Negative abatement costs, explained

Ted Gayer, co-director of economic studies at the Brookings Institution, published a lazily-reasoned and lazily-researched article dwelling on the apparent impossibility of GHG emissions reduction opportunities with negative marginal costs:

But Krugman oversells the affordability claim by linking to a widely cited report by McKinsey & Company. The main point of the McKinsey study is provided in their Exhibit B, which illustrates a rather peculiar finding that there are a significant number of pollution abatement options that can be achieved at “negative cost.” This finding violates the basic principles of economics. If firms (or consumers) could reduce emissions at negative cost, then they would do so. To say otherwise is to say that they are willingly or ignorantly passing up profits...

Gayer's original post has already been picked apart by luminaries such as Avent, DeLong, and Krugman, but in the interest of documenting this information for myself, I'm going to write out the possibilities Gayer acknowledges, followed by the ones he doesn't.

Gayer can think of four reasons why the McKinsey study might find negative costs:

1. The cost estimates are wrong (i.e. incorrect assumptions)
   


2. The cost estimates are incomplete


3. The private discount rate is incorrectly estimated


4. Firms are behaving irrationally (which he goes on to say is not realistic)

Thankfully, the relevant McKinsey report is available online, and had Gayer bothered to skim it, he would have found starting on p24 this helpful list of potential explanations for negative marginal costs, without invoking irrational behavior:

1. Agency issues (e.g. landlord-tenant)


2. Ownership transfer issues


3. Pricing distortions (e.g. via electricity price regulation)


4. Higher hurdle rates (e.g. consumer's cost of capital could be higher than a corporation's)


5. Constrained access to capital


6. Adverse bundling (e.g. efficient washing machine has other "gold-plated" features unaffordable to non-premium customers)


7. Local product and service availability


8. Improper installation or use
   

Beyond these, it is not hard to believe that well-documented behavioral biases (risk aversion, lack of awareness, habit) play a role in some cases.

Krugman also helpfully points out the considerable existing economic literature on principal-agent problems in energy efficiency.

Gayer's response is also a textbook example of responding to weaker criticisms, and ignoring the stronger ones.

For what it's worth, Gayer seems to support a market-based mechanism (cap-and-trade or a carbon tax) and fear a centralized, command-and-control approach by the government, which are both eminently reasonable positions (which I share, for what it's worth). But arguing flimsily against the existence of negative-cost GHG abatement opportunities like he does is an ineffective way of making that case.

CFC reduction: a success story

Also via Tim Haab at Environmental Economics:

CFC reductions are often hailed as an environmental policy success story. The reason CFC reduction policies have been so successful is Congress taxed the bejesus out of CFCs.

The Clean Air Act (Title VI) established caps on most CFC's as agreed upon under the Montreal Protocol, with a complete phase out occurring around the year 2000. The tax on CFC's was $1.37 a pound in 1990 and 1991, about twice the then current product price. Recycled CFC's were exempted from the tax.

The tax was raised in 1990 and again in 1992. The tax raised to $3.10 per pound in 1995 and to $4.90 per pound in 1996, raising the price six fold.

The lesson: economic incentives change behavior.

I find the subsequent news item about taxing overgrown lawns in Jupiter, Florida entertaining, but not very relevant. And unfortunately the "tax the bejesus" strategy isn't as viable for massive emittants (is that a word) like CO2, NOx and methane.

The underrated efficiency of cap-and-trade

I have a lot of respect for Geoff Styles, who is always very thoughtful in his writing on energy and related issues, but I disagree with him here (I'll reproduce my comment):

Hi Geoff,

I don’t think the current state of economic research on cap-and-trade (both theoretical and empirical) would support your statement that, “The problems with this thinking lie in the enormous distortions and unnecessary economic hardship those uneven allocations will create over the next 20-plus years and the opportunity cost of the emissions reductions that could have been achieved more quickly and cheaply.” See for example Robert Stavins here:

It is exceptionally important to keep in mind what is probably the key attribute of cap-and-trade systems: the allocation of allowances – whether the allowances are auctioned or given out freely, and how they are freely allocated – has no impact on the equilibrium distribution of allowances (after trading), and therefore no impact on the allocation of emissions (or emissions abatement), the total magnitude of emissions, or the aggregate social costs...

Generally speaking, the choice between auctioning and freely allocating allowances does not influence firms’ production and emission reduction decisions. Firms face the same emissions cost regardless of the allocation method. When using an allowance, whether it was received for free or purchased, a firm loses the opportunity to sell that allowance, and thereby recognizes this “opportunity cost” in deciding whether to use the allowance. Consequently, the allocation choice will not influence a cap’s overall costs.

I agree there’s a lot of room to argue that the W-M allocation of permits is unfair, and some shareholders will benefit while others will suffer. But the attractive efficiency properties of the cap-and-trade mechanism (as opposed to a carbon tax, incidentally) mean that the aggregate social outcome is likely to be fairly efficient, rather than “enormously distorted” as you worry.

Update 1: bartman on merchant power generation:

R:

If you read the text of Waxman-Markey, you will find that in one key way it invalidates Stavins' statement that you cited.

The key feature is the dynamic allocation of permits to merchant coal plants. Since their annual allocation is based upon their previous year's production, then a merchant coal plant has a greater incentive to use the permits than to trade them, because trading them means that they get a lower allocation the following year. The end result is that, ceteris paribus, there will be more coal burned than in a world with static allocation or full auctioning.

We've had a lot of debate at work whether this is a feature or a bug.

R:

bartman,

Good point that the dynamic allocations for merchant power generation may not uphold optimal efficiency. As you say, TBD whether the dynamic allocations are a feature or a bug. But these are only 5% of total allocated permits, so even in the worst case this won’t drastically compromise the efficiency of the overall outcome (and 95+% efficiency would be a pretty good outcome of the political sausage-making process).

Update 2: Geoff responds:

R,
I'd go further than Bartman and suggest that while Stavins might be right about the "equilibrium distribution after trading", such systems rarely attain equilibrium in the real world. Dr. Stavins has an impressive bio, but I don't see any real-world commodity trading experience in there.

The fundamental problem is that for the first 15-18 years of cap & trade, the highest emitting sector will receive the lion's share of free allowances and have the least incentive either to alter operations or to invest in more efficient hardware. That's compounded by W-M's structural requirement that the utilities pass on the benefits to their customers; without that, they'd have more incentive to reduce their own emissions and sell their free allowances to the poor refiners who already operate at 90% effeciency and have little scope for reducing emissions further. The result of that structural flaw will be higher emissions from the electricity sector than under full auctioning or even-handed allocation, while refiners will be unable to cut enough to compensate for that and will merely have to pay up, perhaps at the penalty level, with the resulting increase in their costs driving up fuel costs for consumers and putting marginal US refiners out of business, leading to higher imports of refined products than otherwise. None of those outcomes looks good for energy security, and they won't do very much for the global climate, either.

R:

Geoff,

Re “such systems rarely attain equilibrium in the real world”, the empirical research I’m aware of shows that cap-and-trade systems have often reached efficient equilibria in the past (e.g. the SO2 allowance trading system) – see this paper for more detail, and please let me know if you’re aware of other evidence contradicting this.

Re “the highest emitting sector will receive the lion's share of free allowances and have the least incentive either to alter operations or to invest in more efficient hardware”, Stavins’ key point is free allowances alone don’t alter operating incentives (in this sense they’re similar to sunk costs). If the allowances market is clearing at $20/ton CO2e, any company (utility, refinery, whatever) with available abatement opportunities at a cost of $15/ton can make money by implementing that project. If it’s a utility with allocated allowances, the benefit will come from selling a credit for $20, whereas the benefit to a refinery without allocated allowances is avoiding the purchase of credit for $20, but the incentives are the same. The case for abatement opportunities at a cost above the market-clearing price is similar – implementing these projects doesn’t make economic sense, whether or not the company was allocated allowances.

So an initial overallocation to utilities will not in and of itself result in high inefficiency. The tricky part is that utilities don’t capture the full benefits of cost-effective abatement, since as you rightly point out these must be passed on in part to consumers. If I remember correctly, states and municipalities have some latitude to regulate the credit pass-through locally, so there is some flexibility to mandate investment in energy efficiency at the local level. I agree that this makes the overall efficiency properties messier. But if utilities capture even a part of the benefits of their own investments, efficiency investments below the market-clearing cost of carbon will still be profitable, and it seems overall efficiency will still be pretty good.

Update 3: The dialogue continues:

R,
I'm sorry, but those arguments would have been much more persuasive had we not been treated to such a massive display of behavioral inefficiencies in markets. Nor am I terribly reassured that SO2 markets might have reached equilibrium, since at scale the GHG markets seem likely to look a lot more like energy commodity markets, which manifestly are rarely in equilibrium, except by sheer coincidence. In that light, it matters very much how those free allowances are allocated. I don't mean to imply that any economic analysis of the situation is irrelevant, but rather that it must be tempered with the fruits of real-world experience.

R,
I'd like to add a thought to my response above. Without invoking the whole sub-prime, CDO/CDS debacle, a much more relevant example comes from the California electricity crisis. Contrary to conventional wisdom that lays all the blame on certain bad actors, the underlying problem--which created the loopholes through which traders drove Mack trucks--was poor market design. From my long experience in real markets, a situation that so massively skews the initial allocation looks like another recipe for inefficiency, distortion and exploitation; i.e., it constitutes poor market design. Good market design requires listening not just to the lobbyists, regulators, economists, and NGOs, but also to folks who have run or participated in actual markets.

R:

Geoff,

I totally agree that good market design is critical, and I appreciate that you're skeptical of markets reaching relative equilibrium based on recent events and your real-world experience (which is totally fair).

I don't understand the mechanism by which you believe initial allowance allocations (a.k.a. unconditional free money) will affect the decisions made by actors like utilities. Are you arguing that they'll behave irrationally, or do you think allowance allocations in some way affect the actual economic incentives they face?

Geoff:

There are many issues and uncertainties here, and it's important to note that behavior that might seem irrational to you could seem perfectly rational in the context of these players. For example:

A. At the foreseeable price of allowances in the early phases of cap & trade the foregone opportunity cost of selling allowances (the benefit of which passes to customers, anyway) is unlikely to be sufficient to compensate utilities for not running their fleets of billion-dollar assets. I.e., for the largest group of allowance recipients, it may be more rational for them to consume the allowances they receive and pass on their notional value to customers than to under-run and sell them to refiners.

B. The impact of the large share of allowances allocated to non-commercial recipients is hard to predict. Will they sit on their allowances until they need revenue, sell them cheaply to middlemen to get funds up front, strike long-term deals with refiners, or choose to retire them in an effort to accelerate reductions? Some of these outcomes could impede market liquidity and leave refiners exposed to penalties or higher costs than anticipated.

R:

In situation A), whether the utility was originally allocated allowances doesn't change their rational action. If the benefit of running their legacy assets exceeds the market price of the necessary allowances (as you describe), a rational utility would run those legacy assets no matter what, buying allowances on the market if necessary.

B) is a valid point, but I'm not sure the danger of strange market behavior would be lessened by giving those allowances to refiners with sophisticated trading operations instead.

Update 4: And... the dénouement:

R,
Your background on your blog indicates you're a strategy consultant. Are you telling me that all of your clients have treated an opportunity cost exactly the same as a cash cost? I'm highly skeptical, because in my own experience businesses treat the two very differently. Anyone else want to weight in on that one?

R:

Businesses don't always treat cash costs and opportunity costs the same in practice, but I would categorize this as irrational behavior (or making bad business decisions, if you prefer).

For example, imagine your exact scenario A), except in the world where the utility wasn't allocated allowances. Running legacy assets would still be the right business decision (and not doing so would be a bad decision). Whether the utility should run its legacy assets depends on the allowance market price, but not the utility's initial allocation of allowances.

My point is not that utilities won't behave irrationally, but rather that A) is not an example of rational commercial incentives leading to inefficient outcomes.

Thank you, by the way, for the spirited but respectful tone of this discussion.

Geoff:

Spirited but respectful is what we aim for. Thanks.

Stavins on cap-and-trade vs. carbon tax

The distinguished Robert Stavins (who is >> Mankiw on cap-and-trade, and probably >>>> Rex Tillerson) takes on the cap-and-trade vs. carbon tax debate.

After first demonstrating the need for a market-based mechanism, he then strongly defends cap-and-trade (and not on the grounds of political pragmatism, which is perhaps the argument most oftenly advanced).

While there are tradeoffs between these two principal market-based instruments targeting CO2 emissions — a cap-and-trade system and a carbon tax – the best (and most likely) approach for the short to medium term in the United States is a cap-and-trade system. I say this based on three criteria: environmental effectiveness, cost effectiveness, and distributional equity. So, my position is not capitulation to politics. On the other hand, sound assessments of environmental effectiveness, cost effectiveness, and distributional equity should surely be made in the real-world political context.

Here is his more detailed argument:

Having said this, there are some real differences between taxes and cap-and-trade that need to be recognized. First, environmental effectiveness: a tax does not guarantee achievement of an emissions target, but it does provides greater certainty regarding costs. This is a fundamental tradeoff. Taxes provide automatic temporal flexibility, which needs to be built into a cap-and-trade system through provision for banking, borrowing, and possibly a cost-containment mechanism. On the other hand, political economy forces strongly point to less severe targets if carbon taxes are used, rather than cap-and-trade – this is not a tradeoff, and this is why environmental NGOs are opposed to the carbon-tax approach.

In principle, both carbon taxes and cap-and-trade can achieve cost-effective reductions, and – depending upon design — the distributional consequences of the two approaches can be the same. But the key difference is that political pressures on a carbon tax system will most likely lead to exemptions of sectors and firms, which reduces environmental effectiveness and drives up costs, as some low-cost emission reduction opportunities are left off the table. But political pressures on a cap-and-trade system lead to different allocations of allowances, which affect distribution, but not environmental effectives, and not cost-effectiveness.

Proponents of carbon taxes worry about the propensity of political processes under a cap-and-trade system to compensate sectors through free allowance allocations, but a carbon tax is sensitive to the same political pressures, and may be expected to succumb in ways that are ultimately more harmful: reducing environmental achievement and driving up costs.

In summary, there is a true trade-off between certainty on emissions reduction and certainty on cost, but in a realistic political environment, the carbon tax is vulnerable to both looser targets, and less efficiency in meeting them. The "picking favorites" criticism directed against legislation such as Waxman-Markey holds equally against a carbon tax, and the political outcomes would be in all likelihood worse from an environmental standpoint.

Update: Environmental Economics highlights a different (but also excellent) passage from Stavins:

The Hamilton Project staff concluded in an overview paper (which I highly recommend) that a well-designed carbon tax and a well-designed cap-and-trade system would have similar economic effects. Hence, they said, the two primary questions to use in deciding between them should be: which is more politically feasible; and which is more likely to be well-designed?

The answer to the first question is obvious; and I have argued here that given real-world political forces, the answer to the second question also favors cap-and-trade. In other words, it is important to identify and design policy that will be “optimal in Washington,” not just from the perspective of Cambridge, New Haven, or Berkeley.

In “policy heaven,” the optimal instrument to address climate-change emissions may well be a carbon tax (largely because of its simplicity), but in the real world in which policy is developed and implemented, cap-and-trade is the best approach if one is serious about addressing the threat of climate change with meaningful, effective, and cost-effective policies.

Exxon CEO on carbon tax

Few people believe that ExxonMobil CEO Rex Tillerson's defense of a carbon tax is not motivated partly by the political difficulty of passing such a tax. In his latest address, he shows that he's cognizant of this criticism:

But when the boss of Exxon attacks climate-change legislation, skepticism is usually in order. “We think it’s fair to view Exxon’s opposition to cap-and-trade – Tillerson’s reasonable critiques notwithstanding – as a tactic meant to delay passage of meaningful legislation,” argue the folks at Green Energy Reporter.

Mr. Tillerson addressed that head on:

Now, some people have suggested that a revenue-neutral carbon tax has no chance of gaining sufficient support in Congress to become law. They say a carbon tax is too politically sensitive and that it is easier and more politically expedient to support a cap-and-trade approach, because the public will never figure out where it is hitting them. They will just know they hurt somewhere in their pocketbook.

I disagree with this assessment. I believe the American people want climate policy to be transparent, honest, and effective. Economists generally agree that achieving a given emissions target costs less under a tax or fee approach than under a cap-and-trade system. I firmly believe it is not too late for Congress to consider a carbon tax as the better policy approach for addressing the risks of climate change. Indeed, there has never been a more opportune time for Congress to pursue this course of action.

Good for him for recognizing, although it doesn't change my opinion of his underlying motivations.

Feature or bug?

Greg Mankiw says that Waxman-Markey’s customer rebates are a bug, not a feature as Paul Krugman claims. But this is easily remedied with a lump-sum rebate, rather than lower prices (a nice trick which has interesting potential elsewhere in the realm of energy efficiency, and also bears some similarity to cap-and-dividend).

Felix Salmon gets it, noting that a flat refundable tax credit avoids the problem. He adds that it's hard to avoid generating any unequal results:

Even that, however, would create inequities: three college roommates sharing a small city apartment would get three times the amount going to a single mother trying to raise three kids in the countryside — someone whose energy consumption would naturally be much higher.

There isn’t a simple and fair way of doing things — even channeling fixed payments through the energy companies themselves would unfairly advantage people who use say electricity and natural gas and heating oil. Instead, I fear that the fair method is going to be complicated, based on ZIP codes and size of household at the very least.

Yes, it's hard to get it perfect, but Mankiw's original argument is easily refuted.

Carbon tax in France

Greg Mankiw is pleased that France has proposed a revenue-neutral carbon tax. Tyler Cowen is less sanguine:

In reality, France’s carbon tax is basically just a gasoline tax—and a tiny one at that. The electricity sector, overwhelmingly powered by emissions-free nuclear power, isn’t part of the plan [TC: Duh!], Prime Minister Francois Fillon told Le Figaro. The tax will basically fall on liquid fuels—raising pump prices 3 euro cents a liter (that’s roughly 15 U.S. cents a gallon).

In theory it will be revenue-neutral but most French voters are nonetheless opposed to the measure.

Incidentally, one argument that carbon tax proponents make against cap-and-trade is that a carbon tax is less subject to loopholes and political manipulation. That argument has always been silly; now it is more demonstrably so:

Large CO2 emitters, such as oil refiners and steel makers, will be exempted from paying the new tax. The government will propose special compensations for fishermen, farmers and truckers...

P.S. Not the first backdoor gasoline tax we have seen proposed this year.

Outsourced responses to cap-and-trade originator

Greg Mankiw and Environmental Capital both pointed quickly to an article in the WSJ citing one of the originators of the cap-and-trade idea, Thomas Crocker, who is now in favor of a carbon tax. But Felix Salmon (my favorite finance blogger, mind you) shows his versatility by turning Crocker's arguments on their head:

Let’s take Crocker’s arguments one by one, with the proviso that they’re coming second-hand, via the WSJ, rather than directly from Crocker himself.

First, Crocker says that a carbon tax “would be easier to enforce” than a cap-and-trade system. But it’s hard to see why that should be the case: both of them involve measuring the same carbon emissions. It’s certainly easier to enforce when you measure upstream rather than downstream, but that applies equally to carbon taxes and to cap-and-trade.

Crocker then gets into the meat of his argument:

Mr. Crocker sees two modern-day problems in using a cap-and-trade system to address the global greenhouse-gas issue. The first is that carbon emissions are a global problem with myriad sources. Cap-and-trade, he says, is better suited for discrete, local pollution problems. “It is not clear to me how you would enforce a permit system internationally,” he says. “There are no institutions right now that have that power.”

Yes, cap-and-trade is better suited for local pollution problems than it is for global pollution problems. But that doesn’t mean that a carbon tax is better for global pollution problems than cap-and-trade is. Indeed, the opposite is true. In theory, once a number of jurisdictions implement a cap-and-trade system, carbon traders will start arbitraging the various different carbon permits, and we will end up with something approaching a global system. Carbon taxes, by contrast, are ever and always local. Crocker is right that a US cap-and-trade system wouldn’t necessarily slow global carbon emissions if China and India refuse to play ball. On the other hand, neither would a carbon tax. But at least a cap-and-trade system has the ability to scale into China and India.

But moving on:

The other problem, Mr. Crocker says, is that quantifying the economic damage of climate change — from floods to failing crops — is fraught with uncertainty. One estimate puts it at anywhere between 5% and 20% of global gross domestic product. Without knowing how costly climate change is, nobody knows how tight a grip to put on emissions.

In this case, he says Washington needs to come up with an approach that will be flexible and easy to adjust over a long stretch of time as more becomes known about damages from greenhouse-gas emissions.

Agreed, 100% — which is exactly why we need a flexible cap-and-trade system rather than an inflexible carbon tax. A cap-and-trade system can be tweaked much more easily than a carbon tax, both in terms of the level of the cap and in terms of the proportion of the permits which is auctioned off rather than given away. Crocker says it’s hard to adjust a cap once it’s in place — but he neglects to mention that it’s harder still to adjust a tax once it’s in place.

John Whitehead at Environmental Economics arrives at the same conclusions, noting sagely that:

Unfortunately, there is no good solution to the global nature of the problem. Protectionism, voluntary agreements and war are three ways to enforce international "policy."

... and the interesting factoid that

Crocker, a 2008 AERE Fellow, won the 2001 AERE Publication of Enduring Quality Award for his work on cap-and-trade.

The versatility of cap-and-trade is such that it's hard to come up with arguments against it that don't apply to a carbon tax as well - other recent failures include admin costs and the distortionary effect of taxes.

Dear Greg, Why Waxman-Markey is minimally distortionary

Greg Mankiw, whose style and wonkiness I love, posts a simple model of how he sees a cap-and-trade bill or carbon tax functioning.

The trick is how to fix the second distortion [inefficiently high carbon intensity of consumption baskets] without making the first one [work/consumption decision distortions via taxes] worse.

This is a fair way of framing the problem.

The basic problem is that a new tax on carbon-intensive products C1 is also an additional tax on consumption C, unless there is some other offsetting tax change.

This is where the Rorschach test comes in. A carbon tax without a compensating income tax cut makes one problem better and one worse. The question then is which problem is bigger.

Stop there - what I think the crucial point he is missing is that Waxman-Markey will effectively function as an income tax cut for the most part. Let's review the Waxman-Markey allowance allocations and split them into those which will effectively function as consumer rebates, and those which will not.

- 46% to electricity generators and local natural gas distribution companies, "which they must use to protect consumers from price increases" (REBATE)
- 15% to low- and moderate-income households to "protect them from other energy cost increases" (REBATE, although implicitly redistributionary)
- 17% to domestic energy-intensive, trade-exposed industries and refining (REBATE unless there is some collusion in pricing to increase prices beyond the cost increase)
- 11-16% to energy efficiency and clean energy technology (NOT A REBATE)
- ~10% to other public purposes like avoiding deforestation, adaptation and international clean tech transfer (NOT A REBATE)

So almost 80% of the "giveaway permits" are effectively consumer rebates, and the remaining ones, which will be largely purchased by oil refiners, amount to a gasoline tax whose proceeds will be used for energy efficiency, clean tech and other public purposes.

So in addition to Robert Stavins' conclusion that “the appropriate characterization of the Waxman-Markey allocation is that more than 80% of the value of allowances go to consumers and public purposes, and less than 20% to private industry,” I would add that ~80% are effectively rebated to the consumers, mitigating the majority of the distortion which Mankiw laments in his post. Waxman-Markey is in fact, for the most part, a carbon tax with a compensating income tax cut.

The cost of administering a carbon tax

According to the CBO, administering cap-and-trade will cost $7.5bn over the next decade (at <$1bn/year, doesn't sound too horrendous). I can't tell if Environmental Capital is trying to be snarky here:

What was it that carbon-tax proponents always say about a tax being the simpler, easier solution to curb carbon emissions?

If not, no worries, but if so, this is just silly. Just like cap-and-trade, a carbon tax would require millions of entities to track their carbon emissions, have those emissions audited and processed by the CO2 IRS. Add that to the fact that any carbon tax that got passed would inevitably be riddled with loopholes and exceptions, it's not at all clear that the administrative burden would be any less than the one we're contemplating now for cap-and-trade