Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Method overview

There are 10 links in the logical chain used to calculate the costs of Southern compliance with the Convention (see box on page 106). Also displayed are the key variables that affect each step. The quantitative goal of the chapter is to estimate a time profile of national greenhouse gas emissions, required reductions, and the costs related to achieving the required reductions. This stream of costs is then present valued, converted to an annuity, and compared with the national or aggregated greenhouse gas reduction obligation-to-pay rankings described in Chapter 6.

Which gases?

For the reasons adduced above and in Chapter 2, only carbon emitted as carbon dioxide from fossil fuel usage is included in this analysis. I have not included costs related to greenhouse gases that are already partly controlled by the Vienna Convention on ozone depletion. Nor do I include costs that will arise from controlling greenhouse gas emissions from agricultural or land use changes. While past estimates and projections have been made for emissions of these gases, such figures are highly speculative and scientifically controversial, making the monitoring and verification of controls on these gases such as methane emissions from paddy fields or cattle virtually impossible. Also, a separate agreement and financing arrangement will likely deal with carbon storage and release in forests, although such was not achieved at Rio in 1992.

Method overview

Key variable is shown in [ ]

1 Select cost elements

• Which gases [CO2 fossil fuel]
• Which losses [coastal protection]

2 Distribute notional carbon sink property rights [science, population, territory]

3 Estimate marginal abatement costs

• Nordhaus/US National Academy of Sciences
• Study estimates [RR ratio or ReqRed/CO2ff Projected, levelized cost abatement $/T C]

4 Project emissions

• Reference scenario [COsff only to IPCC stringent target]
• Efficiency scenario [CO2ff adjusted for autonomous energy efficiency increase]

5 Estimate required reduction

• Efficiency scenario [adoption rate of efficiency, renewables, energy substitution to force CO2ff-ReqRed 2025 to equal target emission]

6 Estimate coastal protection costs [time profile costs, discount rate]

7 Present value and annuitize calculated incremental abatement and coastal protection costs
[Discount rate, limited global and zero local benefits counted]

8 Redistribute according to Obligation to Pay Index

• South's OTP = 7% World incremental Cost
• North's transfer OTP to South = South's Incremental Cost - South's OTP
• North's OTP = North's incremental cost + North's transfer [Obligation To Pay index]

9 Evaluate transfer mechanisms

• Carbon tax [projected versus emitted carbon]
• Tradeable permits [savings from trade S/T abated]
• Regional traceable permits [barriers to trade]

10 Estimate monitoring and verification costs [size, number of sources, gases, science]

A comprehensive agreement that incorporates the relative global warming potential of different greenhouse gases and regulates the net emissions from all sources to achieve the least cost abatement strategy is certainly optimal in the long run. But desirable as it is, achieving such comprehensiveness and flexibility is likely to prove elusive at the outset of the Convention. In addition, the IPCC is revising the radiative forcing equivalence of the different greenhouse gases relative to carbon dioxide. Therefore, I cost only abatement of carbon released during use of fossil fuels which is a substantial portion of the total greenhouse problem. Fortunately, these emissions are not only the most quantifiable; they are also among the least costly. Emissions of fossil carbon are also reasonably reliable ( 5-10 per cent for many big countries although the confidence level is much lower in many small, poor countries) and are estimated regularly through an existing, generally accepted UN office. These emissions are amenable to actions that can be taken and are well understood in the near and medium term, many of which are already economically justified.

Moreover, my modest approach is also conservative in that I estimate only the minimal costs of compliance with the Convention. Later - when the scientific basis improves for the control of greenhouse gases other than carbon dioxide from fossil fuels - so the requisite economic information for computing and allocating related abatement costs will become available. Until then, it is premature to include more than carbon dioxide released from projected fossil fuel usage.


Some economists have argued that carbon emissions in the distant future should be discounted as being less significant than those emitted in the near term. There is, however, no ecological argument for discounting future emissions. Indeed, if the absorptive capacity of the atmosphere is finite and if ecological degradation accelerates in non-linear fashion after system thresholds of change are exceeded, then future emissions may be more significant than past or near term emissions. (This would imply that negative rather than positive discounting should be used.)

In this study, I did not discount future emissions of CO2ff. The economic value of the costs of future abatement, however, is discounted by a real discount rate of 5 per cent throughout. Arguably, a higher discount rate (such as the 10 per cent used by the World Bank for project evaluation) is justified in light of the capital scarcity and high social opportunity costs of capital investments in developing countries. Conversely, as economist William Cline has argued, an appropriate annual discount rate for long and very long term cost-benefit analysis might be as low as 1-2 per cent. Moreover, as the US National Academy of Sciences notes, uncertainty as to the impact of greenhouse emissions may increase rather than decrease as more knowledge accumulates, thereby warranting a decrease in standard discount rates to as low as 3 per cent. In short, if it is difficult to estimate the impact and costs of pollution, then it is even more difficult to quantify our ignorance about these impacts over time - a fact that should be reflected in economic analysis.

Of course, investors and consumers may demand much higher returns than the social discount rate adopted in this study, due to market failures, lack of information, and competing investment opportunities. Indeed, these individual preferences and market failures explain why there is a gap between economically justified greenhouse abatement strategies and private investment in energy efficiency and other abatement options. However, social policy should be based on economic, not financial criteria. By the same token, however, it is apparent that many institutional, regulatory, and pricing policy initiatives are necessary to overcome the 'payback' gap that characterizes socially justified strategies.

Calculating cost

With these analytic steps defined, it is possible to specify more precisely the method for calculating the cost of compliance for a country or grouping of countries with carbon abatement required by my Climate Change Convention scenario. In the algorithm I have used to calculate the costs of the South's compliance (see box), the total incremental cost consists of two elements, calculated separately. These are labelled A, the cost of abating CO2ff emissions in the South during 1995-2025; and B. the cost of constructing coastal protection in countries vulnerable to sea level rise. Each of the sequential steps presented in this box and the previous one are explained in greater detail below.

Incremental cost method

Incremental Cost = Net Present Value P of (A = Incremental Abatement Cost) + (B = Coastal Protection Cost) where:

A = Incremental Abatement Cost = Sum of A given P of lC1,2,3 = Required Reduction Ratio RR x Marginal Abatement Cost = CO2ff Projectedk,t /Required Reductionk.t =

{(Base year Qt,k X Qt/capk.t X Projected Populationk,t x Efficiency Rule 1)}/
(Base year Qt,k X Efficiency Rule 2)] x (Marginal Costk,t,1,2,3 at RRk,t)

B = Global Coastal Protection Cost = Sum of A given P of: (Half estimated 100year coastal protection capital cost)/30y

Key: A = Annuity given, P = Present Value of IC = NPVi,lCk,t

Base year = 1995, CO2ff = carbon dioxide emissions from fossil fuel use, T C/y
IC1,2,3 = Incremental Cost for Marginal Abatement Costs, cases 1,2,3
i = discount rate for present valuing, 5% per year
k = country or grouping of countries (North, East, South, World)

Marginal Costk,t,1,2,3 = Marginal Cost of CO2ff Abatement for country k at year t, $/T C abated-year-1

Case 1 = Nordhaus schedule, high cost
Case 2 = US National Academy of Sciences schedule, low cost
Case 3 = study estimate, medium cost n = range of t, 0-30 years from start 1995 to end 2024
P = Net present value of IC for abatement and coastal protection
Q = CO2ff in year t, T C/year

Efficiency Rule 1 for Projected Qk,t: Base Year Qk,t plus 40 % of Reference Scenario annual growth in Qk,t equivalent to a global 1%/y autonomous energy efficiency increase

Efficiency Rule 2 for Required Reductionk,t: Base Year Qk,t - (94%/Y of Ok,t-1 due to 6%/y adoption of efficiency, renewable or energy substitution options) + (65% of growth in Qk,t-1), iterated until Qk,2025 converges on target)

Efficiency rule 2 varies for South (98%/y and 45%/y versus North/East of 94%/y and 65%/y to achieve convergence
RRk,t = Required Reduction Ratio for country k in year t, 0< RR< 1 t = year between 1995 and 2025

Marginal abatement costs

The first step in the calculation is to estimate the marginal cost curve. After reviewing the problems associated with cost data, I construct three cost curves used in the calculations.

Data problems
Complete, disaggregated supply curves for greenhouse gas abatement do not exist for most non-OECD countries in the world. In Figure 5.1, 1 show earlier (1989) 'indicative' estimates made by the consulting firm Mckinsey and Co for the Dutch government. They estimated that the global marginal cost curve per unit abatement achieved is negative for the first five per cent abatement; hovers around zero for the next 30 per cent abatement; and then becomes positive (rising to about $50/T-C-y-1 abated between 35-45 per cent reduction ratio and steeply thereafter). McKinsey's study, however, gives no empirical basis for its estimates which were reportedly based on consultations with practitioners of energy economics.

Figure 5.1 Global abatement cost curve, McKinsey and Co, 1989

Figure 5.2 US NAS marginal cost estimates

Figure 5.3 Marginal costs and damages, Nordhaus

Figure 5.1 may be compared with Figure 5.2, where I present estimates obtained from the US National Academy of Sciences (NAS) which gives a US-only cost curve for greenhouse gases (in contrast to Nordhaus who gives a global cost curve for greenhouse gas abatement, see below and Figure 5.3).

The NAS estimates are based on detailed end use engineering studies of the potential for existing energy services to be provided from efficiency options, thereby abating greenhouse gas emissions. The resulting supply curve of abatement options and costs is provided in Figure 5.2. (Note that costs in Figure 5.2 are given in $/T CO2-y-1 abated; these are converted to $/ T-y-1 Carbon as CO2 in Figure 5.4 which is used in this study.)

Nordhaus's cost estimates related to reduction of CO2 from use of fossil fuel and other sources and are shown in Figure 5.3.15 This curve in turn provides the basis for case 1 in Figure 5.4, where the cost curves used in this study are shown. Only the curve related to CO2ff in Figure 5.3 is incorporated into Figure 5.4.

Cost is always positive in the Nordhaus study - which is contrary to detailed technology-cost driven estimates such as the NAS study for the United States or those by Amulya K.N. Reddy for the Indian state of Karnataka. This assumption derives from the origins of Nordhaus's estimates, in turn, a regression of prior estimates made by macroeconomic modellers as to tax rates required to dampen demand sufficiently to achieve target reduction levels. Such methods assume that current energy markets are reasonably efficient, which makes the cost of energy futures other than 'business-as-usual' inherently positive - in spite of contrary evidence and the practice of many energy utilities in the OECD.

Figure 5.4 Marginal cost, CO2 reduction (Dollars per tonne of carbon abated per year)

Figure 5.5a Partial country abatement cost curves (-$400 to +400/T-C as CO2)

The NAS curve in Figure 5.2 shows the low, mid-point, and high cost estimates at an implementation rate of 50 per cent of the identified technological potential. In Figure 5.4, case 2 is based on the NAS mid-point cost estimate at a 100 per cent implementation rate (as it was the only one that achieved an 80 per cent reduction level). Also, I excluded the technological options and abatement associated with halocarbons and agriculture as unrelated to CO2 fossil fuel abatement. As is evident, the NAS identified substantial amounts of abatement that should be available at negative cost (up to about 20 per cent reduction level), offsetting the steep rise in marginal cost at higher levels of abatement.

Figure 5.5b Full country abatement cost curves

Key: a = The year and amount of total projected emissions; b = The year of abatement cost estimate; c = The year of dollar price; d = Discount rate; e = Time horizon for cost estimate; f = Source; 9 = Assumptions used in calculations; h = Abatement technology from least cost (left) to highest cost (right).

ST = Short term scenario; MT = Medium term scenario; LT = Long term scenario; n/a not applicable or known; RE = Residential energy IE = Industrial energy; AE = Agricultural energy; CE = Commercial & service sector energy; TE = Transport energy.

EE a: 1 985, 356MT-C; b,c 1988. d.7%, e savings projected for 2005-2025; 9 The share of coal in electricity production decreases from the current 69% to 50% and 30% in 2005 and 2025. f: Chapter 12; h: IE1.Buildings insulation, IE2.Boiler replacement, IE3.Heating improvement, IE4.Cogeneration, IES.lmproved transmission and distribution, IE6.Improved existing industrial equipment, IE7.Ferrous metals, IE8.New electrical motors in industry, IE9.Construction industry improvements. CIS/USSR a: 2005, 1315MT-C, b: 2005, c: 1988. For cost of carbon saved, rubles figures developed by Makarov, Bashmakov, were used and ruble (where $1 = 6 Rubles) were discounted to correspond to early 1992's bank exchange rate in Moscow ($1 - 90 Rubles). d: 7%, e capital cost for 19902005,savings projected for 2005; f: Chapter 12; 9: CIS/USSR refers to former Soviet Union h AE1.Shifting from harvesters to site threshing, IE1.Switching small boilers to high-grade fuels, RE1.lnsulation of steam supply network, IE2.Advanced technologies for industrial heating, AE2.Insulation of cattle breeding buildings, IE3.Automation of heating stations, IE4.Efficient centralized boilers, IE5.Change inefficient ovens to large boilers, IE6.Regulated electric drive, IE7.Control and measurement in energy use, IE8.Low capacity multifuel boilers, IE9.Reduction of electric transmission losses, lE10.Replacing wet cement clinker with dry method, lE11.Gas turbine and combined cycle plants, RE2.Efficient lighting, IE12.Improved brick production, IE13.Improved gas compressors in pipelines.

Thailand a: 2001, 34.96MT-C; b: 2001; c 1991; d n/a; e 1991-2001; f Chapter 11; h RE1.Refrigerators, RE2.Lighting, RE3.Cooling, RE4.Rice Cookers.

India a 1987, 144MT-C; b 1989 (ST), 2000 (MT), 2000 (LT); c 1989,15 28Rs/dollar (ST), 1991, 18 81Rs/dollar (MT,LT); d 5%; e 1989 (ST), 2000-2025 (MT,LT). For ST, investment in 1989 generates benefit in the same year. Far MT and LT, investments during 1 989-2000 generate benefit during 2000-2025. Costs for MT and LT were levelized through 2000-2025. f Prodipto Ghosh, Analysis of Energy and Technology Options for Reducing Global Climate Change Concerns: India Country Report. Tata Energy Research Institute, New Delhi, lndia. November, 1991; Jayant Sathaye & Nina Goldman, CO, Emissions from Developing Countries: Better Understanding the role of Energy in the Long Term. Volume III: China, India Indonesia and South Korea. International Energy Studies Group. Lawrence Berkeley Laboratory. July, 1991. 9: The total CO2 emission level was estimated by taking the average of 133 MT-Cin 1988 (Gosh,1991) and 155MT-Cin 1985(Sayathe & Goldman, 1991). For MT and LT investment between 1989 and 2000 was assumed to generate benefits of energy reduction from 2000 to 2025. h: IE.MT (insulation of waste heat recovery systems, replacement of inefficient boilers, better instrumentation and control systems, and adoption of better technologies). IE.LT (Introduction of cogeneration systems, adoption of new energy efficient technologies, and automation of process control) IE.ST (Improved housekeeping, energy audits, and training of personnel).

Brazil a 2000, 84.6MT-C; b 2000;c 1991;d 12%;e 1991-2000; f Chapter 10;h CE1.Lighting, IE1.Electric ovens & boilers, IE2.Variable speed drivers, IE3.High efficient motors, IE4.House keeping measures, RE2.Air conditioning, IE5.Lighting, IE6.Electrolytic processes, TE1.lmprovement in auto mobiles, TE2.Efficient diesel motors, TE3.Improved urban transportation, TE4.Alternative fuel-natural gas, TE5.Alternative fuel-alcohol, RE3.Efficient refrigeration, CE2.Public illumination, TE6.Highway improvement, RE4.Solar water heating.

China a 1985, 478MT-C; b: 1985; c exchange rate for mid 80's (2.9 yuan/dollar) was used for 1985 cost estimate; d: n/a; e: 1985; f Sathaye & Goldman, 1991 for a; Mark D. Levine et al. Energy Efficiency, Developing Nations and Eastern Europe, A Report to the US Working Group on Global Energy Efficiency, Developing Nations and Eastern Europe, April, 1991 for b,c; h: IE.

Australia a 2005, 52.4MT-C if no change in technology, i.e. 'frozen efficiency'; b 1991 costs of equipment; c 1991 AS N.B. A$1 = US$0.75 approximately; d 8%; e 1991 -2005 (2030). Costing of measures is based an emission and energy savings throughout the whole life of the equipment concerned which in some cases is up to 25 years beyond 2005, i.e. to 2030, f Chapter 13; 9 see Chapter ;3; h CE1.Commercial miscellaneous, CE2.Commercial HVAC, IE1.lndustrial metal processing, IE2.Industrial electric motors and drives, IE3.Industrial high temperature IE4.Smelting, IE5.Industrial electrolysis, RE1.Residential hot water, CE3.Commercial lighting, RE2.Residential refrigeration, RE3.Residential major appliances.

Some 'point' data are also available to supplement and validate the marginal cost curves presented in Figure 5.4. In Figure 5.5,1 summarize the cost curve data provided in the national case studies in Part 111 of this book. Readers are cautioned that these costs were not calculated over the same time frames, nor using the same discount rates, technological repertoires, etc. (The divergent parameters are stated concisely in the notes to the figure.) Moreover, some of the curves are for one sector only (as in Thailand) whereas others are for all sectors. These curves, therefore, are not strictly commensurable. Nonetheless, they give a summary view of the state-of-theart of estimating technological costs of carbon emission abatement in developing countries.

As can be seen, the cost ranges from + $2000/t of carbon abated (in the case of Brazil) and cluster around + $100/t. The reduction level covered by the curve ranges from a few per cent (in India) to about 32 per cent reduction relative to projected emissions (in Australia). The cost curves in these developing and transitional economies are similar in shape to those of the US National Academy referred to above, but extend to only 10-30 per cent reduction levels.

In general, these curves indicate that carbon savings may be obtained in the South at a significant savings and low net cost, provided that the required reductions do not increase much above the twenty per cent level. Much more research is needed to ascertain the true shape of these curves and the extent of possible reductions in southern economies. Although many studies have begun, usable results will not be available in many southern signatories to the Climate Change Convention until about 1995.

Some data are also available for the costs of carbon fixing by reforestation and reversal of deforestation where costs range from a few dollars to upwards of $60 per tonne of carbon fixed (see Figure 5.6); and for country level industrial, building, and transport energy efficiency (with implied costs per abatement of carbon emissions). But virtually without exception, these data are scattered, apply to different years, are incomplete (that is, only some portions of cost, such as capital cost, are given), and are partial (that is, only parts of a sector are covered). In short, these data are almost useless for the purposes of a comprehensive, global analysis. It will take some years before carbon abatement cost data of greater reliability and scope are generated and collated. Until then, the only recourse is to use hypothetical curves supplemented with point data and judgement.

Figure 5.6 Carbon abatement by forestry management

Marginal cost curves
The studies referred to above provide a range of marginal cost estimates drawn from two schools of energy economics that may be termed the top-down, price-driven 'economic pessimists' (high cost) on the one hand, and the bottom-up, end-use oriented, 'technological optimists' (low cost) on the other.

As was noted earlier, I constructed three hypothetical marginal cost curves drawn from this literature to calculate incremental abatement cost (see Figure 5.4). 1 am not suggesting that any of these curves are 'correct'. Rather, I seek to ascertain the impact of each curve on the estimated transfer justified by the obligation-to-pay index.

I simplified the well-known Nordhaus marginal cost schedule to refer to a required reduction level (the range in the ratio 'RR' defined in the box above summarizing the incremental cost method), using a weighted averaging procedure to simplify the original cost schedule into a simple, three-stage step function. This cost estimate is called Case 1, as are subsequent treatments of incremental abatement costs and distribution of costs based on it. It amounts to a high cost estimate in this chapter (although the low cost case is steeper at higher levels of required reduction).

Similarly, I used the NAS high cost estimate at 100 per cent implementation of technological potential, again employing a weighted averaging procedure to simplify the curve to three abatement levels that are very close to (but not exactly the same as) the ranges used in the Nordhaus marginal cost curve. This curve is called Case 2 and like Case 1, denotes subsequent calculations based on it. It amounts to a low cost estimate in this chapter.

Third, I postulated yet another marginal cost curve that combined elements of Cases 1 and 2. Specifically, the curve for the North and the East were taken from NAS and are identical with those used in Case 2. The curve for the South, however, assumes positive cost for the first two abatement levels, but at a level substantially below those given by Nordhaus in Case 1. The cost for the highest abatement level reverts to the same as Case 2; the first level, $30/T C-y-1 I abated, is based on an estimated reforestation option in the South; the second level, $77/T C-y-1 abated, is a hypothetical estimated cost of industrial energy efficiency in developing countries. The World Bank's Global Environment Facility, for example, has published similar estimates that also hold the abatement costs to be mostly positive. The upper level in Case 3 of $294/T C-y-1 abated is the same as in Case 2.

The reason for adopting this composite marginal cost curve, called Case 3 hereafter, will become evident below. But crudely, the curve serves as a medium alternative to the high Case 1 and to the low Case 2.

Three aspects of these marginal cost curves should be kept in mind in subsequent applications. First, it should be noted that the cost in dollars per tonne of carbon abated is an annual, levelized cost of full, life cycle costs including components of capital, operating and maintenance, and salvage cost. Thus, each tonne of carbon abatement paid for in a given year must be paid for again the following year.

Second, the incremental abatement profiles do not account for any benefits that flow from abatement (such as reduced local pollution, preservation of ecosystem values and services, etc.). As noted earlier, it is not possible yet to estimate the benefits of avoided climate change in a quantitatively meaningful fashion. For this reason, it is also highly unlikely that the protocols to the Convention will be based on a global cost-benefit analysis. The analysis herein is therefore a truncated approach that amounts to determining least cost abatement paths for various countries and groupings of countries, and allocating the cost on efficiency and equity criteria.

Third, only technologically defined costs are included here. Many other components of cost are not included. These costs include the impact of carbon emission reductions on trade competitiveness; macroeconomic impacts, especially intersectoral adjustment costs; human resource development costs; and institutional change costs. Little work has been done to measure these categories of cost as of early 1993, especially in developing countries.

A final caveat is in order. No attempt has been made to ensure that the three cost curves used in this study are stated in constant dollars in the same year, using the same discount rate or other underlying assumptions. The curves have been taken 'as given' in the sources and are used in an heuristic exercise to calculate incremental costs. An actual application of this procedure under a protocol of the Convention would have to attend to these methodological details more rigorously.

With the marginal cost schedules defined, it is necessary next to determine the projected emissions that must be abated. I develop two scenarios to this end, a reference scenario, and an efficiency scenario for purposes of abatement cost calculation.

Reference emission projections

Many projections exist for future emissions. These projections are sometimes calculated via large-scale econometric modelling. Simpler methods using projected population, GNP per capita, and emissions per capita are also popular. I used a simple method in this study that applied projected emissions per capita estimated by the IPCC working groups to projected populations in 1995, 2000, and 2025. The intermediate years were then interpolated.

Whichever method is employed contains many variables that are subject to great uncertainty. A GNP-based estimate is subject to enormous variation in projected GNP, even if historical GNP per capita and emissions per GNP rates are employed. Econometrically-driven models simply project the past onto the future (and probably overstate energy demand and thereby emissions).

Simple population-based models are attractive because the emissions per capita can be estimated physically. However, the method suffers from the disadvantage that demographic dynamics are also controversial and projected populations may be wrong; and projected emissions per capita from the IPCC are available only at a regional level (and thus are often too high or too low for a given country over the first five years of the projection).

Figure 5.7 Projected CO2ff emissions, 1995-2025, reference scenario

The projected carbon dioxide emissions from fossil fuel usage in the reference scenario are shown in Figure 5.7. Starting at about 6.6 GT/y of C in 1995, the world reaches about 11.8 GT/y of C in 2028 in the unrestrained reference scenario. This projection is consistent with a variety of other IPCC and OECD projections and is adopted in this study as the reference scenario.


Contents - Previous - Next