Page 836
Appendix R
Description of Economic Estimates of the Cost of Reducing
Greenhouse Emissions
It is useful to compare the estimates of the cost of mitigation derived in this report with those from other studies. One systematic study is a survey (Nordhaus, 1991) that derives from nine different economic modeling families a "cost function" for greenhouse gas emission reductions. The conceptual experiment performed for each model is to estimate the cost of an efficient reduction in CO2 emissions. For models that assume no externalities or other market imperfections, this is equivalent to estimating the response of greenhouse gas emissions to increasingly stringent carbon taxes. The range of carbon taxes is from zero to around $100/t CO2 equivalent.1
The methods employed in the nine modeling studies differ considerably. At one extreme are the "econometric" models, which rely largely on behavioral estimates of supply and demand functions. In these studies, estimates are made of the structure of demand and supply on the basis of observed market data on prices and quantities. The models are often energy models that have been extended to include CO2 emissions. A second generic approach draws on programming or optimization models of the energy sector. In this approach, the energy sector or the economy is represented in terms of technological activities such as demand for space heating or transportation services. By using a mathematical programming or other algorithm, the models then solve for the "optimal" trajectory of prices, outputs, fuel mix, and technologies. It can be shown that, under certain conditions, the optimal trajectory would correspond to the outcome of perfectly competitive markets. Some models are a hybrid of the two approaches.
All studies reviewed here share two important characteristics, however, that differ from the approach of the Mitigation Panel. First, they are all comprehensive energy sector models. That is, they include a consistent accounting of the demand, supply, and resources used in the countries or
Page 837
regions studied. In this respect, they differ from this report, which examines the possibilities for greenhouse gas reductions from individual technologies and attempts to make the estimates mutually consistent by manual calculations.
A second important difference between the Nordhaus (1991) survey and this report is that the survey estimates the cost function for reducing greenhouse gas emissions beginning from the point at which all the "negative-cost" options have been employed. In most economic models, the market equilibrium is this point; in one model, where market failures are allowed, the results have been recast so that the cost estimates begin from the point at which market failures have been allowed for. It is important to note, then, that the negative-cost part of the cost function, should it exist, is excluded from this survey.
The models represented in the survey encompass a wide variety of approaches to energy sector modeling. The studies surveyed (Nordhaus, 1991) were the following, listed roughly in order of their chronological development:
1. A series of mathematical-programming models, developed by Nordhaus and his associates, that use a technological specification for supply and econometric estimates for demand.
2. A purely behavioral or econometric model developed by Nordhaus and Yohe for the 1983 NAS study, Changing Climate, with a simplification for estimates of the impact of different taxes.
3. A series of models developed by Edmonds and Reilly, which are a mixture of technological data on the supply side and behavioral assumptions on the demand side.
4. A series of studies by Manne and Richels, which have much the same analytical structure as study 1 but are generally smaller and provide less detail on the demand side.
5. Estimates by Bodlund and associates for Sweden, using largely a mathematical optimization model with many alternative technologies.
6. Studies by Kram and others using a linear programming model of the Netherlands economy with a structure similar to that in study 5
7. A six-region computable general equilibrium model developed by Whalley and Wigle that includes the energy and nonenergy sectors. The energy sector is purely behavioral and is not econometrically estimated.
8. A series of optimization models developed by the European Community that include both supply and end-use technologies for five European countries. These models allow for market failures in the energy-using sectors.
9. An extension of the Jorgenson approach by Jorgenson and Wilcoxen to include CO2 emissions. The estimates are purely econometric but are based on extensive data and estimation.
Page 838
The results of the survey are shown in Figure R.1, with the individual points representing estimates from one of the nine families of models. Figure R.1 also shows a "high" and "low" frontier from the different models. The range has been constructed to include almost all the models, although some of the extreme estimates have been omitted if they appear to have features that are particularly problematic.
Figure R.2 employs a consensus, derived as the central tendency of the different models for CO2 and alternative estimates for reforestation and CFCs, which shows the "best-guess" estimate of the cost function for reducing greenhouse gas emissions.
Finally, Figure R.3 shows a total cost function derived from the survey. The estimates are the total cost of different reductions of greenhouse gases at 1989 levels of world economic activity and industrial mix.
Note
1. Tons (t) are metric.
Reference
Nordhaus, W. D. 1991. The cost of slowing climate change: A survey. The Energy Journal 12(1):37–65.
Page 839