Copyright 1994, Richard S. J. Tol. All rights reserved. For more information contact author. THE CLIMATE FUND Optimal Greenhouse Gas Emission Abatement Richard S.J. Tol[0] 8 July 1994 INTRODUCTION The greenhouse effect is a hot topic. Thus, it is not surprising that the Dutch ministry of Housing, Physical Planning and Environment commissioned a Dutch National Research Programme: Global Air Pollution and Climatic Change (in short: NOP-MLK). This programme seeks to investigate all impacts, from physical to philosophical, of the presumed global change and the options and measures to counteract this. This interim report is the fourth one written in the context of the project `Socio-Economic Aspects of the Greenhouse Effect: Climate Fund' of the NOP-MLK. This project investigates the possibility of setting up an international climate fund and the way in which this could be arranged. A climate fund is a method to enhance the efficiency and efficacy of the reduction of the increase in the amount of greenhouse gases (GHGs) in the atmosphere; this increase presumably leads to a change in the global climate. The idea behind the fund is simple: those countries which are willing and able to pay for the reduction of the emission of GHGs are not necessarily the countries to achieve this goal in the most efficient and effective way. Economic theory teaches that enlarging the number of feasible options for reduction, e.g., by incorporating international capital transfers, lowers the total economic costs. Therefore, it would be sensible to reallocate the economic, technical, social and natural efforts. The proposed climate fund is a tool to coordinate this. At the same time, the climate fund could be a way to redistribute the damages caused by global warming. This project focuses on the way on the economic and game- theoretic justification of a climate fund. Most existing studies on the economic aspects of the greenhouse effect neglected international capital transfers or took them for granted. This report belongs to both the second and the third stage of the project. In the first stage (Tol, 1993a) we treated the general background of the greenhouse effect and surveyed the large and diverse literature on the costs and benefits of greenhouse gas reduction measures and climate change. Knowledge of costs and benefits is the first step towards sensible proposals for a climate fund. Focus was on the regional distribution of the economic impacts of the presumed global warming and of its countermeasures; we attempted to reach a synthesis on the subject. This synthesis, combined with some new information, forms the base of the second stage, first reported in Tol (1993b). In the second stage, we design a model for the costs and benefits of greenhouse politics in the nine world-regions defined earlier[1]. The model has been named the Climate Framework for Uncertainty, Negotiation and Distribution (FUND). In the first stage, particular attention was paid to the distribution of costs and benefits of GHG abatement measures over the globe. In the second stage, these are modelled in Climate FUND. Tol (1993b) describes FUND, version 1.0; Tol (1994a) describes FUND, versions 1.1 and 1.2, with particular attention to interregional capital transfers. Following Tol (1993c) and Tol (1994b), a number of the assumptions underlying FUND, versions 1.0 to 1.2, need to be revised. This report therefore starts with presenting the revised version of FUND (Chapter 1), with a particular focus on the differences between the present and the past assumptions. Chapter 2 continues with discussing the optimal regional reduction policies Ń after discussing what `optimal' might be Ń both under the assumption of neglecting and incorporating the other regions' abatement measures. In the third stage of the project, the negotiation part of FUND is represented by an analysis of the impact of interregional capital transfers on the efficiency and efficacy of emission reduction policies. The manner in which these capital transfers are modelled as well as the results of these efforts are the subject of Chapter 3. Finally, chapter 4 summarises, concludes and comments the previous chapters. The last stage (of this project) will pay particular attention to the uncertainties, although analytical results of uncertainty analysis were incorporated in earlier stages as well. 1. THE CLIMATE FUND, VERSION 1.3. This first chapter treats the set-up of the Climate Framework for Uncertainty, Negotiation and Distribution, version 1.3, the model on which the results of the later chapters are based. Section 1.1 is on the general features of the model; compared to version 1.2, only minor revisions are made and therefore the elaborations are brief. Section 1.2 treats the revision of the climate module of FUND, which brings it a bit closer to climatological theory. Section 1.3 discusses the damage costs of climate change which underwent major changes compared to version 1.2. Section 1.4, finally, elucidates the (unaltered) costs of greenhouse gas emission abatement. 1.1. General Features The overall set-up of the Climate Framework for Uncertainty, Negotiation and Distribution, version 1.3, is the same as for the earlier versions (cf. Tol, 1993b; Tol, 1994a). Essentially, FUND consists of a set of exogenous scenarios and endogenous perturbations, specified for nine major world- regions. The exogenous scenarios concern the rate of economic growth, the share of agriculture in Gross Regional Product (GRP), the population growth, autonomous energy efficiency improvements, the rate of decarbonisation of the energy use (autonomous carbon efficiency improvements), and non-carbon emissions (expressed in carbon equivalents). The endogenous parts of FUND consist of the climate module (emissions leading to changes in the atmospheric composition leading to changes in the global mean temperature leading to changes in the sea level and the number of hurricanes; cf. section 1.2), the impact of carbon dioxide emission reductions on the economy and the emissions (cf. section 1.4), and the impact of the damages of climate change on the economy and the population (cf. section 1.3). The costs and benefits (read: avoided damage) of emission reduction are weighted against one another according to a utility function (cf. section 2.2). Gross regional product is given by with j and t the region and the time period, respectively, g0 the exogenous growth rate, gT the reduction in growth due to energy taxes, and gE the reduction in growth due to external (outer-region) emission reduction (set to zero for all regions but the Middle East), andwith LT the damage costs of global warming, DT the tangible damages of air pollution, and CR the costs of emission reduction. The exogenous economic growth rates are based on interpolations of the figures of Manne and Richels (1992). The initial GRPs are based on WRI (1991). The other variables are discussed below. The regional populations follow a similar scheme:with gP the exogenous population growth, andwith D the number of additional deaths as a result of the enhanced greenhouse effect, L the number of people leaving a region, and E the number of people entering. The exogenous growth rates are based on interpolations of the Worldbank's long-term population projections (Bulatao et al., 1990). The initial values are based on WRI (1991). Gross agricultural product is an exogenously specified share of gross regional product; this share is assumed to be declining. Energy use is also modelled as a linear function of the GRP; the energy intensity parameter declines with the autonomous and policy-induced energy efficiency improvements. Carbon emissions are linear in energy use; the carbon intensity parameter declines with the autonomous and policy- induced carbon efficiency improvements. CFC emissions are assumed to be phased out. Other greenhouse gas emissions are assumed to grow with the economy and to decline with the autonomous energy and carbon efficiency improvements. All parameters are calibrated using WRI (1991) and Manne and Richels (1992) as the main sources. The main differences between FUND, versions 1.0 to 1.2, and FUND, version 1.3, are (i) that the exogenous scenarios in version 1.3 are linearly interpolated between the decade figures of version 1.0 and (ii) that the expected damages of climate change and sea level rise are replaced by the modal damages in version 1.3. The latter point implies that FUND has lost its hybrid half-stochastic, half-deterministic character; uncertainty analyses will be performed in the next report. 1.2. The Climate Module The climate module of FUND, version 1.3, differs considerably from the one in the versions 1.0 to 1.2. First, the global mean temperature is assumed to depend on the logarithm of the effective atmospheric concentration of greenhouse gases (expressed in carbon equivalents). The effective concentration is the weighted sum of the atmospheric concentrations of the previous twenty years, with exponentially increasing weights for earlier years, that is, the initial response of the temperature to a change in the atmospheric composition is small but exponentially growing to its new equilibrium after twenty years. This preserves the climate's delay in its response to changes in the atmospheric composition of FUND, version 1.0, but improves on its jump- like behaviour. The logarithmic specification is inspired by climatological theory (Shine et al., 1990). The modal temperature rise is set to 2.5ūC for a doubling of atmospheric carbon dioxide equivalents, corresponding to the IPCC's best guess (Houghton et al., 1990, 1992). The level of the sea is assumed to rise linearly and simultaneously with the temperature, at a pace of 17 cm/ūC. The second major change in the climate module is explicitly taking up the number of hurricanes. This number is assumed to stay constant under global warming, which is the best-guess of the climatologists at the time of writing (Olsthoorn and Tol, 1993; Vellinga and Tol, 1993); it is possible, however, that this number might increase drastically (up to 440%; Ryan et al., 1992). This will be taken up later on in the uncertainty and sensitivity analyses. 1.3. The Costs of Climate Change The modelling of the damage costs of climate change underwent major changes in the revision of FUND; part of the new parameterisation is also described in Tol (1993c). Essentially, the structure of the damage costs remains the same. The costs are split into tangible (i.e., marketable) and intangible (i.e., non-marketable) goods. This distinction is made because tangible damages affect the economy directly, and so human welfare indirectly, whereas intangible damages affect human welfare directly, and so the economy indirectly. Ignoring this distinction can have a profound influence on the model results (Tol, 1994b). The costs are attributed to changes in the temperature, sea level or hurricane number. The costs are further split into those due to the change compared to a base level (denoted by a subscript 0) and those due to the rate of change. Parts of the costs are linear in the relevant parameter, other parts are quadratic. 1.3.1. Agriculture The largest change concerns agriculture. At the time of writing the first report of this project (Tol, 1993a), the present author was not aware of any comprehensive study on the probable losses to agricultural production due to climatic change. Because of this lack of knowledge, FUND, version 1.2, sets the direct agricultural losses to 7.5% of Gross Agricultural Product (GAP), depending linearly on the rate of climate change. The indirect losses are set equal to 1.2 times the direct losses plus .01 time the direct losses squared. In the meantime, the study of Rosenzweig et al. (1993; see also Fischer et al., 1993, and Rosenzweig and Parry, 1994) became available. This study contains changes in the production of the four major grain crops (wheat, rice, maize and soybean) under a number of climate change and adaptation levels. It is based on a suite of site-specific crop models and a world food trade model. Although there are still many shortcomings to the approach (cf. Reilly, 1994, but also the original authors), it is the only study of this kind presently available; so, the agricultural losses in FUND are adjusted accordingly. Table 1.1 translates the national figures of Rosenzweig et al. (1993) into an assessment for the nine regions of FUND. Under suitable adaptation, the OECD, Central and Eastern Europe and the former Soviet Union, and Centrally Planned Asia appear to be gaining from a changed climate (see Xia and Wei, 1993, for comments on the positive impacts on China's agriculture). On the other hand, the difference between the adaptation and no adaptation scenarios indicates that most regions, particularly OECD-America and Centrally Planned Asia, are vulnerable to climate change itself. Peculiarly, Europe, the former USSR, Japan, Australia and New Zealand are not susceptible to changes, i.e., adjustment to climate change does not alter agricultural yields substantially. Table 1.1 results in a damage cost function in the following way. All damages are assumed to be tangible. The figures in the third-right column are associated with the damages of a changed climate (note that these can be negative, i.e., benefits), represented by a in equation (1.5); the figures in the rightmost column are associated with a changing climate, represented by § in equation (1.5). The latter costs are assumed to be for three-quarters due to the rate of temperature changes and for one-quarter due to the rate squared. Thus, 1.3.2. Benchmark Climate Change Damages We now discuss the benchmark climate change damages. The most important studies on the socio-economic impacts of climate change are those of Nordhaus (1991), Cline (1992) and Fankhauser (1992, 1993, 1994a), which are reviewed and compared to others by Nordhaus (1993), Fankhauser (1994b) and Tol (1993a). These studies attempt to assess the tangible and intangible damages due to a doubling of atmospheric carbon dioxide on a highly aggregate level, based on the literature on case studies and educated guessing and extrapolation. Their findings for the USA are summarised in Table 1.2. The results of Tol (1993a) are adjusted according to the discussion on agricultural losses described above. Also, the intangible damages associated with the loss of species were, by mistake, an order of magnitude too low; this has been restored. Table 1.1. Agricultural Yield Changes 2xCO2 (percents of Gross Agricultural Product)a modelb UKMO GISS GFDL avg. avg. diff. region\scenarioc 1 2 3 1 2 3 1 2 3 2+3d 1e avg.f OECD-A -20.0 -5.0 -5.0 -5.0 +10.0 +10.0 -5.0 +10.0 +10.0 +5.00 -10.00 -15.00 OECD-E +5.0 +5.0 +5.0 +10.0 +10.0 +10.0 -5.0 -5.0 -5.0 +3.33 +3.33 0.00 OECD-P +7.5 +7.5 +7.5 +7.5 +7.5 +7.5 +7.5 +7.5 +7.5 +7.50 +7.50 0.00 CEE&SU -7.5 -7.5 -7.5 +22.5 +22.5 +22.5 +7.5 +7.5 +7.5 +7.50 +7.50 0.00 M-E -22.5 -22.5 -7.5 -7.5 -7.5 +7.5 -7.5 -7.5 +7.5 -5.00 -12.50 -7.50 L-A -22.5 -22.5 -8.5 -15.0 -15.0 -1.0 - 10.0 -10.0 +4.0 -8.83 -15.83 -7.00 S&SEA -20.0 -20.0 -10.0 -10.0 -10.0 0.0 -10.0 -10.0 0.0 -8.83 -13.33 -4.50 CPA -7.5 +7.5 +7.5 +7.5 +22.5 +22.5 +7.5 +22.5 +22.5 +17.50 +2.50 -15.00 AFR -20.0 -20.0 -20.0 -7.5 -7.5 +7.5 -15.0 -15.0 0.0 -6.67 -14.17 -7.50 Notes: a After Rosenzweig et al. (1993); cf. also Fischer et al. (1993), Rosenzweig and Parry (1994) and Reilly (1994). b The climate change scenarios used are the equilibrium 2xCO2 experiments according to the General Circulation Models of the United Kingdom Meteorological Office, the Goddard Institute for Space Studies and Geophysical Fluid Dynamics Laboratory. c The scenarios concern no adaptation (1), minor shifts (2) and major shifts (3) in behaviour. d The average of adaptation scenarios 1 and 2 over the three models. e The average of the no adaptation scenario over the three models. f The difference between the averages described under notes e and f. Table 1.2. US Climate Change Damage (2xCO2 - 109$(1988))a loss category Fankhauser Cline Nordhaus Tolb coastal defence 0.2 1.0 7.5 1.5 dryland loss 2.1 1.5 3.2c 2.0 wetland loss 5.6f 3.6 - 5.0 species loss 7.4f 3.5 - 5.0 agriculture 7.4 15.2 1.0 18.0 forestry 1.0f 2.9 - - fishery - - - - energy 6.9f 9.0 - - water 13.7 6.1 - - other sectors - 1.5 38.1d - amenity -f - - 12 life/morbidity 16.6 >5.0 - 33.3e air pollution 6.4 >3.0 - - migration 0.5 0.4 - 0.6 natural hazards 0.2 0.7 - 0.25 total USA 68.0f >53.5 50.3 77.5 (% GDP) (1.4f) (>1.1) (1.0) (1.6) world total 304.2f - 220.0 383.7 (% GDP) (1.6f) (1.33) (1.9) a Table adapted from Fankhauser (1994a), Tol (1993a) and Nordhaus (1993). b Including Canada. c Total land loss (dry- and wetlands). d Including those not assessed. e Tol values an American life twice as high as Fankhauser does. f Fankhauser (1992, 1993) guesstimated $8.4 billion for wetland loss, $6.4 billion for species loss, -$1.8 billion for forestry, nought for energy, $6.8 billion for amenity, $64.1 billion in total (1.3% of GDP) and $250.0 for the world total (1.5% of total world product). The four studies grossly agree on the damages for the USA, with Nordhaus on the lower side and Tol on the higher. Major loss categories are sea level rise and agriculture. The damages for the world as a whole differ more, with Fankhauser and Nordhaus in close agreement but with Tol considerably higher, based on the recent literature on agriculture, which reports only limited impacts of the carbon dioxide fertilisation (e.g., Erickson, 1993) and more adverse impacts on agriculture in general (e.g., Fischer et al., 1993). Other reasons are the higher values attached to a human life by Tol (somewhere in the middle of the range reported by Cline, whereas the figures of Fankhauser and Cline are at the lower end), and the (intangible) losses due to people migrating (thrice the income per capita per displaced person; cf. Jansen, 1993). Table 1.3 contains the estimated damages for nine major world regions, after Tol (1993a) and adjusted according to discussions above. Notice that the figures in Tables 1.2 and 1.3 represent the losses for the present economy. Some of the damage costs will grow with the economy and the population, others will decline, such as the agricultural losses in the developing countries, and others will increase, particularly the intangibles. Table 1.3 Climate Change Damage (2xCO2 - 109$(1988))a regionsb 1 2 3 4 5 6 7 8 9 total coastal defence 1.5 1.7 1.8 0.5 0.0 1.0 2.0 0.5 0.5 9.5 dryland loss 2.0 0.5 4.0 1.3 0.0 0.5 1.0 0.0 0.5 9.8 wetland loss 5.0 4.0 4.5 1.3 0.0 1.5 1.5 0.5 0.5 18.8 species loss 5.0 5.0 5.0 2.5 0.0 2.0 1.0 1.0 0.5 22.0 agriculturec 18.0 -5.1 -6.0 -13.51.7 17.4 38.2 -3.2 17.4 74.9 amenity 12.0 12.0 12.0 -1.0 0.1 0.4 1.2 1.0 0.5 38.2 life/morbidityd 33.3 45.0 38.6 22.6 2.6 8.6 20.6 15.5 7.8 194.5 migration 0.6 1.3 0.6 0.5 0.0 2.3 4.7 2.8 1.3 14.1 natural hazards 0.3 0.0 0.8 0.0 0.0 0.1 0.1 0.5 0.4 2.1 total 77.6 64.4 61.2 14.1 4.4 33.8 70.3 18.6 29.9 383.7 (% GDP) (1.6)(1.5)(2.9)(0.5)(13.4)(4.7)(11.3)(5.3)(8.5)(1.9) a After Tol (1993a). b 1 = USA and Canada; 2 = OECD-Europe; 3 = Japan, Australia and New Zealand; 4 = Central and Eastern Europe and the former Soviet Union; 5 = the Middle East; 6 = Latin America; 7 = South and South East Asia; 8 = Centrally Planned Asia; 9 = Africa. c Assuming the base rate of temperature change, i.e., ĘT=0.04ūC/year. d The value of a human life is assumed to be $3,000,000 for an inhabi-tant of the OECD, $1,500,000 million for someone from Central and Eastern Europe, the former Soviet Union and the Middle East, $400,000 for Latin Americans, and $300,000 for the inhabitants of the remaining regions. 1.3.3. Damage Cost Functions We now briefly discuss how the benchmark estimates of subsection 1.3.2 are transformed into eight-parameter cost functions. The parameters themselves are presented in Table 1.4. The cost functions have the same form for all regions, the parameters differ. The costs of wetland loss are assumed to be linear in sea level rise. Half are due to the pace of sea level rise. Also, half of the costs are assumed to be intangible. These choices reflect the ignorance of the present author. The costs of dryland loss and coastal protection are linear, tangible, and due to total sea level rise. Drylands too valuable to be lost are supposed to be protected. The (tangible[2]) costs are for one half due to total sea level rise (linearly) and for the other half due to the pace of the rising sea level (one quarter linearly and the other quadratically); faster implemented protection will lead costs to increase more than proportionally, because of budget and capacity constraints (cf. Jansen et al., 1993). The number of people leaving is assumed to be linear in total sea level rise. The intangible costs of people leaving are set at three times the average income per capita per person. The tangible costs of people entering are assumed partly linear and partly quadratic in the number of people, because of budget and capacity constraints. Note that people who are displaced are counted as both leaving and entering a region. The agricultural losses are assumed to be quadratic in the pace of global warming, although with only a small quadratic term. The agricultural damages are completely tangible. The damages due the enhanced natural disasters are assumed to be quadratic in total hurricane frequency increase. Note that Fankhauser (1993) assumes linearity. However, small changes will incur small damages whereas larger changes might set an adverse chain of events in motion, such as insolvencies in the (re)insurance sector, with spill over costs to the whole of the financial sector and the public funds. One quarter of the damage costs is ascribed to the quadratic term, a conservative choice. The natural hazard damages are split in tangible and intangible losses, the latter represented by the additional loss of human life. The intangible costs of the loss of species due to climate change is assumed to be quadratic in total temperature rise and its pace, with equal shares; the linear parameter represents three-quarters of the costs, as a conservative choice[3]. The intangible damages due to loss in human amenity and increased mortality and morbidity are assumed to be quadratic in the pace (seven- eights) and total (one-eight) global warming; this is inspired by Cline (1992). To summarise, the damage cost functions look like and with DC the damage costs and PI the climatological parameter of interest. The parameters can be found in Table 1.4. More details can be found in Tol (1993b). Table 1.4. Climate Change Damage Costs Functions tangible damage intangible damages parameter PI ĘPI PI2 ĘPI2 PI ĘPI PI2 ĘPI2 of interest coastal defence 0.5 0.25 0.25 sea level dryland loss 1 sea level wetland loss 0.25 0.25 0.25 0.25 sea level species loss 0.5 0.5 temperature agriculturea 1 0.75 0.25 temperature amenity 0.125 0.875 temperature life/morbidity 0.125 0.875 temperatureb emigrationc 1 sea level immigrationd 1 sea level natural hazards 0.75 0.25 hurricanese a Cf. equation (1.5) and Table 1.1. b The presumed increase in the number of hurricanes also induces some additional losses of human life. c This represents the number of migrating people. The costs are set to three times the average income per capita times the number of people leaving. d The number of people entering, N, is assumed to be linear in the total sea level rise. The costs of this supposedly equal bN+0.005N2; following Fankhauser and Cline, b equals $4,000 for the OECD, $1,500 for Central and Eastern Europe, the former Soviet Union and the Middle East, $750 for Latin America and $500 for the remaining regions, reflecting the differences in the services offered to refugees as well as in the costs thereof. e Number of hurricanes. 1.4. The Costs of Emission Abatement The functions which describe the costs of greenhouse gas emission abatement in FUND, version 1.3, are the same as in the earlier versions (see Tol, 1993b, for details). The costs of conventional air pollution, associated with the burning of fossil fuels, is assumed linear in the total emissions amount; half of the costs is assumed to be tangible, the other half intangible. As the study investigates emission reduction, the costs of conventional air pollution are benefits of emission abatement. The costs of energy and carbon saving programmes are assumed to be quadratic in the initial amount of additional (i.e., on top of the autonomous energy and carbon efficiency improvements) energy and carbon saved, with the first part of the savings at a net negative cost. The costs are modelled as dead-weight losses to the gross regional product. The impact of the savings is assumed to slowly fade out after the start of the programme. The costs of fiscal measures are assumed to be quadratic in the amount of the tax, and strictly positive. The costs impact on the economic growth rate, and slowly fade away after the year of instalment as the economy adjusts to the new tax. The impacts on the energy and carbon efficiency are proportional to the square-root of the tax. Fiscal measures and energy and carbon efficiency programmes are assumed to be mutually exclusive. The economic growth rate of the Middle East is lowered proportional to the sum total of the other regions' policy induced energy and carbon savings. The costs of afforestation (stopping deforestation is neglected in FUND) are assumed to be quadratic in the amount of land afforested. These costs are modelled as dead-weight losses to the gross regional product. 2. OPTIMAL EMISSION ABATEMENT STRATEGIES This second chapter treats the optimal greenhouse gas emission abatement strategies without interregional capital transfers. Capital transfers are the subject of the next chapter. The discussion is started by presenting the business as usual scenario according to the assumptions of Climate FUND, version 1.3, and comparing this to the no damage scenario. The chapter continues with some remarks on the definition of optimality, i.e., the utility function, following Tol (1994b). Next the regionally optimal emission reduction strategies under neglect of the other regions' measures are elucidated. Section 2.4 is on the impact on the abatement strategies of taking up this knowledge. Section 2.5 elaborates another new feature of FUND, version 1.3: a cooperative emission reduction game. 2.1. The Business as Usual Scenario This section compares the business as usual scenario to the no damage scenario. The business as usual scenario is defined as the scenario in which no greenhouse gas emission abatement is implemented. The no damage scenario is defined as the scenario in which no abatement is implemented and in which climate change causes no socio-economic impact (or, equivalently, the enhanced greenhouse effect causes no climate change). Comparing the two scenarios yields insight in the nature of the damage costs of climate change as implemented in FUND, version 1.3. Table 2.1 contains some of the initial parameters of FUND. These include the gross regional product, the gross regional welfare (i.e., the natural logarithm of the GRP corrected for the intangible losses Ń see the discussion in section 2.2), the regional population, the carbon dioxide emissions and the emissions of the other greenhouse gases for the starting year 1990. Figure 2.1 depicts the sum of the gross regional products over the OECD and the non-OECD regions for the business as usual and no damage scenarios. It is clear that the impact of climate change is much more profound in the poorer regions. In the no damage scenario, the non-OECD gross product approaches the OECD whereas in the business as usual scenario the difference remains about the same. Note that the non-OECD population grows faster. Figure 2.1. The sum total of the gross regional products of the OECD and the non-OECD regions according to the no damage and the business as usual scenarios. Table 2.2 contains the model's outcomes for the business as usual scenario, summarised with the net present values of the gross regional product and welfare (NPVY and NPVW, respectively), and the GRP, welfare and population in the year 2100 as compared to 1990 (y2100, w2100 and p2100, respectively). A slight economic decline is observed for Africa, and an enormous growth for Centrally Planned Asia; on the other hand, CPA's welfare collapses, i.e., the economic boom results in an ecologic disaster. Overall, the economic growth is larger than the welfare growth. Table 2.1. Initial Parameters region GRP Welfare Population CO2 other GHGs (109 $) (ln 109 $)(mln) (mln tCa) (mln tCb) OECD-A 5,000 8.52 275 600 540 OECD-E 4,250 8.35 340 380 555 OECD-P 2,100 7.65 140 150 155 CEE&SU 2,900 7.97 510 610 285 M-E 32.5 3.48 60 65 50 L-A 720 6.58 450 115 190 S&SEA 620 6.34 1,650 135 195 CPA 350 5.86 1,250 280 135 AFR 350 5.86 650 75 100 a tC stands for tonnes of carbon b expressed in tonnes of carbon equivalents Table 2.2. Summary of the Business as Usual Scenario region NPVY NPVW y2100 w2100 p2100 (1012 $) (ln 109 $)(-) (-) (-) OECD-A 599.64 610.72 2.65 2.52 1.19 OECD-E 566.94 605.40 3.15 2.90 1.18 OECD-P 301.13 560.62 3.24 2.64 1.63 CEE&SU 243.42 386.26 3.35 3.19 1.09 M-E 2.11 166.21 1.78 0.61 3.16 L-A 26.30 176.79 2.64 2.37 1.81 S&SEA 22.98 143.87 3.79 2.05 2.33 CPA 13.77 132.32 12.18 0.00 1.46 AFR 8.52 125.00 0.94 0.80 3.80 Figure 2.2 depicts the sum total of the tangible losses of the OECD and the non-OECD regions due to global warming and sea level rise according to the business as usual scenario. The losses due to global warming rise fast at first, then fall and finally start rising fast again, following the pace of the temperature change. FUND projects a sharp temperature increase for the coming decades; after that, as the economy growth rate declines, the greenhouse damages accumulate and the energy and carbon use gets more efficient, the emissions decline and the rate of climate change falls; but, as the climate change damages fall, and, for some regions, become negative, i.e., benefits, and the energy and carbon efficiencies reach their limit, emissions start growing again and temperature starts to rise rapidly, leading to a sharp increase in the damage costs[4]. The damage costs of sea level rise steadily through time as they largely depend on the absolute sea level rise; the last decades show the same acceleration of damages. Figure 2.2. The sum total of the tangible losses of the OECD and the non-OECD regions due to global warming and sea level rise according to the business as usual scenario. Figure 2.3 depicts the sum total of the intangible losses of the OECD and the non-OECD regions due to global warming and sea level rise according to the business as usual scenario. The damages steadily grow, and accelerate towards the end of the period. The losses due to global warming dominates those due to sea level rise. Table 2.3 compares the business as usual scenario to the no damage scenario, presenting (XND-XBaU)/XBaU, where X represents the same features as in Table 2.2. Table 2.3 reconfirms Table 1.3, pointing out that the poorer regions, particularly Africa, are most vulnerable to climate change, and that the OECD, particularly Europe, is reasonably well off. Central and Eastern Europe and the former Soviet Union even benefit from climate change. The richer regions, particularly the European parts of the OECD, are faced with a large number of immigrants, though. Figure 2.3. The sum total of the intangible losses of the OECD and non-OECD regions due to global warming (left axis) and sea level rise (right axis) according to the business as usual scenario. Table 2.3. Comparison of the Business as Usual and No Damage Scenarios region NPVY NPVW y2100 w2100 p2100 (%) (%) (%) (%) (%) OECD-A 8.14 0.79 12.20 16.05 -5.30 OECD-E 1.01 0.22 1.39 8.15 -10.67 OECD-P 4.83 0.90 10.68 32.59 -4.44 CEE&SU -6.06 -0.53 -14.10 -11.41 -3.70 M-E 16.87 7.69 23.77 251.05 -4.66 L-A 25.77 2.66 112.75 131.91 2.44 S&SEA 44.70 3.83 347.59 580.89 1.03 CPA 61.64 6.50 5.66 -a 1.52 AFR 70.60 6.99 1061.37 1176.27 2.20 a The welfare of Centrally Planned Asia collapses in the business as usual scenario. 2.2. Some Remarks on the Definition of Optimality Before we start calculating what an optimal reduction strategy might be, we first have to define what we mean by `optimal'. Within the neo-classical paradigm, economic agents are assumed to maximise their utility, and the policy which yields the utility maximal is the optimal policy. Many justified objections exist against this maximising behaviour assumption, but it is, for convenience, adopted here. This leaves us the question how to define `utility'. Traditionally, utility U is the discounted sum of the natural logarithm of the present and future consumption or income. However, human behaviour is not solely determined by the consumption of economic goods, other `goods', such a human amenity and nature, play equal parts; these goods are called intangibles. In environmental economics, the intangible goods are expressed in their monetary value by implicitly equating the utility of the intangible good and the utility of a certain amount of money. This does not imply that tangibles and monetised intangibles are similar, but rather that they are cast in a common metric. The utility is then defined as the discounted sum of some function of the present and future consumption or income and the monetised intangibles. The function chosen in FUND, version 1.3, is the natural logarithm of the sum of income and intangibles, implying (i) declining marginal returns for both tangibles and intangibles, and (ii) substitutability of the utility of tangibles and intangibles. Both implications are open to questions, especially the latter. In order to compensate for the latter, still keeping the computational requirements manageable, the monetary value of the intangibles is assumed to grow linearly with the income per capita[5]. That is, if `nature' is valued at one percent of the national income at present, then it is valued at two percent of the income if this doubles. Fankhauser (1994c) uses the same specification, based on Pearce (1980) So, tangibles and intangibles are still substitutable in utility, but substituting tangibles for intangibles will lead to an increase in the value of the latter, and thus pays less. Tol (1994b) discusses the implications this utility specification has for Nordhaus' (1993) DICE model. The final question to be answered is which discount rates to use. As the yearly utility is discounted, the discount rate reflects the pure rate of time preference. On ethical grounds, one could argue that this should be zero, as, apart from the future generations being richer anyhow[6], there is little reason to prefer the present generations' welfare over the future generations' welfare (see Cline, 1992, for a discussion of the discount rate). However, this would also require a full specification of the utility function, rather than the ad hoc approach described above. Moreover, positive rates of pure time preference are common practice, and a zero rate would require an infinite time horizon. Therefore, the discount rate are set at 1% per year for the OECD, 2% for Central and Eastern Europe and the former Soviet Union and the Middle East, 4% for Latin America and 5% per year for the remaining poorest regions. The discount rates are fixed over time in order to reflect the interests of present-day decision makers. 2.3. Regionally Optimal Abatement In this section we discuss what would be the optimal greenhouse gas emission policies for all regions given the assumption that the other regions would behave as under the business as usual scenario. Table 2.4 contains the abatement measures the regions implement under this scenario, enhancing energy and carbon efficiency and carbon absorption while reducing the economic (potential for) growth. The OECD is rather active, for certain compared to FUND, version 1.2, but the Middle East and Africa (!) are also on the move, obviously because of their high vulnerability. Interestingly, also Central and Eastern Europe and the former Soviet Union reduce their emissions, even though they benefit from the business as usual scenario (compared to the no damage scenario); apparently, they can benefit more if the climate changes less than under business as usual. Table 2.4. Regionally Optimal Abatement Measures region 1 2 3 4 5 6 7 8 9 taxa 2.8 2.3 1.3 0 0 0 0 0 0 energyb 0 0 0 0.5 0.5 0.3 0.3 0.3 0.3 carbonc 0 0 0 0.2 0.2 0.2 0.2 0.2 0.2 forestryd 5 10 5 5 0 0 5 15 105 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Table 2.5 compares the business as usual scenario to the regionally optimal abatement scenario in the same manner as in Table 2.3. The American parts of the OECD do better under this scenario than under the no damage scenario, and the Pacific parts of the OECD raise their income compared to the no damage situation. The other regions also benefit from the emissions reduction, but not to that large an extent, except for the Middle East, despite the losses it faces due to the reduced income from oil exports. Compared to FUND, version 1.2, the emission reductions clearly have a more profound impact. Table 2.5. Comparison of the Business as Usual and Regionally Optimal Abatement Scenarios region NPVY NPVW y2100 w2100 p2100 (%) (%) (%) (%) (%) OECD-A 11.35 0.92 25.34 26.66 -0.16 OECD-E 8.98 0.72 20.10 22.41 -0.33 OECD-P 6.62 0.61 14.13 20.50 -0.12 CEE&SU 4.50 0.43 8.20 9.68 -0.11 M-E 6.83 1.67 12.16 72.40 -0.14 L-A 3.03 0.39 6.61 9.62 0.09 S&SEA 3.34 0.41 10.05 35.15 0.04 CPA 4.66 0.67 8.10 -a 0.06 AFR 3.37 0.55 14.32 19.07 0.08 a The welfare of Centrally Planned Asia collapses in the business as usual scenario. As a little sensitivity test, the regionally optimal abatement is calculated also for the climate sensitivities 1.5ūC and 4.5ūC for a doubling of atmospheric carbon dioxide, the IPCC's lower and upper bound, respectively. Tables 2.6 and 2.7 present the results. Table 2.6. Regionally Optimal Abatement Measures, Climate Sensitivity 1.5ūC region 1 2 3 4 5 6 7 8 9 taxa 3.0 2.3 1.9 0 2.3 0 0 0 0 energyb 0 0 0 0.5 0 0.3 0.3 0.3 0.3 carbonc 0 0 0 0.2 0 0.2 0.2 0.2 0.2 forestryd 5 0 5 5 0 25 0 5 5 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Table 2.7. Regionally Optimal Abatement Measures, Climate Sensitivity 4.5ūC region 1 2 3 4 5 6 7 8 9 taxa 1.7 1.2 0 0 0.2 0 0 0 0 energyb 0 0 0.3 0.5 0 0.3 0.3 0.3 0.3 carbonc 0 0 0.2 0.2 0 0.2 0.2 0.2 0.2 forestryd 15 5 5 5 0 20 20 5 105 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Paradoxically, there is an overall tendency to reduce more greenhouse gas emissions if the climate sensitivity is less, although the observed pattern is more diverse. The overall trend can be explained by pointing out that unilateral action is less worthwhile the swifter climate change is: the marginal benefits are rather low and, more important, rather flat. Thus, these results suggest that it is better to enjoy business as usual than to suffer sacrifices for a little mitigation. 2.4. Non-Cooperative Optimal Abatement It is not realistic to assume that the regions act as if they do not know that the other regions will also abate their greenhouse gas emissions, as we did in the previous section. Incorporating knowledge of the other regions' measures might lead a region to abate less, or more. The first possibility is the well known free rider behaviour, reaping the benefits of the others' actions without acting oneselves. The second possibility is also observed in FUND, version 1.0; this might be explained from the fact that unilateral action hardly pays (in terms of reduced damage) whereas multilateral action does. Interactions between the regions' abatement measures are modelled in the following manner. Instead of the Nash-Cournot equilibrium, which assumes full knowledge of the other regions' actions, we introduced a more realistic, though rather ad hoc interactive equilibrium in Tol (1993b). Assume that all regions plan abatement as in section 2.3, and all regions learn about the intended actions of the other regions. Because of this, all regions revise their plans, optimising their own actions assuming that the other regions act as under the old situation. Assume further that again all regions learn of the revised abatement plans, and that again they revise their own plans accordingly. This can be repeated ad infinitum, and might under certain circumstances lead to the Nash-Cournot equilibrium, but we suppose that the regions' decision makers grow tired of this game after the third step. As they do not exactly know what the other regions' actions will be, they assume that the other regions will act halfway between the once and the twice revised abatement plans, and plan and implement abatement accordingly. The equilibrium, thus defined, circumvents the assumption of full knowledge but incorporates some (stochastic) information on the other regions' actions. Note that the equilibrium is non-cooperative as all regions act according to their own benefits. Table 2.8 summarises the implemented actions in the equilibrium defined above. The Pacific part of the OECD and the Middle East, the most vulnerable among the richer regions, abate more than under the regionally optimal scenario, the other regions' actions are equal or less, particularly the European part of the OECD and Africa show free rider behaviour. It should be noted that the optimal African afforestation measures are highly sensitive to the other regions' actions. Table 2.8. Non-cooperative Optimal Abatement Measures region 1 2 3 4 5 6 7 8 9 taxa 2.8 1.8 1.7 0 0 0 0 0 0 energyb 0 0 0 0.5 0.6 0.3 0.3 0.3 0.3 carbonc 0 0 0 0.2 0.2 0.2 0.2 0.2 0.2 forestryd 5 0 0 5 0 0 5 0 10 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Table 2.9 compares the economic and social outcomes of the non-cooperative equilibrium to those of the business as usual scenario. The gains of emission reduction are in the same order of magnitude as in case of the regional optimum, but comparing Table 2.9 to Table 2.5 reveals that, in general, the poorer regions are better off under the regionally optimal abatement scenario than under the non-cooperative equilibrium, whereas the pattern for the richer regions is more diversified. This is due to the fact that the total emission reduction is less in the interactive case. Table 2.9. Comparison of the Business as Usual and Non- cooperative Optimal Abatement Scenarios region NPVY NPVW y2100 w2100 p2100 (%) (%) (%) (%) (%) OECD-A 11.32 0.92 25.56 27.83 -0.14 OECD-E 8.41 0.69 18.00 21.82 -0.30 OECD-P 7.18 0.65 15.97 25.81 -0.10 CEE&SU 4.53 0.43 8.25 10.40 -0.10 M-E 7.04 1.73 13.01 104.13 -0.13 L-A 2.89 0.37 6.57 10.94 0.08 S&SEA 3.14 0.38 10.07 47.54 0.04 CPA 4.34 0.62 9.60 -a 0.06 AFR 3.08 0.50 14.44 21.76 0.08 a The welfare of Centrally Planned Asia collapses in the business as usual scenario. As a little sensitivity test, the regionally optimal abatement is calculated also for the climate sensitivities 1.5ūC and 4.5ūC for 2xCO2. Tables 2.10 and 2.11 present the results. Table 2.10. Non-cooperative Optimal Abatement Measures, Clim. Sens. 1.5ūC region 1 2 3 4 5 6 7 8 9 taxa 3.2 2.1 1.7 0 1.3 0 0 0 0 energyb 0 0 0 0.5 0 0.3 0.3 0.3 0.3 carbonc 0 0 0 0.2 0 0.2 0.2 0.2 0.2 forestryd 30 0 5 0 0 15 15 10 50 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Table 2.11. Non-cooperative Optimal Abatement Measures, Clim. Sens. 4.5ūC region 1 2 3 4 5 6 7 8 9 taxa 1.9 1.6 0 0 0 0 0 0 0 energyb 0 0 0.4 0.5 0.1 0.3 0.3 0 0.3 carbonc 0 0 0.2 0.2 0.2 0.2 0.2 0.1 0.2 forestryd 10 0 0 15 0 5 5 0 105 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Two different patterns of free rider behaviour are observed for the IPCC's lower and upper climate sensitivity bounds. Under the lower bound, the richer regions are inclined to abate a little less and the poorer a little more whereas, under the upper bound, this pattern is reversed. This is directly related to the differences in vulnerability and to the changes in the marginal benefits of emission abatement, as outlined in the discussion on the sensitivity analysis of the regionally optimal abatement strategies. 2.5. Cooperative Optimal Abatement This section discusses another new feature of FUND, version 1.3: a cooperative emission reduction game. The two previous sections treated non-cooperative reduction with and without interactions between the regions. Here it is assumed that all regions cooperate to reach a common goal, but without affecting the own original position. Thus, what is to be optimised is the net present value of the world's welfare, defined as being the sum of the regional net present welfares. Note that, in this definition, all regions are treated alike, and therefore that the individual inhabitants of the different regions are treated differently. This is in line with the assumptions on, e.g., the discount rate and the value of a human life, but, due to the logarithmic transformation, there is a bias towards small regions. The restriction imposed upon the cooperative game is that no region's net present welfare is allowed to be lower than its business-as-usual net present welfare. Table 2.12 contains the optimal policy according to the cooperative game. The OECD regions are assumed to choose energy taxes, the other regions energy and carbon efficiency programmes. Table 2.12. Cooperative Optimal Abatement Measures region 1 2 3 4 5 6 7 8 9 taxa 2.8 2.1 1.8 0 0 0 0 0 0 energyb 0 0 0 0.5 0.6 0.3 0.1 0.4 0.3 carbonc 0 0 0 0.3 0.3 0.2 0.1 0.1 0.2 forestryd 15 5 5 0 0 0 5 10 5 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. Note that the tax is set to zero. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. The greenhouse gas emission reduction in the cooperative game is larger than in the non-cooperative one, as now the mostly positive external effects of the other regions' measures are taken into account. South and South East Asia reduce less, however. Table 2.13 compares the cooperative reduction equilibrium to the business as usual scenario. Central and Eastern Europe and the former Soviet Union and South and South East Asia have a smaller net present welfare under the cooperative game than under the non-cooperative one; the other regions benefit. In the long run, all regions benefit. Table 2.13. Comparison of the Business as Usual and Cooperative Optimal Abatement Scenarios region NPVY NPVW y2100 w2100 p2100 (%) (%) (%) (%) (%) OECD-A 11.38 0.93 26.21 30.61 -0.16 OECD-E 8.79 0.72 19.40 26.40 -0.34 OECD-P 7.31 0.68 16.58 34.39 -0.10 CEE&SU 4.32 0.39 8.89 12.52 -0.11 M-E 7.25 1.86 14.77 175.87 -0.14 L-A 2.95 0.38 7.40 14.79 0.10 S&SEA 2.71 0.34 10.60 75.50 0.05 CPA 4.28 0.62 13.59 -a 0.07 AFR 3.19 0.52 18.00 31.30 0.09 a The welfare of Centrally Planned Asia collapses in the business as usual scenario. Tables 2.14 and 2.15 display the results for the climate sensitivities of 1.5ūC and 4.5ūC, respectively. Table 2.14. Cooperative Optimal Abatement Measures, Climate Sensitivity1.5ūC region 1 2 3 4 5 6 7 8 9 taxa 3.3 2.5 2.2 0 0 0 0 0 0 energyb 0 0 0 0.5 0.7 0.3 0.3 0.4 0.3 carbonc 0 0 0 0.3 0.3 0.2 0.2 0.2 0.2 forestryd 25 5 15 15 0 5 15 15 5 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. Note that the tax is set to zero. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Again, the optimal abatement is higher under the low climate sensitivity scenario than under the best guess, and the best guess reduction is higher than the under the upper bound climate sensitivity assumption. Table 2.15. Cooperative Optimal Abatement Measures, Climate Sensitivity 4.5ūC region 1 2 3 4 5 6 7 8 9 taxa 1.0 2.0 1.0 0 0 0 0 0 0 energyb 0 0 0 0.7 0.3 0.2 0 0.4 0.4 carbonc 0 0 0 0.3 0.3 0.2 0.3 0 0 forestryd 0 0 0 0 0 0 0 15 0 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. Note that the tax is set to zero. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. 3. INTERREGIONAL CAPITAL TRANSFERS This chapter describes the way in which interregional capital transfers are modelled in FUND, version 1.3 (section 3.1), and discusses the impact of allowing capital transfers on the efficiency and efficacy of greenhouse gas emission reduction in both the non-cooperative (section 3.2) and cooperative (section 3.3) games defined above. 3.1. Modelling Capital Transfers The actual `climate fund', i.e., the clearinghouse through which regions can finance abatement measures in other regions, is modelled in the following way. First, all regions are assumed to have implemented the measures according to the interactive games described in chapter 2. This is in the spirit of the Framework Convention on Climate Change which states that measures implemented in a foreign country are not to replace national emission reductions. Each region evaluates additional abatement measures.[7] If a measure shows to be beneficial, i.e., increasing the region's own net present welfare, it is implemented. Otherwise, a price X is attached to it. This price follows from setting the loss in net present welfare W equal to the discounted sum of X times the (present) economic growth, i.e.,leading toActually, X should follow from equating the welfare loss due to the abatement measure and the model-based welfare gain due to the capital received; this would require too much computational effort, though. We will have to do with this short-cut.Next, all measures plus their prices of all regions are offered to the clearinghouse and evaluated by all regions. The evaluation is based on the net present welfare of the investing region which transfers an amount of capital equal to price X to the receiving region which implements the emission reduction. The (hypothetical) auctioneer grants the transfer right to the region with the highest welfare gain. If this gain is positive, the transfer is implemented, otherwise it is not. After evaluating all measures offered in the manner described above, the auctioneer checks whether any capital transfers took place and, if so, calls for another round of offers and bids. 3.2. Non-Cooperative Optimal Abatement with Side Payments If the regions take their measures according to the non- cooperative optimal abatement game to the clearinghouse, two interesting observations can be made. First of all, the fact that all emission abatement programmes have to be reviewed for every additional measure a region offers to the clearinghouse implies that the regions of the OECD, the Middle East and Africa `discover' that they should implement more abatement themselves. The clearinghouse alters the nature of the game from a mixed prisoner's dilemma and assurance game into a genuine assurance game. An assurance game (Sen, 1966, 1967) actors are acting positively as long as the other actors do. The second observation is that the capital transfers, or side payments, themselves are rather limited. The Middle East buys one tenth of a tax-unit of additional taxes in the Pacific part of the OECD, OECD- Pacific buys the same in OECD-America, and OECD-America in turn buys 0.1% additional carbon efficiency improvement in Central and Eastern Europe and the former Soviet Union. Nevertheless, the emission reduction substantially increases as Table 3.1 reveals. Table 3.1. Non-cooperative Optimal Abatement Measures with Side Payments region 1 2 3 4 5 6 7 8 9 taxa 3.1 2.5 2.0 0 0 0 0 0 0 energyb 0 0 0 0.5 0.7 0.3 0.3 0.3 0.3 carbonc 0 0 0 0.3 0.3 0.2 0.2 0.2 0.2 forestryd 5 5 5 5 0 0 5 0 105 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Table 3.2 compares the non-cooperative optimal abatement with side payments scenario to the business as usual scenario. All regions benefit from the clearinghouse compared to the business as usual as well as the regionally optimal abatement and the non-cooperative optimal abatement scenarios. Table 3.2. Comparison of the Business as Usual and Non- cooperative Optimal Abatement with Side Payments Scenarios region NPVY NPVW y2100 w2100 p2100 (%) (%) (%) (%) (%) OECD-A 11.62 0.94 27.75 33.00 -0.18 OECD-E 9.13 0.74 20.99 29.19 -0.39 OECD-P 7.52 0.70 17.47 38.18 -0.12 CEE&SU 4.29 0.39 8.82 12.97 -0.13 M-E 7.50 1.92 16.09 204.26 -0.16 L-A 3.09 0.40 8.13 16.64 0.11 S&SEA 3.34 0.42 13.58 89.08 0.06 CPA 4.89 0.72 15.55 -a 0.08 AFR 3.46 0.57 20.97 36.79 0.10 a The welfare of Centrally Planned Asia collapses in the business as usual scenario. As a little sensitivity test, the non-cooperative optimal abatement with side payments is also calculated for the climate sensitivities of 1.5ūC and 4.5ūC. Table 3.3 presents the results of the lower bound. Table 3.3. Non-cooperative Optimal Abatement Measures with Side Payments, Climate Sensitivity 1.5ūC region 1 2 3 4 5 6 7 8 9 taxa 3.2 2.3 2.0 0 1.6 0 0 0 0 energyb 0 0 0 0.5 0 0.4 0.3 0.3 0.3 carbonc 0 0 0 0.2 0 0.2 0.2 0.2 0.2 forestryd 35 5 0 0 0 20 15 10 50 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Compared to Table 2.10, little additional effort is observed. The actual capital transfers are limited, South and South East Asia buys some additional taxes in the Pacific part of the OECD which in turn sponsors some additional energy efficiency improvement in Latin America. The case of the upper bound climate sensitivity is more interesting. The non-cooperative game without side payments resulted in a limited abatement, but the clearinghouse sets in motion a substantive programme of emission reductions, not only through regional abatement but also through large interregional capital transfers, with the OECD, the Middle East (a little) and Africa (!) as investors, and Central and Eastern Europe and the former Soviet Union as the main receiver. It appears, however, that the richer regions benefit most whereas some of the poorer regions, notably Centrally Planned Asia, are worse off with the interregional capital transfers[8]. Table 3.4. Non-cooperative Optimal Abatement Measures with Side Payments, Climate Sensitivity 4.5ūC region 1 2 3 4 5 6 7 8 9 taxa 3.9 2.8 0 0 1.6 0 0 0 0 energyb 0 0 1.8 2.1 0 0.5 0.4 1.1 0.8 carbonc 0 0 2.3 2.4 0 0.7 2.0 2.4 0.8 forestryd 75 30 95 75 30 210 320 100 135 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Arguably, rational behaviour is closer to assuming the climate sensitivity to be the upper bound of 4.5ūC than to assuming it to be the best guess of 2.5ūC[9]. Therefore, we may tentatively conclude that, in a non-cooperative game, interregional capital transfers do contribute substantially to controlling the climate change problem. 3.3. Cooperative Optimal Abatement with Side Payments The cooperative game with side payments is defined to be played after the game without side payments. The clearinghouse is set-up in a slightly different manner as in the non-cooperative situation. Emission reduction measures are first evaluated without interregional capital transfers. If not found worthwhile, i.e., violating Pareto's condition or not improving the global welfare, all losers are compensated by all winners, proportional to their gain, and the measure is re-evaluated. The bilateral capital transfers of the non-cooperative game are thus transformed to multilateral capital transfers. Table 3.5 contains the outcomes for the base case. The greenhouse gas emission reductions increase substantially compared to the case without side payments, and are also somewhat higher than in the non-cooperative game with side payments. However, the four poorest regions are net investors whereas the OECD, Central and Eastern Europe and the former Soviet Union and, especially, the Middle East are net receivers of capital (note that the definition of welfare is in favour of small regions). Table 3.5. Cooperative Optimal Abatement Measures with Side Payments region 1 2 3 4 5 6 7 8 9 taxa 4.1 2.3 2.5 0 0 0 0 0 0 energyb 0 0 0 0.6 0.7 0.4 0.4 0.7 0.3 carbonc 0 0 0 0.3 0.4 0.2 0.1 0.2 0.2 forestryd 20 5 5 15 0 15 15 10 5 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Table 3.6 compares the outcomes of the cooperative game with side payments to the business as usual scenario. Obviously, all regions benefit. However, compared to Table 2.13, containing the outcomes for the cooperative game without capital transfers, most regions suffer from the capital transfers; the gain in the global welfare is almost completely on account of the Middle East. Comparing the outcomes to Table 2.3, containing the outcomes for the non- cooperative game with side payments, the same conclusions result. An explanation of this is that the global welfare, defined as the sum of the net present value of the gross regional welfare, does not take distributional effects into account; the Pareto constraints concern the business as usual welfares. Therefore, a scheme with Pareto constraints based on the game without side payments results in exactly the same outcomes as the game without capital transfers. Table 3.6. Comparison of the Business as Usual and Cooperative Optimal Abatement with Side Payments Scenarios region NPVY NPVW y2100 w2100 p2100 (%) (%) (%) (%) (%) OECD-A 11.73 0.90 30.67 35.24 -0.22 OECD-E 9.03 0.74 20.22 20.22 -0.46 OECD-P 7.85 0.71 19.66 19.66 -0.15 CEE&SU 4.36 0.38 9.85 9.85 -0.15 M-E 17.75 4.77 17.74 213.90 -0.19 L-A 2.39 0.30 8.41 15.17 0.13 S&SEA 2.96 0.36 13.89 78.89 0.07 CPA 3.31 0.45 16.21 -a 0.04 AFR 2.47 0.42 21.66 33.58 0.12 a The welfare of Centrally Planned Asia collapses in the business as usual scenario. Tables 3.6 and 3.7 display the outcomes of the cooperative game with side payments for climate sensitivities of 1.5ūC and 4.5ūC, respectively. Allowing interregional capital transfers greatly enhances the optimal emission reduction, especially in case of the lower bound, but at the expense of the poorer regions. Table 3.6. Cooperative Optimal Abatement Measures with Side Payments, Climate Sensitivity 1.5ūC region 1 2 3 4 5 6 7 8 9 taxa 3.9 3.4 3.5 0 0 0 0 0 0 energyb 0 0 0 0.6 0.9 0.3 0.4 0.4 0.3 carbonc 0 0 0 0.3 0.3 0.2 0.1 0.3 0.2 forestryd 170 185 105 115 0 235 100 10 5 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. Table 3.7. Cooperative Optimal Abatement Measures with Side Payments, Climate Sensitivity 4.5ūC region 1 2 3 4 5 6 7 8 9 taxa 2.5 2.8 2.0 0 0 0 0 0 0 energyb 0 0 0 0.7 0.7 0.2 0.3 0.6 0.4 carbonc 0 0 0 0.3 0.3 0.2 0.3 0.3 0.1 forestryd 30 15 25 15 5 15 0 15 15 a Expressed in tax-units; one unit of tax reduces the economic growth by 0.1% of GRP in the year it is first levied; one unit of tax stabilises the carbon emissions (reduces emission growth to 1%) in the year it is imposed in the OECD (non-OECD) regions. b Expressed in percents of energy efficiency improvement. c Expressed in percents of carbon efficiency improvement. d Expressed in million tonnes of absorbed carbon. 4. SUMMARY AND CONCLUSION This final chapter summarises the findings of this report (section 4.1), discusses the findings from a broader perspective, and draws some conclusions (section 4.2). 4.1. Summary This report describes the integrated climate economy model Climate Framework for Uncertainty, Negotiation and Distribution (FUND), version 1.3. FUND is a nine-region model of the world economy coupled to modules to calculate greenhouse gas emissions, climate change, climate change damages, and greenhouse gas emission reduction costs. The model evaluates several emission reduction scenarios, and calculates the optimal reduction in a number of game- theoretic settings. The main aim of the model is to investigate the influence of interregional capital transfers on the efficiency and efficacy of emission reduction. The differences between FUND, version 1.3, and the earlier versions are minor. The model is essentially scenario driven, for instance, economic growth, demographic development and technological innovation are exogenously determined. All endogenous interactions are highly parameterised, influencing the exogenous parameters. The climate module has been improved by bringing the global mean temperature's reaction to atmospheric changes closer to climatological theory, and through the explicit introduction of tropical cyclones. The damage costs of climate underwent a number of minor improvements except for the agricultural damages for which substantial new information became available. The costs of emission reduction remained the same. Regarding evaluating policies, the utility function employed takes explicit account of intangible (non-marketable) losses while the value attached to this losses grows with the income per capita. Partly due to this change, the optimal emission reduction are more profound than in the earlier versions of the model though still not substantial. The most aggressive reduction programme results in an atmospheric concentration of greenhouse gases of over 550 ppm carbon equivalents in the year 2100, most reduction programmes, however, result in 2100 concentrations of over 750 ppm. Comparing non-cooperative reduction scenarios in which the regions do or do not know of each others actions reveal a mix of free rider and assurance behaviour. In a prisoner's dilemma type game, actors are tempted to lower their own emission reduction in order to benefit from the other actors' actions at no cost. However, unilateral action hardly pays if it concerns global warming. Therefore, actors are also tempted to enhance their own emission reduction in order to benefit from the synergy with the other actors' actions. This is known as an assurance game. A new feature of version 1.3 of FUND is the cooperative game in which the Pareto equilibrium is calculated. The optimal reduction is higher in this case than in the non-cooperative game. An analysis of the sensitivity of the results to changes in the temperature's reaction to the enhanced greenhouse effect yield the paradoxical result that the higher the climate sensitivity, the lower the optimal emission reduction. However, in case the climate is highly sensitive, emission reduction has little benefits as the climate changes rapidly anyway. Aggressive action is needed in order to counteract this but, under the assumptions of the model, the costs of this action does not weight against its benefits. Aprs nous, le dŽluge. Interregional capital transfers are, in the non-cooperative game, modelled as handled by an auction. Each region sells and buys specific emission reductions at appropriate prices, according to their own evaluation how this trade might affect their welfare. The main conclusion is that the clearinghouse, though little active itself, transforms the mixed prisoners' dilemma and assurance game into a genuine assurance game: reduction is substantially enhanced despite limited interregional capital transfers. This is even more pronounced in case of the climate is highly sensitive to the enhanced greenhouse effect. The reduction achieved in this setting is substantive. In the cooperative game, interregional capital transfer are evaluated according to Pareto's criterion and proportionally spread over the winner and the losers of the reduction considered. In this case also, capital transfers enhance the optimal emission reduction substantially. 4.2 Conclusion This fourth report in the Climate Fund project has carried the FUND model one step further in the process of computational and, more important, conceptual debugging. In its present stage, FUND points at some interesting conclusions. Earlier versions pointed at the complicated interactions between the costs and benefits of emission reductions and between the reduction programmes of several regions. Here this is confirmed. The most outstanding notion in this regard is that the benefits of one region's emission reduction are enhanced by the other regions' reductions, and that therefore more action is called for if others take action. In other words, the danger of free riding is counteracted by the assurance nature of the game. As a consequence, interregional capital transfer, or simply the possibility thereof, strengthens this tendency and so stimulates greenhouse gas emission reduction. The next report will contain a further revision of FUND and perform extensive sensitivity and uncertainty analyses. This report will be followed by the final report, summarising all the findings of the project Socio-Economic Aspects of the Greenhouse Effect: Climate Fund of the Dutch National Research Programme: Global Air Pollution and Climate Change at a level understandable to a broad public. The final report will also contain some remarks on the political feasibility of interregional capital transfers. REFERENCES Bulatao, R.A., E. Bos, P.W. Stephens and M.T. Vu. (1990), World Population Projection Ń Short- and Long-Term Estimates, John Hopkins University Press, Baltimore and London. Cline, W.R. (1992), The Economics of Global Warming, Institute for International Economics, Washington, D.C. Erickson, J.D. (1993), `From Ecology to Economics: The Case against CO2 Fertilization', Ecological Economics, 8, pp. 157- 175. Fankhauser, S. (1992), Global Warming Damage Costs: Some Monetary Estimates, CSERGE Working Paper GEC 92-29, London and Norwich. Fankhauser, S. (1993), `The Economic Costs of Global Warming: Some Monetary Estimates' in Kaya et al. (1993). Fankhauser, S. (1994a), Global Warming Damage Costs: Some Monetary Estimates (revised version of CSERGE Working Paper GEC 92-29), CSERGE, London and Norwich. Fankhauser, S. (1994b), The Economic Costs of Global Warming Damage: A Survey, CSERGE, London and Norwich. Fankhauser, S. (1994c), `The Social Costs of Greenhouse Gas Emissions: An Expected Value Approach', The Energy Journal (forthcoming). Fischer, G., K. Frohberg, M.L. Parry and C. Rosenzweig (1993), `Climate Change and World Food Supply, Demand and Trade' in Kaya et al. (1993). Houghton, J.T., G.J. Jenkins and J.J. Ephraums (eds.) (1990), Climate Change Ń The IPCC Scientific Assessment, Cambridge University Press, Cambridge. Houghton, J.T., B.A. Callander and S.K. Varney (eds.) (1992), Climate Change 1992 ŃThe IPCC Supplementary Report on the IPCC Scientific Assessment, Cambridge University Press, Cambridge. Jansen, H.M.A. (1993), `Are We Underestimating, When Valuing the Benefits of Greenhouse Gas Reduction?' in Kaya et al. (1993). Jansen, H.M.A., Klein, R.J.T., Tol, R.S.J. and Verbruggen, H. (1993), Some Considerations on the Economic Importance of Pro-active Integrated Coastal Zone Management, paper prepared for the World Coast Conference 1993 (International Conference on Coastal Zone Management), 1-5 November, Noordwijk Kaya, Y., Naki_enovi_, W.D. Nordhaus and F.L. Toth (eds.) (1993), Costs, Impacts and Benefits of CO2 Mitigation, IIASA, Laxenburg. Manne, A.S. and R.G. Richels (1992), Buying Greenhouse Insurance Ń The Economic Costs of CO2 Emission Limits, MIT Press, Cambridge. Nordhaus, W.D. (1991), `To Slow or not to Slow: The Economics of the Greenhouse Effect', The Economic Journal, 101, pp. 920-937. Nordhaus, W.D. (1993), Managing the Global Commons: The Economics of Climate Change, The MIT Press (forthcoming). Olsthoorn, A.A. and R.S.J. Tol (1993), Socio-economic Aspects of Changes in the Incidence and Intensity of Extreme (Weather) Events Ń Report of the Amsterdam Workshop, 24-25 June 1993, Institute for Environmental Studies E93/14, Vrije Universiteit, Amsterdam. Pearce, D.W. (1980), `The Social Incidence of Environmental Costs and Benefits' in T. O'Riordon and R.K. Turner (eds.), Progress in Resource Management and Environmental Planning, Vol. 2, Wiley, Chichester. Reilly, J. (1994), `Crops and Climate Change', Nature, 367, 118-119. Rosenzweig, C., M.L. Parry, G. Fischer and K. Frohberg (1993), Climate Change and World Food Supply, Environmental Change Unit Research Report 3, University of Oxford, Oxford. Rosenzweig, C. and M.L. Parry (1994), `Potential Impact of Climate Change on World Food Supply', Nature, 367, 133-138. Ryan, B.F., I.G. Watterson and J.L. Evans (1992), `Tropical Cyclones Frequencies Inferred from Gray's Yearly Genesis Parameter: Validation of GCM Tropical Climates', Geophysical Research Letters, 19 (18), pp. 1831-1834. Sen, A. (1966), `Labour Allocation in a Cooperative Enterprise', Review of Economic Studies, 33, pp. 361-371. Sen, A. (1967), `Isolation, Assurance and the Social Rate of Discount', Quarterly Journal of Economics, 81, pp. 112-124. Shine, K.P., R.G. Derwent, D.J. Wuebbles and J. J. Morcrette (1990), `Radiative Forcing of Climate' in Houghton et al. (1990). Tol, R.S.J. (1993a), The Climate Fund Ń Survey of the Literature on Costs and Benefits, Institute for Environmental Studies W93/01, Vrije Universiteit, Amsterdam. Tol, R.S.J. (1993b), The Climate Fund Ń Modelling Costs and Benefits, Institute for Environmental Studies W93/17, Vrije Universiteit, Amsterdam. Tol, R.S.J. (1993c), The Damage Costs of Climate Change Ń Towards More Comprehensive Calculations, Institute for Environmental Studies W93/18, Vrije Universiteit, Amsterdam. Tol, R.S.J. (1994a), The Climate Fund Ń Interregional Capital Transfers, Institute for Environmental Studies W94/02, Vrije Universiteit, Amsterdam. Tol, R.S.J. (1994b), `The Damage Costs of Climate Change Ń A Note on Tangibles and Intangibles, Applied to DICE', Energy Policy (forthcoming). Vellinga, P. and Tol, R.S.J. (1993), `Climate Change, Extreme Events and Society's Response', Journal of Reinsurance, 1 (2), pp. 59-72. WRI - World Resources Institute (1991), World Resources 1990- 91 Ń A Guide to the Global Environment, Oxford University Press, Oxford. Xia, G. and Wei, Z. (1993), Climate Change and its Technical and Institutional Adaptations: China's Perspective, paper presented at the International Workshop on Integrative Assessment of Mitigation, Impacts and Adaptation to Climate Change, 13-15 October, IIASA, Laxenburg, Austria. Footnotes [0] Institute for Environmental Studies, Vrije Universiteit, De Boelelaan 1115, NL-1081 HV Amsterdam, The Netherlands. The author hereby expresses his thanks to Huib Jansen for his helpful comments and to the Dutch National Research Programme: Global Air Pollution and Climate Change for partly funding this research. [1] That is, the American part of the OECD (OECD-A), the European part of the OECD, including Turkey and Israel (OECD- E), the Pacific part of the OECD (OECD-P), Central and Eastern Europe and the former Soviet Union (CEE&SU), the Middle East, excluding Egypt (M-E), Latin America and the Caribbean (L-A), South and South East Asia (S&SEA), Centrally Planned Asia (CPA) and Africa (Afr). [2] Note that the intangible losses due to protection (such as naturally and historically valuable features lost due to building dykes) are neglected in the studies surveyed in Tol (1993a). [3] It would be more realistic, though less practical, to model biodiversity as a stock variable of which subsequent losses would be valued increasingly higher. [4] Some prelimenary results with the time horizon extended to 2170 indicate that the losses rise so sharply that some economies, e.g., those of OECD-America and Centrally Planned Asia, collapse. Obviously, the model structure is not fit to cope with this kind of effects. [5] And, of course, with the number of people. So, the absolute value of the intangibles are assumed to grow with income squared, as the losses as a propertion of income grow with income per capita times population. [6] Note that this captured by the convex transformation from income to utility. [7] The regions which have implemented a carbon tax evaluate a further tax raise; the regions which have implemented an energy and carbon efficiency programme evaluate additional energy and carbon savings; all regions evaluate additional afforestation. [8] Note that each region optimises its own net present welfare. Other regions' measures might reduce this, and the climate fund has no manner (nor indeed the intention) to compensate for this. [9] At least, in this context. Remind that we neglect all uncertainties in FUND, version 1.3. Given the asymmetrical (right-skewed!) uncertainties and the concave loss and value functions, the damages as perceived in a decision analysis under uncertainty are much higher than those under certainty (cf. also Tol, 1993c).