99
ISBN: 978-92-64-05606-0
The Economics of Climate Change Mitigation
Policies and Options for Global Action beyond 2012
© OECD 2009
Chapter 4
Towards Global Carbon Pricing
This chapter examines ways in which a global carbon price can be built up gradually
to achieve broad-based international pricing of carbon. Important steps include
removal of environmentally harmful fossil fuel energy subsidies and increasing the
use of emissions trading schemes while linking them together. The chapter
investigates the global and regional gains from linking regional emissions trading
schemes, and the harmonisation issues that need to be addressed in the case of direct
linking. It examines indirect linking, for instance through the Clean Development
Mechanism (CDM) or possible sectoral crediting approaches. It concludes with a
discussion of market regulatory issues and the role of financial markets.4.  TOWARDS GLOBAL CARBON PRICING
100   THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Key Messages
• The use of domestic/regional emission  trading schemes (ETSs) is spreading rapidly
internationally. Together with international crediting mechanisms, these are important elements
in the gradual build-up of a global carbon price. Global carbon pricing could help to reduce
mitigation costs overall, and might also  potentially reduce carbon leakage and limit
competiveness concerns.  
• Removing environmentally harmful fossil fuel energy subsidies, especially in non-OECD countries
is an important first step. This would reduce greenhouse gas (GHG) emissions drastically in the
subsidised countries, in some cases by over 30% relative to business-as-usual (BAU) levels by
2050 and it would also raise GDP per capita in most of the countries concerned. A multilateral
removal of energy subsidies would cut GHG emissions globally by 10% by 2050 relative to BAU
and this cut could be increased if developed countries adopt binding emission caps. The removal
of energy subsidies would lower the cost of achieving a given mitigation target.  
• Directly linking domestic or regional ETSs could form a key building block of a global carbon market
by helping a single international carbon price to emerge. Linking reduces the cost of achieving the
joint target and increases carbon markets liquidity. However, linking also raises a number of concerns
including i) the uneven distribution of the gains from linking across countries, ii) the spreading of some
design features of a particular scheme to others and iii) the risk of a dilution of the environmental
integrity of the linked system. The distributional effects of linking can be mitigated through permit
allocation rules and possibly by allocating commitments across countries conditionally on expected
growth or by the use of an intensity target. While the other two concerns could be addressed by
imposing some limits on linking, a more cost-effective approach would be to reach an agreement on
the harmonisation of key ETS design features prior to linking. Such harmonisation would also increase
the liquidity of carbon markets, which would foster the development of derivative markets, thereby
lowering the cost of insurance against carbon price uncertainty.
• Linking domestic or regional ETSs indirectly through a common crediting mechanism like the
CDM would help build up an integrated world carbon market and lower mitigation costs.
Modelling analysis suggests that allowing Annex I (industrialised) regions to meet 20% of their
commitments through emissions reductions in non-Annex I (developing) countries would nearly
halve their mitigation costs. Raising the cap from  20% to 50% would bring further benefits,
especially for those Annex I regions with high  marginal abatement costs and which are most
carbon-intensive. However, in order for these potential gains to be reaped, the current CDM
would have to be carefully reformed.
• Even very large emission reductions in developed countries would not, on their own, suffice to
halt climate change. Sectoral approaches can help to broaden participation to developing
countries. The most effective approach would be to focus on the largest emitting sectors, such as
energy intensive industries and the power sector, and/or key countries. Sectoral approaches would
expand the potential for lower-cost emission reductions and could reduce leakage and
competiveness issues, but they would need to be  based on ambitious sectoral baselines or to
include emission caps in order to ensure net additional emissions reductions.  
• Negotiation and consensus building should be placed at the core of the development of the carbon
market. Inter-governmental institutions that support implementation of the UNFCCC, the Kyoto
Protocol, and others could  help provide a framework in which participating governments can
harmonise and co-ordinate their targets and the design features of emission trading schemes prior
to linking. Compliance mechanisms at the national, regional or international level will also be
needed. A working group of regulators could facilitate exchange of information about regulations
and risks associated with the development of spot and but also derivative carbon markets. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009   101
Introduction
Stabilising GHG concentrations at an ambitious level will be difficult to achieve immediately. It
will require international action by all the main emitters, driven by a cost-effective set of policy
instruments (including a global international carbon price, R&D policies and targeted regulations and
standards). This chapter therefore examines how to achieve broad-based international carbon pricing
gradually in practice. A range of policy instruments is considered:  
• Removing environmentally harmful fossil fuel energy subsidies – as these amount to a negative
carbon price. This would be a first step towards broad-based international carbon pricing and
would free up budgetary resources to target more directly the social objectives supported by the
subsidies. The first Section of this chapter analyses the potential environmental and economic
effects of removing energy subsidies in non-OECD countries.
• The direct linking together of domestic/regional emissions trading schemes. With
domestic/regional ETSs spreading internationally, albeit with a large heterogeneity in terms of
both targets and design features, a global carbon market may gradually build up as
domestic/regional ETSs become linked together. The conditions and implications of this direct
linking of ETSs are explored in Section 4.2.  
• Linking domestic/regional ETSs indirectly through the use of a common crediting mechanism.
This is another way to gradually build up an integrated world carbon market and to lower
mitigation costs. This is analysed in Section 4.3.
• The use of sectoral approaches, which can broaden participation to developing countries. This
will be necessary because even very large emission reductions in developed countries alone
would not be enough to halt climate change. Section 4.4 explores ways of designing sectoral
agreements in order to move towards a global carbon price and to limit carbon leakage and
competitiveness concerns.
1
• The institutions and rules needed to foster the development of carbon markets and to address
likely risks within a linked system of multiple independent and heterogeneous emission trading
schemes. The final Section explores this.
4.1.  Removing environmentally-harmful energy subsidies
2
In many non-OECD countries there are currently large subsidies to the consumption of fossil fuels
which keep fossil fuel use, and hence GHGs emissions, at high levels. Furthermore, as they imply some
decoupling of domestic energy prices from world prices, they prevent the price signals of world energy
markets from affecting domestic markets. Removing these subsidies would be an obvious initial step in
pricing carbon worldwide. In addition to environmental benefits, it would yield economic gains because
of a more efficient allocation of resources in the countries which remove these subsidies. Thus this step
is one of the few “no regret” options for contributing to climate stabilisation.  
Very few studies have attempted to quantify the impact of removing these subsidies in non-OECD
countries. In its 1999  World Energy Outlook, the International Energy Agency (IEA) estimates the
impact of removing energy subsidies in eight non-OECD countries (IEA, 1999, using a partial
equilibrium approach). For this sample of countries, CO2 emissions were estimated to fall by 16% on
average, translating into a reduction of 5% in world emissions. Corresponding gains accruing to these
countries were estimated to amount to 0.7% of their GDP on average, reaching approximately 2% in 4.  TOWARDS GLOBAL CARBON PRICING
102  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
some cases. Using the general equilibrium model GREEN, Burniaux  et. al.  (1992) estimated that
removing all existing distortions on primary fossil fuels would reduce world CO2 emissions by 18%
relative to the baseline in 2050 while generating an average discounted real income increase at the world
level by 0.7% over the period 1990-2050. Again using a general equilibrium approach, the OECD (1999)
found that removing energy subsidies in Annex I countries would reduce the costs of achieving the
Kyoto Protocol, albeit only by a modest amount. Taken together, Annex I countries would benefit from
such subsidy cuts, and the efficiency gains achieved in those that remove their subsidies (Russia and
Eastern European countries) would benefit the others (mainly the United States and the EU), as these
efficiency gains are entirely “exported” through the Annex I-wide ETS assumed in the OECD analysis.  
This Section assesses the potential environmental and economic benefits of removing energy
subsidies using the OECD model ENV-Linkages and a recent dataset assembled and provided by the IEA.
4.1.1. The size of environmentally-harmful energy subsidies
Energy subsidies take many different forms and can involve both direct and indirect subsidies,
with the latter being more difficult to measure. Some subsidies aim at increasing fossil fuel consumption,
while others aim to support domestic production. A common way to subsidise energy consumption is by
exempting some energy consumptions from normal taxation (IEA, 1999). While each of these forms of
subsidies should ideally be modelled explicitly in order to quantify their impact, this approach was not
feasible in this analysis due to lack of data. Instead, it is assumed that different forms of subsidies result
in a lower domestic energy price relative to a reference price. Accordingly, the energy goods are assumed
to be relatively similar, and various forms of subsidies are summarised by a single statistic, the observed
price gap between the energy domestic price and the reference price, differentiated across different types
of end-use consumers (households, power generation, manufactures and services) when data were
available. Although this approach has a number of well-known limitations
3
it is the only one possible
given the information currently available on these subsidies in non-OECD countries.
The IEA has estimated price gaps corresponding to energy subsidies for 2005 and 2007 in 20
non-OECD countries, accounting for about 40% of world energy consumption (IEA, 2009). These gaps
were estimated after adjustments were made to take into account market exchange rates, transportation
margins and domestic taxes (including VAT). For fossil fuels, the reference price is the corresponding
international price. As electricity is rarely traded, the reference price corresponds to an estimation of the
production cost in the country considered (expressed in local currency).
The price gaps estimated for 2007 by energy sources and by countries/regions are significant in a
number of cases (Table 4.1). As the influence of international energy prices on domestic markets is
incomplete,
4
these gaps are likely to have changed following the 2008 oil price spike and then again
when oil prices fell. The first column of the table shows the average gaps for all energy demands that are
effectively subsidised in each country/region, thereby illustrating the magnitude of the gaps (the larger the
number in absolute terms, the greater the subsidy). The second column reports the average gap across all
demands, so that the difference between both columns depends on whether subsidies, for each fossil fuel,
concern some specific demands or are broadly used across all types of demand. Countries not covered in
the IEA database are included in regional aggregates (for instance, the “rest of the world” region) for
which gaps of zero have been assumed. This assumption is fairly conservative as it is likely that some of
these countries do also subsidise part of their energy consumption. The table shows how energy gaps
differ across energy sources and countries/regions. Energy tends to be subsidised more heavily in Russia
(especially natural gas), India and non-EU Eastern European countries. By contrast, the subsidy rates
estimated by the IEA for China are rather moderate. The subsidy rates in oil-exporting countries and the 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    103
rest of the world regional aggregates appear to be relatively low, but they are understated due to the
incomplete country coverage in the IEA estimates.
Table 4.1.  Energy price gaps are found to be significant in a number of non-OECD countries
Energy price gaps in non-OECD countries,
1
2007
% deviation of domestic relative to world prices
Country  Energy  Average subsidy rate over the demands that
are effectively subsidised for each type of fuel
Average subsidy rate over the total
demand for each type of fuel
China
Coal
Gas
Refined oil
Electricity
-18.1
-27.0
-7.1
-22.3
-0.5
-2.8
-2.0
-3.2
India
Coal
Gas
Refined oil
Electricity
0.0
-53.6
-51.8
-19.6
0.0
-28.3
-10.1
-9.1
Brazil
Coal
Gas
Refined oil
Electricity
-40.4
0.0
-14.4
0.0
-8.5
0.0
-2.2
0.0
Russia
Coal
Gas
Refined oil
Electricity
-51.6
-84.7
-23.6
-48.9
-1.2
-26.8
-3.3
-35.0
Oil-exporting countries
Coal
Gas
Refined oil
Electricity
0.0
-18.9
-29.2
-21.9
0.0
-5.9
-22.3
-20.4
Non-EU Eastern
European countries
Coal
Gas
Refined oil
Electricity
-30.0
-39.6
-5.4
-37.4
-4.9
-20.4
-1.8
-20.7
Rest of the world
Coal
Gas
Refined oil
Electricity
-2.1
-25.6
-8.5
-6.7
-0.5
-7.7
-3.4
-5.1
1.  Energy subsidies are approximated by the difference in the domestic energy price and world prices.
Source: IEA (2008a). 4.  TOWARDS GLOBAL CARBON PRICING
104  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
4.1.2. The impact on GHG emissions and mitigation costs of removing
environmentally-harmful energy subsidies
To measure this impact two scenarios are analysed:  
i) The impact on GHG emissions and income of removing existing energy subsidies gradually in
non-OECD countries between 2013 and 2020 assuming that no other mitigation action is
implemented. Two cases are considered: 1a) a unilateral removal of energy subsidies in each
non-OECD country/region (i.e. each country takes the action alone); and 1b) a multilateral
removal of energy subsidies in all non-OECD countries/regions (i.e. all countries act
simultaneously).  
ii) The impact on the mitigation cost of removing energy subsidies in the context of a GHG
mitigation policy. Again, two cases are considered: 2a) a reduction in GHG emissions (of 20%
and 50% respectively in 2020 and 2050 compared to 1990 levels), combined with removing
energy subsidies, with both actions taken in Annex I countries only; and 2b) the
implementation of a world carbon tax designed to reduce global emissions to stabilise overall
GHG concentrations below 550 ppm CO2 equivalent (“550 ppm-base scenario”), combined
with the multilateral removal of energy subsidies in all non-OECD countries.
The gradual multilateral removal of existing energy subsidies in non-OECD countries (Scenario
1b) would lead to quite a substantial drop of GHG emissions from fossil fuel combustion by 2050 in
some countries/regions, amounting to around 30% or more in non-EU Eastern European countries,
Russia and the Middle East (Figure 4.1).
5
However, while GHG emissions from fossil fuel combustion
would fall by 14% in non-Annex countries in 2050, they barely decline in Annex I countries. This is
because reductions in Russia and non-EU Eastern European countries are offset by increases in those
Annex I countries that do not subsidise their energy demand, encouraged by falling world energy prices
induced by the multilateral removal of subsidies. Of the 8.8 GtCO2-eq emission reduction achieved by
removing energy subsidies in non-OECD countries in 2050 (corresponding to a reduction of their
emissions by 16% relative to the baseline), around 16% would be offset by an increase of emissions in
OECD countries. As a result, global GHG emissions would be reduced by 10% in 2050 through this
subsidy removal compared with business as usual.
6
With binding emission caps in OECD countries, the
“leakages” would be contained, and the environmental benefits of subsidy removal would be even larger.
For purposes of comparison, if each non-OECD country/region were to remove existing subsidies in
isolation (unilateral removal, Scenario 1a), emission reductions would be lower than for a multilateral
removal (Figure 4.2).
7
All countries/regions (with the exception of non-EU Eastern European countries
8
) would benefit
from a unilateral removal of energy subsidies and real income gains would range from 0.1% in Brazil to
over 2% in India and Russia in 2050 (Table 4.2).
9
However, these gains would differ in the case of a
multilateral simultaneous removal of energy subsidies in all non-OECD countries. Some of the
non-OECD countries – especially Russia, the Middle East and non-EU Eastern European countries – that
remove their subsidies would no longer enjoy real income gains. This is because the efficiency gains
from improved resource allocation would be more than offset by the terms-of-trade losses associated
with the sharp fall in world energy prices that a multilateral removal of subsidies would induce.
However, energy-importing OECD countries, especially the European Union and Japan, would enjoy
significant terms-of-trade and income gains. Overall, GDP and real income gains at the world level
would be small, amounting to just 0.1% relative to BAU in 2050. This primarily reflects the fact that
demand for energy goods is not very sensitive to price, so that the distortive impact of energy subsidies
and the gain from their removal are limited.
104.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    105
Table 4.2.  Most countries would benefit from unilateral and multilateral removals of energy subsidies
(Household equivalent real income,
1
% deviation relative to the BAU)
Regions
Impact of unilateral removal of energy subsidies  Impact of multilateral removal of energy subsidies
2020 2050 2020 2050
Australia & New Zealand  0.0  0.0  0.0  -0.6
Brazil  0.0 0.1 0.0 0.1
Canada  0.0  0.0  -0.4  -1.5
China  0.0 0.3 0.1 0.7
EU27 & EFTA  0.0  0.0  0.4  0.9
India  1.1 2.2 1.4 2.5
Japan  0.0  0.0  0.4  0.9
Oil-exporting countries  -1.1  1.0  -2.1  -4.5
Non-EU Eastern European
countries
0.5  -1.8  -2.0  -15.2
Rest of the world  0.0  0.2  -0.1  0.0
Russia  1.3  2.2  0.1  -3.7
United States  0.0 0.0 0.1 0.1
Annex I  0.2  0.1
Non-Annex I      -0.2  0.0
World  0.1  0.0
1.  Hicksian "equivalent real income variation"defined as the change in real income (in percentage) necessary to ensure the
same level of utility to consumers as in the baseline projection.
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
106  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.1.  A multilateral removal of energy subsidies would lower GHG emissions in non-OECD countries
(% deviation in GHG emissions relative to BAU)
-50.0
-40.0
-30.0
-20.0
-10.0
0.0
10.0
20.0
Non-EU Eastern
European countries
Russia
Oil-exporting
countries
India
China
Rest of  the World
Brazil
Australia &
New Zealand
Canada
United States
Japan
EU27 & EFTA
Annex I
Non-Annex I
World
%
2020
2050
Source: OECD, ENV-Linkages model.
In the context of efforts to reduce their emissions as part of a post-2012 agreement, Annex I
countries might also aim to remove their energy consumption subsidies. The cost savings associated with
this action are estimated to be small, in line with the results reported in OECD (1999). Figure 4.3 reports
the estimated economic costs of Annex I countries cutting their emissions by 20% and 50% by 2020 and
2050 respectively (relative to 1990 levels), with and without a removal of their energy subsidies
(scenario 2a as described above). Overall, the cost savings are concentrated in Russia and the European
Union, reflecting some efficiency gains and a terms-of-trade improvement, respectively.
Table 4.3.  The treatment of energy subsidies in the business-as-usual and policy scenarios matters for the
global GDP cost of a 550 ppm CO2eq concentration stabilisation
Scenarios  World GDP loss in 2050 (% difference from BAU)
“550 ppm-base” scenario with energy subsidies in place  -3.4
“550 ppm-base” scenario with complete removal of energy
subsidies in non-OECD countries
-3.2
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    107
The impact of removing energy subsidies is then assessed in the presence of a world carbon tax
policy to stabilise overall GHG concentration below 550 ppm CO2eq (“550 ppm-base” scenario11
or
scenario 2b as described above). Assuming that energy subsidies are kept in place, the stabilisation effort
is estimated to reduce world GDP by 3.4% in 2050, compared with the business as usual scenario
(Table 4.3).
12
The removal of energy subsidies generates a slight GDP gain and reduces emissions,
making the stabilisation target easier to achieve. As a result, there is a lower world GDP loss in 2050
from mitigation action when energy subsidies are also removed (-3.2% as opposed to -3.4% when
subsidies are in place). India and, to a lesser extent, China and OECD countries, benefit from the subsidy
removal, while mitigation costs increase in energy-exporting countries (Figure 4.4). Cost savings at the
world level are relatively small mainly because the economic distortion from energy subsidies is
limited.
13
In conclusion, removing existing energy subsidies in non-OECD countries would help to reduce
world emissions, and lead to GDP gains in these countries. But part (almost one-fifth) of the
environmental benefit of this reform would be lost unless emissions are capped in OECD countries. As
theory suggests, removing these subsidies would generate real income gains in the countries where this
reform is applied, as well as in OECD countries. As a result, incorporating the removal of energy
subsidies into a global mitigation action will reduce the economic costs of this action.  
Another important step in developing a global carbon market is then analysed – the linking
together of emission trading schemes.
Figure 4.2.  Unilateral removal of energy subsidies in non-OECD countries would lower GHG emissions but
to a lesser extent than in the case of a multilateral removal
(% deviation in GHG emissions relative to BAU)
-30.0
-25.0
-20.0
-15.0
-10.0
-5.0
0.0
Russia
Non-EU
Eastern European
countries
Oil-exporting
countries
India
China
Rest of  the world
Brazil
%
2020
2050
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
108  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.3.  Removing energy subsidies in Annex I countries would slightly lower mitigation costs
in these countries
Mitigation costs under a 20% cut by 2020 and a 50% cut by 2050 relative to 1990 levels in each Annex I region, household
equivalent real income1
, % deviation relative to BAU
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
Canada
Australia &
New Zealand
Oil-exporting
countries
United States
Non-EU Eastern
European countries
EU27 & EFTA
Japan
Rest of  the World
China
Brazil
India
Russia
Annex I
Non-Annex I
World
%
2020
With energy subsidies removal
Without energy subsidies removal
-20.0
-18.0
-16.0
-14.0
-12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
Non-EU Eastern
European countries
Canada
Australia &
New Zealand
Russia
Oil-exporting
countries
EU27 & EFTA
United States
Japan
Rest of the World
Brazil
China
India
Annex I
Non-Annex I
World
% 2050
With energy subsidies removal
Without energy subsidies removal
1.  Hicksian “equivalent real income variation” defined as the change in real income (in percentage)
necessary to ensure the same level of utility to consumers as in the baseline projection.  
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    109
Figure 4.4.  A multilateral removal of energy subsidies would lower the global mitigation costs of stabilising
GHG concentration at 550 ppm CO2eq
Mitigation costs under the "550ppm-base" scenario1
, household equivalent real income2
, % deviation relative to BAU
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Oil-exporting
countries
Non-EU Eastern
European countries
China
Canada
Russia
India
Brazil
Australia &
New Zealand
Rest of the World
United States
EU27 & EFTA
Japan
Annex I
Non-Annex I
World
%
2020
With energy subsidies removal
Without energy subsidies removal
-30.0
-25.0
-20.0
-15.0
-10.0
-5.0
0.0
Russia
Non-EU Eastern
European countries
Oil-exporting
countries
China
Canada
Australia &
New Zealand
India
Brazil
Rest of the World
United States
EU27 & EFTA
Japan
Annex I
Non-Annex I
World
% 2050
With energy subsidies removal
Without energy subsidies removal
1.  The pathway of emissions corresponds to a stabilisation below 550ppm (all gases) identical to
the "550 ppm-base" case described in Chapter 1, Table 1.2.
2.  Hicksian “equivalent real income variation” defined as the change in real income (in
percentage) necessary to ensure the same level of utility to consumers as in the baseline
projection.  
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
110  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
4.2.  The direct linking of emission trading schemes
Several domestic/regional GHG emission trading or cap-and-trade schemes (ETSs) are already in
place or are emerging. These place a cap on GHG emissions from a number of sectors and allocate rights
to emit amongst the firms in these sectors. The total amount of emission rights or permits cannot exceed
the cap, limiting total emissions to that level. The participating firms are then allowed to buy or sell
emission rights amongst themselves. Firms with opportunities for relatively cheap emissions reductions
implement these and sell their emission rights to those for which emission reductions would be more
expensive.  The ETSs in place or emerging all vary significantly in terms of their target, size, and other
design features. At present there are virtually no direct links between them, other than the link between
the EU and Norwegian ETSs. Yet, as more ETSs are expected to emerge in the future, direct linking is
likely to gain prominence and could form a key building block of a global carbon market and thus reduce
global mitigation policy costs. However, it does raise a number of concerns that will have to be
addressed to ensure environmental effectiveness (Section 4.2.3). Different ETSs can be linked either
directly, or indirectly through access to a common crediting mechanism that allows for emission
reductions to take place in countries not covered by an ETS. This Section focuses on direct linking, while
the indirect link through crediting mechanisms is discussed in Section 4.3.  
The effects of linking different domestic ETSs are mainly illustrated here using the OECD model
ENV-Linkages to run a “benchmark” scenario. Under this scenario, each Annex I region is assumed to
use an ETS to cut its GHG emissions unilaterally below 1990 levels by 20% by 2020 and by 50% by
2050. On its own, this commitment would be insufficient to achieve ambitious climate objectives. World
emissions would still rise by about 20% and 50% by 2020 and 2050 respectively, versus about 85% by
2050 in a baseline scenario with no further mitigation policy action. It would, therefore, need to be fairly
rapidly tightened and/or supplemented with further action, including in non-Annex I countries.
Nevertheless, the illustrative benchmark scenario raises a number of lessons about the cost-effectiveness
and competiveness impacts of linking.
4.2.1. The benefits of linking
Improving cost-effectiveness
Direct linking occurs if the tradable permit system’s authority allows regulated entities to use
emission allowances from another ETS to meet their domestic compliance obligations. Direct linking can
be “two way” if each system recognises the others’ allowances, or “one way” if one system recognises
the other system’s allowances but the other does not reciprocate. Linking ETSs directly tends to lower
the overall cost of meeting their joint targets by allowing higher-cost emission reductions in one ETS to
be replaced by lower-cost emission reductions in the other. Once ETSs are linked, this cost-effectiveness
is achieved regardless of the magnitude of the initial emission reduction commitment across countries or
regions; the distribution of emission reductions is determined through market mechanisms. The potential
gains from linking are greater the larger the initial difference in carbon prices – and thereby in the
marginal costs of reducing emissions – across individual ETSs.  4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    111
Figure 4.5.  Linking regional Annex I emission trading schemes would affect where
emission reductions take place
(Under a 20% cut by 2020 and a 50% cut by 2050 relative to 1990 levels in each Annex I region)
0
100
200
300
400
500
600
700
Russia United
States
Non-EU
Eastern
European
countries
EU27 &
EFTA
Japan Australia &
New Zealand
Canada Annex I
Carbon price USD/t CO2 eq
Panel A. Carbon  prices prior to and after  linking
2020, prior to  linking
2050, prior to  linking
Carbon
price
after
linking
-70
-60
-50
-40
-30
-20
-10
0
10
20
30
Russia United
States
Non-EU
Eastern
European
countries
EU27 &
EFTA
Japan Austral ia &
New Zealand
Canada
% change relative to 2005
Panel B. Emission reductions  prior to and after  linking
2020, emission reductions after linking
2050, emission reductions after linking
2020, emission reductions prior to linking
2050, emission reductions prior to linking
1.  There is no crediting mechanism in these simulations,  i.e. all emission reductions must be
achieved in Annex I regions only.
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
112  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
As an illustration, the gains from linking together the domestic ETSs of Annex I regions are
assessed in the benchmark scenario above.  This is done by considering another scenario in which the
same GHG emission reduction at the Annex I level is achieved through a linked system of ETSs.
Concretely, an Annex I-wide ETS is assumed to be established, under which each participating region is
allocated emission rights corresponding to a -50% individual emission reduction target by 2050
(compared to 1990 levels), as in the benchmark scenario. In the benchmark scenario, meeting their
domestic caps alone was found to cost Annex I regions about 1.5% and 2.8% of their income on average
by 2020 and 2050, respectively. Linking is found to enhance emission reductions in those schemes which
had lower marginal abatement costs before linking (especially Russia, Figure 4.5 Panel B), but to weaken
emission reductions in the others (Figure 4.5 Panel B). However, the associated reduction in overall
mitigation costs for Annex I countries is just under 10%, or about 0.2 percentage points of income by
2050 (Figure 4.6, Panel B). This reduction in mitigation costs if fairly limited in part because in the
illustrative benchmark scenario, carbon price differences prior to linking are estimated to be relatively
small across the larger Annex I economies who account for the bulk of Annex I GDP (Figure 4.5
Panel A). Larger gains from linking would be found under more heterogeneous emission reduction
commitments by Annex I countries than considered here.
In addition to improving the overall cost-effectiveness of the linked ETS system, linking is
expected to benefit each participating region (e.g. Jaffe and Stavins, 2007 and Figure 4.6). The larger the
change in the carbon price after linking, the larger the income gain, other things being equal. In turn, the
carbon price level prior to linking depends on the size of the country’s commitment, as well as on the
availability of cheap abatement opportunities. In the illustrative scenario considered here, countries with
lower pre-linking carbon prices (mainly Russia) gain because the equilibrium price of the linked system
exceeds their marginal abatement costs, enabling them to abate more and sell the saved permits with a
surplus while countries with higher pre-linking carbon prices (Australia and New Zealand, Canada,
Japan) benefit from the lower carbon price (Figure 4.7).
14
While basic economic theory suggests that permit trading among Annex I regions should benefit
all participants, modelling does not always produce this result in practice. This reflects the so-called
Dutch disease effects in the presence of various market imperfections. Because of this, non-EU Eastern
European countries – which together form the “Rest of Annex I” region of the ENV-Linkages model –
are found to lose from linking by 2050 (Figure 4.6). This is because their large permit export flows lead
to a real exchange rate appreciation, which in turn results in a fall in the exports and output of their
manufacturing sector, where scrapping capital entails costs. Nevertheless, these “Dutch disease” effects
should be discounted because the ENV-Linkages model exacerbates them, partly due to a lack of explicit
modelling of the international capital market. For instance, the real exchange rate appreciation could be
smoothed in practice if some of the revenues from permit sales were recycled in international capital
markets and if linking were to occur progressively.
15
Linking schemes can also improve cost-effectiveness by increasing the size and liquidity of carbon
markets. In the scenario presented above,  i.e. when Annex I ETSs are linked, the size of the market is
projected to reach 2.5% of Annex I GDP in 2020. As idiosyncratic shocks are shared across regions
under linking, a larger market size tends to dampen the impact of such shocks, thereby lowering overall
carbon price volatility and enhancing incentives for firms to make emission reduction investments.
16
Furthermore, transaction costs are expected to be smaller in a larger, more liquid market, especially if
some regional schemes are too small to foster the development of institutions for reducing such costs.
Larger market size also reduces problems that may arise if some sellers or buyers have market power
(Hahn, 1984). Finally, market liquidity can lower the cost of insuring against uncertainty by fostering the
development of derivative markets (Section 4.5).  4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    113
Figure 4.6.  Linking regional Annex I emission trading schemes would affect the distribution of mitigation
policy costs across countries for a 50% emission cut in Annex I
(Under a 20% cut by 2020 and a 50% cut by 2050 relative to 1990 levels in each Annex I region)
-7.0
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
Canada Australia &
New Zealand
Non-EU
Eastern
European
countries
Russia United
States
Annex I Japan EU27 &
EFTA
Mitigation cost (income equivalent variation  
relative to baseline, in %)
Panel A. 2020
With linking
Without linking
-25.0
-20.0
-15.0
-10.0
-5.0
0.0
Non-EU
Eastern
European
countries
Russia Canada Australia &
New
Zealand
Annex I EU27 &
EFTA
United
States
Japan
Mitigation cost (income equivalent variation
relative to baseline, in  %)
Panel B. 2050
With linking
Without l inking
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
114  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.7.  Projected geographical distribution of permit buyers and sellers under a 50% emission cut in
Annex I
1
(Under a 20% cut by 2020 and a 50% cut by 2050 relative to 1990 levels in each Annex I region)
Panel A. 2020
United
States
71%
Canada
7%
Australia
& New
Zealand
10%
Japan
12%
Buyers
Non-EU
Eastern
European
countries
16%
Russia
67%
EU27 &
EFTA
17%
Sellers
Panel B. 2050
Canada
31%
Australia
& New
Zealand
33%
Japan
10%
Non-EU
Eastern
European
countries
6%
EU27 &
EFTA
20%
Buyers
United
States
21%
Russia
79%
Sellers
1. This simulation assumes there is no crediting mechanism, i.e. all emission reductions are achieved in Annex I regions
only.
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    115
Mitigating competitiveness concerns  
Another advantage of linking is its ability to reduce competitiveness concerns in regions with
higher pre-linking carbon prices, by allowing carbon prices to converge across linked schemes. Full
convergence in prices will be achieved provided the recognition of allowances is mutual and there are no
limits on trading. One-way linking (when system A recognises system B’s allowances but the latter does
not) ensures that the price in system A never exceeds the price in system B, and hence, would only limit
competitiveness concerns for firms belonging to system A.
17
However, although competitiveness
problems are reduced through linking, the losses in output of energy intensive industries would still be
unevenly distributed across countries with countries with the lowest marginal abatement costs (Russia
and non EU Eastern European countries, Figure 4.5, Panel A) facing the largest losses (Figure 4.8,
Panel A). Furthermore, competitiveness problems would remain for regions in which GHG emissions are
not priced or curbed by other means.
In the absence of strategic behaviour, linking across two ETSs does not affect the total emissions
of the linked schemes since the number of permits is simply the sum of those issued under each system.
However, it can still affect the overall environmental effectiveness of the scheme indirectly through its
effect on carbon leakages towards uncapped countries, which occurs when emission reductions in one set
of countries are partly offset by increases in countries elsewhere. If linking lowers the carbon price in the
region that faces the highest leakage rate, then leakage towards uncapped countries is reduced. By the
same token, if linking raises the carbon price in the region that faces the lowest leakage rate, then leakage
towards uncapped countries is increased. In the illustrative benchmark scenario examined here, model
simulations suggest that overall, linking among Annex I regions slightly reduces leakage (Figure 4.8,
Panel B).  
Applying the principle of “common but differentiated responsibilities and respective
capabilities”  
Compared with a global emissions trading system, it has been argued that a linked system of
regional ETSs may be an easier way to reflect “common but differentiated responsibilities and respective
capabilities” across regions, and thereby to extend participation to developing countries (Jaffe and
Stavins, 2007). Permit allocation rules make it possible to differentiate across regional commitments and
costs under a top-down approach. However, such differentiation can also be achieved through regions’
own assessment of their responsibilities and reflecting their specific national circumstances – as revealed
de facto by  their target choice – under a bottom-up approach. The gains from linking Annex I ETSs to
potential non-Annex I country ETSs would be larger than those achieved through linking within Annex I
only, if the heterogeneity in pre-linking carbon prices (and hence in commitments or actions) between
Annex I and non-Annex I is higher than within Annex I. 4.  TOWARDS GLOBAL CARBON PRICING
116  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.8.  Linking Annex I regional emission trading schemes would affect the distribution of
energy-intensive industries output losses across regions and would lower carbon leakage rates in 2020
(Under a 20% cut by 2020 and a 50% cut by 2050 relative to 1990 levels in each Annex I region)
-15.0
-13.0
-11.0
-9.0
-7.0
-5.0
-3.0
-1.0
1.0
3.0
5.0
Australia &
New Zealand
Japan United States Canada Non-EU
Eastern
European
countries
EU27 & EFTA Russia
Output (relative to baseline, in %)
Panel A. Output of energy-intensive industries in Annex I regions
Without linking
With linking
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
Without linking With linking
% Panel B. Carbon  leakage rate3 (Annex I as a whole)
1. Energy intensive industries include chemicals, metallurgic, other metal, iron and steel
industry, paper and mineral products.
2. There is no crediting mechanism in these simulations, i.e all emission reductions must be
achieved in Annex I regions only.
3. The carbon leakage rate is calculated as: [1-(world emission reduction in
GtCO2eq)/(Annex I emission reduction objective in GtCO2eq)]. It is expressed in per cent.
When the emission reduction achieved at the world level (in GtCO2eq) is equal to the
emission reduction objective set by Annex I (in GTCO2eq), there is no leakage overall,
and the leakage rate is 0.
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    117
4.2.2. Potential risks of linking
Though linking ETSs yields a number of benefits, it also raises environmental and income
distribution concerns.
18
These mainly arise from differences in the design features of ETSs prior to
linking. Their impact on the gains from linking are reviewed below and summarised in Table 4.4
(Sterk et al. 2006; Baron and Bygrave, 2007; Flachsland, et al. 2008a, 2008b; Haites and Mullins, 2001).
These issues would need to be addressed in order to reap the full gains from linking, and to avoid
potential risks.
The distributional impacts of linking and the implications of differences in allocation rules
Although linking ETSs tends in general to lower the mitigation cost of each of the participating
regions, it affects the distribution of costs, both across and within schemes. After linking, the shared
carbon price settles somewhere between the pre-linking levels in the two regions considered (Figure 4.5).
The larger the gap between the pre- and post-linking carbon price levels, the larger the gain to one region
and the smaller to its foreign counterpart, other things being equal. Furthermore, within the region where
linking leads to a carbon price increase, permit sellers gain while permit buyers lose, (and  vice versa
within the regions where linking results in a carbon price decline). The magnitude of these distributional
effects depends on the extent to which domestic carbon prices are affected by linking, which in turn is
determined by several factors, including the respective sizes of the permit markets, the difference in
pre-linking targets and the steepness of the supply and demand curves for permits. In particular, the
larger the difference between regions in the stringency of targets prior to linking, the stronger the
distributional impacts, other things being equal. Likewise, the larger the relative size of the market to
which the domestic ETS links, the larger the distributional effects are expected to be.
However, these distributional effects are similar in nature to those of international trade. Some of
their political economy problems can be reduced through permit allocation rules if necessary; for
instance through allowing transitory grandfathering19
upon linking in the region with the lower
pre-linking carbon price. While differences in allocation rules across linked schemes sometimes raise
competitiveness concerns (Jaffe and Stavins, 2007), these effects would typically have existed before the
schemes were linked and, therefore, would not be exacerbated by linking.
20
Furthermore, in reasonably
competitive goods and services markets, the opportunity cost of free permits is reflected in firms' output
prices, and therefore allocation rules have no effect on output and competitiveness.  
The spread of cost-containment measures and the risks for environmental effectiveness
Linking would automatically lead to the spreading across regions of some design features specific
to one particular scheme. As a result, governments in the linked regions would lose control over several
features of their existing ETS. In particular, provisions to contain the cost of mitigation
(cost-containment measures), such as carbon price caps ("safety valves"), or provisions for credits to be
banked for or borrowed from future commitment periods, would be spread through linking (Ellis and
Tirpak, 2006; Jaffe and Stavins, 2007; Flachland et al. 2008a, 2008b; Table 4.5).  
The spread of cost-containment measures can undermine the environmental effectiveness of the
overall system. For example, the spread of the safety valve implies that the overall target is relaxed once
the safety valve is reached. Likewise, linking to an offset credit system whose environmental integrity is
weaker than that of an ETS could also raise environmental concerns in countries that have more
restrictive policies for the use of offsets. Partly for these reasons, the EU directive on linkage currently
forbids linking to a scheme featuring a safety valve.  4.  TOWARDS GLOBAL CARBON PRICING
118  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Table 4.4.  Different design features in pre-linked emission trading schemes would affect the performance of
the linked system
Impact of linking on:
ETS design feature
differences in:
Technically
feasible?
Cost-
effectiveness
Environmental
effectiveness
Competitive
distortion
Distribution of
costs and gains
within each
scheme
Suggested
solution
Emission target  Yes  Improved if the
regions with less
stringent targets
prior to linking
are those with
the lowest
abatement costs
Possibly affected
because of leakage,
strategic behaviour
and the “hot air”
issue2 but the
overall effect is
undetermined
No effect  Depends on the
gap between
pre- and
post-linking
carbon price
levels
Common cap or at
least a procedure
to set the cap
Link with another
scheme  
Yes, but the link
with the scheme
also applies to
other schemes
Improved if
marginal
abatement cost
is lower in the
offset scheme
Possibly reduced if
the environmental
integrity of the offset
scheme is weak
No effect  Affected  Find agreement on
decisions to link to
another scheme
and define
procedures to
decide future
linking.  
Allowance
allocation rule
Yes  No effect unless
permits are
grandfathered on
a regular basis
No effect  No effect
unless permits
are
grandfathered
on a regular
basis
Determines who
will be the
winners and the
losers within
each scheme
Specific allocation
rules can be used
to avoid the
resistance of some
sectors, firms,
households to
linking
Absolute versus
intensity target
Yes, but an
intensity target
makes little
sense and is
difficult to
achieve within a
linked system
No effect  Affected, and can
be reduced if the
allocation rule is
adjusted to growth  
No effect  No effect  Harmonisation is
desirable. If one
scheme has an
intensity target, the
allocation rule
should preferably
be made prior to
linking rather than
being frequently
updated
Safety valve (i.e.
carbon price cap)
in one of the
schemes  
Yes, but the
safety valve1 to
other schemes
will be spread
No effect  Reduced  Reduced since
all firms have
access to the
safety valve
under the
linked system
No effect  Harmonisation is
desirable
Banking and
borrowing
Yes, but
borrowing and
banking will also
apply to the
other schemes
Improved  No effect, unless
there is a “hot air”
problem
Reduced since
all firms have
access to these
provisions
No effect  None
Heterogeneity in
sector and/or gas
coverage,
downstream versus
upstream schemes
Yes  Improved   No effect  No effect  No effect  Need to adopt a
common measure
in terms of CO2
equivalent if gas
coverage differs
Table 4.4. continued on next page. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    119
Table 4.4.  Different design features in pre-linked emission trading schemes would affect the performance of
the linked system
(continued)
Impact of linking on:
ETS design feature
differences in:
Technically
feasible?
Cost-
effectiveness
Environmental
effectiveness
Competitive
distortion
Distribution of
costs and
gains within
each scheme
Suggested
solution
Lifetime of permits  Yes  Can be undermined
through a decrease
in market liquidity
No effect   No effect  No effect  Harmonisation is
desirable
Compliance period  Yes Can be undermined
through a decrease
in market liquidity
No effect  No effect  No effect  Harmonisation is
desirable. Avoid
attaching permits to
compliance periods
Monitoring,
reporting,
enforcement
provision
Yes  Can be undermined  Can be
undermined
No effect  No effect  Harmonisation is
desirable. Use a
common
centralised
institution to verify
offsets
Notes:  This table shows the implications of differences in design feature on the overall performance of the system after linking.
It does not show impacts that would occur anyway without linking. For instance, two schemes with different sectoral coverage
would raise some competitive distortion concerns even if the two systems are not linked.
1.   Provisions to contain the cost of mitigation, such as carbon price caps.
2.   A “hot air issue” can occur if the aggregate emissions cap in one of the schemes exceeds its business-as-usual emissions.
In the absence of linking, and if this scheme does not allow banking, the surplus allowances would be lost after the trading
period. However, in a linked scheme, the surplus is sold to relax the cap of the scheme with the binding target, thereby
allowing increased emissions from the overall system.  
Table 4.5.  Most cost-containment measures would be spread through linking
Type of cost-containment measure
Cap level  Safety-valve  Offsets  Banking  Intensity targets
Cap relaxed to lower
the allowance price
Buy-out provisions
to cap allowance
prices
Import credits from
non-capped
sources
Allowances carried
over for compliance in
future periods
Define obligations in terms of
emissions per unit output
Implications for the newly linked-in system without cost containment provisions
The allowance price
declines but the
country/region gains
from exporting
credits
Allowance price
decreases to
ceiling
Imported credits
implicitly accepted
Banking indirectly
available
No implication if allowance
allocation is set beforehand.
Possible increase in price stability
if allowance allocation is updated
with output
Source: EPRI (2007) and OECD. 4.  TOWARDS GLOBAL CARBON PRICING
120  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Creating a link between an ETS and a carbon tax
To some extent, similar concerns would be raised by creating a link between a domestic ETS price
and a foreign carbon tax. This would imply allowing domestic firms to pay the foreign carbon tax rather
than purchase a domestic permit. The effect would be to introduce a safety valve equal to the foreign
carbon tax, or even allow a switching to the foreign carbon tax if it is lower than the domestic permit
price.
21
The region with the carbon tax might find it easier to allow firms to switch from the carbon tax to
ETS permits, although fiscal revenues would be lost in that firms that buy (cheaper) foreign permits are
exempted from the tax. Under this one-way linking, the region with the carbon tax would gain if its tax
was higher than the foreign permit price. The region with the ETS would benefit from permit sales to
foreign firms, which would more than offset the higher mitigation costs arising from permit prices rising
towards the foreign carbon tax level.  
A one-way link between a carbon tax system and an ETS could also emerge if a government took
on binding commitments to reduce emissions, set a carbon tax to achieve these reductions, and then
bought permits at the end of the compliance period if the emission reductions turned out to be
insufficient to reach its commitment (or is allowed to sell permits if emission reductions were going to be
higher). Such a link would benefit each participant and enhance the cost effectiveness of the global
system. In theory, the overall gain would be maximised if the carbon tax were set at a level that would
prevail if the country had instead opted for an ETS that was linked to the foreign ETS. This would mimic
a full linking of two ETSs, but may have little practical feasibility or relevance.
Linking absolute level and intensity target schemes
Another way to contain costs is to adopt "intensity targets", which are expressed in terms of
emissions per unit of output as opposed to absolute targets, which are expressed in terms of emissions
(Box 4.1). Such targets may be seen as an insurance policy against the risk of high costs in case of
higher-than-expected GDP growth. The impact of linking an ETS with intensity targets to an ETS with
absolute targets depends on the permit allocation rules. If the cap on emissions in the system with the
intensity target is set  ex ante on the basis of projected GDP growth, then that scheme is  de facto
equivalent to an ETS with an absolute cap, and linking does not affect overall emissions.
22
By contrast, if
the permit authority of the intensity target scheme regularly adjusts the supply of permits in order to meet
its intensity target, overall emissions will fluctuate. Within a linked system, such adjustments need to be
more frequent and larger than within an independent one because emissions from any particular region
are determined endogenously by market forces, and depend on all shocks to the system – including those
specific to other regions (e.g. a cold winter in a large participating region). Furthermore, the impact of
permit supply adjustments on domestic emission intensity will depend in part on the extent to which
some of the newly emitted permits are bought by foreign firms.  4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    121
Box 4.1  Intensity targets and their implications for linking
Why intensity targets?
Uncertainty about future economic growth can be an obstacle to adopting emission caps because it means
greater uncertainty about compliance costs. Intensity targets, under which permit allocations would be linked to
future GDP, and would automatically adjust to unexpected growth shocks, have been proposed to reduce carbon
price and mitigation cost uncertainty (Marcu and Pizer, 2003; Kolstad, 2006; Gupta et al. 2007; Jotzo and Pezzey,
2007). They may be more acceptable than absolute caps for developing countries that are experiencing strong but
nevertheless uncertain long-term economic growth prospects (Fischer and Morgenstern, 2008). This is the case
even though intensity targets increase mitigation costs when GDP is unexpectedly low and lower mitigation costs
when GDP is unexpectedly high, which might be seen as an undesirable insurance property. Also, ways would
need to be found to make such targets compatible with the need for fast-growing emerging economies to take on
more (rather than less) stringent mitigation action as they catch up with developed countries.  
Intensity targets would deal with uncertainty about future GDP, but not with uncertainty about future
emission intensity or about structural abatement costs (Jotzo and Pezzey, 2007, Marschinski and Lecocq, 2006).
Furthermore, the extent to which intensity target can limit mitigation cost uncertainty depends on the share of
GHG emissions that are linked to GDP. Intensity targets are likely to be an effective insurance device in countries
where emissions mainly come from fossil fuel combustion, and so are strongly correlated with GDP. They would
be less effective for countries where a significant proportion of emissions come from land-use and land use
changes. It has been shown that there is an  optimal degree of indexation of emission targets to GDP, which
depends positively on the share of emissions linked to GDP and on the stringency of the target.
Under intensity targets, uncertainty is shifted to some extent away from costs onto emissions levels, since
the overall amount of emission reductions is not fixed (Dudek and Golub 2003). However, the overall
environmental performance of an intensity target scheme may not necessarily be weaker than under an absolute
cap, for three reasons (Jotzo and Pezzey, 2007). First, because they lower mitigation costs, intensity targets can
induce countries to take on more stringent commitments. Second, the impact on long-term emissions depends on
the nature of shocks. Prevalence of unexpected positive (negative) growth shocks would increase (lower)
emissions ceteris paribus, but insofar as such shocks are uniformly distributed around an average, they would be
expected to at least partly offset each other over a long time period. Finally, if a limited number of emitters is
involved in the scheme, temporarily relaxing the target as a result of medium-term indexation of emissions to
GDP would only have a small effect on the stock of GHGs in the atmosphere.
Ways of implementing intensity targets
There are basically two ways of implementing an intensity target: An ex ante permit allocation rule, under
which the amount of emission credits is set once and for all in order to meet an intensity target, conditional on a
pre-specified projected GDP path; and an updating allocation rule, under which the amount of emission credits is
adjusted over time based on the actual GDP path, in order to meet the intensity target ex post. While the former
rule is ultimately equivalent to an absolute target, the latter is not since the cap is adjusted upward in order to
allow firms to emit more when GDP is higher than projected. In practice, there would necessarily be some delay
between GDP growth developments and the adjustment of emission credits under the latter rule, making it
difficult to meet the intensity target strictly.  
Box 4.1 continued on next page. 4.  TOWARDS GLOBAL CARBON PRICING
122  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Box 4.1  Intensity targets and their implications for linking
(continued)
Implications for linking
From a technical perspective, linking a scheme with an absolute cap to another with an intensity target is
feasible under both types of allocation rule since in both cases governments will have to translate the target into a
fixed quantity of assigned units in order to allow emission trading (Philibert, 2005).
1
From an environmental
perspective, the impact of having one system with an intensity target depends on the allocation rule. Insofar as the
intensity target scheme has an ex ante rule, linking to other schemes with absolute targets should not affect total
emissions. The only risk would be that, if GDP growth turns to be lower than projected, the emission quota may
turn out to exceed business as usual emissions, in which case the surplus would be sold to absolute target
schemes, thereby increasing total emissions. However, this problem is in fact similar to a “hot air issue”, and
could also be encountered with an absolute target.
If the amount of emission credits is updated with GDP growth so as to meet an intensity target within a
linked system, governments would have to intervene frequently in order to adjust the supply of permits. This is
because once systems are linked, the distribution of emissions reductions across participating regions is
endogenously determined and is affected by any shocks to the system (e.g. a cold winter in a large participating
region). In the case of a positive growth shock, a larger number of permits will have to be emitted under a linked
system than under a system in which ETSs are not linked, since some of the newly-emitted permits would be
bought by foreign firms. This would tend to stabilise the carbon price in the linked system at the cost of more
uncertain emissions. One way to ensure the predictability of total emissions within the linked system would be for
participating regions with absolute targets to agree to adjust their caps so as to offset changes in the supply of
allowances from regions with intensity targets. Under such an arrangement, the latter regions would in effect
transfer some of the risk of unexpected changes in their mitigation costs to the former regions, which would then
lose the carbon price stabilisation gains from linking to an intensity target scheme.
When applied at the firm – or even possibly at the sector, but probably not at the national – level, and under
grandfathering, intensity targets may yield perverse incentives for firms to increase their output and thus, their
emissions, in order to obtain more credits. This incentive is reinforced by linking, since firms then have the
possibility to export these permits. One possible answer to such concerns is to introduce a “gateway” in order to
limit the net permit sales from the intensity rate-based programme, as is the case in the UK system, although such
restrictions also impose economic costs on the overall system.
1.  With an absolute target, the number of credits is determined directly by the cap. With an intensity target, the government
must set the intensity target, derive the corresponding emissions on the basis of a GDP projection, and compute the
amount of credits accordingly.
Permit supply adjustments by one region to meet its intensity target would increase carbon price
stability within the linked system, at the cost of greater uncertainty about overall emissions. Overall
environmental performance does not have to be undermined if emissions merely fluctuate around the
level that would prevail under absolute caps, but it could be affected if positive growth shocks prevail, or
if the intensity target system creates an incentive to increase production and emissions in order to obtain
additional credits. One way to ensure the predictability of total emissions within a linked system would
be for participating regions with absolute targets to agree to adjust their caps so as to offset changes in
the supply of allowances from regions with intensity targets. Under such an arrangement, the latter
regions would in effect transfer some of the risk of unexpected changes in their mitigation costs to the
former regions, which might otherwise lose the carbon price stabilisation gains from linking to an
intensity target scheme.
Linking may raise some environmental concerns that would need to be addressed by appropriate
institutions and regulations (Section 4.5):  4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    123
• The region with the lower carbon price ex ante has a further incentive to relax its cap in order
to generate additional revenue from exporting allowances – and a larger gain from linking more
broadly once systems are linked (Helm, 2003; Rehdanz and Tol, 2005). In order to alleviate
this, the region with the higher carbon price may also relax its target, thereby triggering a "race
to the bottom". This problem may be most acute for countries that face only limited damages
from climate change (Helm, 2003).
23
• Another source of environmental concern associated with linking is the “hot air” issue. It arises
if the aggregate emissions cap in one of the schemes exceeds its business as usual emissions. In
the absence of linking, and if this scheme does not allow banking, the surplus of allowances
would be lost after the trading period. By contrast, with linking, the surplus is sold to relax the
cap in the scheme with the binding target, thereby increasing the emissions of the overall
system.  
Addressing differences in sectoral, gas and time coverage
Differences in sectoral, gas and time coverage across linked schemes increase the complexity of
the overall system, but are no cause for concern in general:
• Different rules for a given sector could raise competitiveness concerns; firms covered by a
more stringent scheme might compete with firms that are exempt from binding commitments in
the other. However, this problem already exists regardless of linking. The same holds for the
risk of double taxation associated with linking a scheme in which the carbon price is applied at
an “upstream” level of the energy supply chain to one in which the carbon price is applied at a
downstream level. However, this can be addressed by exempting fossil fuel imports from the
upstream ETS from carbon pricing in downstream ETSs.  
• Differences in the lifetime of permits (the period during which they can be used for
compliance) would be expected to be reflected in price differentials; the market is likely to put
a higher price on credits with a longer lifetime, especially as future permit issuances are
expected to be low and uncertainty is high.
• Differences in the compliance period are unlikely to be a source of institutional incompatibility,
but if permits cannot be banked between compliance periods, they may unnecessarily multiply
permit vintages (that can be used for different compliance periods), lower market liquidity and
thereby increase price volatility (Section 4.5).  
4.2.3. Addressing the concerns about environmental integrity and the spread of design
features
There are some instrumental approaches that have been put forward to prevent linking from
weakening the environmental integrity of the overall system and to prevent one particular scheme's
design features from spreading to the others (Box 4.2). However, major drawbacks of such limits to
trading between schemes are that they hinder the full convergence in carbon prices and lower the
cost-effectiveness of linking. Both countries lose: the "protectionist" country because it has to achieve
more of its emission reduction domestically at a higher price; and the other country because permit
export revenues fall. Moreover, these instruments would lower market liquidity and increase its
complexity. They would also only contain, but may not fully prevent, the importation of the foreign
scheme's design features. The only way to do so would be one-way linking, under which trade with one
of the domestic schemes is not allowed, but again this would limit the cost-effectiveness gains from 4.  TOWARDS GLOBAL CARBON PRICING
124  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
linking. Finally, limits to international trading may run the risk of triggering retaliation by affected
countries.  
Box 4.2  Some instruments for maintaining environmental integrity in linked schemes
Three main strategies have been put forward to prevent linking from weakening the environmental integrity
of the overall system and to prevent the spreading of one particular scheme's design features to the others
(Rehdanz and Tol, 2005):  
• Discounting permit imports from the less restrictive schemes, meaning that an imported emission credit
normally worth one unit of CO2 counts for less than that in the purchasing country.  
• Setting permit import quotas,  i.e. requiring that a certain amount or percentage of emission reduction
must be achieved domestically.  
• Applying tariffs on permit imports; unlike quotas, these would bring fiscal revenues into the importing
country.  
By limiting trading, these instruments have the potential to reduce the exposure to shocks in other schemes,
and they can also limit the spreading of specific foreign scheme design features, such as safety valves or links to
offset credit systems. Other instruments to maintain the environmental integrity of the linked schemes, such as a
system of buyer liability, are discussed in Section 4.5.
In light of these drawbacks, the instruments in Box 4.2 may best be seen as emergency measures
rather than as permanent provisions. A more cost-effective approach to addressing concerns about
environmental integrity and the spread of design features would be for regions to agree on key issues
prior to linking, notably:
• The level of, or the procedures for, setting emission caps in future compliance periods. This
would remove governments' incentives to adjust future domestic caps in a way that maximises
domestic gains from linking, improve the environmental integrity of the overall scheme, and
provide greater certainty and incentives for clean investment by market participants.
• Whether to include a safety valve, given the spreading of this feature.
• The type of overall target (absolute versus intensity) and the allocation rule. In the long run, as
fast-growing emerging economies catch up with developed countries in terms of income levels,
they may switch from intensity to absolute targets as the latter would have lower environmental
uncertainty. In the context of a global emissions trading  scheme based on absolute targets, one
way to reflect economic development concerns would be to allocate commitments across
countries conditional on expected economic growth rates, and to adjust them over time.
• Procedures for assessing the future expansion of linking (to emission trading and offset
systems). Given the potentially large distributional and environmental impacts of an expansion
of the linked system, the rules applying to future linking of one participating region to other,
non-participating schemes should if possible be set in advance.  4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    125
4.3.  The role of emission crediting mechanisms and related challenges  
Linking can also occur “indirectly” when an ETS allows part of a region’s emission reductions to
be achieved in countries outside the ETS, for example through a common crediting mechanism such as
the Clean Development Mechanism (CDM), which is one of the flexibility mechanisms of the Kyoto
Protocol.
4.3.1. The potential gains from well-functioning crediting mechanisms
The CDM allows emission reduction projects in non-Annex I countries – i.e. developing countries,
which have no GHG emission constraints – to earn certified emission reduction (CER) credits, each
equivalent to one tonne of CO2eq. These CERs can be purchased and used by Annex I countries to meet
part of their emission reduction commitments. In principle, assuming that developing country emitters do
not take on binding emission commitments in the near future, well-functioning crediting mechanisms
could play four important roles: i) improve the cost-effectiveness of GHG mitigation policies in
developed countries, both directly and indirectly through partial linking of their ETS; ii) reduce carbon
leakage and competitiveness concerns by lowering the carbon price in developed countries; iii) boost
clean technology transfers to developing countries; and (iv) facilitate the implementation of explicit
carbon pricing policies in developing countries at a later stage by putting an opportunity cost on their
GHG emissions.  
Well-functioning crediting mechanisms appear to have very large potential for saving costs,
reflecting the vast low-cost abatement potential existing in a number of developing countries, particularly
China. To illustrate this, the same hypothetical “benchmark” scenario as in the previous Section is
considered. Under this scenario, each Annex I region of the ENV-Linkages model is assumed to establish
a regional ETS that caps GHG emissions at 20% and 50% below 1990 levels by 2020 and 2050,
respectively. As stressed earlier, this scenario is purely illustrative and, at best, based on a transitory
arrangement that is not in itself compatible with meeting ambitious climate change mitigation targets.
Compared with that benchmark scenario, allowing Annex I regions to meet 20% of their commitments
through reductions in non-Annex I countries is estimated to nearly halve their mitigation costs
(Figure 4.9). Raising the cap on offsets allowed from 20% to 50% would bring further benefits. Cost
savings are found to be largest for those Annex I regions that otherwise face the highest marginal
abatement costs – and, therefore, the highest carbon price levels (Figure 4.10) – and/or are most
carbon-intensive. Australia, New Zealand, and Canada fall into both categories, while Russia falls into
the latter. Non-Annex I regions would enjoy a slight income gain from exploiting cheap abatement
opportunities and selling them profitably in the form of offset credits. In this illustrative scenario, China
would be by far the largest seller and the United States the largest buyer in the offset credit market,
accounting for about half of worldwide sales and purchases by 2020, respectively (Figure 4.11).  4.  TOWARDS GLOBAL CARBON PRICING
126  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.9.  Mitigation policy costs under a 50% emission cut in each Annex I region can be cut by allowing
access to a well-functioning crediting mechanism
(Under a 20% cut by 2020 and a 50% cut by 2050 relative to 1990 levels in each Annex I region)
-7.0
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
Canada
Australia &
New Zealand
Non-EU
Eastern European
countries
Russia
United States
Annex I
Japan
EU27 & EFTA
Mitigation cost (income equivalent variation relative to
baseline, in %)
Panel A. 2020
Without crediting mechanism
With crediting mechanism, 20% cap on use of offset credits
With crediting mechanism, 50% cap on use of offset credits
-20.0
-18.0
-16.0
-14.0
-12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
Non-EU
Eastern European
countries
Russia
Canada
Australia &
New Zealand
Annex I
EU27 & EFTA
United States
Japan
Mitigation cost (income equivalent variation   relative to
basel ine, in  %)
Panel B. 2050
Without crediting mechanism
With crediting mechanism, 20% cap on use of offset credits
With crediting mechanism, 50% cap on use of offset credits
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    127
Figure 4.10.  A well-functioning crediting mechanism would lead to a convergence in carbon prices
Carbon prices under a 50% emission cut by 2050 relative to 1990 levels in each Annex I region separately, with and without
crediting mechanisms
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
160.0
180.0
200.0
Non-Annex I
Russia
Non-EU
Eastern
European
countries
EU27 & EFTA
United States
Canada
Australia &
New Zealand
Japan
USD/ t CO2 eq
Panel A. 2020
Without crediting mechanism
With crediting mechanism, 20% cap on use of offset credits
With crediting mechanism, 50% cap on use of offset credits
0.0
100.0
200.0
300.0
400.0
500.0
600.0
700.0
Non-Annex I
Russia
United States
Non-EU Eastern
European
countries
EU27 & EFTA
Japan
Australia & New
Zealand
Canada
USD/ t CO2 eq
Panel B. 2050
Without crediting mechanism
With crediting mechanism, 20% cap on use of offset credits
With crediting mechanism, 50% cap on use of offset credits
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
128  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.11.   Geographical distribution of offset credit buyers and sellers by 2020 under a 50% emission cut
in each Annex I region separately
Panel A. With a 20% cap on use of offset credits in Annex I countries
Brazil
4% India
10%
China
49%
Oil-
exporting
countries
7%
Rest of  
the world
30%
Sellers
United
States
53%
Canada
7%
Australia
& New
Zealand
7%
Japan
7%
Non-EU
Eastern
European
countries
6%
Russia
3%
EU27 &
EFTA
17%
Buyers
Panel B. With a 50% cap on use of offset credits in Annex I countries
Brazil
4% India
10%
China
52%
Oil-
exporting
countries
7%
Rest of  
the world
27%
Sellers
United
States
53%
Canada
7%
Australia
& New
Zealand
7%
Japan
7%
Non-EU
Eastern
European
countries
6%
Russia
3%
EU27 &
EFTA
17%
Buyers
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    129
Box 4.3  CDM baselines, credits and carbon leakage
In order to receive credits under the current CDM, several steps must be undertaken in the project registration and
issuance process. This is designed to ensure that approved CDM projects generate “real, measurable and verifiable
emission reductions” compared to a baseline, which UNFCCC (2005) defines as “the scenario that reasonably represents
the emissions by sources of GHGs that would occur in the absence of the proposed project activity”. The Marrakech
Accords consider three types of baseline, based on either actual or historical emissions, technology type or emissions
from previous, similar project activities.  
The choice of the baseline against which certified emission rights (CERs) are granted does not only have an impact
on the volume of credits generated, but also matters for carbon leakage. This is because an emissions baseline established
before the project is implemented depends on the assumptions made about policies and projects in other sectors and
regions, and their effect on output and emissions within the project boundary. Three approaches can be identified in
setting a baseline:
1. Accounting for the impact of all other CDM projects on the project’s expected emissions. If these other projects
lower the international carbon price and thus reduce leakage from countries covered by binding emission caps within
the project boundary, they should lower the project’s emission baseline. This would be the “theoretically correct”
baseline under current UNFCCC guidelines. However, implementing this approach is complex and costly, and would
likely remain so even under a scaled-up CDM.  
2. Excluding the impact of all other projects on the project’s expected emissions. Because of the complexity and cost of
the first approach, the CDM Executive Board and our model take as the baseline the project’s emission level under a
scenario where some countries – currently most of the Annex I countries – have emission commitments while the rest
of the world does not. This therefore does not account for the effect of other projects on the output from, and
therefore credit generated by, the CDM project. Implicitly, this assumes that all individual CDM projects have a
marginal effect on the world economy.
3. Setting the baseline as the BAU emission level in a hypothetical “no world action” scenario where no country has
binding emission commitments. In this case, CDM projects would receive fewer credits than under approach 1, as
they would be required to more than offset any leakage within the project boundary resulting from action in other
sectors and regions. However, this approach would imply a lower credit volume than the approach in current
UNFCCC guidelines, and indeed it does not appear to fit current practice whereby “market leakage” is not taken into
account in CDM baselines. For example, the CDM Executive Board does not quantify the impact of the EU-ETS on
emissions in non-Annex I countries.
Under all three approaches, Kallbeken (2007b) find that the CDM would lower the carbon price differential
between countries that face binding emissions caps and other countries, and would thereby reduce leakage (all other
factors being equal). However, these leakage reductions are typically smaller under the second, and perhaps most
plausible, approach above. This is because the approach does not account for the fact that implementing all other CDM
projects together reduces international carbon prices, leakage and thereby the projected emissions of any other project
considered. Consequently, the volume of credits that would be granted for each project is higher than both in the
“theoretically correct” approach and – to an even greater extent – in the “no world action” approach. As a result, “too
many” CERs are granted, and the more so the higher the number of projects that are implemented,  i.e. the larger the
share of recipient countries’ emissions that benefits from CERs.
1
In the absence of any constraint on the use of CERs in
Annex I countries, the effect of the CDM under such a baseline would be simply to reallocate emissions between
Annex I and non-Annex I countries, without addressing the fact that actions in Annex I boosted the emissions of
non-Annex I countries before implementation of the CDM.
  In practice, the effects of these alternative approaches have been found to be limited under moderate mitigation
action scenarios – and, therefore, fairly low carbon prices and leakage – and limited CDM use (Kallbeken 2007b).
2
There
are some remaining questions which call for further research. Firstly, is this still true given the current boom in CDM
projects? Secondly, would this still be true under more stringent commitments and a scaled up CDM, especially if fairly
lax caps are put on CDM use in countries covered by binding emission commitments?
1.  Alternatively, if the total volume of credits is fixed (e.g. because of demand constraints), this will imply that fewer projects
can be undertaken.
2. Vöhringer et al. (2006) stress the difference between the marginal impact of an individual project and the combined effect
of all projects together, which can lead to significant leakage. They propose attributing leakage proportionally to
individual projects. 4.  TOWARDS GLOBAL CARBON PRICING
130  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.12.   The gains from direct linking across Annex I emission trading schemes when they are already
linked through a crediting mechanism would be limited
(Under a 50% emission cut in each Annex I region separately by 2050 relative to 1990 levels,
with a 50% cap on use of offset credits)
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
Canada
Non-EU
Eastern European
countries
Oi l-exporting
countries
Russia
Austral ia &
New Zealand
United States
Total all regions
Rest of the world
EU27 & EFTA
Japan
Brazi l
China
India
Mitigation cost (income equivalent variation   relative to
baseline, in %)
Panel A. 2020
Indirect linking through the crediting mechanism and
no direct linking
Full linking and access to  the crediting mechanism
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
Non-EU
Eastern European
countries
Russia
Canada
Austral ia &
New Zealand
Oil-exporting
countries
Uni ted States
EU27 & EFTA
Total all regions
Japan
Rest of the world
Brazi l
China
India
Mitigation cost (income equivalent variation relative to
baseline, in  %)
Panel B. 2050
Indirect linking through the crediting mechanism and
no direct linking
Full linking and access to  the crediting mechanism
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    131
Figure 4.13.   A well-functioning crediting mechanism would have a limited effect on carbon leakage rate but
would lower output losses of energy-intensive industries in Annex I countries
50% emission cut by 2050 relative to 1990 levels in each Annex I region separately, with and without crediting
mechanisms, 2020
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
Without crediting
mechanism
With a 20% cap on
use of offset credits
With a 50% cap on
use of offset credits
% Panel A. Carbon  leakage  rate2 (Annex  I as a whole)
-15.0
-13.0
-11.0
-9.0
-7.0
-5.0
-3.0
-1.0
1.0
3.0
5.0
Australia &
New Zealand
Japan United States Canada Non-EU
Eastern
European
countries
EU27 &
EFTA
Russia
Output (change relative to baseline, in %)
Panel B. Output of energy-intensive  industries in Annex  I regions
Without crediting mechanism
With a 20% cap on use of offset credits
With a 50% cap on use of offset credits
1.  Energy intensive industries include chemicals, metallurgic, other metal, iron and steel
industry, paper and mineral products.
2. The carbon leakage rate is calculated as: [1-(world emission reduction in
GtCO2eq)/(Annex I emission reduction objective in GtCO2eq)]. It is expressed in per cent.
When the emission reduction achieved at the world level (in GtCO2eq) is equal to the
emission reduction objective set by Annex I (in GTCO2eq), there is no leakage overall, and
the leakage rate is 0.  
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
132  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
However, these cost saving and trade flow estimates should be seen as upper bounds, because they
assume a crediting mechanism with no transaction costs and no uncertainty on delivery, as is apparent
from the very low projected offset prices in these simulations (Figure 4.10). In practice, there are
numerous market imperfections and policy distortions which may prevent some of the non-Annex I
abatement potential from being fully reaped. These include transaction costs and bottlenecks,
information barriers, credit market constraints, and institutional and regulatory barriers to investment in
host countries (Chapter 6).
24
This well-functioning crediting mechanism that is modelled here is largely
equivalent to an international (asymmetric) ETS covering all non-Annex I countries, in which each of
them is assigned a target equal to their baseline emissions.
Crediting mechanisms also indirectly link the ETSs of countries covered by binding emission caps
if credits from a single mechanism (e.g. the CDM) are accepted in several different ETSs. Indeed, they
result in partial convergence of carbon prices and marginal abatement costs across the different ETSs,
which improves their cost-effectiveness as a whole. In the illustrative scenario run here, the variance in
carbon prices across Annex I regions is found to decline dramatically as the cap on the use of offsets is
relaxed, becoming fairly small for instance under a 50% cap (Figure 4.10). As a result, once schemes are
indirectly linked through crediting mechanisms, the additional gains from direct linking are smaller than
discussed in Section 4.2. They depend on the degree of carbon price convergence already achieved
through indirect linking, which in turn depends in part on limits to the use of offset credits. The looser
the constraints on the use of credits, the stronger the indirect linkage between systems, and the smaller
the additional gains from explicit linking. For instance, ENV-Linkages simulations suggest that if
Annex I regions are allowed to meet up to 50% of their domestic commitments through the use of
offsets, the overall additional gain from direct linking would be close to zero, although some countries
would still benefit significantly (Figure 4.12).
25
4.3.2. Challenges with crediting mechanisms generally, and the current CDM specifically
In theory, crediting mechanisms may also reduce carbon leakage and mitigate competitiveness
concerns. Compared with a situation where countries covered by binding caps cut their emissions
unilaterally, the availability of credits lowers the differences between carbon prices in participating and
non-participating countries. These differences are an important driver of leakage. However, whether
leakage is actually reduced partly hinges, in practice, on the definition of the baseline against which
credits are granted (Kallbekken et al. 2007a). The baseline used in the modelling work for this chapter
already incorporates some leakage, i.e. it is “too high” (Box 4.3). This baseline corresponds to the BAU
level in non-Annex I countries under emission reduction action in Annex I – an approach close to that
followed by the CDM up to now. As a result, the reduction in leakage from emission crediting could turn
out to be small, or even non-existent (Box 4.3, and Kallbekken 2007b).
26
For instance, in the illustrative
50% Annex I emission cut scenario above, while simulated carbon price levels in Annex I countries fall
drastically when emitters are allowed to meet part of their commitments through offsets, the leakage
rate
27
barely declines (Figure 4.13, Panel A) – although leakage is estimated to be small to start with, in
line with most other existing models.
28
Under this baseline definition, crediting mechanisms primarily
reallocate emissions between Annex I and non-Annex I countries, without addressing the fact that
leakage boosted the baseline emissions of non-Annex I countries in the first place. Nevertheless, by
lowering the carbon price in most Annex I countries, crediting is found to be an effective way to mitigate
the competitiveness and output losses of their energy-intensive industries (EIIs) (Figure 4.13, Panel B).
294.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    133
More importantly, in its current form the CDM raises a number of issues which, if not addressed,
will undermine its ability to deliver the expected benefits:
• Additionality, transaction costs and bottlenecks. The so-called additionality criterion is key to
ensuring the environmental integrity of the CDM. Under this, only emission reductions that can
be attributed to the carbon project give rise to certified emission rights or carbon credits
(CERs). Otherwise, CERs would amount to a mere income transfer to recipient countries
without reducing GHG emissions. However, the transaction costs and bottlenecks associated
with ensuring that CERs are indeed “real, additional and verifiable” are large and well
documented (Capoor and Ambrosi, 2008; Ellis and Kamel, 2007). They have increased as the
CDM has become a victim of its own success, with more than 4 000 projects currently in the
pipeline. Despite these costs, it has been argued that a large share of CDM projects do not bring
about actual reductions in emissions (ICCP, 2007; Schneider, 2007; Wara and Victor, 2008).
These strains on the system have emerged even though the supply of CERs remains lower than
it would under future increasingly stringent emission targets in Annex I countries. For instance,
the annual volume of CERs issued was about 0.25 Gt of CO2eq in 2008. Based on the number
of projects currently in the pipeline and planned, it is expected to reach about 1.4 Gt of CO2eq
in 2012 (Figure 4.14). Under the illustrative scenario  presented above, of a 50% emission cut
in each Annex I region by 2050, the simulated supply is estimated to reach over 3 Gt of CO2eq
in 2020 if up to 50% of domestic commitments can be made through the use of offset credits.  
• Perverse incentives to raise emissions. The CDM is asymmetric in that it rewards emission
reductions but does not penalise increases. As such, the CDM comes close to an emission
reduction subsidy. This makes it subject to the “dynamic inefficiency issue” (Baumol and
Oates, 1988). By reducing firms’ total expected investment costs, the CDM can create perverse
incentives to raise initial investment and output in carbon-intensive equipment, so as to get
emission credits for reducing emissions later, depending on expectations about how future
baselines will be set.
30
The larger the gap between the market price of CERs and the abatement
cost, the greater such perverse incentives would be. This may be seen as a form of
“intertemporal leakage”, whereby expected action tomorrow increases emissions today.
• Reduced incentives for non-Annex I countries to take ambitious mitigation action. Another
incentive problem is that the large financial inflows from which developing countries may
benefit under a future CDM could undermine their willingness to take on binding emission
commitments at a later stage. This is because most of them would earn more under a
well-functioning crediting mechanism than they would under most rules for allocating emission
rights in a world ETS, except for the most favourable rules. For example, non-Annex I
countries as a whole gain more in the illustrative benchmark scenario (50% emission reduction
in Annex I, with Annex I countries allowed up to 50% of offsets to meet their commitments),
than in a scenario where the same world emission reduction (in Gt CO2eq) is achieved by
granting every human being the same amount of allowance (global ETS with per capita
allocation rule). China in particular would lose from moving to a world ETS like this
(Figure 4.15).  4.  TOWARDS GLOBAL CARBON PRICING
134  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.14.  CERs issued under the CDM: trends and breakdown by type of project to 2012
0
200
400
600
800
1000
1200
1400
1600
31-12-2004 31-12-2006 31-12-2008 31-12-2010 31-12-2012
Million CERs
Panel A. Volume of CERs
Already issued
Projected
HFCs, PFCs &
N2O reduction
26.0%
Renewables
35.2%
CH4 reduction &
Cement & Coal
mine/bed methane
18.8%
Supply-side energy
efficiency
11.1%
Fuel switch
7.0%
Demand-side
energy efficiency
1.3%
Afforestation &
Reforestation
0.4%
Transport
0.2%
Panel B. Decomposition  of expected  CERs  in 2012 by type of project
(based only on CERs already issued)
Note: A project has to go through four stages before the associated CERs can be issued:
validation, verification, registration and issuance. The future volume of CERs is projected based
on assumptions over the 2009-2012 period regarding the number of projects that will be
submitted to validation, the share of projects currently at the validation stage that will be
validated, the share of projects that will successfully go through the registration stage, and the
amount of CERs registered projects will effectively generate (issuance success).
Source: UNEP RISØ Center. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    135
Figure 4.15.   A crediting mechanism may lower the incentive for non Annex I countries to join a world ETS
Mitigation policy costs under a world ETS and equivalent Annex I mitigation action with a crediting mechanism
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
Canada
Non-EU Eastern
European countries
Oil-exporting
countries
Russia
Australia &
New Zealand
United States
Annex I
Non-Annex I
Rest of the world
EU27 & EFTA
Japan
Brazil
China
India
Mi tigation cost (income equivalent variation relative to
basel ine, in %)
Panel A. 2020
Under Annex I mitigation action only with a
50% cap on use of offset credits
Under a world ETS  with per capita
allocation rule
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
Non-EU Eastern
European countries
Russia
Canada
Australia &
New Zealand
Oil-exporting
countries
Annex I
United States
EU27 & EFTA
Non-Annex I
Japan
Rest of the world
Brazil
China
India
Mitigation cost (income equivalent variation   relative to
baseline, in  %)
Panel B. 2050
Under Annex I mitigation action only
with a 50% cap on use of offset credits
Under a world ETS with per capita
allocation rule
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
136  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
• The permanence problem. Some types of carbon projects might not involve permanent
emissions reductions. In other words, CERs may be issued for emission reductions made
initially, but these might be offset by increases in the more distant future. This issue could be
significant if sequestration projects such as carbon capture and storage or avoided deforestation
become eligible for the CDM.
31
For instance, carbon storage capacities might not be
maintained,
32
and deforestation might be simply delayed rather than permanently avoided. This
problem could be compounded by the difficulty for firms to commit, and for insurance
companies to cover the risk of non-compliance, over very long periods.
4.3.3. CDM reform options
Despite these drawbacks, crediting mechanisms have the potential to significantly lower the future
mitigation costs incurred by regions covered by emission caps. Therefore, in the absence of a global
permit trading architecture involving all the main emitters, the CDM should be scaled up. This would
require a move from a project-by-project to a wholesale approach in order to reduce transaction costs and
bottlenecks drastically (e.g. Bosi and Ellis, 2005). A number of proposals have been made for how to do
that. These approaches are not mutually exclusive, although potential overlap – in particular risks of
double counting – would need to be carefully addressed. They may also complement, rather than replace
the project-by-project approach, which may have to continue in sectors with dispersed emission sources
(e.g. agriculture), or where emission reductions are clearly additional (e.g. CCS, or some non-CO2
projects such as N2O destruction activities, whose only revenues would be the CERs). The three main
CDM scaling-up options are:
• Bundling and “programmes of activities”. These two forms of scaling up have been eligible
under the CDM since a 2005 decision (4/CMP.1) at the Meeting of the Parties to the Kyoto
Protocol (COP/MOP1) on “further guidance to the CDM”.
33
Bundling involves bringing
together several small-scale CDM project activities to form a single CDM project activity or
portfolio without losing the distinctive characteristics of each project. Credits are obtained for
bundled projects. Under the “programme of activities” approach, credits may be granted for a
range of projects that differ in timing or geographical location (e.g. Hinostroza  et al. 2007).
This may be especially useful in the area of energy efficiency, where the CDM is currently
under-developed.
34
Bundling together small dispersed projects which alone would have
prohibitively high transaction costs could ultimately lead to large emission reductions. It may
also help expand CDM use to geographic regions where it is currently negligible partly due to
the relatively small scale of potential projects, such as in Africa.
• Sectoral crediting mechanisms would further scale up the CDM by allowing emission
reductions at the sector level to yield credits after validation against a pre-defined baseline
(e.g. Baron and Ellis, 2006). This “sectoral CDM” would require setting up sectoral baselines
for selected industries in each potential recipient country. This would raise a number of
methodological issues. In particular, using a standardised baseline for an industry across
countries may not be appropriate. There are good economic reasons for cross-country
differences in emission levels and intensity within a given industry (e.g. differences in goods
and/or production processes, factor prices, or natural resource endowments), including EIIs and
the power sector to some extent (Baron and Ellis, 2006).
35
Intensity baselines (emissions per
output) are often considered to be easier to establish than absolute baselines. However, they
would be more complex to monitor and enforce as they would require measures of both output
and emissions. This approach is also discussed below in the broader context of sectoral
approaches (Section 4.4).  4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    137
• Policy CDM would allow specific government policies to deliver CERs (e.g. Aldy and Stavins,
2008). Eligible policies could be sectoral, in which case they would be equivalent to sectoral
crediting mechanisms. Or they could be cross-sectoral, and might include for instance
renewable energy standards (e.g. a policy of installing energy-efficient light bulbs), building
codes or even possibly the implementation of carbon taxes or a removal of energy subsidies.
One advantage of a policy CDM is that additionality may be easier to check for. However, this
approach would share the drawbacks of technology standards,  i.e. it would run the risk of
mandating the use of specific technologies that could eventually turn out to be costlier than
alternatives, and might also undermine innovation incentives. Furthermore, setting a baseline at
a “policy” level and, especially, monitoring and verifying the emission reductions achieved
from a policy could raise major methodological difficulties and affect the environmental
integrity of the scheme. Also open to question is whether electorates in developed countries
would support the large, transparent payments to developing countries that would likely be
involved if that option were to be used extensively.
While these options could achieve drastic cuts in transaction costs and thereby vastly increase the
volume of credits issued, they would not specifically address the deeper problems of additionality,
leakage and perverse incentives. One way to reduce these concerns might be to negotiate baselines today
for the largest possible number of sectors for a sufficiently long time period (e.g. a decade), and to set
these baselines below BAU emission levels expected without further world mitigation action efforts.
Establishing long-term baselines would address the perverse incentive issue by ruling out the possibility
that any future increase in emissions might, if offset by subsequent reductions, deliver CERs. It would
also minimise the risk of leakage, especially as the number of countries and sectors covered would be
large. Setting baselines below BAU levels might be seen as an insurance against the risk of
over-estimating baseline emissions and the excess supply of CERs, but it may mean that some potential
low-cost abatement opportunities are lost. The main weakness of this approach is that estimating and
negotiating baselines simultaneously across a wide range of countries and sectors would mean
overcoming significant methodological and political obstacles.  
An international agreement on CDM reform could also incorporate built-in “graduation
mechanisms”. This would encourage developing countries to take on increasing GHG mitigation actions
or commitments over time as their income levels converge with those of developed countries, and/or to
stop hosting crediting projects under certain conditions or after a given period of time. Even if the latter
is not agreed internationally, it is likely that developed countries will unilaterally limit CDM offsets to
projects in only some partner countries (e.g. lower income developing countries) and/or some sectors.
Such graduation measures would: i) address environmental integrity concerns; ii) reduce the disincentive
for recipient countries to take on binding commitments once scaled-up CDMs are in place; and iii) help
put world emissions on a path towards meeting ambitious long-run global targets. For instance, the
sectoral and/or country baselines negotiated in the context of a scaled-up CDM might be gradually
tightened, along with some relaxation of restrictions on offset use in countries covered by ETSs, where
additionality would be less of a concern. This would induce some convergence between “hard” permit
and credit prices, albeit at some cost to developed countries. Over the longer run, the tighter baselines
might in turn be converted into binding emission caps, which could then be gradually lowered
(Section 4.4).  
As a radical alternative, some have suggested moving away from a strict accounting,
“tonne-for-tonne” emission reduction logic towards direct support to actions that create progress toward
mitigation in developing countries (Keeler and Thompson, 2008). However, relaxing the additionality
criterion may only make it harder to gather political support in developed countries for the large and
transparent international financial transfers that would be associated with a policy CDM.  4.  TOWARDS GLOBAL CARBON PRICING
138  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
4.4.  The potential and limitations of sectoral approaches
On their own, even large emission reductions in Annex I countries cannot stop climate change
(Section 3.1). Sectoral approaches are being proposed as a way to broaden participation to developing
countries, and therefore to expand the potential for emission reductions and/or lower their cost. They are
also expected to mitigate leakage and competiveness issues.
4.4.1. Forms of sectoral approaches
A range of sector-based mitigation policies has been discussed in the policy debate and the
literature (Baron et al. 2009). They can be classified into three main groups, depending on the level of
commitment they impose on countries:
• Binding sectoral targets. Quantitative emission targets would be negotiated at a national or
international level for specific sectors.
36
A cost-effective way to achieve them would then be to
set up national or international sectoral emission trading systems, under which allowances
would be allocated to firms on the basis of the usual rules (e.g. auctioning or grandfathering).
37
Allowances might be traded within one or several sectoral markets, and possibly between them
and some economy-wide ETSs. The system would involve a cap-setting process, as well as
monitoring, reporting and verifying (MRV) procedures. Binding sectoral targets could also be
achieved through the development and transfer of technologies.  
• Sectoral crediting mechanisms. Another option is to establish sectoral emission baselines
(e.g. for EIIs) at a national or international level; sectors which reduced their emissions below
this baseline would generate credits that could be sold in international carbon markets. This
option would involve a baseline-setting process, MRV procedures, and a crediting mechanism
for verified emission reductions.  
• Non-binding technology-oriented approaches.  The focus here would be on voluntary
agreements to promote more efficient or cleaner technologies, with no reward from the
international community for emission cuts achieved. However, unlike the previous two options,
this approach would put neither a price nor an opportunity cost on carbon. As a result, it would
be unlikely to provide emission reduction incentives to firms in the sectors covered, and for this
reason it is not considered further in our analysis.  
While sectoral approaches could, in principle, be applied across a wide range of sectors and
countries, special emphasis might be placed in practice on the largest emitting sectors and, within those,
possibly on key country players. The argument is that a narrowly-focused agreement covering firms that
share some characteristics and compete among themselves may be easier to achieve than broader
agreements. Indeed, a relatively small number of sectors account for a large share of world emissions. In
particular, EIIs and the power sector account for almost half of current world GHG emissions (excluding
emissions from land use, land use change and forestry), over half of which are in non-Annex I countries
(Figure 4.16). A sectoral approach could also be useful in the international shipping and air transport
sectors due to their significant contribution to world emissions and their transnational character.
384.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    139
Figure 4.16.   Energy-intensive industries and the power sector are responsible for a significant
share of world GHG emissions
Contribution of energy intensive industries
1
and the power sector to world GHG emissions
2
, 2005
Annex I EIIs
6%
Annex I Power
sector
16%
Non-Annex I EIIs
6%
Non-Annex I Power
sector
17%
Annex I Other
27%
Non-Annex I Other
28%
1.  Energy-intensive industries include chemicals, metallurgic, other metal, iron and
steel industry, paper, mineral products.
2.  Excluding emissions from Land Use, Land-Use Change and Forestry.
Source: OECD, ENV-Linkages model.
4.4.2. Increasing emission reductions and lowering mitigation costs through sectoral
approaches
The OECD model ENV-Linkages is used to illustrate the impact on overall mitigation costs and
emission reduction potential of sectoral approaches covering developing countries. Two main types of
scenarios are explored:  
• Scenario 1. Binding sectoral cap with different linking scenarios. A 50% emission cut in each
Annex I region by 2050 from 1990 levels, with full linking across their ETSs (Section 4.2), is
now assumed to be supplemented with a binding sectoral cap in EIIs and the power sector in
non-Annex I countries. Under this, emissions are reduced just under 10% in 2020 and 20% in
2050 relative to 2005 levels.
39
Three versions of this scenario are considered: Scenario 1(a): no
linking,  i.e. each non-Annex I country has to achieve its sectoral target alone; Scenario
1(b): direct linking across non-Annex I regions,  i.e. international sectoral permit trading is
allowed within the non-Annex I area; and Scenario 1(c): full linking,  i.e. permit trading is
allowed within the non-Annex I area as well as between non-Annex I and Annex I countries (a
single ETS).
• Scenario 2. Sectoral crediting mechanism. Here a sectoral crediting mechanism for EIIs and the
power sector is assumed to be introduced in non-Annex I countries. Credits are granted for
emissions reductions against a baseline corresponding to the BAU level in a scenario where
only Annex I countries cut their emissions (by 50% in 2050 relative to 1990). Annex I
countries are allowed to achieve up to 20% of their emission target by buying these credits.  4.  TOWARDS GLOBAL CARBON PRICING
140  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Scenario 1 results: binding sectoral cap and linking options
A binding sectoral cap covering EIIs and the power sector in non-Annex I countries could
substantially reduce emissions worldwide.
40
Owing to the fast emission growth expected in non-Annex I
countries, a 20% emissions cut in these sectors in non-Annex I countries would achieve a larger
reduction in world emissions (compared to a scenario under which the world took no action) than a 50%
cut in Annex I countries by 2050. The emission reductions under the former scenario would be 24% and
30% higher in 2020 and 2050 respectively than under the 50% cut in Annex I countries. Nevertheless,
world emissions would barely decline by 2020 relative to 2005 levels, and would rise slightly beyond
2020. This indicates that in order to achieve more ambitious global emission reductions, either targets
would need to be more ambitious or an agreement would need to include more sectors, which would be a
far cheaper option (Figure 4.17, Panel A).
41
Binding sectoral caps would entail costs which would vary across non-Annex I countries. The
costs involved would depend on the level of emission reductions demanded by the cap, the availability of
cheap abatement options (the shape of the marginal abatement cost curve), the carbon intensity of output,
and whether international permit trading was allowed. For instance, in the illustrative scenario considered
here (Scenario 1 above), India is found to incur larger mitigation costs than China (Figure 4.18), mainly
due to its faster projected BAU emission growth. However, that gap would be reduced substantially if
international permit trading (internal linking) was allowed across non-Annex I regions (compare the
scenarios “with no linking’’ and “with direct linking within non-Annex I” in Figure 4.18). Despite facing
a smaller emission reduction relative to BAU than Annex I countries (-25% versus -30% by 2020 and
-40% versus -60% by 2050) and benefiting from their larger potential to reduce emissions more cheaply,
non-Annex I countries would incur larger costs (more than 3% of their joint income in 2020, compared
to less than 1.5% for Annex I countries), reflecting their higher carbon intensity, particularly by 2020.
Linking sectoral ETSs in non-Annex I countries to economy-wide ETSs in Annex I countries
could also generate aggregate gains by exploiting the wide heterogeneity of (marginal) abatement costs
between the two areas, as long as carbon prices differ sufficiently prior to linking. At the same time, such
linking could have significant redistributive effects across countries. Therefore, allocation rules may need
to be adjusted upon linking to ensure that the gains from linking are shared widely across participating
countries. However, in the scenario considered here (Scenario 1c “full linking”), there is virtually no
aggregate gain. This is because the initial difference in carbon prices across the schemes happens to be
low (2005 USD 75 per tonne of CO2eq in Annex I countries, versus USD 98 in non-Annex I countries in
2020) because they are both quite similar in their stringency. The general rule – that permit sellers in the
market with the lower pre-linking carbon price gain, while permit buyers lose (and vice versa) – applies
here for countries when two region-wide ETSs are linked, rather than for firms. Non-Annex I countries
(India, oil-exporting countries) which bought permits from China before linking with Annex I, lose from
the increase in the permit price after linking. On the other hand, Annex I countries (the United States,
Russia) which sold permits to the rest of Annex I lose from the price decline induced by competition
with China.4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    141
Figure 4.17.  The impact of sectoral approaches in non-Annex I regions on world emission reductions and
mitigation costs
Panel A. 50% emission cut in Annex I by 2050 and binding sectoral cap (20% cut by 2050) in
non-Annex I covering EIIs and the power sector
-10
-8
-6
-4
-2
0
2
4
6
8
10
No linking Direct linking
within non-
Annex I
Full linking
Change   relative to 2005, %
World emissions
2020
2050
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
No linking Direct linking
within non-
Annex I
Full linking
Income equivalent variation relative to baseline, in  %
World mitigation  costs
2020
2050
Panel B. 50% emission cut in Annex I by 2050 and sectoral crediting mechanism covering EIIs and the power sector
1
0
10
20
30
40
50
60
70
80
90
100
Baseline Without
sectoral
crediting
mechanism
With sectoral
crediting
mechanism
Change   relative to 2005, %
World emissions
2020
2050
-1.8
-1.6
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
Without sectoral
crediting mechanism
With sectoral crediting
mechanism
Income equivalent variation relative to basel ine, in  %
World mitigation  costs
2020
2050
1.  With a 20% cap on use of offset credits.
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
142  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.18.   Mitigation costs under an international ETS in Annex I and binding sectoral caps in
non-Annex I regions
50% cut in Annex I regions and 20% cut in EEIs and power sector in non-Annex I regions by 2050
-10.0
-9.0
-8.0
-7.0
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
India
Oil-exporting
countries
Canada
Australia &
New Zealand
China
Non-EU Eastern
European countries
Non-Annex I
Russia
United States
Annex I
Rest of the World
Brazil
EU27 & EFTA
Japan
Mi tigation cost (income equivalent variation   relative to
baseline, in %)
Panel A. 2020
Without any  linking
With direct  linking within non-Annex I sectoral
schemes
With full linking between Annex I economy-
wide and non-Annex I sectoral schemes
-25.0
-20.0
-15.0
-10.0
-5.0
0.0
Non-EU Eastern
European countries
Russia
Oil-exporting
countries
Canada
Australia &
New Zealand
India
Non-Annex I
China
Annex I
EU27 & EFTA
United States
Rest of the World
Japan
Brazil
Mitigation cost (income equivalent variation   relative to
baseline, in  %)
Panel B. 2050
Without any  linking
With direct  linking within non-Annex I sectoral
schemes
With full linking between Annex I economy-wide
and non-Annex I sectoral schemes
Note: All scenarios combine a 50% emission cut in Annex I (relative to 1990 levels) and a 20% cut in EIIs
and the power sector in non-Annex I (relative to 2005 levels) by 2050.
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    143
Figure 4.19.   The impact of sectoral crediting on mitigation costs in Annex I and non-Annex I regions
(Under a 20% cut by 2020 and a 50% cut by 2050 relative to 1990 levels in each Annex I region)
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
Canada
Australia &
New Zealand
Non-EU Eastern
European countries
Oil-exporting
countries
United States
Russia
Annex I
EU27 & EFTA
Japan
Non-Annex I
Rest of  the World
China
Brazil
India
Mitigation cost (income equivalent variation relative to
baseline, in %)
2020
Without sectoral crediting mechanism
With sectoral crediting mechanism
covering EIIs and power sector(1)
-25.0
-20.0
-15.0
-10.0
-5.0
0.0
5.0
Non-EU Eastern
European countries
Russia
Canada
Australia &
New Zealand
Oil-exporting
countries
Annex I
EU27 & EFTA
United States
Japan
Non-Annex I
Rest of the World
Brazil
China
India
Mitigation cost (income equivalent variation relative to
baseline, in  %)
2050
Without sectoral crediting mechanism
With sectoral crediting mechanism covering
EIIs and power sector(1)
1.  With a 20% cap on use of offset credits.
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
144  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Scenario 2 results: sectoral crediting mechanism
The second scenario, the sectoral crediting mechanism, would affect emissions only through a
small decrease in carbon leakage. Thus, it has a limited effect on world emissions, which are found to
rise sharply relative to 2005 (Figure 4.17, Panel B).
42
Like any other crediting mechanism, however,
sectoral crediting in developing countries can lower the cost of achieving an emission reduction target in
developed countries (Section 4.3). Because it lowers mitigation costs, sectoral crediting might still
indirectly help achieve more ambitious targets by encouraging Annex I countries to adopt more stringent
objectives. It would also be expected to reduce considerably the transaction costs and bottlenecks
experienced in the current CDM, since credits would be granted on the basis of a sector-wide baseline
rather than on a project-by-project basis.
43
Compared with a scenario where Annex I meets its 50%
emission reduction objective alone, and despite the fairly restrictive 20% limit on offset use, a
well-functioning sectoral crediting mechanism appears to lower Annex I mitigation costs (in
income-equivalent terms) from 1.3% to 0.8% in 2020, and from 2.6% to 1.5% in 2050 (Figure 4.19). A
model simulation (not reported) in which this crediting mechanism is expanded to all sectors of
non-Annex I economies finds costs to decline only marginally further.  
4.4.3. Impact of sectoral approaches on carbon leakage and competitiveness concerns
44
Binding sectoral targets and – to a lesser extent, depending on their design – sectoral crediting
mechanisms have the potential to reduce carbon leakage. Since leakage fundamentally results from
incomplete coverage of binding mitigation action, sectoral targets for EIIs and the power sector address it
automatically. Leakage would then only be likely to occur in non-covered sectors in developing countries
or to a more limited extent in sectors with very lax targets. In Scenario 2, sectoral crediting in EIIs and
the power sector is found to lower carbon prices in Annex I countries, thereby also reducing leakage to
other (non-covered) sectors in non-Annex I countries, and overall leakage rates (Figure 4.20). In practice,
the extent to which sectoral crediting reduces leakage depends in part on the baseline against which
credits are granted (Section 4.3). In particular, if firms ultimately receive the proceeds from credit sales,
they would benefit from a surplus that could lower output prices and, therefore, increase local demand,
output and emissions.
45
  This would apply to industries sheltered from international competition, such as
the power sector in many developing countries. If agreed sectoral baselines are set in a way that does
acknowledge this effect, they might be “too high”, in which case sectoral crediting could increase rather
than reduce leakage (Bollen  et al.  1999, 2005). No such problem arises in the above scenarios, as
baselines are assumed to be set before the start of the whole compliance period, i.e. before sectoral
crediting is implemented.  
Sectoral approaches may also help to reduce competitiveness concerns in developed countries by
“levelling the playing field” in internationally competitive industries (e.g. Sawa, 2008). A binding
sectoral cap in EIIs in major developing countries can be expected to curb and, depending on its
stringency, even possibly reverse the market share and output losses of firms in Annex I countries by
pricing the emissions of their non-Annex I competitors. For instance, in the illustrative 50% Annex I
emission reduction scenario explored previously, the output loss of EIIs in Annex I countries appears to
be significantly reduced when a sectoral cap is put on EIIs and the power sector in non-Annex I countries
(Figure 4.21). This is especially the case if Annex I and non-Annex I markets are linked and carbon
prices fully converge. By reducing the carbon price, sectoral crediting is also found to limit the output
losses of EIIs in Annex I countries. Whether sectoral crediting is more effective than a sectoral cap in
addressing competitiveness problems depends in part on which approach achieves the strongest degree of
convergence in carbon prices between developed and developing countries. While linking economy-wide
ETSs in developed countries to sectoral ETSs in developing countries can achieve full convergence,
sectoral crediting cannot – at least while there are constraints on offset credit use. As simulated here, 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    145
sectoral crediting puts a relatively low opportunity cost on carbon in non-Annex I countries, and thereby
has smaller effects on the output of EIIs in Annex I countries than a sectoral cap approach. However, the
two simulations cannot be readily compared as they achieve different world emission reductions.
Figure 4.20.   Sectoral crediting would lower carbon leakage rates
Carbon leakage rate for Annex I as a whole, under a scenario where emissions in Annex I regions are cut by 20% by 2020 and
50% by 2050 relative to 1990 levels
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
50% emission reduction in Annex I by 2050, with
sectoral crediting mechanism (2)
50% emission reduction in Annex I by 2050,
without sectoral crediting mechanism
%
2020
2050
1.   The carbon leakage rate is calculated as: [1-(world emission reduction in GtCO2eq)/(Annex I emission reduction
objective in GtCO2eq)]. It is expressed in per cent. When the emission reduction achieved at the world level (in
GtCO2eq) is equal to the emission reduction objective set by Annex I (in GTCO2eq), there is no leakage overall, and
the leakage rate is 0.  
2.   With a 20% cap on use of offset credits.
Source: OECD, ENV-Linkages model. 4.  TOWARDS GLOBAL CARBON PRICING
146  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Figure 4.21.  Impact of sectoral approaches on the output of energy-intensive industries
(% deviation relative to baseline)
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
Without sectoral
crediting
With sectoral
crediting
Without any
linking
With ful l
linking
Without sectoral
crediting
With sectoral
crediting
Without any
linking
With ful l
linking
% 50% cut  in
Annex I
50% cut  in Annex  I
and 20% binding
sectoral cap in
non-Annex I
50% cut  in
Annex I
50% cut in Annex  I
and 20% binding
sectoral cap in
non-Annex I
Non-Annex I
regions
Annex I regions
1.  Energy-intensive industries include chemicals, metallurgic, other metal, iron and steel industry, paper, mineral
products.
2.  The scenario "50% cut in Annex I" and "50% cut in Annex I and 20% binding sectoral cap in non-Annex I" are
not directly comparable as they do not acheive the same emissions reductions, with the latter one being more
stringent than the former one.
Source: OECD, ENV-Linkages model.
4.4.4. Limits of sectoral approaches and options for the future
While sectoral crediting would reduce transaction costs and bottlenecks, it may not necessarily
address major concerns over the current CDM regarding additionality, perverse incentives to raise
emissions, and, to some extent, leakage (Section 4.3). A sectoral crediting mechanism would also raise
the question of how to transfer the carbon price signal to firms (Baron et al. 2009). In practice, under this
system, it is generally expected that governments in developing countries would receive the credits
generated from below-baseline sectoral emissions in their countries, and would then need to find ways to
induce firms to effectively reduce their emissions. In principle, this could be achieved by: i) setting up a
domestic carbon tax; or ii) a firm-level crediting mechanism under which local firms in the sector would
be assigned baselines (reflecting the overall sectoral baseline) and would receive credits for emission cuts
below those baselines. This might prove difficult in practice, however, especially for countries with weak
institutions. A weaker price signal for firms would reduce emission reduction incentives and sectoral 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    147
crediting would achieve lower cuts than expected. The simulations discussed above implicitly assume
that the price signal is fully transmitted to firms.  
In any event, sectoral crediting is likely to have to evolve eventually into more binding
arrangements such as sectoral caps, which raise smaller environmental integrity problems, but potentially
allow larger world targets to be adopted. While, in principle, both sectoral crediting and absolute sectoral
emission caps can be designed to achieve similar emission cuts, they imply very different distributions of
mitigation costs across countries. Given the fast rate at which emissions are projected to rise in most
developing countries under a BAU scenario, meeting ambitious world targets through sectoral crediting
alone would mean that developed countries would need to have negative emission level objectives (or to
store more CO2 than they are emitting)  by 2030-2040 (Chapter 6). There would also have to be either lax
or no constraints on offset credit use so that these targets could be met. This would impose large
economic costs on developed countries, while developing countries would gain from this world
mitigation framework. Assuming that this arrangement is implausible, a sectoral crediting approach
would therefore have to evolve into binding caps, at least for key developing country emitters.
Simulation results suggest that even relatively moderately stringent caps on EIIs and the power sector
have the potential to generate large emission reductions relative to the baseline. Over the lifetime of the
sectoral crediting scheme, baselines could be progressively tightened –  i.e. set further below BAU
emission levels – from one commitment period to the next.  
One option that has sometimes been put forward in the policy debate would be to start with
sectoral intensity targets – in which the allowable amount of emissions is a function of future output –
rather than absolute targets. At the sectoral level, emission intensity tends to be driven by technology
choices and energy efficiency, rather than by output. This makes it easier to identify the changes in
advance, and thereby the costs needed to meet an intensity target, while the costs of absolute targets are
more uncertain as they depend on future output (Bradley et al. 2007). However, one of the challenges for
emission trading is to transform the intensity target into an absolute amount of emission rights that can be
traded (Section 4.2). This would require an initial allocation of permits based on a projected output path,
followed by regular adjustments of permit supply to reflect (unexpected) output growth developments. If
permits can be traded between sectors, achieving the intensity target in a given sector would not be
straightforward, in part because newly emitted permits could be bought by firms from another sector
(Box 4.3 above). An alternative option is to assign permits for a sufficiently long time period based on a
projected sectoral output path, without adjusting permit supply before the next commitment period. In
this case, the intensity target ultimately boils down to a particular rule for allocating permits under an
absolute cap, and sectoral schemes could be easily linked to other sectoral or economy-wide systems.
However, this approach does not provide insurance against the risk of mitigation cost increases that
would result from higher-than-expected output growth.  
4.5.  Regulatory issues and the role of financial markets
Carbon markets will naturally develop as more and more countries undertake mitigation actions.
These carbon markets are expected to become large. They are likely to reach 1% of world GDP by 2050
if Annex I countries alone reduce their emissions by 50% (relative to 1990 levels), and 5% of world GDP
under a global carbon price scenario that stabilises overall GHG concentration below 550 ppm CO2eq.
46
Institutions and rules will be needed to foster their development and to address risks that are expected to
emerge within a linked system of multiple independent and heterogeneous emission trading schemes.
This Section discusses these risks and outline the institutions needed to overcome them. 4.  TOWARDS GLOBAL CARBON PRICING
148  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
4.5.1. The risks associated with the development of carbon markets
The development of carbon markets raises four main risks (Table 4.6):
• An environmental risk, which may be the major risk associated with a linked system of several
independent and heterogeneous ETS and crediting mechanisms (Sections 4.2 and 4.3).
• Lack of market liquidity. Liquid primary markets foster the emergence of derivative
instruments (futures, forwards, options, and swaps) that would lower the cost for firms to insure
against future carbon price uncertainty.
47
Liquid markets would also reduce the opportunities
for market manipulation. Markets can fail to become liquid for several reasons. If the treatment
of two types of allowances differs across two systems – e.g. due to limits on the use of the other
system's allowances, or the use of a discount rate for allowances whose environmental quality
is perceived to be low – the allowances would be imperfect substitutes and traded at different
prices in spot markets, thereby leading to market fragmentation. Likewise, if some credits entail
specific risks, as is the case under the current CDM, they will be traded at different prices to
those of other units in the primary market.
48
For instance, the price difference between future
EU permits (EU allowances, EUA) and high-quality – i.e. those with the higher price – forward
CER contracts on the primary market was around EUR 10 in 2008 (World Bank, 2008). These
differences disappear in secondary markets after the CER has been traded once. Differences in
permit design features in terms of lifetime or banking possibilities would also reduce the
liquidity of markets. Such differences would also affect futures and other derivatives in carbon
markets. Investors might hold different expectations for possible programme changes in the
various emission trading systems, thereby making allowance futures imperfect substitutes and
fragmenting the derivatives market. Nonetheless, market liquidity concerns could be reduced in
the future as the size of carbon markets increases with broader country participation and more
stringent objectives.  
• Risk associated with the development of derivative markets. Speculative trading is expected to
play an important role in the development of carbon derivative markets. This is because, unlike
in other commodity markets, if permits are mainly auctioned, most regulated firms will tend to
hedge the cost of their compliance obligation by buying allowance futures and financial traders
will have to take most of the opposite position, selling allowance futures, thereby taking part of
the net risk.
49
The role of speculative trading is expected to be reflected in futures prices in the
form of financial traders' risk premia.
50
• A counterparty risk. With 70% of carbon trading in Europe being conducted through bilateral
negotiations between participants – the "over-the-counter" (OTC) markets – and with most of
this trading being for deferred delivery (Point Carbon, 2007), the counterparty risk, the risk to
each party of a contract that the counterparty will not live up to its contractual obligations, is
significant in current carbon markets and could lead to market dysfunction and reduced
cost-effectiveness. As markets develop and if transactions continue to operate mainly through
OTC markets, this risk could increase. However, it could be reduced if organised exchanges
expand, for example through the intermediation of clearinghouses that verify trade orders and
net out offsetting contracts by the same clearing member. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    149
4.5.2. Institutions and regulations to address carbon market risks
Policies aiming at environmental integrity
Given the various incentives for countries to free ride and to increase their emissions, it has
sometimes been suggested that an independent international institution acting as a central bank could
help achieve mitigation targets at least cost, while preserving environmental integrity and anchoring
carbon price expectations (e.g. Grubb and Neuhoff, 2006; Yohe, 2007 and Edenhofer  et al. 2007).
Indeed, the comparison between emission credits and money has often been made. In practice, however,
countries are unlikely to accept such a transfer of their power, and attempting to create a new global
institution would run the risk of further delaying mitigation action (Mc Kibbin, 2007). Therefore, the
following discussion tries, as far as possible, to build on existing institutions and rules.  
Table 4.6.  Regulations to address carbon market risks
Carbon market risks  Consequences  Provisions/recommendations
Environmental risk  The system (international or emerging from
regional initiatives) does not achieve its
emission target
Agree within a centralised institution (e.g. UNFCCC
body) on emission targets, emission trading schemes
design features, cost-containment measures, links to
other systems, and MRV procedures
Use complementary compliance mechanism, such as
a system of performance bonds  
Market liquidity risk  Lack of liquidity would imply:
• Larger carbon price volatility
• Higher transaction costs
• Risk of market power problems
• Higher cost of derivatives
Allow regular spot sales of short-term permits
Allow banking  
Ensure credible commitments on future mitigation
policies (to foster the development of derivative
markets)
Derivative markets risk    Could make financial markets unstable  Harmonise regulations on position limits for financial
traders  
Eventually introduce limits on banking for financial
traders  
Identify financial market authorities responsible for
carbon markets  
Counterparty risk  Market dysfunction, reduced cost-effectiveness  Extend the access to clearinghouses or introduce
penalties for performance failure in contracts
Ensure credible commitments to future mitigation
policies (to limit the risk of imbalances between
allowances supply and demand)
Given the difficulty of enforcing international rules in sovereign states, negotiations and consensus
building must be at the core of the development of the carbon market. In order to facilitate future linking
and to maximise its market liquidity benefits, participating governments should seek to agree on their
targets and the emission trading schemes design features that will need to be harmonised prior to linking,
including cost containment measures (Section 4.2), decisions to link to another system, and how to
co-ordinate monitoring, reporting and verification efforts (Haites and Wang, 2006). However, this has
not happened so far in practice, as existing ETSs have very different rules on allocation methodologies,
non-compliance provisions, and allowable offsets (e.g. Ellis and Tirpak, 2006; Reinaud and 4.  TOWARDS GLOBAL CARBON PRICING
150  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Philibert 2007). Centralised institutions that support implementation of the UNFCCC, the Kyoto
Protocol, and any future protocol could help by providing a framework to discuss issues of linking
national and regional ETSs. Already, a centralised permit registry has been created under the UNFCCC,
and the Kyoto Protocol’s project-based mechanisms (CDM and Joint Implementation) are managed
centrally.
A number of options exist to reduce the risks of non-compliance and to relax the target in future
compliance periods once systems are linked. However, none of them would fully address the issue of
enforcing international commitments by sovereign governments, which will ultimately require adequate
participation incentives to be provided (Chapter 6). As a general principle, longer-term agreements on
well-defined emission caps would help limit the need and room for frequent re-negotiations. One
possible complementary compliance mechanism could be the emission of performance bonds. Under this
approach, governments would put some of their own bonds before the start of a compliance period into
the hands of a compliance committee (e.g. a UNFCCC body), which would then have the right to either
sell those bonds in the open market if the country fails to meet its commitment, or return them to
governments if they do. International agreement on such a system may be hard to reach. Nevertheless, it
could be an improvement over the penalty system embedded in the Kyoto protocol (which involves
making up excess emissions in the next period plus a 30% penalty), which has proved to be weak in the
absence of a long-term framework. A reserve, such as the existing “commitment period reserve” under
the Kyoto protocol, which requires governments to keep a certain percentage (90%) of their assigned
units in a fund, would also limit the risk of "overselling" by participants that do not meet their targets.
However, this could come at the cost of imposing limits on trading and weakening the cost-effectiveness
of the scheme (OECD, 2001). Yet another alternative would be to use trade sanctions, but these have
been found to be costly and might trigger trade retaliation, rather than greater country co-operation
(Chapter 3).  
An alternative to explicit enforcement mechanisms would be to give governments incentives to
comply, e.g. through a system of buyer liability, under which buyers would be liable for the poor quality,
in terms of environmental performance, of the permits or offsets they hold (Baron, 2000; Victor, 2001;
Keohane and Raustiala, 2008).
51
The validity of permits or offsets emitted by a given country would be
assessed on a regular basis, and if some permits of the country are found to be of a poor quality, all of
them would then be discounted on a national basis. Under this system, (net) selling countries would be
induced to improve the quality of their permits so as to increase export revenues, while buyers would
have similar incentives to gather information in advance,  and possibly buy low-quality permits at a
discount price, as on the international sovereign bond market. One major weakness of this proposal,
however, is that quality checks would probably have to be performed by an independent agency,
52
which
may thus face the same problems as the current CDM executive board. Furthermore, improved
environmental integrity would come at the cost of reduced market liquidity, because permits would have
different prices according to their quality and country of issuance. Also, the system ultimately rests on
the willingness of (net) buying countries to enforce penalties on their domestic emitters. If buyers were
reluctant to endorse such an approach, they could instead apply a penalty later, in the form of discounts
on all permit imports from the selling country concerned (Box 4.2).
Reaching agreement on standards and procedures for validating and verifying the domestic and/or
international offset credits accepted within ETSs is also essential, since the use of such offsets has effects
across linked systems, even if only one of these formally recognises them (Sections 4.2 and 4.3). The
easiest way to achieve agreement would be through the use of a common centralised institution to
manage the offset verification programme, as happens for the CDM. An alternative would be to share
full information about standards and procedures so that mutual confidence is built in offsets’
environmental integrity. Reforming the CDM to reduce the additionality issue (Section 4.3) and lower 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    151
transaction costs and bottlenecks would help to make allowances from ETS and CERs closer substitutes,
thereby also increasing market liquidity.  
Policies to improve market functioning
Some permit design features can foster the development of derivative markets and lower the cost
of hedging. Permits are usually valid for only one single tonne of carbon (“short-term” permits), and the
lifetime during which they can be used is either fixed (a particular year or compliance period), or
undefined if permits can be banked. Liquid spot markets (through regular spot sales of short-term
permits), banking and credible commitments on future mitigation policies may be the most effective way
to foster the development of least cost derivatives. Some imperfect substitutes for futures contracts have
also been proposed to help the formation of future prices, such as the emission of future vintages of
allowances (if permits cannot be banked), or of long-term permits that grant the holder the right to emit
one tonne of carbon annually (e.g. McKibbin and Wilcoxen, 2006).
53
However, each of these instruments
has its own characteristics, risks and price. If different types of instruments are allowed to multiply for
future delivery, this can also fragment the market and possibly reduce liquidity in particular instruments.
This could occur for instance if credits cannot be banked between compliance periods and multiple
vintages are traded in spot markets alongside derivative contracts. For these reasons, short-term permits
that can be banked between compliance periods seem preferable. Long-term permits can be considered to
show governments’ commitment and to create a constituency (permit holders) that supports mitigation
policies, but they also run the risk of fragmenting the market. Therefore they may be primarily
considered in cases where the credibility of the scheme could be weak otherwise, or once carbon markets
are sufficiently large.  
Carbon market regulation will need to strike the right balance between environmental integrity,
stability and liquidity objectives. In particular, one open issue is whether position limits on financial
traders in spot and derivative commodity markets should also be set in permit markets in order to limit
the risk of sudden or unwarranted carbon price fluctuations. One option might be to impose limits on
banking for financial traders, to prevent them from possibly banking large amounts of allowances in
response to changes in expectations and thereby being able to manipulate and cause large fluctuations in
spot markets. Such restrictions would have to be weighed against the fact that financial traders will
provide liquidity and firms will need derivative markets to hedge against price uncertainty. Since within
a linked system, limits on the positions of financial traders in derivatives markets would spread across
schemes, national and/or regional regulators would also have to co-ordinate and possibly harmonise
regulatory frameworks. The supervision of carbon markets will typically fall under existing financial
market authorities, but for countries with multiple financial market authorities, those responsible for
carbon markets will have to be clearly identified. Broad regulations that already exist or will be
developed in response to the recent financial crisis will also apply, and thus should incorporate the
carbon market dimension.  
In the short run, it will remain a major challenge to devise regulations that generate confidence
among participants and that protect against sources of future market fragility, while not impeding
innovation and market development. Given the wide variety of possible permit design and regulatory
features in emission trading schemes, best practice may emerge only gradually and harmonisation is
likely to take time. During that transitory period, in order to avoid the risk of significant market
disruptions caused by linking, limits to trading between schemes could be set up that would then be
gradually phased out as knowledge about best practice builds up and scheme design features are
harmonised. The creation of a carbon market working group made up of international regulators, perhaps
as part of the Financial Stability Board, could facilitate exchange of information about regulations, risks
and harmonisation needs.  4.  TOWARDS GLOBAL CARBON PRICING
152  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
Finally, there are several ways to address the counterparty risk that arises from the fact that a large
proportion of transactions take place in over-the-counter markets. Contracts could specify penalties for
performance failures and allow the seller or buyer to complete the transaction with another party after a
specified short number of days. It is also possible to let participants access a clearinghouse even though
transactions are over-the-counter – as is the case under the EU ETS for small players – or to require
transactions above a certain size to be passed through clearinghouses (Point Carbon, 2007). If there is
concern that delivery failures may occur because the number of allowances is smaller than the demand
coming from financial traders with short futures positions, limits on futures positions could be tightened,
although this may also increase the cost of hedging. Limiting the uncertainty on long-term emission
reduction commitments – and the associated amount of allowances – as well as extending ETSs to other
countries will also be crucial to address the counterparty risk.
Notes
1.  The issue of carbon leakage and competiveness concerns of climate change mitigation policies is
also analysed in Chapter 3.
2.  This Section focuses on subsidies affecting the  demand for fossil fuels – including indirectly
through electricity subsidies – in non-OECD countries only. Though OECD countries do provide
subsidies to energy production and consumption, these are generally indirect and often not
transparent, which makes it very difficult to estimate their magnitude. Direct subsidies to fossil fuel
consumptions in OECD countries are small in  comparison to non-OECD countries. Subsidies
targeted to renewable energy sources in OECD countries are not covered in the analysis here.
3.  In particular, the estimated gap may reflect market imperfections other than subsidies, such as
differences in quality or uses, imperfectly competitive behaviour or a lack of representativeness of
the chosen reference price. Although this approach treats all subsidies as if they were consumer
subsidies, removing a producer subsidy will have, in practice, a different impact from removing a
consumer subsidy, even if both types of subsidies ultimately contribute to lower the price and
increase the consumption of the corresponding energy source. IEA (1999) – discussion of the pros
and cons of the “price gap” approach.
4.  This is the case for instance when energy prices are set administratively.
5.  This analysis assumes that the subsidy wedges remain constant in the business as usual (BAU)
scenario. Alternatively, assuming some decoupling of domestic energy prices relative to world
energy prices would lead to an increase in subsidy wedges over time, given the projected rise in
world energy prices in the BAU scenario. The impact of subsidy removal on emissions would
therefore be larger under that alternative assumption.
6.  These reductions are comparable with those reported in IEA (1999), taking into account that the
sample of countries for which subsidy data have been collected here covers a much larger share of
world energy demand.
7 .  In the case of a multilateral removal of energy subsidies, oil and gas prices fall by more than coal
prices, mainly because coal is generally less heavily subsidised. This change in relative prices
would induce a substitution from coal to oil and gas, which are less polluting, and would therefore
reduce GHG emissions. This effect would not occur in the case of a unilateral removal, since no
country acting alone has a sufficiently large impact on world relative prices of coal, oil or gas.  
8.   This regional aggregate consists of very different economies (Armenia, Azerbaijan, Belarus,
Croatia, Georgia, Kazakhstan, Kyrgyzstan, Moldova, Tajikistan, Turkmenistan, Ukraine,
Uzbekistan), where the removal of the subsidies induces a dramatic shift in the economic structure 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    153
towards low-productivity sectors. The resulting overall productivity loss more than offsets the
welfare gains from the subsidy removal.
9.   It is often argued that subsidies are justified on equity grounds, for instance to alleviate poverty. In
this simulation, the budgetary saving obtained from the subsidy removal is entirely refunded to
households through a lump-sum (non-distorting) way. This implies that subsidies for the
consumption of fossil fuel are replaced by a direct and larger transfer to households. Alternatively,
this transfer could be used to reduce other distorting taxes, which would increase the real income
gain from subsidy removal, or to reduce poverty in a more targeted and efficient way than through a
uniform subsidy to fossil fuel consumption.
10.   An additional explanation for the small magnitude of world real income gains is that the fall in
world fossil fuel prices induces producers to reduce their supply, leaving more of their reserves in
the ground. This leads to a GDP loss, all other things being equal.
11.  Chapter 1, Table 1.2  –  emission path corresponding to the “550 ppm-base” scenario.
12 .  This world GDP cost estimate is lower, by  0.4%, than the one presented in Chapter 1. This is
because, here, energy subsidies are explicitly incorporated in the BAU scenario, which lowers the
mitigation cost because of so-called “second-best effect”, Burniaux et al. (2009), Box 2.  
13 .  Another reason is that he “second best” GDP gain coming from the reduction of the share of sectors
that are subject to distortion because of energy subsidies is then lost, Burniaux  et al. (2009),
Section 3.
14.  For regions with lower pre-linking carbon prices, the gains from exporting permits more than offset
the additional economic costs from the increase in carbon prices, in the absence of market
imperfections. Likewise, for regions with higher pre-linking carbon prices, the reduction in
mitigation costs associated with the decline in carbon prices more than offsets the cost of importing
permits.  
15.  More broadly, the OECD ENV-Linkages model incorporates many market imperfections and
distortions and, therefore, the impact of permit trading on each participating region has to be
interpreted in a second-best context. As countries sell permits abroad, imports must rise and/or other
exports must decline in order to satisfy the exogenous balance-of-payments constraint (see
Annex 2). Restoring the external balance requires an appreciation of the real exchange rate, which
triggers costly reallocation of capital across sectors, reduces aggregate output and, in some cases,
lowers income and welfare. These model features tend to exacerbate Dutch disease effects.  
16.  Carbon price volatility may still increase in one of the two schemes if the other is subject to larger
and/or more frequent shocks, and is large enough to have significant influence on the overall carbon
price after linking.
17.  However, under one-way credits to firms in system linking, firms in system A would be penalised
by not being allowed to sell B.
18.  By changing the distribution of emission reductions, linking will affect the co-benefits of mitigation
policies in terms of reduced local air pollutant emission. This calls for pricing of local air pollution
benefits separately through local, targeted (e.g. transport and electricity) taxes.
19.  Grand-fathering consists in allocating permits for free on the basis of historical emissions.
20.  One exception may be if firms – especially those in the region with the higher initial carbon price –
expect permits to continue to be grandfathered in the future. This would undermine their incentives
to reduce emissions compared with one-off grandfathering which generates only windfall gains. 4.  TOWARDS GLOBAL CARBON PRICING
154  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
21.  It should be noted that under a system that combines a carbon tax and an ETS, regardless of whether
systems are linked or not, the overall amount of emissions is not fixed, unless the cap of the ETS
and/or the carbon tax are adjusted in order to achieve a given joint target.
22.  Linking may still increase emissions if the cap is set on the basis of a strongly overestimated GDP
and turns not to be binding ex post, as extra permits could then be sold to other schemes upon
linking. This problem is a “hot air” issue.  
23.   Once schemes are linked, there is also an incentive for net permit seller countries to issue more
permits. However, this can be addressed through an appropriate regulatory framework (Section 4.5).
24.   For instance, under a USD 20 carbon price in Annex I countries, Bakker et al. (2007) tentatively
estimate that the amount of emissions abated through crediting projects in non-Annex I countries
might be reduced by a factor of up to two if these barriers were taken into account.
25.  Full linking between ETSs implies that the 50% cap on offsets applies to the Annex I region as a
whole rather than to each country individually. Full linking in addition to indirect linking through
the CDM generates a marginal gain for regions taken as a whole. However, some countries would
lose, notably Australia and New Zealand, because  they benefit from a smaller amount of offset
when the cap on offset is set at the Annex I level as a whole and allocated across countries in a
cost-effective manner.
26.   At a more basic level, the CDM may increase leakage if changes in the whole supply chain of the
project considered are not fully taken into account when granting the emission credits. For example,
although a company might use a more efficient technology, it might be built from materials that are
produced in a highly energy-intensive way. In this case the net impact of the project on emissions is
less than the gross impact. For a discussion of such leakage effects and a methodology to account
for them, Geres and Michaelowa (2002).
27.  With the CDM, the computation of carbon leakage rates needs to account for the fact that the CDM
substitutes emission reductions in non-Annex I countries for increases in their Annex I counterparts.
Therefore the carbon leakage rate is calculated here  as: [1-(world emission reduction in
GtCO2eq)/(Annex I emission reduction objective in GtCO2eq)]. When the emission reduction
achieved at the world level is equal to the emission reduction objective set by Annex I (in
GtCO2eq), the leakage rate is zero.  
28.  Hourcade and Shukla (2001), or Kallbekken (2007b). One exception is Babiker (2005), who finds
large leakage rates when assuming increasing (rather than constant) returns to scale in
energy-intensive industries and homogenous (rather than heterogenous) goods.
29.  EIIs exposed to international trade competition include here non-ferrous metals, iron and steel,
chemicals, fabricated metal products  (excluding machinery and equipment), paper and paper
products, and non-metallic minerals (including cement). See Annex 2 for details.
30.  As an extreme example, Schwank (2004) estimates that producers of chlorodifluoromethane
(HCFC-22) in non-Annex I countries could expand output indefinitely, give that output away, and
still make a profit simply by implementing process changes to reduce emissions of trifluoromethane
(HFC-23) – a very potent GHG and a by-product of HCFC-22 manufacture – and selling the CERs
for that greenhouse gas.
31. Reforestation/afforestation projects are eligible under the CDM, but avoided deforestation is not.
32.  For discussion of the permanence and liability issues associated with incorporating CCS in the
CDM, and how these could be addressed, Philibert et al. (2007). While additionality would not be
an issue in general, one possible exception is enhanced oil recovery projects. 4.  TOWARDS GLOBAL CARBON PRICING
THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – ©  OECD 2009    155
33.  For some discussion of differences and possible overlap between bundling and “programmes of
activities”, Ellis (2006).
34.  For evidence of this under-development, e.g. Arquit Niederberger and Fecher (2006). For details on
how to set up programmatic CDMs to enhance energy efficiency, Figueres and Philips (2007).
35.  One additional problem is the existence of linkages across different activities. For instance, in
industries where the emission-intensive  component of the production process can be outsourced
(e.g. cement), the whole supply chain may have to be considered in order to avoid leakage.
36.  International caps might be easier to negotiate in more concentrated sectors, such as aluminum.
37.  Another cost-effective way to achieve binding sectoral targets would be through a carbon tax, but
this option has not gained much interest in practice.
38.  For instance, air transport is planned to be included in the EU ETS by 2011.  
39.  The base year for sectoral emission reductions in non-Annex I countries is assumed to be 2005
rather than 1990 partly due to existing uncertainties around 1990 emission levels in a number of
developing countries, particularly for non-CO2 emissions.
40.  This is true under the three alternative linking scenarios.
41.  Chapter 1 underlines the gains from broadening the sectoral coverage of mitigation action beyond
EIIs and the power sector. While a 9% cut in world emission levels relative to 2005 was found to
cost 1.7% of world GDP (compared to the baseline) in 2050 (Chapter 1, Table 1.2), here a 2% rise
in emissions appears to lower world GDP by over 3% in 2050, (Figure 4.18).  
42.  As with the CDM, the cap on Annex I countries is unchanged with sectoral crediting (-50% by 2050
relative to 1990 levels in the simulations considered here), but part of the emission reductions are
bought and achieved in non-Annex I countries.  
43.  As all general equilibrium models used to assess the economic effects of alternative climate policies
are either economy-wide or – like ENV-Linkages – sectoral models, they cannot assess the
implications of moving from a project-by-project to a sectoral approach. As simulated here, sectoral
crediting replaces rather than complements the current CDM, which might imply a reduction in the
sectoral coverage of emission crediting mechanisms, and thereby a more limited access to cheap
abatement opportunities. However, restriction to EIIs and the power industry does not appear to be
binding in practice, because these sectors encompass a sizeable share of low-cost abatement
opportunities in non-Annex I countries.
44.  See Chapter 3 for estimates of the magnitude of carbon leakages and competiveness concerns.
45.  For each tonne of CO2 abated, the surplus would equal the difference between the credit price and
the abatement cost. It arises because the price paid for credits in a reasonably competitive world
market would be determined by the abatement cost of the marginal project which, due to the
convexity of the abatement cost curve, exceeds the abatement costs of all other (cheaper) projects.  
46.  By comparison, for instance, in 2007 the US sub-prime mortgage market (total outstanding amount
of sub-prime loans) amounted to about 9.5% of US GDP, or about 3% of world GDP at current
exchange rates (OECD, 2007c).
47.  In particular, carbon price uncertainty discourages investment (Jamet, 2009).
48.  These risks include registration risks, risks coming from the financial situation of the project leader
and its access to credits, and several host country risks (Point Carbon, 2007). 4.  TOWARDS GLOBAL CARBON PRICING
156  THE ECONOMICS OF CLIMATE CHANGE MITIGATION – ISBN: 978-92-64-05606-0 – © OECD 2009
49.  The sellers of allowances (governments) have no interest in hedging and therefore, hedging is
expected to be one-sided in the carbon market. In most other commodity markets, hedging is
two-sided (producers of the commodity hedge by selling futures while processing firms hedge by
buying futures), hedging demands are at least partly offsetting and speculative trading is less needed
for the development of derivative markets. If permits are allocated for free, the market will be more
two-sided, since firms undertaking investment to reduce emissions would hedge against the risk of a
decrease in the future carbon price.  
50  Other types of hedging costs (fees, bid-ask spreads, and the costs of maintaining margin) should be
comparatively small.
51.  Under this system, buyers would be liable to make up the difference between invalid permits that do
not represent the full amount of carbon reduction their face value implies, and valid ones. In the
case of offsets, the additionality would have to be assessed. In the case of allowances, the validity
could be assessed on a broader range of criteria including the validity of monitoring, reporting and
verification (MRV) procedures and the stringency of the country’s allocation of permits during the
compliance period.
52.  In the international bond market, the quality of sovereign bonds is priced by markets. No
independent international agency is needed to rate bonds because buyers have a clear financial
incentive to assess the risk of default in advance. By contrast, in the international permit/credit
market, neither buyers nor sellers would have a  financial incentive to assess the environmental
quality of permits or credits in advance, unless an independent authority has the power to enforce a
penalty at the end of a project if there have been environmental integrity problems.  
53.  Compared with a future contract, future vintages of allowances or long term allowances have the
advantage of being more flexible, as they can be used to meet compliance obligations if needed,
before the future contract maturity date.