Carbon taxes and greenhouse gas emissions trading systems: what have we learned?

ABSTRACT

Systematic evidence relating to the performance of carbon pricing – carbon taxes and greenhouse gas (GHG) emissions trading systems (ETSs) – is sparse. In 2015, 17 ETSs were operational in 55 jurisdictions while 18 jurisdictions collected a carbon tax. The papers in this special thematic section review the performance of many of these instruments over the 2005–2015 period. The performance of existing carbon taxes and GHG ETSs can help policy makers make informed choices about whether to introduce these instruments and to improve their design. The purpose of carbon pricing instruments is to reduce GHG emissions cost effectively. Assessing their performance is difficult because emissions are also affected by other policies and exogenous factors such as economic conditions. Carbon taxes in Europe prior to 2008 and in British Columbia reduced emissions from business-as-usual but actual emissions continued to rise. Since 2008 emissions subject to European carbon taxes have declined, but in most countries, other mitigation policies have probably contributed more to the reductions than the carbon taxes. Emissions subject to ETSs, with the exception of four systems without emissions caps, have declined. The ETSs contributed to the emissions reductions, but their share of the overall reduction is not known. Most tax rates are low relative to levels thought to be needed to achieve climate change objectives. Few jurisdictions regularly adjust their tax rates. All ETSs have accumulated surplus allowances and implemented measures to reduce these surpluses. The largest ETSs now specify annual reductions in their emissions cap several years into the future. Emissions trading system allowance prices are generally lower than the tax rates.

Key policy insights

• Theoretical discussions usually portray carbon taxes and GHG ETSs as alternatives. In practice, a jurisdiction often implements both instruments to address emissions by different sources.

• Designs of ETSs have evolved based on experience shared bilaterally and via dedicated institutions.

• Carbon tax designs, in contrast, have hardly evolved and there are no institutions dedicated to sharing experience.

• Every jurisdiction with an ETS and/or carbon tax also has other policies that affect its GHG emissions.

KEYWORDS:

• Greenhouse gases
• carbon taxation
• emissions trading

1. Introduction

Economists argue that putting a price on greenhouse gas (GHG) emissions – carbon pricing – is the most cost-effective regulatory approach to reducing those emissions (High-Level Commission on Carbon Prices, 2017).¹ A price on GHG emissions provides an incentive to implement emission reduction options whose mitigation costs are lower than the price.

Carbon pricing can be implemented using either of two instruments – a carbon tax or a GHG emissions trading system (ETS). With a tax, the government sets the tax rate and specifies the sources subject to the tax. The emission reduction achieved depends upon the response of the affected sources to the imposition of the tax. With an ETS the government sets a limit on GHG emissions by specified sources and distributes allowances approximately equal to the limit.² Allowances are tradable. Each source must surrender allowances equal to its actual emissions to the government. Emissions cannot exceed the limit, but the compliance cost depends on the market price of allowances.

There is an extensive literature on the relative merits of these two instruments (e.g. Partnership for Market Readiness (PMR), 2017; Pollitt, 2016). The debate is waged almost exclusively on theoretical grounds because of the limited information available on the actual performance of these instruments.

The first carbon taxes were implemented in 1990 while the first mandatory GHG ETS became operational in 2002.³ The number of jurisdictions with a carbon tax and/or a GHG ETS has increased steadily in recent years (Métivier, Postic, Alberola, & Vinnakota, 2018). This special thematic section focuses on the actual performance of the instruments in effect at the end of 2015. The focus is on 2005–2015. That provides a common time period with at least some operational data.

This paper provides a synthesis of the papers in the special thematic section. Deng, Li, Pang, and Duan (2018) document the implementation of the seven ETS pilot schemes in China. Wakabayashi and Kimura (2018) review the performance of the Tokyo Metropolitan ETS. Narassimhan, Gallagher, Koester, and Alejo (2018) assess other ETSs and Mascher (2018) reviews the carbon taxes, ETSs and hybrid systems implemented by Canadian provinces.⁴

This paper also draws on other studies for lessons learnt from the implementation of carbon taxation and emissions trading, but does not claim to provide a comprehensive review of the extensive literature that exists on carbon pricing.

The theoretical design of each instrument is summarized in the next section. The instruments implemented differ from the theoretical design in ways described in section 3. An overview of the taxes and ETSs in operation at the end of 2015 is provided in Section 4. Assessment criteria are discussed in Section 5. Evidence relating to the performance of carbon taxes and GHG ETSs is presented in Sections 6 and 7, respectively. Section 8 concludes with lessons learned from the performance of these instruments and suggestions for further research.

2. Theoretical design

The theoretical design of each instrument is simple. Actual designs, as discussed in the next section, are more complex.

The theory of a carbon tax is that the government designates the emissions sources subject to the tax and sets the tax rate per unit of emissions (PMR, 2017). Economic efficiency is improved by increasing the share of the jurisdiction’s total GHG emissions subject to the tax. GHGs other than CO₂ can be expressed in tonnes of CO₂ equivalents (tCO₂e) to increase coverage.⁵ Inevitably some emissions are excluded because they are too difficult or costly to measure, the quantity is small or for other reasons.

The tax rate can be set in various ways (PMR, 2017). One option is to set the tax rate equal to the estimated benefit of reducing GHG emissions by 1 tonne CO₂, the so-called social cost of carbon. Another option is to set the tax rate at a level that yields a target emission reduction based on economic modelling. Other options are to set that tax rate at a level similar to that of other jurisdictions or at a level that generates a desired amount of tax revenue. Regardless of the level selected, the tax rate may be phased in over time.

To maintain its effectiveness, or generate further emission reductions, the tax rate needs to be adjusted regularly for the effects of inflation, increases in real income, technological change and other factors such as changes in fossil fuel prices. Most GHG emission reduction measures involve an investment. Tax rates specified several years into the future facilitate those investment decisions.

The most prevalent form of ETS, a ‘cap-and-trade’ system, imposes a government-established limit on aggregate GHG emissions by specified sources, distributes tradable allowances (usually 1 tCO₂e each) approximately equal to the limit and requires regulated emitters to surrender allowances equal to their actual emissions. Allowances are distributed free based on allocation rules and by auction. The number of allowances distributed may be less than the established limit because some allowances may be withheld for purposes such as allocations to new entrants or because they are not purchased at allowance auctions. Allowances in excess of the cap may be distributed as transitional mechanisms or to limit price increases when allowance prices exceed specified thresholds.

A less common form of ETS, a ‘baseline-and-credit’ system, specifies an emissions limit for each participant and allocates tradable credits to emitters whose actual emissions are less than their limit.⁶ Emitters that exceed their limit can purchase credits to cover the excess emissions from other participants. The term ‘allowance’ will be used to include the credits in such systems.

Under both forms, the market determines the price for allowances and thus the price per tCO₂e emitted. An ETS must be large enough to create a competitive market for allowances; that means numerous participants with none large enough to dominate the market.

In a cap and trade system, the government-established aggregate cap is usually a reduction from the recent emissions level or trend for the sources covered. In a baseline and credit system, the emissions limit for each participant can be absolute or based on production and an emissions intensity.⁷ In effect, the emissions limit is a free allowance allocation to the participant and the aggregate cap is the sum of the participants’ emissions limits.⁸ As with a tax, economic efficiency is improved by increasing the share of total GHG emissions covered by the ETS, but inevitably some emissions are excluded. To generate further reductions, the aggregate cap must decline over time.⁹

Combustion of fossil fuels – coal, oil and natural gas – is a major source of GHG emissions. The emissions can be determined accurately from the carbon content of the fuel, which is well defined for each fuel. This means that the emissions due to combustion of fossil fuels can be regulated based on the carbon content of each fuel as it enters the jurisdiction’s economy rather than at the point of combustion. That reduces the number of entities covered by the tax or ETS dramatically.

A tax and an ETS both generate revenue; the latter mainly from auctioned allowances.¹⁰ The revenue is generally used for one or more of the following purposes: financing mitigation and adaptation measures, addressing inequitable impacts of the instrument on low income groups, providing rebates to energy-intensive, trade-exposed (EITE) firms, reducing existing taxes and general revenues (Carl & Fedor, 2016; Métivier et al., 2018). Commitments relating to the use of the revenue may be important for gaining public support for the instrument. When evaluating the impacts of a tax or ETS, some analyses include the use of the revenue, while others ignore the revenue-related impacts.

GHGs have some properties that can affect the implementation of a tax or ETS. Most GHGs have long atmospheric residence times – 100 years or more for many of the gases – and are ‘stock’ pollutants – meaning that the impacts depend on the cumulative stock in the atmosphere rather than current emissions (Heal, 2017). That means current emissions have global impacts for 100 years or more, which complicates estimation of the social cost of carbon. And to stabilize the atmospheric stock of GHGs, global emissions must be reduced to virtually zero (IPCC, 2014).

Under highly restrictive assumptions, a carbon tax and an ETS yield identical results for equivalent emission reductions (Weisbach, 2010; Weitzman, 1974). These assumptions include perfect foresight (no uncertainty relating to the future), perfect competition in all markets, no interaction with other policies and universal coverage (all sources of GHG emissions). In practice, the assumptions needed for theoretical equivalence do not hold; so the real-world performance of a carbon tax and an ETS is likely to differ.

3. Practical designs

In a given jurisdiction, various factors, including constitutional provisions, international commitments, the emissions profile, the ability to create a competitive allowance market, compatibility with the instruments of neighbouring jurisdictions, the influence of large emitters, public opinion and the government’s need to raise revenue, can affect the choice and design of an instrument. Thus, the designs of carbon taxes and GHG ETSs implemented vary across jurisdictions.

All jurisdictions that implement a tax or GHG ETS include provisions to protect EITE firms against adverse economic impacts from competitors in jurisdictions with no, or less costly, GHG emission regulations. Failure to protect EITE firms may lead to lower output and loss of employment in the jurisdiction with the tax or ETS and higher emissions elsewhere – a problem known as leakage (Branger & Quirion, 2014; Zhang & Zhang, 2016). In the case of a tax, protection takes the form of tax exemption, a lower tax rate or rebates. In the case of an ETS, it takes the form of free allowances or rebates.

Every jurisdiction with a tax or ETS also has other policies that directly or indirectly affect the GHG emissions of sources subject to the instrument. Some policies, such as incentives for renewable energy and energy efficiency standards for appliances, buildings, equipment and vehicles, have a direct impact on GHG emissions. Other policies, such as regulations to limit SO₂ emissions by coal-fired generating stations and revenue-generating taxes on energy, have incidental effects on GHG emissions. Other policies can distort the impacts of an ETS (Schmalensee & Stavins, 2017) and make it difficult to assess the effectiveness of a carbon pricing instrument.

Every carbon tax and GHG ETS excludes some emissions sources. Some jurisdictions, almost exclusively ETS jurisdictions, allow specific excluded sources to earn credits for verified emission reductions – for example, reduced emissions due to changes in agricultural practices or carbon sequestration through forestry. Such credits can be used for compliance by sources subject to the ETS, but typically with quantitative limits; less than 10% of the compliance obligation for example. Eligible credits may be restricted to domestic sources or include those from other jurisdictions.

It can be difficult to distinguish a carbon tax from other taxes on fossil fuels, such as excise taxes.¹¹ Almost all jurisdictions impose some taxes on fossil fuels. Since the carbon content of each fuel is well specified, any tax on a fossil fuel can be expressed as a tax per tonne of CO₂.¹² A carbon tax typically imposes the same tax per tonne of CO₂ on multiple fuels and/or emitting activities.

The instruments implemented are sometimes hybrids, incorporating features of a tax and an ETS. For example, an ETS with an allowance price equal to the specified minimum or maximum is similar to a carbon tax.

A GHG ETS raises several design issues in addition to the level of the cap. Allowances are distributed using a mix of free allocation and auctions. With experience the share of allowances auctioned has increased, free allocations tend to be limited to EITE participants and those free allocations tend to be based on performance standards and actual output rather than historic emissions.¹³ To encourage early mitigation action, almost all ETSs allow surplus allowances to be banked for future compliance use. To assuage fears of very high allowance prices, most ETSs incorporate price stability provisions. These can include increased use of credits and release of additional allowances at specified levels up to a price ceiling.

4. Existing carbon taxes and GHG ETSs

The carbon taxes and GHG ETSs operational as of the end of 2015 listed in Table 1 are the focus of the analysis here because data on the performance of more recently implemented instruments are not available.¹⁴ Most of these instruments are discussed in the papers of this special thematic section (Deng et al., 2018; Mascher, 2018; Narassimhan et al., 2018; Wakabayashi & Kimura, 2018). Seventeen ETSs were operational in 55 jurisdictions – 34 national and 21 sub-national – while 17 national and 1 sub-national jurisdiction collected a carbon tax.¹⁵ Thus, ETSs are more common, even though they are pre-dated by several carbon taxes.

Table 1. Carbon taxes and GHG emissions trading systems in operation at the end of 2015.

Carbon taxes					GHG emissions trading systems
Jurisdiction	Year launched	Price August 2017 US$/tCO2	Share of jurisdiction’s emissions covered (%)	Emissions covered by the instrument MtCO2e	Jurisdiction	Year launched	Price August 2017 US$/tCO2	Share of jurisdiction’s emissions covered (%)	Emissions covered by the instrument MtCO2e
Finland	1990	69–73	36	21	European Union	2005	6	45	1939
Poland	1990	<1	4	15	Alberta	2007	24	45	123
Norway	1991	4–56	60	32	New Zealand	2008	13	51	41
Sweden	1991	140	42	23	Switzerland	2008	7	11	17
Denmark	1992	27	45	22	RGGI (USA)	2009	4	21	86
Slovenia	1996	20	24	2	Tokyo	2010	14	20	10
Estonia	2000	2	3	4	Saitama	2011	14	18	7
Latvia	2004	5	15	1	California	2013	15	85	375
British Columbia	2008	24	70	43	Quebec	2013	15	85	68
Liechtenstein	2008	87	26	0	Beijing	2013	8	45	6
Switzerland	2008	87	35	5	Guangdong	2013	2	60	388
Ireland	2010	24	33	20	Shanghai	2013	5	57	170
Iceland	2010	12	55	3	Shenzhen	2013	6	40	30
Japan	2012	3	70	926	Tianjin	2013	1	55	100
United Kingdom	2013	24	25	127	Chongqing	2014	<1	40	126
France	2014	36	40	185	Hubei	2014	2	35	324
Mexico	2014	1–3	46	332	South Korea	2015	18	68	470
Portugal	2015	8	26	18
Average		$12.09	45.54%	Total	Average		$7.86	48.10%	Total
Range		<1 to 140	3–70%	1778	Range		<1 to 24	18–85%	4280

Sources: World Bank, Ecofys and Vivid Economics (2017), Métivier et al. (2017), UNFCCC Greenhouse Gas Inventory Data (2015), ICAP (2017) and Environment and Climate Change Canada (2017).

Notes: August 2017 prices are from World Bank, Ecofys and Vivid Economics (2017). Shares of emissions covered are from Métivier, Postic, Alberola, and Vinnakota (2017). For the EU and national jurisdictions, the emissions covered by the instrument are calculated from national emissions excluding LULUCF for 2015 reported by the UNFCCC Greenhouse Gas Inventory and the share of emissions covered by the instrument. For sub-national GHG ETSs, the emissions covered are from the ICAP fact sheets for the respective systems. For the BC carbon tax, the emissions covered are calculated from the BC emissions for 2015 excluding LULUCF reported by Environment and Climate Change Canada and the share of emissions covered by the tax.

Of the 6058 million tCO₂e (MtCO₂e) annual emissions covered by a carbon price, ETSs cover 4280 MtCO₂e (71%) while carbon taxes cover 1778 MtCO₂e (29%). The number of jurisdictions with a carbon pricing instrument has doubled over the past decade. Including instruments implemented after 2015, they cover 20–25% of global GHG emissions (Métivier et al., 2018). The average share of a jurisdiction’s emissions covered by an ETS (48.1%) is a little higher than the average for a carbon tax (45.5%).¹⁶ The share of emissions covered ranges from 18% to 85% for ETSs and from 3% to 70% for carbon taxes.

As shown in Table 1, as of August 2017, the average tax rate (US$ 12.10) is over 50% higher than the average ETS allowance price (US$ 7.86).¹⁷ The tax rates range from less than US$ 1 to US$ 140 tCO₂e while ETS prices range from less than US$ 1 to US$ 24 tCO₂e. The revenue generated by these instruments in 2013 was almost US$ 20 billion; US$ 13 billion by carbon taxes and US$ 6.5 billion by ETSs (Carl & Fedor, 2016).¹⁸ By 2017 it had increased to US$ 32 billion; US$ 21 billion by taxes and US$ 11 billion by ETSs (Métivier et al., 2018). Use of the revenue varies widely by jurisdiction. Overall, tax revenue is used for ‘green’ spending (14%), general revenue (46%) and tax cuts and direct rebates (38%) (Carl & Fedor, 2016).¹⁹ The corresponding shares for ETS auction revenue are 70% green spending, 21% general revenue and 9% tax cuts and rebates.

It is interesting to note that apart from Japan and Mexico all of the jurisdictions with a carbon tax also have an ETS, although they usually cover different emissions sources. Thirteen of the countries with a carbon tax participate in the EU ETS. Switzerland and Alberta also have both a tax and an ETS.²⁰ British Columbia has established a trading system to complement its carbon tax, but currently, there are no emitters subject to the trading system (Mascher, 2018).

5. Assessing performance

The purpose of a carbon tax or GHG ETS is to reduce GHG emissions cost effectively. Thus, these instruments should be assessed on the basis of the emission reductions achieved and the cost per tCO₂e reduced. Tax rates and allowance prices are readily available as shown in Table 1, but the impact on emissions must be estimated.

Other aspects of the performance of these instruments include low price volatility; price commitment into the future; harmonization of marginal costs across jurisdictions; revenue raised; potential cross-jurisdiction revenue flows and administrative issues (Haites et al., in press). In addition, a jurisdiction implementing a carbon tax or GHG ETS may be concerned about, inter alia, employment, income distribution, regional impacts, economic growth and technology development. While those are interesting and important issues, the primary focus here is on emission reductions achieved.

Virtually all published estimates of emission reductions achieved by carbon taxes and GHG ETSs calculate the reduction from business-as-usual emissions projected using econometric methods, models or control groups. Although ex post projections of business-as-usual emissions can incorporate actual weather, fuel prices, economic activity and other factors, they remain projections; so the calculated reductions are only estimates of the actual reductions.

A more stringent criterion proposed by one study is a reduction from the actual emissions during the first year the instrument was in force (Haites et al., in press). The rationale is that most jurisdictions that have implemented a carbon tax and/or GHG ETS have an emissions target that requires a reduction of actual emissions.²¹ Since emissions subject to the tax or ETS are verified, the actual emissions during the first year are more reliable baseline than an estimate of business-as-usual emissions.

Implementation of energy efficiency measures, switching to less emissions-intensive fuels and shifting to less emissions-intensive products are expected and desired responses to a carbon tax or GHG ETS. Other policies, such as energy efficiency programmes, also promote such adjustments perhaps differentially across emissions sources subject to the pricing instrument. Some policies, such as fossil fuel subsidies, may inhibit the desired adjustments by some sources. Thus, it is often difficult to disentangle the impact of the carbon pricing instrument from the effects of other policies.

6. The performance of carbon taxes

Estimates of the emissions reductions achieved by carbon taxes are available for the BC carbon tax and European carbon taxes.

At the time of writing, seven estimates of the impact of the BC carbon tax had been published.²² All relate to the 2007–2012 period when the tax rate was rising at C$ 5/tCO₂ (about US$ 4.75 in 2008) per year. The effect of the tax was to reduce fuel consumption and GHG emissions 5 to 15% from business-as-usual (Murray & Rivers, 2015).²³ Actual emissions subject to the tax increased during this period and continued to increase during the next five years while the tax rate was constant at C$ 30/tCO₂ (Haites et al., in press).

At the time of writing, five main studies had estimated the emissions reductions achieved by carbon taxes in one or more European countries for different periods prior to 2008 (Andersen, 2010; Andersson, 2017; Bruvoll & Larsen, 2004; Lin & Li, 2011; Mideksa & Kallbekken, undated). The studies differ in terms of the countries considered to have a carbon tax;²⁴ all of them include Denmark, Finland, Norway and Sweden while some studies include Austria, Germany, Italy, Netherlands, Slovenia and the United Kingdom and others do not.²⁵ The studies find that the carbon taxes yielded only small reductions – up to 6.5% over several years – from business-as-usual emissions. The GHG emissions of those countries continued to rise. In a few countries, the introduction of the carbon tax was accompanied by a reduction of existing energy taxes. The small impact reflects extensive tax exemptions/reductions for EITE sources, the low incremental tax burden on fuels and relatively inelastic demand for fuels in the sectors where the tax was implemented.²⁶

Fourteen countries that participate in the EU ETS have a carbon tax while 16 do not.²⁷ Since the carbon taxes are applied almost exclusively to non-ETS emissions, changes in non-ETS emissions can be compared for countries with and without a tax.²⁸ Between 2008 and 2015, non-ETS emissions declined more rapidly in countries without a carbon tax (−2.59%/year) than in those with a carbon tax (−2.08%/per year) (Haites et al., in press). This suggests other policies may have contributed more than carbon taxes to reducing non-ETS emissions. Data on taxed emissions are available for seven countries. Only in Denmark and Switzerland have taxed emissions declined more rapidly than non-ETS emissions.²⁹ In short, actual emissions subject to carbon taxes have declined in European countries since 2008, but with a few possible exceptions, much of the decline is likely due to non-tax policies.

Why have carbon taxes not been more effective? There are at least two reasons: the tax rates are too low and tax rate changes are uncertain.

Tax rates are low relative to the social cost of carbon and relative to the prices of the taxed fuels. Estimates of the social cost of carbon vary widely for a variety of theoretical and practical reasons (Smith & Braathen, 2015). Only the tax rates in Finland, Liechtenstein, Sweden and Switzerland exceed the carbon price thought to be needed in 2020 to hold the global average temperature increase to well below 2°C above pre-industrial levels, in line with the Paris Agreement (High-Level Commission on Carbon Prices, 2017; World Bank, Ecofys and Vivid Economics, 2017). In most European countries, the carbon taxes have been less than 15% of the retail price of diesel fuel and have been less than fluctuations in the retail price; so the tax may be lost in the ‘noise’ of price changes (Haites et al., in press).

Apart from periods of three to five years when a tax is introduced, few jurisdictions regularly adjust their tax rates (Haites et al., in press). Carbon taxes are applied mainly to transportation fuels and fossil fuels used for residential, commercial and institutional heating. The taxes are intended to provide an incentive to reduce consumption and adopt less emissions-intensive technologies. But vehicles and equipment have lives of 10–20 years; so the adjustment is gradual.³⁰ Regular adjustments to the tax rate are needed to sustain the incentive to invest in lower-emitting technologies.³¹

7. The performance of GHG ETSs

Data on aggregate emissions are available for many GHG ETSs. The exceptions are the seven Chinese pilot programmes that do not disclose emissions data³² and the Korean system where results for the first compliance period are not yet available (Narassimhan et al., 2018). Québec reports compliance with the cap for its 2013–2014 compliance period. Data on emissions reductions during its 2015–2017 compliance period, which features significantly broader coverage, are not available.

California achieved annual emission reductions of −4.81% during 2013–2014 and −2.79% for the first two years of its 2015–2017 compliance period (Haites et al., in press).³³ To meet the state’s emission reduction target, a number of policies, including a renewable portfolio standard, low carbon fuel standard and energy efficiency measures, were implemented in conjunction with the ETS. Together those policies cover about 80% of the emissions subject to the second phase of the ETS. The ETS was intended to be a ‘backstop’ in case the other policies did not achieve the anticipated reductions (Bang, Victor, & Andresen, 2017). Estimates of the share of the reductions attributable to the ETS are not available.

The Tokyo and Saitama ETSs claim emission reductions, but actual emissions may have increased. The ETSs assume a constant carbon intensity for electricity which accounts for a large share of the emissions covered. The great east Japan earthquake and tsunami of 2011 led to the gradual shutdown of all nuclear generating stations and greater reliance on fossil-fired generation. The effect of the increased carbon intensity of electricity on actual emissions cannot be calculated.³⁴

The New Zealand ETS has achieved only minimal reductions from business-as-usual emissions due to regulatory uncertainty and the ability to use unlimited quantities of low cost imported units for compliance (Ministry for the Environment, 2016). Emissions under Alberta’s Specified Gas Emitters Regulation increased between 2007 and 2015 due, in part, to higher output by participants and an increase in the number of installations subject to the regulation (Haites et al., in press).

Actual emissions by electric power plants covered by the Regional Greenhouse Gas Initiative (RGGI) in the USA have declined faster than the emissions cap with the consequence that about 40% of the allowances available at auctions between June 2012 and December 2012 were not sold at the specified minimum price (Ramseur, 2017). Lower natural gas prices, other mitigation policies and the 2008–2009 recession contributed to the reductions. Among the other mitigation policies are energy efficiency and renewable energy initiatives funded by RGGI allowance auction revenue.³⁵ The ETS is estimated to be responsible for about half of the emission reductions by fossil-fired generating plants in RGGI states between 2009 and 2012 (Murray & Maniloff, 2015). Part of the reduction may have been offset by higher emissions in neighbouring states (Fell & Maniloff, 2018).

Despite an abundant supply of, and consequent low prices for, allowances, emissions by sources subject to the Swiss ETS have declined (Zurich University of Applied Sciences, School of Management and Law, 2017). The situation is aggravated by uncertainties related to banking of surplus allowances and the possible link with the EU ETS (now agreed).

It is not surprising that the EU ETS, the oldest and largest GHG ETS, has been the subject of numerous evaluations (e.g. Fujiwara et al., 2017).³⁶ The studies use a variety of methodologies and span different periods. Some apply to the ETS as a whole, while others are limited to a single country. The studies indicate that the ETS reduced emissions by a total of 130–247 MtCO₂ relative to business-as-usual during its first (2005–2007) phase (Anderson & Di Maria, 2011; Ellerman & Buchner, 2008; Ellerman, Convery, & de Perthuis, 2010). The reductions appear to have been achieved mainly through fuel switching and efficiency measures in the power sector (about 60% of the emissions covered) with limited contributions by the industrial sectors.

Emission reductions during the second (2008–2012) phase were larger with industrial facilities contributing to the results. The impacts of other policies and economic developments make it very difficult to estimate the share of the reductions attributable to the ETS during this period. In 2008 the EU implemented policies to achieve a 20% share of total energy consumption from renewable energy and a 20% improvement in energy efficiency by 2020. The recession in 2008–2009 and changes to the relative prices of coal and natural gas also affected emissions.

The only attempt to estimate the share of the emission reductions attributable to the ETS is inconclusive (Gloaguen & Alberola, 2013).³⁷ Analyses that compare emissions of ETS participants with those of smaller plants in the same country not covered by the ETS, suggest the ETS contributed to the 6% emission reduction achieved over the 2008–2012 period.³⁸ The renewable energy policies implemented by EU member states reduced power sector emissions thus reducing the price of allowances and increasing the supply available to industrial sources.

In summary, actual emissions have declined in the jurisdictions that have ETSs for which data are available with the exception of Alberta and New Zealand, which do not cap emissions, and possible small increases in Tokyo and Saitama due to the nuclear plant shutdowns. However, only rough estimates are available of the respective contributions of the ETS, the 2008–2009 recession, changes in relative fuel prices and other emission reduction policies.

In every ETS, emissions have fallen faster than the cap, leading to the accumulation of a bank of surplus allowances. The banks range in size from 15% to over 500% of the annual compliance obligation (Haites et al., in press). Large banks depress allowance prices. To reduce the sizes of their banks, most ETSs have cancelled allowances and restricted the use of offset credits and banked allowances. As well, many ETSs now have declining annual emissions caps and market stability reserves. These reserves withdraw allowances from the market when the supply is abundant and release allowances if shortages arise. They are relatively new features and their effectiveness has yet to be demonstrated.

8. Lessons learned and further research

Carbon taxes and GHG ETSs have become increasingly popular as policies to regulate GHG emissions and they now cover over 20% of global emissions. ETSs have expanded more rapidly than carbon taxes. The choice of instrument and its design are affected by circumstances specific to the jurisdiction, including its emissions profile, constitutional provisions and political considerations. Many of the carbon taxes are in European countries that also have an ETS generally covering different emissions sources.

Carbon taxes and GHG ETSs have both demonstrated that they can reduce emissions from business-as-usual. Reductions in actual emissions subject to a tax or ETS are less common and are attributable, at least in part, to the impacts of other mitigation policies and exogenous factors such as changes in fossil fuel prices and economic conditions.

The BC carbon tax and European carbon taxes prior to 2008 reduced emissions from business-as-usual but actual emissions continued to rise. After 2008 emissions subject to European carbon taxes have declined, but in most countries, other mitigation policies probably contribute more to the reductions than the carbon taxes.

Emissions subject to ETSs, with the exception of the four systems without emissions caps – Alberta, New Zealand, Saitama and Tokyo – have declined. While there is evidence that the ETS has contributed to the emissions reductions, the share that can be attributed to the ETS as compared to other mitigation policies and external factors is not clear.

The emission reductions by ETSs are achieved with allowance prices that are generally lower than the tax rates. Almost all current tax rates are low relative to the social cost of carbon and as a share of fuel prices. Other mitigation policies and external factors contribute to the difference between tax rates and allowance prices. They do not affect tax rates, but they reduce emissions subject to ETSs, thus leading to allowance surpluses and lower allowance prices.

The smaller emission reductions with higher marginal costs for carbon taxes may reflect the sources to which carbon taxes and GHG ETSs are applied. Carbon taxes tend to apply to fossil fuels with limited coverage of energy and industrial sources while the opposite is true for ETSs (Métivier et al., 2018). The low price elasticity of demand for fossil fuels means that reduction of the associated emissions may require higher carbon prices and take longer than energy and industrial emission reductions.

Uncertainty relating to the future of the instrument or the carbon price reduces the effectiveness of these instruments. Most mitigation actions involve an investment. A predictable price signal several years into the future facilitates those investment decisions. Several ETSs now specify annual reductions to the emissions cap, and where applicable to the floor price, for periods of three to eight years in legislation or regulations. Apart from a three to five year period when a carbon tax is introduced, governments rarely specify future tax rates. Thus, ETSs now often create a more stable investment environment than carbon taxes.

Designs of ETSs have evolved based on experience while there has been virtually no change to the design of carbon taxes. ETSs have increased the share of allowances auctioned, adopted declining emissions caps, specified future caps and floor prices, shifted to benchmarking for free EITE allowances, reduced the accessibility to foreign offset credits and established market stability reserves. Experience has been shared bilaterally and via dedicated institutions including the International Carbon Action Partnership (ICAP), Western Climate Initiative and Partnership for Market Readiness (PMR) (Wettestad & Gulbrandsen, 2017). In contrast, carbon tax jurisdictions still struggle with setting and adjusting the tax rate and there is little evidence of shared learning based on experience.

Experience suggests that it may be more appropriate to view carbon taxes and GHG ETSs as components of a portfolio of mitigation policies rather than as alternative ‘first best’ policies. Virtually every jurisdiction with a tax also has an ETS and every jurisdiction with a tax or ETS also has other policies to reduce emissions by sources subject to the tax/ETS. There are sound theoretical and practical reasons for using multiple price and non-price instruments to reduce GHG emissions.³⁹ But the use of multiple instruments increases the compliance cost and creates complex interactive and distributional effects. Research into the composition of effective policy portfolios to reduce GHG emissions is needed.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 Carbon dioxide (CO₂) is the most common of the greenhouse gases.

2 As discussed below an ETS may have a ‘cap and trade’ or a ‘baseline and credit’ design.

3 The New South Wales (NSW) Greenhouse Gas Reduction Scheme which was replaced in 2012 by Australia’s carbon pricing scheme. The EU ETS, launched in 2005, is the oldest system still operating.

4 Subsequent to the publication of this paper, a new government in Ontario has announced its intention to terminate the provincial emissions trading system.

5 The Intergovernmental Panel on Climate Change (IPCC) reports the global warming potentials (GWPs) of greenhouse gases over different timeframes. The 100 year GWPs are generally used to calculate CO₂ equivalence for policy purposes.

6 The Alberta and Tokyo ETSs use this design. See Mascher (2018) and Wakabayashi and Kimura (2018), respectively.

7 Emission limits and intensities for participants can be based on historic levels or trends or performance standards.

8 The number of designated sources changes over time due to new entrants and closures. In a baseline and credit system, a change in the number of entrants changes the aggregate limit up or down. In a cap and trade system, a new entrant may get free allowances from a new entrant reserve that is part of the cap while closures typically do not lower the cap.

9 To reduce aggregate emissions in a baseline and credit system, participant limits must be reduced enough to offset any growth in the number and output of participants.

10 Some systems, including Alberta and New Zealand, have a fixed price compliance option that generates revenue.

11 E.g. India is identified as having a carbon tax by PMR (2017) but not by World Bank, Ecofys and Vivid Economics (2017).

12 OECD (2018) compares taxes on fuels in terms of €/tCO₂.

13 ETSs differ in terms of their operational definition of an EITE firm. The performance standards or baselines typically are set so as to create an incentive to reduce emissions.

14 Métivier et al. (2018) provides an overview of existing and planned carbon taxes and GHG ETSs as of the end of 2017.

15 The EU ETS covers 31 national jurisdictions and RGGI covers 9 American states.

16 These are weighted averages; the share for each jurisdiction is weighted by the emissions covered by the instrument in that jurisdiction. The average tax rates and allowance prices are calculated in the same way.

17 The price reported in Table 1 for Alberta is the fixed price compliance option. Credits trade at lower prices.

18 These amounts exclude Australia’s carbon pricing mechanism which was subsequently cancelled.

19 These shares do not include Australia’s carbon pricing mechanism. The tax shares do not sum to 100 because the UK components do not equal the total.

20 See Mascher (2018) for a description of Alberta’s carbon levy and revised intensity-based trading system.

21 For example, British Columbia has a target of a 33% reduction from 2007 emissions by 2020. Quebec has a target of a 20% reduction from 1990 GHG levels by 2020. The EU target is a 14% reduction from 2005 emissions by 2020 with the ETS achieving a 21% reduction and non-ETS sectors a 10% reduction. Tokyo has a target of a 25% reduction from 2000 GHG levels by 2020. RGGI states have a regional target of a more than 50% reduction of CO2 emissions from electricity generation from 2005 levels by 2020.

22 Murray and Rivers (2015) presents six estimates plus Lawley and Thivierge (2018).

23 Lawley and Thivierge (2018) has a similar result.

24 The fossil fuel levy in the UK was imposed on the market value of fossil fuels, so the tax rate per tCO₂e varied by fuel. Some analysts consider it a carbon tax while others do not because the tax rate per tCO₂ e varies.

25 Andersson (2017) (Sweden) and Bruvoll and Larsen (2004) (Norway) assess only a single country while the others cover multiple countries. Differences in the countries considered to have a carbon tax lead to differences in the countries included in the control group. Austria and Italy, for example, are considered by one study to have a carbon tax while other studies include them in the control group.

26 Speck (2008) discusses changes in the tax rates and coverage in Denmark, Germany, Sweden and the UK prior to 2008. He describes the German tax as an energy tax rather than a carbon tax. The Fossil Fuel Levy in the UK was an ad valorem tax so the tax per tCO₂e differed by fuel.

27 Denmark, Estonia, Finland, France, Iceland, Ireland, Latvia, Liechtenstein, Norway, Poland, Portugal, Slovenia, Sweden, United Kingdom, Iceland, Liechtenstein and Norway are not part of the EU. Croatia is not included because it did not join the EU ETS until 2013.

28 Although Switzerland has a tax and an ETS, it is excluded because it is not part of the EU ETS.

29 The change in the scope of the tax and ETS in 2013 might distort the result for Switzerland.

30 The price elasticity of demand for fossil fuel uses is highly inelastic in the short run and elastic in the long run (Labandeira, Labeaga, & López-Otero, 2017).

31 Energy efficiency measures also are affected by a variety of market failures that inhibit their response to price signals (Gerarden, Newell, & Stavins, 2017).

32 Deng et al. (2018) assess the performance of these pilot ETSs.

33 Relative to emissions during the year prior to each phase.

34 Entity-specific emissions data for 985 facilities covered by the Tokyo ETS from 2010 through 2015 suggest an increase of 0.35% per year in actual emissions due to the change in the generation mix (see Wakabayashi & Kimura, 2018).

35 About half of the auction is used to fund such programmes (Ramseur, 2017).

36 See Haites et al. (in press) for a summary.

37 The study attributes a maximum of 10% of the reductions to the carbon price, but the ETS may also impact the energy efficiency variable, which accounts for a further 10–20% of the reductions.

38 See Wagner, Muûls, Martin, and Colmer (2014) for France; Petrick and Wagner (2014) for Germany and Klemetsen, Rosendahl, and Jakobsen (2016) for Norway.

39 The ‘Tinbergen Rule’ that each policy objective requires at least one policy instrument (Tinbergen, 1952). Additional policies also are justified if the price established by the tax or ETS is too low (Hoel, 2012). See also the High-Level Commission on Carbon Prices (2017).