Equitable Provision of Long-Term Public Goods The Role of Negotiation Mandates Franck Lecocq, Jean-Charles Hourcade World Bank, Development Research Group Centre International de Recherche sur l'Environnement et le Développement (CNRS, EHESS, ENGREF, ENPC) Abstract In a one-period model, whether or not individual weights in the welfare function are based on initial endowments dictate who provides public goods. But with long-term public goods, banning wealth redistribution still allows for several equilibriums depending on Parties' willingness to acknowledge changes in negotiating powers over time, and on whether or not they care only for their own descendants. "Adaptative" and "universal" mandates lead to far more robust equilibrium. In all cases, a simple rule of thumb for allocating expenditures at first period emerges, independent of both the optimal level of public goods and the second-period distribution of expenditures. JEL Classification: D63, H41, Q25 Keywords: Public Goods, Equity, Negotiation, Climate Change World Bank Policy Research Working Paper 3180, December 2003 The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Policy Research Working Papers are available online at http://econ.worldbank.org. 1. Introduction To provide transnational, long-term and uncertain public goods such as biodiversity, a solid ozone layer, or a preserved global climate the international community must confront inter and intra generations distributional issues simultaneously. The economic literature addresses this issue by, inter alia, extending the Bowen-Lindhal- Samuelson (BLS) conditions to the intergenerational case (Sandler and Smith, 1976). However, negotiations on transnational and long-term public goods are often uniquely guided by `ethical intuitions' such as, in the climate change case, the "common but differentiated responsibilities" principle, per capita distribution of emission rights (Agarwal and Narain, 1991), or the grandfathering scheme. On both sides of the Atlantic, players in these negotiations (Bodansky, 2001, Hourcade, 2000) have showed how reluctance to put some economic insights in the discussion have made it difficult to control the political vagaries of the process, let alone find a compromise. Economists may be partly responsible for their own lack of influence, because of their reflex of keeping ethics separated from economics. This paper builds on the opposite advice, i.e. that "there is something in the methods standardly used in economics, related inter alia with its engineering aspect, that can be of use to modern ethics as well" (A. Sen, 1987, p.9). To do so, using climate as an empirical case, it interprets the benevolent planner metaphor as capturing the behavior of the chairman of a Conference of the Parties1 presenting a take or leave proposal in the final hours of the negotiation (Grubb et al., 1999). Within a two period framework, we define four mandates that can be given to the planner. These mandates combine assumptions about: - diplomatic attitudes: we distinguish a status-quo approach, whereby current balances of power are used to shape long-term policy, and an adaptative approach whereby evolutions in the 1The COP is the negotiating body of the U.N. Framework Convention on Climate Change (UNFCCC). 1 distribution of economic income and power are acknowledged and accounted for; - visions of intergenerational solidarity in the face of climate risks: we distinguish between dynastic solidarity, whereby Parties are concerned by the welfare of their future citizens only, and universal solidarity, whereby Parties consider the welfare of all future individuals, regardless of where they live. These mandates are analyzed under a no redistribution constraint, because countries are not likely to let climate policies--or any other international treaty of that sort--be the occasion of large-scale wealth redistribution across nations. We first focus on the burden sharing principles which emerge from these mandates, and we question their political viability. We then examine their implications for the level of provision of the public good. At each step, we show the specific role of uncertainty. We conclude with some policy implications, and a discussion of the ethical pre-requisites for regimes aiming at managing long-term public goods in an unequal word. 2. A Generic Model with Four Alternative Programs Let us start from a generic model similar to the one developed by Sandler and Smith (1976). The world is divided in N countries, and there are two periods, present and future, the latter indexed by superscript f. At first period, the representative individual2 of the li inhabitants of country i allocates his income yi between ci the consumption of a composite private good chosen as numeraire, and ai his abatement expenses. yi = ci + ai (1) Let x (resp. xf) be the amount of greenhouse gases (GHG) emissions abated worldwide compared to business-as-usual. We use x+xf as an index of GHG concentration in the atmosphere,3 and denote di(x+xf) 2We will not address the internal distribution of revenue in each country. 3This (inversed) index is a simplification of the dynamics of GHG accumulation in the atmosphere, but it suffices in capturing the stock externality character of climate change. 2 the per capita level of damages incurred in country i at second period at any given level of GHG concentration. Since x+xf aggregates avoided tons of GHG emissions, functions di(.) are decreasing. Thus, second period budget equations are as follows. yfi - di(x+xf) = cfi + afi (2) We assume that abatement expenses are used efficiently and denote C(x) (resp. Cf(xf)) the worldwide abatement cost function. The total level of abatement at each period is thus given by:4 liai= C(x) lfi afi = Cf(xf) (3) i i At the beginning of the first period, the planner/chairman of the COP is charged with proposing an abatement level for each country at both periods. This one-shot model is arguably at odds with the sequential nature of the real climate regime, where targets are set for five-year periods only. But, climate change being a stock externality, the planner cannot but make assumptions about future actions when computing present ones. Second period abatements can thus be interpreted as plans which may, or may not, be carried out. To come up with a proposal with reasonable chances of being accepted, the planner maximizes a weighted sum of the representative individuals' utilities, and selects weights in function of the mandate he receives from the Parties. If we assume, despite its controversial character from an ethical point of view,5 that wealthiest Parties impose a no redistribution constraint, according to which climate policies shall 4Let xi be national abatement levels, and Ci(xi) the national abatement cost functions. Then C(x) = Min { Ci(xi) xi = x}. This can be interpreted as a carbon fund, i i provisioned by all countries, which reduces emissions worldwide where it is cheapest to do so. 5See e.g. Azar (1999, p.254): "The global welfare function is a normative, not an empirical question, and few would contest that the world would actually be a much better place if the huge differences income were reduced. A situation where the richest billion people live in abundance, and the poorest billion suffer from chronicle hunger, can by no reasonable standards be considered a global welfare maximum." 3 not be the occasion of large-scale wealth transfer from developed to developing countries, then this collective welfare function must meet the following two conditions. - national contributions ai and afi must be non negative, as no Party will accept to abate more in order to endow another Party with emissions rights higher than its baseline prior to any carbon trading.6 This condition is seemingly trivial, but we will show that it plays a role in the second period equilibrium. - Second, the weights attached to representative individuals' utility functions must be such that the initial distribution of wealth (yi) is welfare maximizing.7 Negishi (1960) tells us that these weights are unique--up to a scale factor--and equal to the inverse of the marginal utility of initial income. If utility functions are logarithmic and if first and second period consumption are separable, these weights are proportional to per capita income.8 However, the set of welfare functions which meet these restrictions is still rather large because there are various ways of interpreting the no redistribution imperative at second period, and various attitudes vis-ŕ-vis climate damages. With regard to the no redistribution constraint, modelers (e.g., Nordhaus and Yang, 1996) often consider that it applies separately at each period. The Negishi weights are thus made time varying so that the projected distribution of income (yfi) is also welfare maximizing at second period. But, by doing so, one makes a strong assumption about the political economy of the negotiation, namely that Parties agree to ask the planner to anticipate changes in income distribution. In other 6The excess quota allocated to Russia and Ukraine by the Kyoto Protocol is obviously a pure tactical concession. A milder approach to this first constraint is that no Party shall benefit from climate policy as a whole; thus the sum of contribution and damages shall be non negative (afi+ dfi >0). 7To avoid any misunderstanding, let us make clear that this technical trick capturing political constraints does not imply a substantive value judgment on the equity of the current state of the World. 8Were these weights all set to 1, total wealth should be redistributed so as to achieve equal per capita income. 4 words, this presupposes a consensus on the legitimacy--or the ineluctability--of changes in economic balances, which contradicts diplomatic traditions where negotiating positions are governed by prevailing balances of power. It is not implausible that Machiavelli's qualification of States as "cold monsters" will remain valid in the 21st century. The richest countries may well not accept the ineluctable decline of their share in world's wealth, or may at least tend to use their current superiority to slow down this decline. They may then be tempted by a status-quo mandate, in which they force the planner to calibrate the collective welfare function at both periods based on current income distribution. Regarding the interplay between the assessment of climate damages and intergenerational equity, two polar attitudes are again possible. The first derives from the observation that negotiating teams, defending national interests, and speaking on behalf of both their present and unborn fellow citizens, tend to follow a dynastic solidarity conduct and primarily consider the damages falling on their own country. A polar option, supported by many NGOs, is that decision- makers should adopt a universal solidarity ethics, and should be concerned by the welfare of all future individuals, regardless of where they live, and regardless of where damages fall.9 These alternatives can be translated analytically by making second period utilities dependent, or not, on damages in other countries. 9We will discuss later the ethical rationale and political likelihood of this mandate. For the time being, we treat it as a pure logical possibility. 5 Four possible programs can be derived by combining these two sets of hypothesis. If we denote Ui (resp. Ufi) the representative individuals' utility functions, i and i the first and second period weights attached to these functions, and the utility discount factor,10 they are: - "Dynastic solidarity" and "status-quo" mandate: W = liiUi(ci) + lfi i Ufi(cfi) (4) i i i = li -1 U'i(yi) with = U'i(yi) (5) i - "Dynastic solidarity" and "anticipative" mandate: W = liiUi(ci) + lfi i Ufi (cfi) (6) i i i = li -1 U'i(yi) with = U'i(yi) (7) i i = lfi -1 Ufi'(yfi) with = Ufi'(yfi) (8) i - In "Universal solidarity" mandates, damages falling on other countries should enter into the computation of the utility of the representative individual of country i, in addition to those falling directly on the country i. "Universal solidarity" and "status-quo" or "dynastic" mandates are thus obtained by substituting Ufi (cfi, d1,...,di-1,di+1,...,dN) to Ufi(cfi) in equations (4) f f f f and (6) respectively. 10We make the following technical assumptions. First, present and future consumptions are assumed separable. Second, individual utility functions are all twice differentiable, with U'>0 and U"<0. Third, the sum of weights over all individuals in all country is equal to one, i.e. that li i = 1, and li i = 1. Fourth, all Parties have the same pure time i i preference. This still allows for differentiated discount rates across countries, as utility functions and growth rates might differ. 6 3. Burden Sharing at First Period: Towards an Easy Rule of Thumb? In all four mandates, solving the planner's program yields the same result at first period: abatement expenses should be allocated so as to equate after abatement weighted marginal utilities of consumption across countries (see Appendix 1 for full derivation of this result, which expresses the BLS condition in the context of our model). 1 U1'(y1-a1) = ... = N UN' (yN-aN) (9) Since by virtue of the no redistribution constraint, before abatement weighted marginal utilities are also equal, the optimal distribution of abatement costs decreases weighted marginal utilities by the same amount. 1 U1'(y1) - 1 U1'(y1-a1) = ... = N UN'(yN) - N UN'(yN-aN) (10) Figure 1 provides a geometric illustration of this result, picturing two regions differing only in income. Since preferences are the same, the poor region has a higher marginal utility of consumption (B) than the rich one (A). To comply with the no redistribution constraint, the planner chooses poor (normalizing rich to 1) so that the weighted marginal utilities of consumption in both regions are equal. The weighted marginal utility of the poor region is thus C instead of B. To preserve this equality in the post abatement equilibrium, it suffices to find the horizontal line intersecting with both the marginal utility function of the rich (continuous line) and the weighted marginal utility function of the poor (dotted line), such that apoor+arich is equal to the total desired level of abatement. How do contributions arich and apoor compare? Geometrically, apoor is lower than arich if the slope of the weighted marginal utility function is steeper at point C than the slope of the marginal utility function is at point A. An analytic condition can be derived when contributions are all assumed to remain small compared with initial revenues. In that case, equation (10) can be approximated by: 7 U" U" -U'(ypoor) apoor - U'(yrich)arich (11) And apoor is lower than arich if and only if U" U" -U'(ypoor) > - U'(yrich) (12) The latter condition holds (see Appendix 2) for any ypoor < yrich in a large class of utility functions, including inter alia logarithmic U = ln(c) and exponential U = ca (0 0 ai = 0 i U'i(yi-ai) - i = with i > 0 if ai = 0 (a9) Since weighted marginal utilities of consumption before abatement are equal (a6), there is a solution to (a9) where all ai are strictly positive, and corresponding Lagrange multipliers i all equal to zero. Since second derivatives of all individual utility functions are negative, this solution is in fact the global maximum. At optimum, derivation of L with regard to afi yields: L Ufi i = 0 if afi+di > 0 afi = 0 i c (yfi-afi-di(x+xf),...) - i = µ with i > 0 if afi+di = 0 (a10) In adaptative mandates, weights i are such that the vector yfi is welfare maximizing. Provided residual damages are not too high in any country, there exists again a solution where all abatement expenditures afi are positive, with Lagrange multipliers i equal to zero. On the other hand, if some residual damages are too high, then constraint (a5) becomes binding in these countries, and the corresponding abatement level afi is zero. In status-quo mandates, weights i are not likely to be such that the vector yfi is welfare maximizing. In that case, the optimal plan is to allocate abatement expenditures to the country which has the lowest weighted marginal utility of consumption before abatement, until optimal provision of public goods is reached, or until the second lowest weighted marginal utility level is reached, in which case both countries contribute, and so on. 23 Abatement Levels At optimum, derivation of L with regard to xf yields: L ' Ufi Ufi-1' xf = 0 Cf'(xf) = - lfi i di(x+xf) -lfi c i dj dj(x+xf) i i ji (a11) With i = i Ufi µ c (yfi-afi-di(x+xf),...) (a12) Weights i are ratios between the weighted marginal utility of consumption at optimum, and the shadow price of carbon (µ). In adaptative mandates when none of the residual damages are too large, all weighted marginal utility of consumption are equal to the shadow price of carbon µ. (a12) can be simplified in: Cf'(xf) = -lfi di(x+xf) - ' lfi c Ufi Ufi-1 dj dj(x+xf) ' (a13) i i ji This is standard optimal provision of public goods: public goods should be provided up to the point where the last unit costs as much to produce as the marginal benefits it creates. The second term of the sum captures the fact that, in "universal" mandates, these benefits include avoiding damages abroad on top of at home. If residual damages in some countries are too high, then (a10) states that the weighted marginal utility of consumption in these countries is higher than the shadow price of carbon µ. For these countries, weights i are thus higher than unity. The same occurs in status-quo mandates, but this time only a few countries have weights equal to one (those who contribute to abatement expenditures). All the others have weighted marginal utilities of abatement higher than the shadow price of carbon, and their weights i are also higher than one. The difference with the previous case is that most countries, a the few most impacted ones, are likely to be in this situation. 24 Derivation of L with regard to first-period abatement level x yields: L Ufi dj Ufi x = 0 C'(x) = - lfi i c d'i(x+xf) - lfi i d'i(x+xf) i i ji (a14) Since Lagrange multiplier is equal to the weighted marginal utility of consumption at first period (a9), this equation can be written: C'(x) = - lfi i Ufi'(yfi - afi - di(x+xf)) d'i(x+xf) i U'i (yi-ai) - lfi i Ufi'(yfi-afi-dfi)d'j(x+xf) Ufi/dj (a15) i ji Where i = i i (a16) Marginal abatement costs at first period take the general form of a discounted sum of future marginal benefits of abatement. The value of the discount factors, however, depends on the mandate. In status-quo mandates, coefficients i are equal to one. The discount factors become Ufi'(yfi - afi - di(x+xf)), U'i (yi-ai) which are exactly country-level consumption discount factors at the margin of the (post abatement) growth path. The discount factors are thus likely to be lower for countries with higher growth rates, thereby reducing the weight attached to their damages in (a15). In adaptative mandates, on the other hand, the discount factors becomes Ui'(yfi) Ui'(yi) Ufi'(yfi - afi - di(x+xf)). f U'i (yi-ai) If abatement expenses and residual damages remain small with regard to baseline revenues, then the last two terms cancel out, and all discount factors are roughly equal to a common value , which can be interpreted as a population-weighted average discount factor amongst countries. 25 Appendix 2: Domain of Validity of Property (12) Let U be a twice differentiable utility function defined over +, with U'>0 and U"<0. We are looking for the conditions under which the following property is valid: (P1) For all x>0 and all y > 0, x < y U"(x) U"(y) U'(x) < U'(y) (a17) For property P1 to hold, U' must be sufficiently convex.23 We show here that if U"/U' is monotonous, and if U is unbounded, then P1 holds. Proof: Let us assume U unbounded. If U"/U' were decreasing, then we would have (U"/U')' = [ln(U')]" 0 over [1,+[. Let G be the twice differentiable function such that G(1) = U'(1), [ln(G)]'(1) = [ln(U')]'(1), and [ln(G)]' constant over [1,+[. G exists, and is uniquely defined. Precisely, G(c) = eac+b with a + b = U'(1) and a = [ln(U')]'(1) <0. Since G(1) = U'(1), [ln(G)]'(1) = [ln(U')]'(1), and [ln(U')]" 0 while [ln(G)]"=0, we have U'(c) G(c) for all c in [1,+[. c c But G(x) dx is bounded, and thus so is U'(x) dx, which 1 1 contradicts the initial assumption that U is not bounded. C.Q.F.D. Appendix 3: Model Resolution under Uncertainty To model uncertainty, we assume the planner faces a finite set of possible scenarios indexed by j{1,2,...,M}. Each set is characterized by climate change impacts dij, second-period baseline income yij, and f future abatement costs Cfj. The planner also knows that full information about the true state of the world will be revealed at the 23In the literature on attitudes towards risk, P1 is equivalent to decreasing absolute risk aversion. 26 beginning of second period. But at the beginning of the first period, the planner only has a set of subjective probabilities pjattached to each possible future state of the world. Assuming the planner's utility function is Von-Neumann, the optimization problem becomes: Max liiUi(yi-ai) i + pjlfi ij Ufi(yij-aij-dij(x+xfj),d1j,...,di-1 ,d ,...,dNj) f f f f f f j i+1 j (a18) j i liai= C(x) (a19) i lfi aij = Cfj(xfj) f (a20) i ai 0 (a21) aij 0 f (a22) i = li -1 U'i(yi) with = U'i(yi) (a23) i i in status-quo mandates ij = j 1 -1 Ui'(yij) f f in adaptative mandates with j = U'i(yij) f (a24) i The Lagrangean becomes L = li i Ui(yi-ai) + lfi pj ij Ufi(yij-aij-dij(x+xfj),d1j,...,di-1 ,d f f f f f j i+1 j i ij ,...,dNj) + [ li ai - C(x)] + f pj µj [ lfi aij - Cfj(xfj)] + f liiai i j i i + lfi ij aij f (a25) ij And first-order conditions are now L ai = 0 i U'i(yi-ai) - i = (a26) 27 L Ufi f f aijf= 0 ij c (yij-aij-dij(x+xfj),...) - ij = µj (a27) L Ufi ' Ukf xfj = 0 µj Cfj' (xfj) = - lfi ij c dij(x+xfj) - lfi ij dkj dkj'(x+xfj) i i ki (a28) L Ufi x = 0 C'(x) = - pjlfi ij c dij'(x+xfj) j i f - pjlfi ij Uk dkj dkj'(x+xfj) (a29) j i ki Appendix 4: Numerical Illustration We consider two regions, called "North" and "South" respectively. "North" comprises high-income countries, as per World Bank (2002) definition, and "South" low and middle income ones. First period is 2000-2050, and second period 2050-2100. First-period income and population data are given by World Bank (2002).24 In the baseline scenario, economic growth in the North is assumed to be 2.5% per year, against 3% in the South. World population is assumed to grow by 2 billions people, all of them in the developing world. Table 4 summarizes key parameters of the baseline scenario. Without action, carbon dioxide emissions are assumed to reach 513 GtCO2 during the first period, and 688 GtCO2 during the second one, as in the IPCC IS92a scenario. Abatement costs at first and second period are assumed quadratic with respect to total abatement expenditures. We assume that marginal costs of a zero-carbon economy is $1,500/tC during the first period, dropping to $1,000/tC during the second period. The abatement cost functions thus become: 24 For simplicity's sake, we use 2000 and 2050 data respectively as averages for the two periods. 28 x = 513 1 - 2.89 ln an + ls as ln yn + ls ys (a30) f f f f xf = 688 1 - 5.91 ln an + ls as (a31) ln yn + ls ys f f f f Damages are assumed to be cubic with the total amount of carbon emitted in the atmosphere x+xf. x+xf 3 dfi(x+xf) = i 1200 (a32) We will use several values for coefficients . All utility functions are assumed to be logarithmic in consumption. The utility discount rate is set at 1% per year. Appendix 5: Proof of Property in Section 4.1 Let U be a twice differentiable utility function defined over +, with U'>0 and U"<0. Let c1,...,cn,r1,...,rn be strictly positive real numbers with r1 > ri for all i 2. We want to explore under which conditions the following holds: U'(r1c1) U'(rici) (P2) U'(c1) < U'(ci) for all i 2 (a33) We give two partial answers to that question. First, let us note that (P2) holds for all utility functions such that U'(rc) = r-k U'(c) (k>0). Those include, in particular, classical utility functions such as ln(c), and ca with 0 -U'(c2)c2g2 (a35) 29 (P2) thus holds--locally at least--if -c U"/U' is constant. The result is less clear otherwise. When -c U"/U' is decreasing with consumption, then (P2) remains valid if the country which grows at the fastest rate is also the country with lowest initial wealth level (c1g2). When -c U"/U' is increasing with consumption, then the result is ambiguous.25 25 In the literature on attitudes towards risk, -c U"/U' is the relative risk aversion. The property holds for constant relative risk aversion, and decreasing risk aversion functions. It is ambiguous for increasing risk aversion ones. 30 References Agarwal, A., Narain, S., 1991. Global Warming in an Unequal World, a case of environmental colonialism. Center for Science and Environment, Delhi. Azar, C., 1999. Weight factors in cost-benefit analysis of climate change. Environmental and Resource Economics 13(3), 249-268. Bodansky, D., 2001. Bonn Voyage. Kyoto's Uncertain Revival. The National Interest, Fall, 45-55. Bruce, J., Lee, H., Haites, E. (eds.), 1996. Climate Change 1995: Economic and Social Dimensions of Climate Change. Cambridge University Press, Cambridge. Chao, H.-P., Peck, S., 2000. Greenhouse gas abatement: how much? and who pays? Resource and Energy Economics 22(1), 1-20. Chichilnisky, G., Heal, G., Starrett, D., 2000. Equity and efficiency in environmental markets: global trade in carbon emissions. In Chichilnisky, G., Heal, G. (eds.). Environmental markets, Colombia University Press, pp.46-67. Grubb, M., Vrolijk, C., Brack, D., 1999. The Kyoto protocol, a guide and an assessment, Royal Institute of International Affairs, London. Guesnerie, R., 1995. The genealogy of modern theoretical public economics: from first best to second best. European Economic Review 39(3/4), 353-381. Hourcade, J.-C., 1994. Economic issues and negotiation on global environment: some lessons from the recent experience on greenhouse effect. In Carraro C. (Ed.) Trade Innovation and Environment. Klüwer Academics Publishers, Dordrecht, pp.385-405. Hourcade, J.-C., 2000. Le climat au risque de la négociation internationale. Le Débat. Hourcade, J.-C., Richels, R., Robinson, J., 1996. Estimating the costs of mitigating greenhouse gases. In Bruce, J., Lee, H., and Haites, E. (eds.). Climate Change 1995: Economic and Social Dimensions of Climate Change. Cambridge University Press, Cambridge, pp.263-296. 31 Lecocq, F., Crassous, R., 2003. International Climate Regime beyond 2012: Are Quota Allocation Rules Robust to Uncertainty? Policy Research Working Paper 3000. World Bank, Washington DC. McCarthy, J.J., Canziani ,O.F., Leary, N.A., Dokken, D.J., White, K.S. (Eds), 2001. Climate Change 2001: Impacts, Adaptation and Vulnerability. Cambridge University Press, Cambridge. Naki enovi , N., Swart, R. (Eds), 2000. Special Report of the Intergovernmental Panel on Climate Change on emission scenarios. Cambridge University Press, Cambridge. Negishi, T., 1960. Welfare economics and existence of an equilibrium for a competitive economy. Metroeconomica 12, 92-97. Nordhaus, W.D., Yang, Z. A Regional Dynamic General-Equilibrium Model of Alternative Climate-Change Strategies. American Economic Review 86(4), 741-765. OECD, 1993. International economic instruments and climate change. OECD, Paris. Rawls. J., 1971. A theory of justice. The Belknap Press of Harvard University Press, Cambridge. Sandler, T., Smith, K.V., 1976. Intertemporal and intergenerational Pareto efficiency. Journal of Environmental Economics and Management 2, 151-159. Schelling, T., 1995. Intergenerational discounting. Energy Policy 23(4/5), 395-401. Sen, A. K., 1987. On ethics and economics. Basil Blackwell Publishers, Oxford. Toth, F., Mwandosya, M., et alii, 2001. Decision-making Frameworks. In Metz, B., Davidson, O., Swart, R., Pan, J. (Eds.). Climate Change 2001: Mitigation, Contribution of Working Group III to the Third Assessment Report to the IPCC. Cambridge University Press, Cambridge, pp.601-690. World Bank, 2002. World Development Indicators. Washington DC. World Bank, 2003. Global Economic Prospects, Washington DC. 32 Table 1: Second-Period Expenditures in Adaptative ­ Dynastic Mandates Scenario Optimal Mitigation Policy Damage Damage Abatement Residual Total Abatement Residual Total maximum maximum Expenditures Damages climate Expenditures Damages climate North South N (aN) N (dN) bill N bill S (aN+dN) S (aS) S (dS) (aS+dS) a 5% 5% 1.01% 1.24% 2.25% 1.01% 1.24% 2.25% b 4% 6% 1.09% 1.06% 2.15% 0.55% 1.60% 2.15% c 3% 7% 1.18% 0.86% 2.04% 0.04% 2.00% 2.04% d 2% 8% 1.09% 0.61% 1.70% 0% 2.47% 2.47% All figures are percentage of second period income yf. Source: Authors' calculation. See Appendix 4 for calibration details. 33 Table 2: Total Abatement Level in Adaptative Dynastic Mandate Scenario Optimal Mitigation Policy Second- Second- Damage Damage Total Period Period maximum N maximum S Emissions x+xf Climate bill Climate bill N S a 5.0% 5.0% 754 2.25% 2.25% b 4.5% 6.2% 754 2.25% 2.25% c 4.0% 7.4% 754 2.25% 2.25% d 3.5% 8.6% 754 2.25% 2.25% e 3% 9.8% 754 2.17% 2.44% f 2% 12.2% 753 1.93% 3.02% g 1% 14.7% 751 1.69% 3.59% h 0.5% 15.9% 750 1.58% 3.87% j 0.0% 17.1% 749 1.46% 4.15% In all scenarios, the aggregate damage function is the same. All figures are percentage of second period income yf. Source: Authors' calculation. See Appendix 4 for calibration details. 34 Table 3: Optimal Emission Levels in All Mandates (First Period, Second Period) Diplomatic Attitude Status-Quo Adaptative Solidarity with future generations Dynastic (488 , 358) (484 , 380) Universal (482 , 282) (477 , 310) Source: Authors' calculation. Baseline emissions (513, 688), damages up to 2% of revenues in N, against 5% of GDP in S, universal utility functions of the form U = ln(c) 1 - 0.01 ds+dn dmax+dmax s n all other assumptions in Appendix 4. 35 Table 4: Economic and Population Assumptions First Period (2000) Second Period (2050) li (billions) yi (1995 US$) lfi (billions) yfi (1995 US$) North 0.95 26,750 0.95 91,943 South 5.11 1,160 7.11 5,085 36 Figure 1: Optimal abatement levels at first period for two regions differing only by income B Utility Unweighted Marginal Utilities (Identical) Marginal C A apoor arich Weighted Marginal Utility of the Poor ypoor yrich First-Period Revenue yi 37 Figure 2: Optimal abatement levels at second period for two regions differing only by income Utility Initial weighted marginal Unweighted utility of rich higher than Marginal Utilities poor's (Identical) Marginal Weighted Marginal Utility of the Poor Expenditures poor should support before rich starts contributing f f Ypoor Yrich Second-Period Revenue yfi 38