Learning from Thirty Years of Experience with Cap-and-Trade Systems

“Those who do not remember the past are condemned to repeat it.”

The implication of this famous line (often misquoted as “those who do not learn history are doomed to repeat it”) from philosopher George Santayana’s 1905 book, The Life of Reason, Volume I – Reason in Common Sense, is that we are wise to learn from our mistakes.  This is undoubtedly true, as is the parallel recommendation that we are wise to learn from our successes.


China is expected to launch later this year the world’s largest (CO2) emissions trading system; the European Union is in the process of extending and strengthening its CO2 cap-and-trade system; California has just extended and strengthened its CO2 cap-and-trade system; and earlier this week, nine New England and Middle Atlantic U.S. states announced their plan to extend and strengthen the Regional Greenhouse Gas Initiative.  With such developments in place and on the horizon, this is an important time to think carefully and critically about the history of cap-and-trade, and identify lessons that can be learned from three decades of prior experiences – both successes and failures.

That is precisely what Richard Schmalensee (Howard W. Johnson Professor of Economics and Management, Emeritus, at the Massachusetts Institute of Technology, and Dean Emeritus of the MIT Sloan School of Management) and I sought to do in an article which recently appeared in the Review of Environmental Economics and Policy (REEP) (“Lessons Learned from Three Decades of Experience with Cap and Trade,” Review of Environmental Economics and Policy, volume 11, issue 1, Winter 2017, pp. 59-79).  I encourage you to read the full article, which – in keeping with the style of the Review of Environmental Economics and Policy – is brief and broadly accessible.

In the hope that you may be stimulated to read the full article, in today’s blog essay I draw on the article to provide the historical context of our analysis, and to review some of our conclusions (for the actual analysis of individual cap-and-trade systems, and the justifications for our conclusions, you will need to see the article).

The Historical Context

Thirty years ago, many environmental advocates argued that government allocation of rights to emit pollution legitimized environmental degradation, while others questioned the feasibility of such an approach.  At the time, virtually all pollution regulations took a command-and-control approach, specifying the type of pollution-control equipment to be used or setting uniform limits on emission levels or rates.

Today, it is widely recognized – at least among students of economics – that because emission reduction costs can vary greatly, the aggregate abatement costs under command-and-control approaches can be much higher than under market-based approaches, which establish a price on emissions – either directly through taxes or indirectly through a market for tradable emissions rights established under a cap-and-trade policy.  Because market-based approaches tend to equate marginal abatement costs rather than emissions levels or rates across sources, they can achieve aggregate pollution-control targets at minimum cost.

In the REEP article, Dick Schmalensee and I examined the design and performance of seven of the most prominent emissions trading systems that have been implemented over the past 30 years in order to identify key lessons for future applications.  We focused on systems that have been important environmentally and/or economically, and whose performance has been well documented.  We excluded emission-reduction-credit (offset) systems, which offer credits for emissions reductions from some counterfactual baseline, because while emissions can generally be measured directly, emissions reductions are unobservable and often ill-defined.

The seven emissions trading systems we examined were:

  • the U.S. Environmental Protection Agency’s (EPA’s) phasedown of leaded gasoline in the 1980s;
  • the U.S. sulfur dioxide (SO2) allowance trading program under the Clean Air Act Amendments of 1990;
  • the Regional Clean Air Incentives Market (RECLAIM) in southern California;
  • the trading of nitrogen oxides (NOX) in the eastern United States;
  • the Regional Greenhouse Gas Initiative (RGGI) in the northeastern United States;
  • California’s cap-and-trade system under Assembly Bill 32; and
  • the European Union (EU) Emissions Trading System (ETS).

All of these programs except the first are textbook cap-and-trade systems.

In the article, we reviewed the design, performance, and lessons learned from each of the seven systems (and briefly discussed several other cap-and-trade systems).  In this blog essay, however, I turn immediately to our summary of key lessons.

Lessons from Thirty Years of Experience

Overall, we found that cap-and-trade systems, if well designed and appropriately implemented, can achieve their core objective of meeting targeted emissions reductions cost-effectively.  This is not something that was taken for granted in the past, and is still not accepted in some quarters.  That said, the devil is in the details, and design as well as the economic environment in which systems are implemented are very important.  Moreover, as with any policy instrument, there is no guarantee of success.  Based on the numerous specific lessons we identified in our analysis, several design and implementation features of cap-and-trade programs appear critical to their performance.

Key Features for System Design and Implementation

First, it is important not to require prior approval of trades.  In contrast to early U.S. experience with emissions offset systems, transactions costs can be low enough to permit considerable efficiency-enhancing trade if prior approval of trades is not required.

Second, it is clear from both theory and experience that a robust market requires a cap that is significantly below BAU emissions.

Third, to avoid unnecessary price volatility, it is important for final rules (including those for allowance allocation) to be established and accurate data supplied well before commencement of a system’s first compliance period.

Fourth, high levels of compliance in a downstream system can be achieved by ensuring there is accurate emissions monitoring combined with significant penalties for non-compliance.

Fifth, provisions for allowance banking have proven to very important for achieving maximum gains from trade, and the absence of banking provisions can lead to price spikes and collapses.

Sixth, price collars are important.  A changing economy can reduce emissions below a cap, rendering it non-binding, or a growing economy can increase emissions and drive allowance prices to excessive levels.  Price collars reduce price volatility by combining an auction price floor with an allowance reserve.  The resulting hybrid systems will generally have lower costs (as more stable prices facilitate investment planning) at the expense of less certain emissions reductions.

Finally, economy-wide systems are feasible, although downstream, sectoral programs have been more commonly employed.

Political Considerations that Affect Cap-and-Trade Design

Experiences with cap-and-trade also indicate the importance of political considerations for the design of cap-and-trade programs.

First, because of the potentially large distributional impacts involved, the allocation of allowances has inevitably been a major political issue.  Free allowance allocation has proven to help build political support. Under many circumstances, the equilibrium allowance distribution, and hence the aggregate abatement costs of a cap-and-trade system, are independent of the initial allowance allocation (Montgomery 1972; Hahn and Stavins 2012).  This means that the allowance allocation decision can be used to build political support and address equity issues without concern about impacts on overall cost-effectiveness.

Of course, free allowance allocation eliminates the opportunity to cut overall social costs by auctioning allowances and using the proceeds to cut distortionary taxes.  On the other hand, experience has shown that political pressures exist to use auction revenue not to cut such taxes, but to fund new or existing environmental programs.  Indeed, cap-and-trade allowance auctions can and have generated very significant revenue for governments.

Second, the possibility of emissions leakage and adverse competitiveness impacts has been a prominent political concern in the design of cap-and-trade systems.  Virtually any meaningful environmental policy will increase production costs and thus could raise these concerns, but this issue has been more prominent in the case of cap-and-trade instruments.  In practice, leakage from cap-and-trade systems can range from non-existent to potentially quite serious.  It is most likely to be significant for programs of limited geographic scope, particularly in the power sector because of interconnected electricity markets.  Attempts to reduce leakage and competitiveness threats through free allocation of allowances do not per se address the problem, but an output-based updating allocation can do so.

Third, although carbon pricing (through cap-and-trade or taxes) may be necessary to address climate change, it is surely not sufficient.  In some cases, abatement costs can be reduced through the use of complementary policies that address other market failures, but the types of “complementary policies” that have emerged from political processes have instead addressed emissions under the cap, thereby relocating rather than reducing emissions, driving up abatement costs, and suppressing allowance prices.

Identifying New Applications

Cap-and-trade systems are now being seriously considered for a wide range of environmental problems.  Past experience can offer some guidance as to when this approach is most likely to be successful.

First, the greater the differences in the cost of abating pollution across sources, the greater the likely cost savings from a market-based system – whether cap-and-trade or tax — relative to conventional regulation (Newell and Stavins 2003).  For example, it was clear early on that SO2 abatement cost heterogeneity was great, because of differences in ages of plants and their proximity to sources of low-sulfur coal (Carlson et al. 2000).

Second, the greater the degree of mixing of pollutants in the receiving airshed (or watershed), the more attractive a market-based system, because when there is a high degree of mixing, local hot spots are not a concern, and the focus can thus be on cost-effective achievement of aggregate emissions reductions.  Most cap-and-trade systems have been based on either the reality or the assumption of uniform mixing of pollutants. However, even without uniform mixing, well-designed cap-and-trade systems can be effective, as illustrated by the two-zone trading system under RECLAIM, at the cost of greater complexity.

Third and finally, since Weitzman’s (1974) seminal analysis of the effects of cost uncertainty on the relative efficiency of price versus quantity instruments, it has been well known that in the presence of cost uncertainty, the relative efficiency of these two types of instruments depends on the pattern of costs and benefits.  Subsequent literature has identified additional relevant considerations (Stavins 1996; Newell and Pizer 2003).  Perhaps more importantly, theory (Roberts and Spence 1976) and experience have shown that there are efficiency advantages of hybrid systems that combine price and quantity instruments in the presence of uncertainty.

Implications for Climate Change Policy

Two highly relevant lessons from thirty years of experience with cap-and-trade systems stand out.  First, cap-and-trade has proven itself to be environmentally effective and economically cost-effective relative to traditional command and control approaches. Moreover, less flexible systems would not have led to the technological change that appears to have been induced by market-based instruments (Schmalensee and Stavins 2013) or the induced process innovations that have resulted (Doucet and Strauss 1994).

Second, and equally important, the performance of cap-and-trade systems depends on how well they are designed.  In particular, it is important to reduce unnecessary price volatility, and hybrid designs can offer an attractive option if some variability of emissions can be tolerated, since substantial price volatility generally raises costs.

All of this suggests that cap-and-trade merits serious consideration when regions, nations, or sub-national jurisdictions are developing policies to reduce greenhouse gas (GHG) emissions.  And, indeed, this has happened.  However, because any meaningful climate policy will have significant impacts on economic activity in many sectors and regions, proposals for such policies have often triggered significant opposition.

In the United States, the failure of cap-and-trade climate policy in the Congress in 2010 was essentially collateral damage from a much larger political war that decimated the ranks of both moderate Republicans and moderate Democrats.  Nevertheless, political support for using cap-and-trade systems to reduce GHG emissions has emerged in many other parts of the world.  In fact, in the negotiations leading up to the Paris climate conference in 2015, many parties endorsed key roles for carbon markets, and broad agreement emerged concerning the value of linking those markets (codified in Article 6 of the Paris Agreement).

It is certainly possible that three decades of high receptivity to cap-and-trade in the United States, Europe, and other parts of the world will turn out to have been only a relatively brief departure from a long-term trend of reliance on command and control environmental regulation.  However, in light of the generally positive experience with cap-and-trade, there is reason for optimism that the tarnishing of cap-and-trade in US political debates will itself turn out to be a temporary departure from a long-term trend of increasing reliance on market-based environmental policy instruments.  Only time will tell.


Trying to Remain Positive

With inauguration day in the United States just two weeks away, it is difficult to harbor optimism about what the Trump presidency will mean for this country and for the world in realms ranging from economic progress to national security to personal liberty (as I wrote in this space one month before the November election – This is Not a Time for Political Neutrality, October 9, 2016).  In the wake of the election, expectations are no better, including in the environmental realm (as I wrote shortly after the election – What Does the Trump Victory Mean for Climate Change Policy?, November 10, 2016).  And since then, the President-elect’s announced nominations for key positions in the administration have probably eliminated whatever optimism some progressives may have been harboring.

Remarkably, the least worrisome development in regard to anticipated climate change policy may be the nomination of Rex Tillerson to become U.S. Secretary of State.  Two months ago it would have been inconceivable to me that I would write this about the CEO of Exxon-Mobil taking over the State Department (and hence the international dimensions of U.S. climate change policy).  But, think about the other likely candidates.  And unlike many of the other top nominees, Mr. Tillerson is at least an adult, and – in the past (before the election) – he had led his company to reverse course and recognize the scientific reality of human-induced climate change (unlike the President-elect), support the use of a carbon tax when and if the U.S. puts in place a meaningful national climate policy, and characterize the Paris Climate Agreement as “an important step forward by world governments in addressing the serious risks of climate change.”

It’s fair to say that it is little more than damning with faint praise to characterize this pending appointment as “the least worrisome development in regard to climate change policy,” but the reality remains.  Everything is relative.  Of course, whether Mr. Tillerson will maintain and persevere with his previously stated views on climate change is open to question.  And if he does, can he succeed in influencing Oval Office policy when competing with Scott Pruitt, Trump’s pick to run EPA, not to mention Rick Perry, Trump’s bizarre choice to become Secretary of Energy?

In the face of all this (and much else), is it possible to offer any statement of optimism or at least hope?  The answer may be found in the reality that U.S. policy – in many issue areas – consists of much more than the policies of the Federal government.  In a variety of policy realms, the states play an exceptionally important role.  One might not normally think about this in the context of addressing a global commons problem, such as climate change, but these are not normal times.

And so I will try to rescue myself from my current mental state – at least temporarily – by focusing today on policy developments in the State of California.  To do this, I offer an op-ed I recently wrote with Professor Lawrence Goulder of Stanford University, which was published in the Sacramento Bee a week before the November election.  Good policy developments at the state level are, of course, even more important now than they were then.


Sacramento Bee

October 30, 2016

New emissions targets make cap and trade the best low-cost, market-based approach

By Lawrence H. Goulder and Robert N. Stavins

This is a critical time for California’s climate policies. Recently, Gov. Jerry Brown achieved his hope of extending California’s action beyond 2020, the termination date of Assembly Bill 32. Whereas AB 32 called for reducing the state’s greenhouse gas emissions to 1990 levels by 2020, the newly signed Senate Bill 32 and AB 197 mandate an additional 40 percent reduction by 2030.

Unless these ambitious goals are pursued with the most cost-effective policy instruments, the costs could be unacceptably high. The governor’s targets make it especially important to use a low-cost, market-based approach: cap and trade.

Unfortunately, rather than increasing cap and trade’s role, recent proposals emphasize the use of less efficient, conventional policies. The environmental justice lobby supports this change, contending that emissions trading hurts low-income and minority communities by causing pollution to increase.

In fact, abandoning cap and trade would harm these communities by raising costs to businesses and thereby prices to consumers. With cap and trade, the sources able to reduce emissions least expensively take on more of the pollution-reduction effort. This lowers costs and prices.

When the environmental justice community worries about cap and trade, their concern is not about the greenhouse gas emissions that cause climate change: These gases spread evenly worldwide and have no discernible local impact. Rather, it’s about “co-pollutants,” such as nitrogen oxides, carbon monoxide and particulates, which often are emitted alongside greenhouse gases.

By reducing California’s greenhouse gas footprint, cap and trade lowers concentrations of these co-pollutants. Still, it’s possible – in theory – for co-pollutant emissions to increase in particular localities. The best defense against this possibility is to tighten existing laws that limit local air pollution. This would prohibit any trades that would violate such limits.

The environmental justice lobby’s concerns about local air pollution are justified: A new report by the U.S. Commission on Civil Rights acknowledges that low-income and minority communities face disproportionately high air pollution. The best response to this situation is to strengthen existing local pollution laws rather than abandon cap and trade.

Moreover, it is not clear that cap and trade shifts local air pollution toward low-income communities. One recent report from the University of Southern California identified emission increases and blamed them on cap and trade. But increased emissions have been due mainly to economic and population growth. And although emissions from some sources did increase, they decreased at 70 percent of facilities, according to mandatory reporting to the Air Resources Board.

The key question, however, is not how emission levels changed, but rather how cap and trade contributed to the change. Without cap and trade, it is likely that any increases in emissions would have been even greater.

Beyond the environmental impacts, it’s important to consider economic impacts on these communities. Reducing greenhouse gas emissions tends to raise costs of energy and transportation. Because low-income households devote greater shares of their income to energy and transportation than high-income households, virtually any climate policy places greater burdens on those households. Cap and trade minimizes these costs.

Further, cap and trade offers the government a powerful tool for compensating low-income communities for such economic burdens. Most emission allowances are auctioned and pursuant to SB 535, 25 percent of the proceeds go to projects that provide benefits to disadvantaged communities. This has already amounted to over $158 million.

Cap and trade serves the goal of environmental justice better than the alternatives, and it deserves a central place in the arsenal of weapons California uses to address climate change. Rather than step away from this progressive policy, the state should increase its reliance on this progressive, market-based approach.

Lawrence H. Goulder is a professor in environmental and resource economics at Stanford University and former chair of the AB 32 Economic and Allocation Advisory Committee to the California Air Resources Board. Contact him at goulder@stanford.edu.

Robert N. Stavins is a professor of business and government at the Harvard Kennedy School of Government, and contributed to assessment reports to the Intergovernmental Panel on Climate Change. Contact him at robert_stavins@harvard.edu.


Market Mechanisms in the Paris Climate Agreement: International Linkage under Article 6.2

The Harvard Project on Climate Agreements hosted a research workshop in Cambridge, Massachusetts, on July 14–15, 2016, the purpose of which was to identify options for elaborating and implementing the Paris Climate Agreement, and to identify policies and institutions that might complement or supplement the United Nations Framework Convention on Climate Change (UNFCCC) process.  We were motivated by our recognition that while the Paris Agreement sets forth an innovative and potentially effective policy architecture for dealing with global climate change, a great deal remains to be done to elaborate the accord, formulate required rules and guidelines, and specify means of implementation.

Participants in the workshop – International Climate Change Policy after Parisincluded twenty-one of the world’s leading researchers focusing on climate-change policy, representing the disciplines of economics, political science, international relations, and legal scholarship. They came from Argentina, Belgium, China, Germany, India, Italy, Norway, the United Kingdom, and the United States.  (A list of workshop participants is here, biographies here, and the agenda here.)

The Harvard Project will next focus on communicating the ideas, insights, and recommendations of workshop participants to climate negotiators and policy makers, in the expectation that they might prove useful in elaborating and implementing the Paris Agreement. Each participant is preparing a brief—based largely on her or his presentation during the workshop. These briefs, together with a workshop summary, will be conveyed to participants in the Twenty-Second Conference of the Parties (COP-22) of the UNFCCC in Marrakech, Morocco in November 2016.  This will be done in meetings with negotiators representing UNFCCC member governments and in a side-event panel at COP-22.

Today I wish to share with readers just one of these draft briefs – namely, my own – on the topic of “International Linkage under Article 6.2 of the Paris Agreement.”

A Key Challenge for Sustained Success of the Paris Agreement

For sustained success of the international climate regime, a key question is whether the Paris Agreement with its Intended Nationally Determined Contributions (INDCs), anchored as they are in domestic political realities, can progressively lead to submissions with sufficient ambition?  Are there ways to enable and facilitate increased ambition over time?

Linkage of regional, national, and sub-national policies can be part of the answer. By “linkage,” I mean connections among policy systems that allow for emission reduction efforts to be redistributed across systems. Such linkage is typically framed as being between two (or more) cap-and-trade systems, but national policies will surely be highly heterogeneous under the Paris climate regime.  Fortunately, research – by Gilbert Metcalf of Tufts University and David Weisbach of the University of Chicago – indicates that linkage between pairings of various types of domestic policy instruments may be feasible.

Linkage and the Paris Agreement

Experience indicates that linkage will bring both merits and concerns in most applications.  To begin with the good news, linkage offers a number of important advantages. First, it offers the possibility of achieving cost savings if marginal abatement costs are heterogeneous across jurisdictions, which they surely are. In addition, linkage can improve the functioning of individual markets by reducing market power, and by reducing price volatility, although we should recognize that price volatility will also be transmitted from one jurisdiction to another by linkage. Finally linkage can allow for the UNFCCC’s important principle of Common but Differentiated Responsibilities (CBDR), but do so without sacrificing cost-effectiveness.

The possibility of linkage also raises concerns, including that there will be distributional impacts within jurisdictions, that is, the creation of both winners and losers. Also, linkage can bring about the automatic propagation from one jurisdiction to another of some design elements, in particular, cost-containment mechanisms, such as banking, borrowing, and price collars. In this and other ways, linkage raises concerns about decreased autonomy.

Linkage under Article 6.2 of the Paris Agreement

It was by no means preordained that the Paris Agreement would allow, let alone encourage, international linkage.  Fortunately, the negotiations which took place in Paris in December, 2015, produced an Agreement that includes in its Article 6.2 the necessary building blocks for linkages to occur.

Under Article 6.2, emissions reductions occurring outside of the geographic jurisdiction of a Party to the Agreement can be counted toward achieving that Party’s Nationally Determined Contribution (NDC) via Internationally Transferred Mitigation Outcomes (ITMOs).  This enables both the formation of “clubs” or other types of coalitions, as well as bottom-up heterogeneous linkage.  Such linkage among Parties to the Agreement would provide for exchanges between compliance entities within the jurisdictions of two different Parties, not simply the government-to-government trading (of Assigned Amounts or AAUs), as was the case with the Kyoto Protocol’s Article 17.

Linkage among Heterogeneous Nationally Determined Contributions

There are three types of heterogeneity which are important in regard to linkage under Article 6.2 of the Paris Agreement. First is heterogeneity among policy instruments. As demonstrated by Metcalf and Weisbach (see above), not only can one cap-and-trade system be linked with another cap-and-trade system, but it is also possible to link a cap-and-trade system with a carbon tax system. In addition, either a cap-and-trade system or a tax system can be linked (via appropriate offsets) with a performance standard in another jurisdiction.  (Linkage with systems employing technology standards are not feasible, however, because such systems are not output-based.)

A second form of heterogeneity that affects linkage and is potentially very important under the Paris Agreement is heterogeneity regarding the level of government action of the relevant jurisdictions. Although the Paris Agreement has as Parties both regional jurisdictions (in the case of the European Union) and national jurisdictions, sub-national jurisdictions are also taking action in some parts of the world. In fact, linkage has already been established between the state of California in the United States and the provinces of Québec and Ontario in Canada.

A third form of relevant heterogeneity is with regards to the NDC targets themselves.  Some are in the form of hard (mass–based) emissions caps, while others are in the form of rate-based emissions caps, either emissions per unit of economic activity, or emissions per unit of output (such as per unit of electricity production). There are also relative mass-based emissions caps in the set of existing NDCs, such as those that are relative to business-as-usual emissions in a specific future year.  Beyond these, there are other parties that have put forward NDCs that do not involve emission caps at all, but rather targets in terms of some other metric, such as the degree of penetration of renewable energy sources.

Combinations of various options under these three forms of heterogeneity yield a considerable variety of types of potential linkages, which may be thought of as the cells of a three-dimensional matrix.  Not all of these cells, however, represent linkages which are feasible, let alone desirable.

The Path Ahead – Key Issues and Questions

There are a substantial number of issues that negotiators will eventually need to address, and likewise, there are a set of questions that researchers (including within the Harvard Project on Climate Agreements) can begin to address now. Among the key issues for negotiators will be the necessity to develop accounting procedures and mechanisms. Also, it will be important to identify means for the ITMOs to be tracked in order to avoid double-counting of emissions reductions. And a broader question is whether and how the UNFCCC Secretariat or some other designated institution will provide any oversight that may be required.

For research, three questions stand out.  First, among pairings from the (3-D matrix) set of instrument–jurisdiction–target combinations that emerge from the three types of heterogeneity identified above, which linkages will actually be feasible?  Second, within this feasible set, are some types of linkages feasible, but not desirable? And third, what accounting treatments and tracking mechanisms will be necessary for these various types of linkages?  Future research will need to focus on these and related questions in order to achieve the potential benefits of Article 6.2 of the Paris Agreement.  Please stay tuned as this work develops.