December 18, 2015

Examining Cato’s Case Against a Carbon Tax



Serious policy analyses highlight the strongest arguments offered up in the conversation. They wrestle with those arguments directly. They are frank about limitations of knowledge and, consequently, about those issues that are difficult to resolve. And they employ strong, internally consistent arguments in the course of making their case. By those criteria, the Cato Institute’s “The Case Against a Carbon Tax,” falls short.

In that paper, the authors argue:

  • The social cost of carbon is low;
  • A U.S. carbon tax will have no significant impact on global temperatures;
  • The costs of holding atmospheric warming to no more than 2 degrees Celsius above pre-industrial levels are greater than the benefits;
  • Policymakers should ignore the possibility of catastrophic climate outcomes;
  • Using revenue derived from a carbon tax to cut other taxes would not produce a net gain for the economy; and
  • A carbon tax in the real world is no more market-friendly than the regulatory alternatives.

I will consider each of their arguments in turn and briefly review the literature that was overlooked or misconstrued in their various narratives. In so doing, I hope to provide a better sense of the real issues in play.

Social Cost of Carbon

The authors spend a great deal of their paper arguing that the social cost of carbon (SCC)—that is, the monetized damages associated with emitting a ton of greenhouse gas into the atmosphere—is less than advertised by mainstream economists. Let’s review their contentions.

(1)  The SCC estimates the global impact of emissions, not the national impact. True enough. Do the authors believe that it’s OK for U.S. emitters to harm third parties as long as they live abroad? They don’t answer that question directly, but the clear implication seems to be “yes.” This is a curious position for libertarians to take in that they do not usually qualify their support for property rights based on where those rights are held.

(2) 20 studies published since 2011 find that the climate is not as sensitive to greenhouse gas concentrations as the SCC calculations suggest. True, given their curated data set. But why are we counting from 2011? Are studies published before 2011 now irrelevant or by definition of lower quality than those published more recently?

Regardless, there have been more than 20 studies published since 2011 on this matter, and the ignored studies, for the most part, find higher climate sensitivity than found by the studies highlighted by the authors. As explained recently by my colleague Joseph Majkut, a survey of all of the recently published studies suggests that the SCC calculations are premised on an accurate read of the underlying literature.

(3) Satellite temperature data says that climate models are warming too fast in the troposphere, therefore model projections of precipitation change are systematically biased. Whether this is true or not is unclear. Given that different analyses of the same principal datasets are showing different results, one cannot say anything about this issue with great confidence. My colleague Joseph Majkut has already written a bit about this and will be writing more shortly, so I won’t get into this complex matter here.

Even so, this discussion has no direct bearing on calculations regarding the social cost of carbon.

(4) There is a lot of educated guesswork and subjective analytic assumptions that are employed in SCC calculations. This is absolutely correct, and it is primarily a manifestation of the profound uncertainties associated with the details regarding how the climate operates and what the economy will look like decades hence. But uncertainty cuts both ways. Statistical analysis by Lewandowsky et al. finds that, contrary to our intuition, “increasing uncertainty is necessarily associated with greater expected damages from warming, provided the function relating warming to damages is convex.” A related paper from the same authors employ a Monte Carlo simulation to demonstrate that greater uncertainty regarding climate outcomes translates into a greater likelihood that mitigation efforts will fail to limit global warming to targeted levels. None of these issues are referenced, much less explored, in the Cato paper.

While the Cato authors may be unfamiliar with literature, they are almost certainly aware of the argument. That’s because the case against marshaling uncertainty as a rationale for inaction is made by the very same authority they cite (MIT economist Robert Pindyck) to make their case that uncertainty exists, and in the very same paper at that:

My criticism of IAMs [Integrated Assessment Models] should not be taken to imply that, because we know so little, nothing should be done about climate change right now, and instead we should wait until we learn more. Quite the contrary. One can think of a GHG abatement policy as a form of insurance: society would be paying for a guarantee that a low-probability catastrophe will not occur (or is less likely). As I have argued elsewhere, even though we don’t have a good estimate of the SCC, it would make sense to take the Interagency Working Group’s $21 (or updated $33) number as a rough and politically acceptable starting point and impose a carbon tax (or equivalent policy) of that amount. This would help to establish that there is a social cost of carbon, and that social cost must be internalized in the prices that consumers and firms pay. (Yes, most economists already understand this, but politicians and the public are a different matter.) Later, as we learn more about the true size of the SCC, the carbon tax could be increased or decreased accordingly.

The authors also ignore an issue they themselves bring up as another source of subjectivity: discount rates. How much should we value losses incurred by future generations? The authors acknowledge that, if we place nearly as high a value on the wellbeing of future generations as we do on present generations, the social cost of carbon emissions is very large. If we place a low value on the wellbeing of future generations, we get a low SCC.

If, as the authors concede, preferences on this matter are subjective, why shouldn’t we place a high value on future wellbeing? The authors do not say.

Perhaps the authors believe that discount rates for future losses, when applied in a policy context, should follow what we observe in markets. A discount rate of 7% would make sense in that case because, as the OMB notes, “This rate approximates the marginal pre-tax return on an average investment in the private sector in recent years.” In short, if we can get better returns by investing our resources in things unrelated to greenhouse gas mitigation, we should.

Alas, it’s not quite as simple as that. First, observed discount rates are driven by individual consumers making decisions about their own future consumption. Climate change, however, is about transferring consumption across different people and different generations, which is a social matter. Second, we do not observe consumer preferences with regards to assets with maturities similar to those of climate impacts, so we can only infer what market preferences might be in this context. Third, observed market discount rates reduce into near nothingness the present discounted value of distant future events and trivializes even truly enormous distant future impacts.

All of this explains why most economists, when confronting climate change, embrace discount rates much lower than 7%. The considerations that drive those lower discount rates are not dissimilar to the considerations that drive lower discount rates in other, non-climate contexts. For instance, when Cato analysts have wrestled with the appropriate discount rate for the future costs associated with entitlement spending, they use the interest rate on the government’s longest maturity (30 year) Treasury securities, which works out to 3.68 percent.

Unfortunately, we don’t know what the authors think the discount rates ought to be in SCC calculations because they do not explore the matter. This is curious given their acknowledgement that the issue is critical in the course of producing an SCC figure.

The (Environmental) Futility of a U.S. Carbon Tax

The authors argue that, by itself, a U.S. carbon tax will not reduce global temperatures enough to justify its economic cost. That is correct. This implies that action must be global, something that no one disputes. But it is also true that U.S. action against greenhouse gas emissions is a necessary condition for a global agreement.

Even without a global agreement, the imposition of a domestic carbon tax—coupled with taxes on imports to reflect the carbon taxes that would have paid had the imported product been produced or manufactured in the United States—would, per Joseph Stiglitz (chapter 6), provide a powerful incentive for countries to impose their own carbon taxes to capture the revenue that would otherwise go to the U.S. Treasury. A variation on this theme would have the United States and like-minded industrial powers establish an agreement (called a “Climate Club” by William Nordhaus) to do the same.

Do the authors of the Cato paper disagree with this line of thinking, which is often forwarded for why a unilateral national carbon tax might well induce global carbon taxation even absent an international agreement? We don’t know because the issue is unexplored in their paper.

Cost & Benefits

The authors contend that the costs of preventing a 2 degrees Celsius warming above pre-industrial levels are greater than the benefits (at least through 2100). But given all of the uncertainties regarding the social cost of carbon emissions—uncertainties the authors themselves take great pains to point out earlier in their paper—how can they, or anyone else, possibly know this? As Robert Pindyck demonstrates, the climate models projecting future warming and associated environmental impacts are crippled by what we don’t know about a host of things, including—most importantly—the feedback loops that might produce catastrophic outcomes. The economic models designed to translate those outcomes into monetized damages are, as Pindyck notes, even more uncertain.

Accordingly, while we can make educated guesses about such things, the error bars associated with regards to both the costs and the benefits are too large to make definitive pronouncements. The most we can say is that, given all that we know, it would appear that the benefits of a robust carbon tax would exceed the costs.

Even if the authors are correct, however, why would analysts at the Cato Institute care that the costs of action exceed the benefits? If party A (greenhouse gas emitters) is harming party B (those harmed by greenhouse gas emissions), the contention that protecting the property rights of B will cost A more than it will benefit B should not encourage principled libertarians at the Cato Institute to green-light rights violations. There is an argument for utilitarian calculations such as these, but they are arguments that sit uncomfortably with libertarianism, a point well made by analysts elsewhere at the Cato Institute.

Regardless, even if the authors are right, they are making an argument against a very stiff carbon tax, which is what is required to hold warming to 2 degrees Celsius. And most carbon tax proposals do not have rates high enough to meet this test … or bear this criticism.

Catastrophic Risks

Returning to Pindyck, are we wrong to worry that SCC calculations do not appropriately consider the possibility of catastrophic climate outcomes— which could increase the SCC by more than 200%? The authors wrestle with this matter in the course of discussing the work from Harvard economist Martin Weitzman. Weitzman pointed out that, when we consider the full range of possible outcomes from climate change, the outlying possibilities on the high-cost side of the range of possible outcomes do not go quickly to zero, which implies far greater risks—and a far greater likelihood of damages—than conventional analyses would suggest.

To illustrate the implications, Weitzman notes that, when we run the climate models, a normal (“thin-tailed”) distribution of impacts from a doubling of atmospheric greenhouse gas concentrations above pre-industrial levels produces a median temperature change of 3 degrees Celsius with a 15% probability of a warming above 4.5 degrees Celsius. The same exercise incorporating a fat-tailed distribution (that is, with the range of possible outcomes on the high side that marches slowly to zero) finds that the probability of temperatures rising more than 8 degrees Celsius is nearly ten times greater. Weitzman argues that “at least in theory,” incorporating fat tailed distributions in our calculations is “capable of swamping the outcome of any BCA [Benefit-Cost Analysis] that disregards this uncertainty.”

Before we go further, it’s worth noting that we don’t need to buy Weitzman’s arguments about fat-tailed distributions of climate outcomes to be concerned with catastrophic climate change. As Pindyck argues:

We don’t need a fat-tailed probability distribution to determine that ‘‘climate insurance’’ is economically justified. All we need is a significant (and it can be small) probability of a catastrophe, combined with a large benefit from averting or reducing the impact of that catastrophic outcome … depending on parameter values, the specific damage function, and the welfare measure, the justification for ‘‘climate insurance’’ could well be based on a probability distribution for climate outcomes that is thin tailed.

The authors of the Cato paper, however, do not dispute the existence of a fat-tailed probability distribution. They argue instead that:

(1)  Policy should be calibrated to address the most likely outcome from warming, not the full distribution of possible outcomes, because the latter is too uncertain an exercise. That’s very hard to maintain. Similar uncertainties are confronted by risk managers in financial markets and market actors—rightly—demand that those risks be considered and priced to the best of our ability. To do otherwise would be to argue that the best allocation of investment dollars in any given year is to put 100 percent of one’s money into equities.

In other contexts, moreover, Cato analysts have forcefully rejected the idea that deep uncertainties should be ignored when considering the future costs of public policy. For instance, economist Jagadeesh Gokhale confronted this exact argument in the course of his work for Cato discussing the implications of heavily borrowing to pay for entitlement spending:

A common criticism of calculations of fiscal imbalances in perpetuity is that high uncertainty associated with very long-term projections renders such calculations less useful. However, that should be a call for estimating the size of that uncertainty rather than simply ignoring the calculations.

Moreover, if the authors wish to argue that we should defer to observed market behavior when it comes to discount rates (something implied in their paper but not explicitly stated), why should we not do so when it comes to observed risk aversion in markets? As Bob Litterman argued in the pages of Cato’s Regulation magazine, climate change is a nondiversifiable risk. That’s because we cannot invest in things that would “pay off” were catastrophic climate change to occur. That’s important because, as Litterman observes, “The primary evidence we have about how society prices nondiversifiable risk is the equity risk premium—the long-run excess return of equities relative to government bonds. It is puzzlingly large, reflecting an implausibly high degree of societal risk aversion.”

(2)  Worrying about low probability catastrophe scenarios takes us down a costly rabbit hole. To quote the authors in full:

Another problem with Weitzman’s approach—as Nordhaus, among other critics, has pointed out—is that it could be used to justify aggressive actions against several catastrophic risks, including asteroids, rogue artificial intelligence developments, and bio‐weapons. After all, we can’t rule out humanity’s destruction from a genetically engineered virus in the year 2100, and what’s worse we are not even sure how to construct the probability distribution on such events. Does that mean we should be willing to forfeit 5 percent of global consumption to merely reduce the likelihood of this catastrophe?  

Not necessarily. Weitzman himself brought up and addressed that question in his seminal 2009 paper. He argued that, while these other concerns might well deserve analysis, there are reasons to think that the non-climate catastrophes referenced by the authors of the Cato paper are of a different magnitude and nature than those posed by climate change and thus, not directly comparable.

And William Nordhaus—the source the authors cite for their criticism—agrees. Contrary to what the authors suggest, Nordaus (who, they neglect to mention, supports a robust carbon tax) does not reject Weitzman’s concerns about fat-tailed catastrophic outcomes and does not think that embracing Weitzman’s argument necessary leads us down a path of expensive hedges against extreme scenarios. In fact, he argues exactly the opposite in the very paper the authors cite:

Martin Weitzman’s Dismal Theorem holds that, under limited conditions concerning the structure of uncertainty and preferences, society has an indefinitely large expected loss from high-consequence, low-probability events. Under such conditions, standard economic analysis cannot be applied. The analysis in the present study concludes that Weitzman makes an important point about selection of distributions in the analysis of decision-making under uncertainty. However, the conditions necessary for the Dismal Theorem to hold are limited and do not apply to a wide range of potential uncertain scenarios.

What do the Cato authors have to say about the conditions upon which Weitzman’s “dismal theorem” may compel action and how they may or may not apply in the various scenarios they’ve brought up? Nothing.

Moreover, to label Nordhaus a “critic” of Weitzman is to misrepresent Nordhaus’ position. It conflates Nordhaus’ greater optimism regarding our ability to respond to emerging catastrophic threats and his various methodological criticisms with fundamental disagreement.

This can be most readily appreciated by reading Nordhaus’ book review of Gernot Wagner’s and Martin Weitzman’s Climate Shock: The Economic Consequences of a Hotter Planet in The New York Review of Books. The Wagner & Weitzman book is a popularization of the arguments that Weitzman has forwarded in the academic literature about catastrophic risks and other matters. In his book review, however, Nordhaus offered not one single word of criticism. In fact, he called the book “a witty, far-ranging, and literate set of observations, but—unlike many books on climate change—it is always informed by a deep understanding of the complexities of economics and particularly the difficulties of reaching international environmental agreements.” He spends a great deal of time discussing their arguments about catastrophic outcomes and referenced it (along with several other of their arguments) as “a singular contribution” to the climate debate that is “worth careful study.” In fact, Nordhaus’ The Climate Casino: Risk, Uncertainty, and Economics for Warming World goes on at great length about the need to consider catastrophic scenarios in the course of establishing climate policy.

Even so, the observation that Weitzman’s logic might suggest a greater degree of public attention to non-climate catastrophe scenarios does not by itself say anything about the merits of his argument.

Double Dividends

The Cato authors argue that using the proceeds from a carbon tax to cut other taxes will not produce a net improvement in economic welfare if we divorce that conversation from any consideration of environmental benefits. They offer two reasons. First, it’s unlikely that Congress would dedicate the vast bulk of the proceeds to the most economically attractive tax cuts, which is the only condition under which such a “double dividend” (meaning, a carbon tax will produce both an environmental dividend and an economic dividend) could theoretically arise. Second, because a carbon tax is actually a tax on labor—and because labor taxes are already extremely inefficient—the compounded inefficiencies associated with a carbon tax will be large enough to offset the gains associated with tax cuts elsewhere.

They are likely correct on both fronts. But it matters little. If a carbon tax makes sense as a precautionary measure to reduce the risk of catastrophic climate change, then we need not go any further in the analysis. A carbon tax is worth embracing on its own terms and what we do with the revenues is an entirely separate, unrelated matter.

While using tax revenues from a carbon tax to cut taxes elsewhere will not produce a free lunch (that is, carbon constraint with no price tag whatsoever), it still might manage—depending on program design—to substantially reduce the cost of the lunch, an observation that can be readily seen from the authors’ own analysis. In the words of economist Lawrence Goulder, who the authors cite in the course of making their argument, “Under these conditions, the carbon tax is not a free lunch. However, even if it’s not a free lunch, it still might be a meal worth paying for!”

Goulder, incidentally, thinks that a carbon tax is a lunch worth paying for, although one would not know that from reading the Cato paper. One can see Goulder make exactly the case I made above in this simple slide show, this lecture before the Colorado School of Mines, or in this paper for the National Bureau of Economic Research. In short, Goulder himself is not persuaded that his arguments about tax interaction effects are a good reason to reject carbon taxation or offsetting tax cuts.

Policy Failure

Finally, the authors argue that a carbon tax is not a market friendly response to climate change. They offer two reasons.

First, they argue that a genuine market friendly response is to introduce property rights to the atmosphere. While they don’t discuss what this reform might look like, property rights imply some sort of governmental agent to protect the same. Without regulation (which the authors oppose), this leaves it to the judiciary to decide when rights violations occur via climate change and to provide the appropriate remedies.

That approach has standing in some corners of the libertarian community. But as both libertarian political scientist Jeffrey Friedman and libertarian philosopher Matt Zwolinski have argued, it is unworkable in the real world and, accordingly, not a viable option. Even so, common law action against greenhouse gas emitters via appeals to property rights violations has been tried in AEP v Connecticut. This would appear to be exactly the sort of approach to the climate change problem that the authors suggest would represent a “genuine” market-oriented policy. Alas, the Cato Institute filed an amicus brief in the case rejecting that entire approach root-and-branch.

Second, the authors argue that, in the course of implanting a carbon tax, the political class will do the job so inefficiently that many of the theoretical gains associated with carbon pricing would be lost. No doubt. But a review of studies regarding carbon pricing schemes from around the world by economist Tom Tietenberg concludes that “they typically find that the cost savings from shifting to [taxes or cap-and-trade] are considerable, but less than would have been achieved if the final outcome had been fully cost effective.” Previous experience suggests that net gains can indeed be realized.

Rather than examining the entire universe of carbon tax regimes around the world, they examine only two: Australia’s retired carbon tax experiment and the carbon tax in British Columbia.

They are right to scorn the implementation of carbon taxation in Australia. That experience offers a useful cautionary tale regarding what how good policy ideas can be lost in political translation. But one can tell similar stories of how good policy initiatives have been mangled by politicians in nearly every single arena of public policy. If libertarian analysts were to reject meritorious reform projects in response to past political disappointments, libertarian analysts would need a new line of work. The same, of course, holds for analysts whatever their ideological persuasion might be.

The discussion regarding British Columbia is a different matter. The authors argue that the carbon tax there has had little impact on fossil fuel consumption, suggesting that carbon taxes cannot achieve their advertised end. But this is nonsense. Emission reductions following from a carbon tax in British Columbia tell us little about market response to the tax because there are only six coal mines in the province and coal-fired electricity is nonexistent in the British Columbia. While demand for coal is quite responsive to price, demand for oil—the main product effected by the province’s carbon tax—is much less responsive to price, and very much less so in transportation markets. This is well understood by economists and not the slightest bit controversial. Discovering that a carbon tax has had minimal impact on greenhouse gas emissions in British Columbia is akin to discovering that night follows day.

Even so, it’s not wrong for the authors to think about political considerations. The authors’ discussion of political reality, however, is rather compartmentalized. The position they seem to embrace—that no policy response to climate change is warranted—has failed politically. Consequently, we now face:

  • A load of command-and-control regulation in the states, much of which orders growing percentages of green energy production;
  • Federal command-and-control regulations in growing sectors of the economy;
  • Expensive state and federal subsidies for green energy and energy conservation;
  • Large regional cap-and-trade regimes established by the states; and
  • An incipient national cap-and-trade regime as a consequence of the EPA’s Clean Power Plan.

In a sense, we already have a carbon tax. Unfortunately, it is imposed indirectly via subsidy and regulation and it is stealthy, inefficient, and ineffective. It is also growing.

All this will be undone … how, exactly? To preemptively reject a carbon tax plan that would replace the bulk of those interventions is to continue with a heretofore failed political project—blanket opposition to intervention—that has little public support and limited political support. The authors are right to suspect that it will be a heavy lift to replace the bulk of those interventions with a carbon tax. Their political strategy, however, has proven to be an impossible lift.

Conclusion

Given the above, the authors fail to persuade. Even so, it’s worth noting that their arguments do not add up to an argument against carbon taxation. One could agree with everything the authors say and still conclude that:

  • Given that warming will likely be at the lower end of the range expected in the most recent IPCC report, a low carbon tax is preferable to a high carbon tax;
  • The tax should be global, which means the United States should redouble its efforts to produce an international agreement to that end;
  • Because of tax interaction effects, we should prepare ourselves for the fact that the tax will have an economic cost; and
  • In the course of adopting the tax, we will likely lose many (but not all) of its theoretical benefits.

The case for a carbon tax remains unscathed.