Working Group III: Mitigation

Other reports in this collection

8.3.1 International Emissions Quota Trading Regimes “Where Flexibility”

Table 8.7 synthesizes marginal abatement costs for the USA, Japan, OECD-Europe, and the rest of the OECD (CANZ) calculated by 13 world T-D models co-ordinated by the Energy Modeling Forum. It also includes the results obtained with the POLES model, which provides a multiregional partial equilibrium analysis of the energy sector, and two other studies of the economic impacts of Kyoto conducted by the US Government, the Administration’s Economic Analysis (Council of Economic Advisors, 1998), and a study by the Energy Information Admini-stration (1998). These results cannot be directly compared with those of the B-U analysis reported in Section, because they incorporate feedback on energy demand, oil prices, and macroeconomic equilibrium. They give, however, an idea of the assumptions on technical abatement potentials retained for each region in these exercises, the main difference with B-U analysis being that these exercises do not explicitly consider negative cost potentials (they are implicit in most optimistic baselines).

Table 8.7: Energy Modelling Forum main results; marginal abatement costs (in 1990 US$/tC; 2010 Kyoto target)
Model No trading Annex I trading Global trading
  USA OECD-E Japan CANZ      
ABARE-GTEM 322 665 645 425   106 23
AIM 153 198 234 147   65 38
CETA 168         46 26
Fund           14 10
G-Cubed 76 227 97 157   53 20
GRAPE   204 304     70 44
MERGE3 264 218 500 250   135 86
MIT-EPPA 193 276 501 247   76  
MS-MRT 236 179 402 213   77 27
Oxford 410 966 1074     224 123
RICE 132 159 251 145   62 18
SGM 188 407 357 201   84 22
WorldScan 85 20 122 46   20 5
Administration 154         43 18
EIA 251         110 57
POLES 135.8 135.3 194.6 131.4   52.9 18.4
Source: cited in Weyant, 1999; Council of Economic Advisors, 1998; EIA (Energy Information Administration), 1998; Criqui et al., 1999.


Despite the wide discrepancies in results across models, the robust information is that, in most models, marginal abatement costs appear to be higher in Japan than in the OECD-Europe. CANZ and the USA have comparable results, approximately two-thirds the European one, and much lower than in Japan.

This means that Kyoto targets are likely to be unequitable. This risk is confirmed by uncertainty analyses based on existing models which provide a pretty wide range of outcomes that can be interpreted as covering the uncertainties prevailing in the real world. This can be shown in the results of domestic cost of carbon: from US$85 to US$410 in the USA, US$20 to US$966 for the OECD-Europe, US$122 to US$1074 for Japan, US$46 to US$423 for CANZ. The variance remains significant if the extreme values:

  • from US$76 to 236/tC for the USA if one excludes GTEM, Merge 3, and Oxford;
  • from US$159 to US$276/tC for the OECD-Europe and from US$145 to US$250 for CANZ if one excludes Worldscan, GTEM, and Oxford; and
  • a continuum from US$122 to US$645/tC for Japan if Oxford is excluded.

In terms of GDP losses, the ranking of impacts differs because of the various pre-existing structures of the economy and of the energy supply and demand in various countries and because these studies do not consider the domestic policies targeted to tackle these pre-exisiting conditions; the GDP losses are from 0.45% to 1.96% for the USA, from 0.31 to 2.08 for the EU, from 0.25 to 1.88 for Japan. This variation is reduced under emissions trading; 0.31 to 1.03 for the USA, 0.13 to 0.73 for the OECD-Europe, from 0.05 to 0. 52 for Japan.

This discrepancy in results reflects differences in judgements about parameters such as technical potentials, emissions baselines, how the revenues of permits are recycled, and how near-term shocks are represented. Another important source of uncertainty is the feedback of the carbon constraint on the demand for oil; a drop in oil prices requires indeed higher prices of carbon to meet a given target since the signals not conveyed by oil prices as to be passed through price of carbon which leads to a totally different incremental cost of the carbon constraint.

These uncertainties about mitigation costs are reflected in the net welfare losses. The preceding discussion in Section 8.2 demonstrated the many sources of a wedge between total abatement costs and welfare losses, including the double dividend from fiscal reforms and the very structures of the economy (share of carbon intensive activities) and of the energy system.

The wide range of cost assessments, far from resulting from purely modelling artefacts, help to capture the range of possible responses of real economies to emissions constraints and to appreciate the magnitude of uncertainties that governments have to face13. They demonstrate that without emissions trading, the Kyoto targets lead to a misallocation of resources, a non-equitable burden-sharing (notwithstanding its mitigation through double-dividend domestic policies analyzed in Section and distortions in international competition. Even in the most optimistic models regarding abatement costs such as Worldscan, trading offers the potential for countries with high domestic marginal abatement costs to purchase emissions permits in countries with low marginal abatement costs and hence a way of minimizing total abatement costs and of hedging against risks of a too high and unequitable burden.

Table 8.8: Energy Modelling Forum main results; GDP loss in 2010 (in % of GDP; 2010 Kyoto target)

  No trading Annex I trading Global trading
ABARE-GTEM 1.96 0.94 0.72 1.96   0.47 0.13 0.05 0.23   0.09 0.03 0.01 0.04
AIM 0.45 0.31 0.25 0.59   0.31 0.17 0.13 0.36   0.20 0.08 0.01 0.35
CETA 1.93         0.67         0.43      
G-CUBED 0.42 1.50 0.57 1.83   0.24 0.61 0.45 0.72   0.06 0.26 0.14 0.32
GRAPE   0.81 0.19       0.81 0.10       0.54 0.05  
MERGE3 1.06 0.99 0.80 2.02   0.51 0.47 0.19 1.14   0.20 0.20 0.01 0.67
MS-MRT 1.88 0.63 1.20 1.83   0.91 0.13 0.22 0.88   0.29 0.03 0.02 0.32
Oxford 1.78 2.08 1.88     1.03 0.73 0.52     0.66 0.47 0.33  
RICE 0.94 0.55 0.78 0.96   0.56 0.28 0.30 0.54   0.19 0.09 0.09 0.19


The full global trading scenarios presented in Tables 8.7 and 8.8 assume non-restricted trade within Annex I and ideal CDM implementation that can exploit all cost effective options in developing countries with unlimited trading. Beyond the fact that the price of carbon is drastically reduced, it is remarkable that the variance of results is far lower than in the no-trade scenarios (between US$15/tC and US$86/tC). Uncertainty about costs persists, but this lesser variance arises because uncertainty is higher on each regional cost curve than on the aggregation of the same regional cost curves, which is exploited in the case of full trading.

In the case of Annex I trading (without considering the CDM) the price of permits ranges from US$20 to US$224/tC instead of US$15 to US$86/tC in the full trade case, which represents a far greater variance. This is mainly from the amount of so called “hot air” 14 retained in simulations. Some countries in Eastern Europe and the former Soviet Union have had a decline in emissions in the 1990s, resulting from the economic dislocations associated with restructuring. As a result, their emissions during the first commitment period are projected to be lower than their negotiated target. If trading is allowed within Annex I, these excess emissions quota may be sold to countries in need of such credits. Hence, the assumption regarding the availability of “hot air” is important. This, of course, will be governed in part by the rate of economic recovery, but also by the role of energy efficiency improvements and fuel switching during the restructuring process.

The main lessons from the above studies using T-D approaches (namely that trade has a marked, beneficial effect on costs of meeting mitigation targets), are confirmed by a series of recent studies using B-U approaches. These provide a more detailed information on the potentials for CDM projects. The MARKAL, MARKAL-MACRO, and MESSAGE models have been adapted and expanded to facilitate such multicountry studies. In North America, Kanudia and Loulou (1998) report MARKAL results for a three-country Kyoto study (Canada, USA, India). The total cost of Kyoto for Canada and the USA amounts to some US$720 billion with no trade, versus US$670 billion when North American emissions and electricity trading is unimpeded, and only US$340 billion when India is added to the permit trading. MARKAL studies in the Nordic states (see Larsson et al. (1998) for Denmark, Sweden, and Norway, and Unger and Alm (1999) for the same plus Finland) show the considerable value of trading electricity and GHG permits within the region when severe GHG reductions are sought. Another MARKAL study computes the net savings of trading GHG permits between Belgium, Switzerland, Germany, and the Netherlands (Bahn et al., 1998) at about 15% of the total Kyoto cost without trading. Another study (Bahn et al., 1999a) shows that Switzerland’s Kyoto cost may be reduced drastically if it engages in CDM projects with Columbia, in which case the marginal cost of CO2 drops to US$12/tC. This type of B-U analysis has also been extended to the computation of a global equilibrium between Switzerland, Sweden, and the Netherlands, using MARKAL-MACRO (Bahn et al., 1999b), with the conclusion that GDP losses resulting from a Kyoto target are 0.2% to 0.3% smaller with trade than without. More ambitious current research aims at building worldwide B-U models based on MARKAL (Loulou and Kanudia, 1999a) or on MARKAL-MACRO (Kypreos, 1998).

Other reports in this collection

IPCC Homepage