Supporting Information for ‘Postponing emission reductions from 2020 to 2030
increases climate risks and long-term costs’
1. Description of the modelling framework IMAGE, TIMER and FAIR
For the analysis of the mitigation scenarios, we used parts of the integrated
assessment modelling framework IMAGE 2.4 (Bouwman et al., 2006)1. The IMAGE
model consists of a set of linked and integrated models that together describe
important elements of the long-term dynamics of global environmental change, such
as air pollution, climate change, and land-use change. We used the FAIR model and
the global energy model, TIMER, as part of the IMAGE model. The latter describes
the primary and secondary demand and production of energy and related greenhouse
gas emissions and regional air pollutants (van Vuuren et al., 2007). A more detailed
description of the models TIMER and FAIR is provided below.
The FAIR 2.1 model
The integrated modelling framework FAIR (den Elzen et al., 2008; den Elzen and van
Vuuren, 2007) is used for the quantitative analysis of emission reductions and
abatement costs at the level of 26 regions. The FAIR–SiMCaP model is a combination
of the abatement costs model of the FAIR model and the SiMCaP model (den Elzen
and Meinshausen, 2006; den Elzen et al., 2007; den Elzen and van Vuuren, 2007).
The FAIR cost model distributes the difference between baseline and global emission
pathways following a least-cost approach using regional Marginal Abatement Costs
(MAC) curves for the different emissions sources (den Elzen et al., 2007). The model
uses baseline emissions of GHGs from the IMAGE land-use model and TIMER
energy model. The aggregated emission credits demand-and-supply curves are
derived from marginal abatement costs curves (MAC) based on the same sources.
More specifically, the MAC curves for energy- and industry-related CO2 emissions
were determined with the TIMER energy model (van Vuuren et al., 2007) by
imposing a carbon tax and recording the induced reduction of CO2 emissions. The
earlier work has been further improved by now including four instead of two different
carbon tax profiles. We now capture the full range of possible carbon tax paths that
represent early action and highly delayed action. The MAC curves for carbon
plantations were derived using the IMAGE model (Strengers et al., 2008). MAC
curves from the EMF21 project (Weyant et al., 2006) were used for non-CO2 GHG
emissions. These curves have been made consistent with the baseline used here and
made time-dependent to account for technology change and removal of
implementation barriers (Lucas et al., 2007). It should be noted that, while CO2
emission reductions from the energy system were described accounting for dynamic
processes such as induced learning and limited capital turnover rates, reductions for
non-CO2 gases were based on simple MAC curves (that change over time, but do not
limit reduction potential based on a vintage structure). Furthermore, in FAIR, a
maximum reduction rate of 3% was included based on the TIMER experiments (for
the BECS scenarios 4%), reflecting the technical (and political) inertia that limits
emission reductions, avoiding premature replacement of existing fossil-fuel-based
capital stock.
1
The model names are acronyms. IMAGE = Integrated Model to Assess the Global Environment; TIMER = The
Targets IMage Energy Regional model; FAIR = Framework to Assess International Regimes for the differentiation
of commitments
The emission credits demand-and-supply curves were used to determine the carbon
price in the international trading market, its buyers and sellers, and the resulting
domestic and external abatements for each region. The abatement costs for each
scenario were calculated based on the marginal abatement costs and the actual
reductions. They represent the direct additional costs due to climate policy, but do not
capture the macroeconomic implications of these costs.
We calculated the abatement costs (in 2005 USD) by assuming use of the flexible
Kyoto mechanisms, such as international emission trading (IET) and CDM, and
calculated the cost-effective distribution of reductions for different regions, gases and
sources. For countries with reduction targets and participate in IET, all their
abatement potential is fully available on the market, while for countries that only
participated in CDM, a limited amount of the abatement potential was assumed to be
operationally available on the market, because of the project basis of the CDM and
implementation barriers. Consistent with studies by Criqui (2002), den Elzen and De
Moor (2002), and Jotzo and Michaelowa (2002), this so-called CDM accessibility was
set at 20% for 2020. This meant that only 20% of the total supply would be available
for offsetting reductions not achieved by Annex I countries.
For the cost calculations, we assumed that Annex I regions begin or continue with
emission reductions in 2012, and all fully participate in emission trading. For the nonAnnex I countries, we considered three groups of countries; advanced developing
countries, other developing countries and least-developed countries (see Table A.1).
The advanced developing countries will join the carbon market in 2020 and
participate in emission trading. The other developing countries only participate in
CDM, and join the carbon market after 2020. For the late entrants in the carbon
market, we also assumed there to be a transition period before they are fully exposed
to the global carbon price. In fully participating regions (such as the Annex I
countries, including the United States), carbon prices are equal. Non-participating or
CDM regions have a zero carbon price. For regions in transition from no to full
participation, the carbon price grows from zero to the level of the participating regions
during the transition period. A linearly growing proportion of the regions’ mitigation
potential is exposed to the global carbon price, and the regional price is the price at
which the exposed mitigation potential is fully implemented, until the global carbon
price is reached (van Vliet et al., 2009). In the high-income and middle-income
countries, this leads to carbon prices of 90 and 60%, respectively, of the international
carbon market price.
Table A.1. Assumptions on participation in international emission trading (IET) and CDM and
the calculated fraction of the global carbon price.
Advanced
developing
countries
(ADCs)
Other
developing
countries
Leastdeveloped
countries
Mexico, Rest Central America, Brazil, Rest South
America, South Africa, Kazakhstan region, Turkey,
the Middle East, Korean region and China:
Reduce below baseline emissions and can
participate in IET
Northern Africa, the Middle East, India, Rest South
Asia, Indonesian region, Rest Southeast Asia:
Reduce below baseline emissions and can
participate in CDM
Western Africa, Eastern Africa and Rest of South
African region:
Follow baseline emissions and can participate in
CDM
IET (90%)
IET (60%)
CDM (20%)
Other main assumptions for the cost calculations were:
 The transaction costs associated with the use of the Kyoto mechanisms were
assumed to consist of a constant 0.55 USD per tonne CO2 eq emissions plus
2% of the total costs

Most Parties propose targets that do not include international bunker fuels,
except for the EU2. Therefore, the emission and cost calculations exclude
international bunker fuels emission projections and costs of reducing these
emissions.
Carbon credits from forest management were included, based on a conservative, low
estimate taken from an extension of the Marrakesh Accords.
For the global climate and GHG concentration calculations, we used the simple
‘climate’ model, MAGICC 4.1 (Wigley, 2003; Wigley and Raper, 2001; 2002). The
model framework calculates emissions of all greenhouse gases, ozone precursors
(volatile organic compounds, CO, and NOx), and sulphur aerosols (SO2) from energy
and land-use related sources, atmospheric concentrations, radiative forcing, and
resulting climate change.
The TIMER model
The energy system simulation model (TIMER) describes the long-term dynamics of
the production and consumption of about 10 primary energy carriers for 5 end-use
sectors in 26 world regions. The model is a system-dynamics model, and its behaviour
is mainly determined by substitution processes of various technologies on the basis of
long-term prices and fuel preferences. These two factors drive multinomial logit
models that describe the investments into new energy production and consumption
capacity3. It should be noted here that the demand for new capacity is limited by the
assumption that capital is only replaced after the end of the technical lifetime is
reached. The long-term prices that drive the model are determined by resource
depletion and technology development. Resource depletion is important both for fossil
fuels and for renewables (for which depletion and costs depend on annual production
rates). Technology development is determined by learning curves or through
exogenous assumptions. Emissions from the energy system are calculated by
multiplying energy consumption and production flows with emission factors. A
carbon tax can be used to induce a dynamic response, such as increased use of low or
zero-carbon technologies, energy efficiency improvement, and end-of-pipe emission
reduction technologies. The carbon tax required to achieve a certain climate policy
target is calculated by the FAIR model, taking into account baseline emissions and
cost curves for CO2 and non-CO2 emission reductions.
2. Maximum reduction rate
To explore how the rate of emission reduction varies across models, we used a
database of a large number mitigation scenarios as assessed for AR4 (Nakicenovic et
al., 2006) and expanded this with recent studies from EMF22 (Clarke et al., 2009), the
ADAM project (Knopf et al., 2009) and work for the European Commission (Rao et
2
For the EU, the -20% unilateral target includes the emissions from aviation, making the target more stringent. For
instance, when including emissions from aviation, EU emissions in 2005 would have decreased by only 6.8%,
compared to 1990. When excluding these emissions, however, EU emissions for that year decreased by 7.9%.
3
A multinomial logit model assigns market shares to fuel or technologies on the basis of their relative costs. Low cost
options receive a large market share; high cost options a low (or even zero) market share.
al., 2009). Only part of these studies cover all greenhouse gases. Therefore, we
focused on the reduction of CO2 emissions from the energy system (that is included in
all models). As this emission category represents 60 to 70% of all emissions during
the 21st century, these reductions rates are reasonable representative for the total
greenhouse gas reduction rate.
The rate of emission reductions is bound by several factors, such as capital turnover
rates, the production rate of new technologies, and implementation barriers. It should
be noted, however, that the maximum rate cannot be determined unambiguously, as
many of these factors are context dependent (there is no ‘law’). In the literature, there
is a wide variety of models. In some of these models, maximum reduction rates have
been included implicitly. In other models, attractive reduction rates emerge as a result
of the time optimisation that has been applied (and the related costs of too high
reduction rates). Looking at the range of model results, therefore, could provide some
insight into the range of reduction rates considered feasible by models.
Figure 1 shows the 10-year average emission reduction rates in the 2010-2050 period
for only the lowest scenario category used in AR4. The figure shows that, in literature,
few scenarios can be found with a reduction rate of beyond 3 to 4% annually, from
2000 emission levels, over a 10-year period. There are only a few scenarios that show
faster reduction rates. In other words, emission reduction rates beyond 3% can be
regarded as extreme, based on the scenario literature. Most models have used some
form of optimal pathway for emission reductions over time (either formal
optimisation or more ad-hoc rules).
Figure 1. The global reduction rates of mitigation scenarios (energy-related CO2
emissions) for the highest ambition targets
References:
Bouwman, A. F., Kram, T. and Klein_Goldewijk, K. (2006). Integrated modelling of
global environmental change. An overview of IMAGE 2.4. Bilthoven, The
Netherlands: Netherlands Environmental Assessment Agency, available at:
www.mnp.nl\en.
Clarke, L., et al. (2009) International climate policy architectures: Overview of the
EMF 22 International Scenarios. Energy Economics 31(SUPPL. 2).
Criqui, P. (2002). GECS Final Report Section 6: Detail report: CNRS-IEPE,
Grenoble. GECS - Research Project N° EVK2-CT-1999-00010, Thematic
Programme : Environment and Sustainable Development, DG Research Fifth
Framework Programme.
den Elzen, M. G. J. and de Moor, A. P. G. (2002) Evaluating the Bonn-Marrakesh
Agreement. Climate Policy 2(1), 111-117.
den Elzen, M. G. J., Lucas, P. and van Vuuren, D. P. (2008) Regional abatement
action and costs under allocation schemes for emission allowances for
achieving low CO2-equivalent concentrations. Climatic change 90(3), 243–
268.
den Elzen, M. G. J. and Meinshausen, M. (2006) Meeting the EU 2°C climate target:
global and regional emission implications. Climate Policy 6(5), 545–564.
den Elzen, M. G. J., Meinshausen, M. and van Vuuren, D. P. (2007) Multi-gas
emission envelopes to meet greenhouse gas concentration targets: costs versus
certainty of limiting temperature increase. Global Environmental Change
17(2), 260–280.
den Elzen, M. G. J. and van Vuuren, D. P. (2007) Peaking profiles for achieving longterm temperature targets with more likelihood at lower costs. Proceedings of
the National Academy of Sciences USA (PNAS) 104(46), 17931–17936.
Jotzo, F. and Michaelowa, A. (2002) Estimating the CDM market under the
Marrakech Accords. Climate Policy 2, 179-196.
Knopf, B., et al. (2009). The economics of low stabilisation: implications for
technological change and policy. In M. Hulme andH. Neufeld (Eds.), Making
climate change work for us.
Lucas, P., van Vuuren, D. P., Olivier, J. A. and den Elzen, M. G. J. (2007) Long-term
reduction potential of non-CO2 greenhouse gases. Environmental Science &
Policy 10(2), 85-103.
Nakicenovic, N., Kolp, P., Riahi, K., Kainuma, M. and Hanaoka, T. (2006)
Assessment of emissions scenarios revisited. Environmental Economics and
Policy Studies 7(3), 137-173.
Rao, S., et al. (2009). IMAGE and MESSAGE Scenarios Limiting GHG
Concentrations to Low Levels: IIASA en PBL. Framework Contract
ENV.C5/FRA/2006/0071
Strengers, B. J., van Minnen, J. and Eickhout, B. (2008) The role of carbon
plantations in mitigating climate change: potentials and costs. Climatic
Change 88, 343–366.
van Vliet, J., den Elzen, M. G. J. and van Vuuren, D. P. (2009) Meeting radiative
forcing targets under delayed participation. Energy Economics 31(SUPPL. 2),
S152-S162.
van Vuuren, D. P., et al. (2007) Stabilizing greenhouse gas concentrations at low
levels: an assessment of reduction strategies and costs. Climatic Change 81(2),
119-159.
Weyant, J. P., De la Chesnaye, F. C. and Blanford, G. (2006) Overview of EMF - 21:
Multi-gas Mitigation and Climate Change. Energy Journal, Multi-Greenhouse
Gas Mitigation and Climate Policy(Special Issue #3), 1-32.
Wigley, T. M. L. (2003). MAGICC/SCENGEN 4.1: Technical Manual: UCAR Climate and Global Dynamics DivisionBoulder, CO.
Wigley, T. M. L. and Raper, S. C. B. (2001) Interpretation of high projections for
global-mean warming. Science 293(5529), 451-454.
Wigley, T. M. L. and Raper, S. C. B. (2002) Reasons for larger warming projections
in the IPCC Third Assessment Report. Journal of Climate 15(20), 2945-2952.