed just to give back the carbon tax receipts back to the citizens in the form of lump-sum rebate checks. The point of the tax wasn’t to raise revenue, but rather to change the marginal incentives for emissions. The carbon tax—as we discussed in the previous section—was part of environmental policy, not tax policy per se.

Yet once economists realized how much revenue would be flowing from an “optimally” calibrated Pigovian tax on carbon emissions, they had an exciting thought: Rather than returning the money back to taxpayers in lump-sum rebate checks, they could instead use it to reduce the marginal rates of other taxes. This would have a much more beneficial impact.

It’s important for the reader to understand this point. As any student of “supply-side” economics knows, it makes a difference if the government gives a tax cut in the form of a flat rebate check sent to every household, versus a scenario in which the government cuts the marginal income tax rates by an amount that would result in the same loss of revenue, if behavior stayed the same. In practice, behavior won’t stay the same. If the government gives, say, what seems to be a $100 billion tax cut by lowering rates accordingly, this actually won’t cost the Treasury the full $100 billion. At the lower marginal rates, people will work longer and firms will produce more, expanding the total tax base. Thus the drop in revenue might only be, say, $90 billion. By implementing the tax cut via a marginal rate reduction, rather than just mailing people lump-sum checks, the Treasury can reduce the “deadweight loss” caused by distortionary income taxes.

Now back to the carbon tax debate: If the government is going to levy a $20/ton carbon tax for environmental purposes anyway, then it can achieve even more benefits by taking that revenue and using it to reduce marginal tax rates elsewhere in the code. Now we seemingly have two benefits: The environmental gain from lower emissions, plus the ability to make the tax code more efficient (i.e. less distortionary in its punishment of labor and saving).

To be sure, the carbon tax still harms conventional economic output, but now it seems that that loss is counterbalanced by two separate benefits. So if it made sense to impose a $20/ton carbon tax with no original tax baseline, surely now it makes even more sense to do it, right?

Putting this conclusion a bit differently: Back when we weren’t worrying about other taxes, and were just considering the negative externality in a vacuum, we concluded that the “optimal” carbon tax was $20/ton, set to equal the social cost of carbon. In that scenario, the government took the revenue and mailed it lump-sum to everybody.

So now, if we more realistically take into account pre-existing distortionary taxes, and are planning to use all of the carbon tax revenues to reduce the rates on these other taxes, shouldn’t we be even more aggressive with our carbon tax? Shouldn’t we impose a carbon tax of, say, $30/ton, because now we’re getting the benefits not just of a better environment, but also we’re reducing the drag of the tax code?

Indeed, some economists were so excited by this prospect, they speculated that the improvement to the efficiency of the tax code would be greater than the economic cost of the carbon tax. This is the so-called “double dividend.” It means that even if we completely ignored the fact that lower emissions will reduce future climate change damage, it would still make society richer by implementing a 100% revenue-neutral carbon tax swap. Even conventional measures of GDP and labor income would rise, following the introduction of such a carbon tax.

It is this mentality that motivates the small but vocal band of policy analysts and economists claiming that a carbon tax is a “conservative” and “pro-growth” policy. To them, it seems perfectly obvious that taxing a “bad” (carbon emissions) and using the revenues to reduce taxes on a “good” (working or investing) can only make society richer, even in conventional terms and not considering the environment. But as I’ll show in the final section, this intuition leaves out something crucial, which (most economists in the field believe) completely overturns their argument.

The Tax Interaction Effect

What the above analysis doesn’t consider is that imposing a new carbon tax will exacerbate the harms of a pre-existing (and inefficient) tax code. By imposing a carbon tax, the prices of goods and services in general will increase. Thus, the paycheck of a worker, or the dividend check paid to a corporate shareholder, will not “go as far” in the marketplace. Thus the new carbon tax represents an implicit tax increase on workers and investors.

The more severe the original distortion on the economy from taxes on workers and investors, the more harmful the incremental distortion from the new carbon tax will be. Now it’s true, if the carbon tax revenues are used to reduce labor and capital taxes, then that move helps the situation. But the point is, the result can still be worse than the status quo.

Generally speaking, economists who publish in this area think that this “tax interaction effect” is large enough to swamp the benefits of the tax swap. The intuition they give is that taxes on carbon have a “narrow” base compared to the much wider scope for taxes on labor and capital. Consequently, to raise $100 billion in revenue from a new tax on carbon, in order to reduce labor and capital taxes by $100 billion, would end up causing more distortion in the economy.

Once we take into account the likely magnitude of the tax interaction effect, the “optimal” carbon tax is lower than the Pigovian amount. In our example, the government should set the tax less than $20/ton—perhaps it should only be $10/ton (if receipts are returned to citizens in lump sum fashion). This result occurs because, in addition to the original direct damage to the economy from the carbon tax, we now have the indirect damage to the economy from the carbon tax’s interaction with pre-existing labor and capital taxes. So on the margin, it’s not worthwhile to reduce emissions as much as the textbook calls for, because the environmental benefit now has to compete with the direct economic damage of the carbon tax, as well as the indirect economic damage coming from the tax interaction effect.

Choosing the Relevant Baseline

At this point in the discussion, something may be troubling the reader. The above results seem completely counterintuitive. Surely it has to be true that using carbon tax revenues to reduce other income taxes is good for the economy, right?

Yet to answer such a question, we have to be clear on what baseline we’re using. Yes it is “good for the economy” to levy a new carbon tax and use the receipts to reduce income taxes, compared to a policy of levying a new carbon tax and then spending the new money, or even just returning it lump sum to citizens. For example, we have already said that the tax interaction effect might reduce the “optimal” Pigovian tax from $20/ton down to $10/ton, if the receipts are distributed in lump sum form back to citizens. If instead, the carbon tax receipts were used to reduce marginal income tax rates, then perhaps the new “optimal” rate would be $15/ton, which of course is higher than $10/ton. Yet either way, the “optimal” rate is still less than the textbook Pigovian rate, which would normally set the carbon tax equal to the “social cost of carbon” which we assumed was $20/ton in our example.

What does all this mean? To say that even with 100% offset where the carbon tax receipts are used to reduce income taxes, the “optimal” tax is lower than the social cost of carbon, implies that the pre-existing distortionary tax code makes it economically efficient to allow more carbon emissions than would be the case if there had been no pre-existing taxes. So the result is the complete opposite of the “pro-growth carbon tax” claim.

In particular, if a person doubts the warnings of Al Gore et al., and wants to disregard the potential environmental benefits (in the form of mitigated climate change), then the “optimal” carbon tax is precisely $0/ton. This isn’t the answer that just the Heritage Foundation would give, this is the answer coming from the standard, peer-reviewed literature. The tax interaction effect is widely assumed to be strong enough to offset the economic benefits of using carbon tax receipts to cut other taxes. Thus, a new carbon tax will hurt conventional economic growth, period; the only way to justify a new carbon tax is to claim that the environmental benefits will outweigh the harms on the conventional economy.

Conclusion

The tax interaction effect (according to the general treatment in the literature) is powerful enough that only environmental benefits could possibly justify a carbon tax. If self-described conservatives want to admit that mitigating climate change is more important to them than conventional economic growth, and that Americans should accept a poorer economy in exchange for a cooler planet, then that is their prerogative. But those who continue to claim that a “carbon tax swap” would promote conventional economic growth are ignoring the standard results in the peer-reviewed literature.

Finally, keep in mind that the above results apply to an idealized carbon tax where the receipts are truly devoted 100% to offsetting other taxes. In reality, any new carbon tax will be accompanied by massive increases in government spending. Thus the drag on the economy will be even worse, and the “pro-growth” claims are even further from the truth.

Both in theory and in practice, a new carbon tax will reduce conventional economic growth. People who argue otherwise are ignoring the standard results in the peer-reviewed economics literature.


View Comments
Back to top