Why the CBO likes a carbon tax
Posted by Daniel Hall on February 14, 2008
The Congressional Budget Office released a report yesterday examining greenhouse gas emissions mitigation policy. Terry Dinan at the CBO authored the report, Policy Option for Reducing CO2 Emissions. The report looks at using 3 forms of incentive-based regulation: a carbon tax, and two variants on a cap-and-trade program, referred to as “inflexible” and “flexible” cap and trade. The report concludes that a carbon tax will be the most efficient policy instrument.
The reported has ignited a series of blog entries: the CBO director’s blog summarized the report, and Mankiw promptly excerpted the key result. Environmental Economics has already put out a series of posts this morning. For my money the most interesting conversation is happening over at Felix Salmon, where you must read the comments as well.
I want to use the discussion as a springboard to explain why the CBO says a carbon tax is more efficient, and I’m going to try to do so in a way that delves into the economics literature but remains grounded in relatively plain language. I’m also going to explain at the end why I still think cap-and-trade is the way to go.
First, Felix highlights a relevant excerpt from the report:
Analysts generally conclude that a tax would be a more efficient method of reducing CO2 emissions than an inflexible cap. The efficiency advantage of a tax stems from the contrast between the long-term cumulative nature of climate change and the short-term sensitivity of the cost of emission reductions. Climate change results from the buildup of CO2 in the atmosphere over several decades; emissions in any given year are only a small portion of that total. As a result, limiting climate change would require making substantial reductions in those emissions over many years, but ensuring that any particular limit was met in any particular year would result in little, if any, additional benefit (avoided damage). In contrast, the cost of cutting emissions by a particular amount in a given year could vary significantly depending on a host of factors, including the weather, disruptions in energy markets, the level of economic activity, and the availability of new low-carbon technologies (such as improvements in wind-power technology).
Relative to a cap-and-trade program with prespecified emission limits each year, a steadily rising tax could better accommodate cost fluctuations while simultaneously achieving a long-term target for emissions. Such a tax would provide firms with an incentive to undertake more emission reductions when the cost of doing so was relatively low and allow them to reduce emissions less when the cost of doing so was particularly high. In contrast, an inflexible cap-and-trade program would require that annual caps were met regardless of the cost, thereby failing to take advantage of low-cost opportunities to cut more emissions than were required by the cap and failing to provide firms with leeway in years when costs were higher.
The efficiency advantage of a tax over an inflexible cap depends on how likely it is that actual costs will differ from what policymakers anticipated when they set the level of the cap. Given the uncertainties involved, such differences are likely to be large–and, therefore, analysts generally conclude that the efficiency advantage of a tax is likely to be quite large. Specifically, available research suggests that in the near term, the net benefits (benefits minus costs) of a tax could be roughly five times greater than the net benefits of an inflexible cap. Put another way, a given long-term emission-reduction target could be met by a tax at a fraction of the cost of an inflexible cap-and-trade program.
Felix acknowledges this is a good argument, but states why he prefers cap and trade:
As I say, this is a good argument. But it’s also a bit weird: in my mind, the whole point of using a cap-and-trade system rather than a carbon tax is that no one knows what the actual costs of carbon emissions reduction are going to be. If those costs turn out to be much higher than anticipated, a carbon tax will simply fail, since it will have been set too low. A cap-and-trade system, by contrast, is dynamic: it can be adjusted in real time to reflect new information about the costs and benefits of certain levels of carbon emissions.
As is getting hashed out in the comments, this paragraph is not quite right, or at least not quite clear. There are two values in question here: the costs we pay to reduce emissions, and the benefits we receive from reducing them.
The reason the CBO prefers a carbon tax goes back to a seminal 1974 paper by Marty Weitzman, Prices vs. Quantities. Weitzman showed that if 1) the marginal benefits curve for abatement is relatively flat and 2) the costs of abatement are uncertain, then a price policy (i.e., tax) is more efficient.
Whoa, econ jargon! What does this mean in English and how does it apply here?
Imagine for the moment that we know that the total cost to society — summed across the entire future and discounted to today — from the next metric ton of CO2 emissions is $20. This is synonymous with saying that the benefit to society of eliminating (abating) that ton of emissions is $20. What about the next ton? Traditionally, we think that the benefit of reducing that ton is also $20. In fact, we think that the benefit of reducing quite a lot of tons would all be pretty close to $20 per ton. This is what is meant by “the marginal benefits curve for abatement is relatively flat.” You could eliminate a significant fraction of the world’s current CO2 emissions — currently running at around 30 billion metric tons, if I recall correctly — and each ton would save you about the same amount in future damages. ($20 in my example.) Dinan summarizes the rationale for thinking this is the case when she says,
Climate change results from the buildup of CO2 in the atmosphere over several decades; emissions in any given year are only a small portion of that total. As a result, limiting climate change would require making substantial reductions in those emissions over many years, but ensuring that any particular limit was met in any particular year would result in little, if any, additional benefit (avoided damage).
Ok, so the benefit of reducing a ton of emissions is a relatively constant figure. What about this “uncertainty of abatement costs”?
Well, abating emissions costs society money: we have to reduce how much energy we use, shift how it is made, etc., etc., and all those changes have costs. Ideally, we’d like to eliminate every ton of emissions we can when it costs less than $20, while still allowing those that cost more. (In this way we balance the costs and benefits of emissions reductions.) The traditional problem faced by regulators is that they don’t have perfect knowledge about how much reductions will cost. Thus the second half of Weitzman’s formulation about the uncertainty of abatement costs also applies.
Why do these two conditions imply that a tax policy will be more efficient? Well, if we set a tax on emissions equal to the benefit from abatement (still $20 in this example), then all the disparate actors in the economy — firms, individuals, etc. — will shift their behaviors in response to the new prices. These shifts will squeeze out emissions where it is cheapest to do so and up to the point where the net cost to society is $20 per ton of emissions, but won’t eliminate more high-value activities where the associated emissions would be costlier to reduce. This process will happen without central coordination, and without anyone needing to know beforehand exactly how it should be done. As I’ve quoted from Hayek before, “The mere fact that there is one price for any commodity… brings about the solution which (it is just conceptually possible) might have been arrived at by one single mind possessing all the information which is in fact dispersed among all the people involved in the process.”
Hayek’s quote also illustrates that it’s theoretically possible for a cap-and-trade program to achieve the same outcome. We could set the quantity of emissions allowances (the cap) at a level such the market price of allowances would be $20. This would balance the costs and benefits of reductions. The problem is that actually setting the cap at such a level requires that “one single mind possessing all the information.” Regulators must guess at what shifts will happen, an impossible task when dealing with something as complex as the entire economy. If regulators set the cap and it turns out allowances cost $50 per ton then we will be reducing emissions more than is optimal (because we’re paying $50 for something that only gets us $20 in benefit). And if they’re wrong the other way, and prices are quite low, say $10, then we get fewer reductions than would be best for society, because there are still a bunch of tons of emissions that are doing us $20 in damage and could be eliminated for less than that. With the tax you don’t have to worry about this: once you know the value of emissions reductions (i.e., the total damage from emissions) you can just set the tax equal to that value.
That’s the theoretical reason why the CBO says that a carbon tax is more efficient than a cap-and-trade system for mitigating greenhouse gas emissions.
Why do I still think a cap-and-trade system is the way to go, then? There are a whole host of reasons:
1. Note that this analysis assumes that a carbon tax will be set to equal the marginal damages from emissions. But there is tremendous uncertainty about what these damages actually are. Thus any program is going to have to respond to new information about the future costs of climate change. I think it is going to be more politically feasible to adjust a cap than change a tax rate. When was the last time we changed the gas tax? And cap-and-trade markets may be able to respond to new information before regulators even act.
2. Further, even if we knew what the costs of future damages were, this would not be how a tax rate — or cap level — was set. Either way it is going to happen through a politically negotiated process, meaning that neither a tax nor a cap-and-trade program has much chance of being “optimal” in a “balancing marginal costs and benefits” sort of way.
4. It’s possible to make make a cap-and-trade program mimic many of the efficiency features of a tax program. This is what’s meant by the CBO’s “flexible cap-and-trade” program; for example, allowing banking and borrowing of allowances prevents unimportant annual fluctuations in things like the weather (and hence energy demand) from causing large spikes and falls in the price of allowances. As we quoted Billy Pizer and Ian Parry in the previous link from their article in Regulation,
…the key distinction is not really between CO2 taxes and emissions trading systems per se. Rather, it is between policies that raise revenues — and use revenues wisely — and have limited price variability (i.e., CO2 taxes or auctioned permits with safety valves and emissions trading over time), versus nonrevenue-raising instruments with no provisions to limit price variability (i.e., traditional permit systems).
5. I think it’s interesting that “pure econ” guys (e.g. Mankiw) tend to support a carbon tax, while, at least in my experience, finance guys like Felix think more favorably about cap and trade. I think finance guys are comfortable that a flexible cap-and-trade program that allows banking and borrowing will have an extremely efficient market for allowances. Fundamentally, emissions allowances will be just another commodity that will have an entire range of financial implements — calls, puts, futures, derivatives — that will let entities manage their compliance obligations pretty efficiently. Now, you can probably write some of this enthusiasm off as self-interest — I’m sure there are plenty of traders salivating at the thought of a several hundred billion dollar market getting created overnight — but I also think they’re on to something here.
In summary, the choice we face isn’t between a first-best carbon tax and a cap-and-trade system. At best it’s between a very imperfect carbon tax and an also imperfect — but somewhat flexible — cap-and-trade program. And in my estimation it’s most likely that the choice is between an imperfect cap-and-trade system and nothing at all. I think we can do significantly better than nothing.
Addendum: Here’s a longer piece from Pizer and Parry that discusses taxes, cap-and-trade, and “command-and-control” standards.