There are many facets to the debate over manmade climate change, and what policies (if any) the U.S. federal and state governments should take in response to it. In recent years, a few writers have made the interesting argument that even if conservatives and libertarians don’t buy into the catastrophic warnings, they should still support a carbon tax so long as its revenues are used to reduce other taxes. What is fascinating about this argument is that the peer-reviewed literature rejects it, and yet these writers keep making the pitch, apparently unaware that their position is untenable. Let me give a very recent example of what I mean, and then review the state of the literature.

Scott Sumner Makes an Intuitive, and Totally Wrong, Appeal

On the classical liberal blog EconLog, Chicago-trained free-market economist Scott Sumner recently wrote:

I seem to be one of the relatively few right-of-center intellectuals that worry about global warming. In previous posts I’ve argued that if the GOP were smart (no jokes please) they would propose the following policy:

1. Global warming is a crisis for our planet, and it’s time to stop playing politics with the issue. Therefore we suggest that Congress pass the sort of policy that experts believe is the most effective solution…

2. It’s clear that experts view a carbon tax as the most efficient solution.

3. This tax should be completely revenue neutral…

4. Therefore the carbon tax should be offset by reductions in our most distortionary taxes, especially those that bias us toward consumption…

I’ve suggested that this is a win-win for the GOP. First, it’s possible (indeed likely) that the concerns over global warming are valid. In that case a revenue neutral carbon tax is clearly beneficial. And second, even if scientists are wrong about global warming, our current tax system is so grotesquely inefficient that it would be easy to find taxes far more distortionary than the carbon tax, which could then be reduced to offset its impact. Thus it’s probably a sound public policy, even if global warming is not a problem at all. [Bold added.]

Sumner’s arguments have a superficial plausibility. After all, with so many natural scientists worried about it, surely climate change might be a decent problem, right? And economists know that the current U.S. tax code is awful, because it destroys a lot more than $1 in potential private sector output for every $1 in revenue it raises for the government.

In this context, then, one would think that levying a carbon tax and using the revenue solely to reduce the rates on other productive activities (such as working and investment) would have to be a good idea. As the popular slogan says, “Tax bads, not goods.” And indeed, Sumner does think this.

Yet as I’ve argued over the years, such reasoning is wrong. Generally speaking, the pre-existence of a distortionary tax code is a reason for levying a smaller carbon tax than one would initially want to do, in light of the underlying assumed environmental harm. Because of this, if one were agnostic about the threat of manmade climate change from carbon dioxide emissions, then it would be clearly harmful to the economy to levy a carbon tax, even if it were revenue neutral.

The “Tax Interaction Effect”

The specific concept that throws a monkey wrench into the apparently commonsense reasoning of writers like Sumner is called the “tax interaction effect.” I’ve written about this extensively in other posts (such as here and here). But here’s the basic intuition, which the leaders in the field use to explain the effect:

Taxes on labor and capital income are harmful because they drive a “wedge” between how much work and investment produce, versus how much take-home earnings the workers and capitalists get to keep. In this context, a carbon tax amplifies this wedge, because it makes energy more expensive and drives up prices. Thus the after-tax, real earnings of workers and capitalists are reduced even further with the imposition of a carbon tax. Even if the revenues of a new carbon tax are fully recycled and used to reduce, dollar for dollar, pre-existing taxes on labor and capital, it is still possible (and indeed empirically likely) that the economy ends up with more total distortion than before.

It may help to understand the big picture by the following consideration: Levying a tax on carbon-intensive activities violates a fundamental principle of efficient taxation, because it strikes at a narrow base. Yes, taxing all labor income is distortionary, but taxing all businesses that start with the letter “J” would be even worse. This is why levying a tax of trillions of dollars on carbon dioxide emissions would be very harmful to standard measures of economic growth.

Estimating the Size of the Tax Interaction Effect

The tax interaction effect is not a mere curiosity; estimates of its size can be very large. For example, in a pioneering 1994 article, Bovenberg and Goulder ran a simulation of the U.S. economy and tax code, and presented these results:

Source: Table 1 from my EconLib article, available here.

In the table above, we see just how powerful the tax interaction effect can be. For example, if we assume the “social cost of carbon” is $50 per ton, then the “optimal” textbook carbon tax is $50 per ton—assuming a blank slate with the rest of the tax code.

But look what happens if we now assume that prior existence of the U.S. tax code, circa 1994. If the proceeds of the carbon tax are returned lump-sum to citizens (meaning it is revenue neutral, and doesn’t fuel new government spending), then the optimal carbon tax falls to $0 per ton. Even stipulating a large environmental “negative externality” of $50 per ton, if the proceeds are simply returned in lump-sum fashion back to households, then even a $1 per ton carbon tax would cause more total economic damage than it would spare in climate change. Thus the government should “do nothing” if its only option were to enact a carbon tax and return all the revenue back to households in lump-sum fashion.

However, we can do better than that. Specifically, we can take all of the carbon tax receipts and use them to reduce the rates charged on the personal income tax. This alters incentives, and gives individuals a reason to work more and thus generate more income.

However, notice that even in this scenario, Bovenberg and Goulder estimate the “optimal” carbon tax at merely $27 per ton. The prior existence of the distortionary income tax code has led to an “optimal” carbon tax that is close to half of the value suggested by the raw negative externality. This is the opposite of the intuition from writers like Sumner.

A Modern Estimate

These qualitative results have held up over time. Consider for example a chart from a 2013 Resources for the Future (RFF) study (source and analysis here)—and keep in mind that RFF as an organization is favorable to a carbon tax.

Source: My earlier IER article.

In the above chart taken from the 2013 RFF study, we see that if a new carbon tax is used to fully fund any of (a) lump-sum rebates to households (purple line), (b) a reduction in sales taxes (green line), or (c) a reduction in payroll taxes (red line), then GDP will be stifled. So we see that a revenue-neutral carbon tax is not sufficient to spare the economy.

Now in fairness, Sumner could point to the one thing that would salvage his point: The RFF study estimated that if all of a new carbon tax’s receipts—which would be more than one trillion dollars over the first decade, for standard ranges of the tax—were fully devoted to reducing taxes on capital (the blue line), then we could see a boost to conventional GDP growth, as well as any environmental benefits.

Yet even here, the effect isn’t that significant. In the chart, we can see that the blue line just breaks 1% on the positive side, while the green line goes lower into negative territory. This means that the economic damage from a revenue-neutral carbon tax with all receipts being used to reduce sales taxes (the green line) causes more harm, than a revenue-neutral carbon tax with all receipts devoted to reducing capital taxes would yield in economic growth.

Furthermore, as the chart shows, the purple line (where receipts are returned to households in a lump-sum fashion) causes triple the damage that the blue line yields in benefits. And remember—these lines are all plotting the effects of a fully revenue-neutral carbon tax. If we realistically allow for the fact that government spending on “green” projects will go up, then the outcome would be much worse.

I don’t think Sumner (let alone the average EconLog reader) realized just how tenuous his case was. Most people assume that if you are already vaguely worried about climate change, then a revenue-neutral carbon tax tied to, say, a combination of payroll and corporate income tax cuts, plus rebates for poor households to help them deal with higher energy prices, would be a no brainer. And yet the standard models in this literature show just the opposite. Writers urging libertarians and conservatives to consider a carbon tax should review the literature before picking up their keyboards.


In this post, I have set aside all of the practical and political problems with a U.S. carbon tax. In reality, we are certainly not going to get a “revenue neutral” carbon tax—just look at what the progressives themselves are saying about it.

However, for the sake of argument, we can stipulate a revenue-neutral carbon tax, and even allow that its receipts would be used to reduce distortionary personal income tax rates. Even in this contrived setting, standard models in the literature show that one would set the “optimal” carbon tax well below the level corresponding to the assumed environmental problem.

Therefore, if conservatives and libertarians doubt that there really is an impending catastrophe from manmade climate change, then they should have nothing to do with a “carbon tax deal”—assuming they want to foster U.S. economic growth.

Print Friendly, PDF & Email