Economy and the Environment PDF

Economy and the Environment 1 Observation: Huge cross-country differences in income per capita…. 2 … largely due to significant differences in growth over the last 200 years…...

Economy and the Environment 1 Observation: Huge cross-country differences in income per capita…. 2 … largely due to significant differences in growth over the last 200 years… 3 … on average the people of the past were many times poorer than we are today but progress creates inequality! 4 In most places people today are much richer than our ancestors 5  Empirical work in economic growth has successfully decomposed these differences in levels and growth rates into contributions from physical capital, human capital and technology. But why do these factors vary so much across countries? o Some possible fundamental causes  Geography  Institutions  Culture  Luck 6  Sources of economic growth An economy’s output of goods and services depends on the quantities of available inputs, such as capital and labor, and on the productivity of those inputs. The relationship between output and inputs is described by the production function Y = AF(K,N) Which relates total output (Y), to the economy’s use of capital (K), and labor (N), and to productivity (A) If inputs and productivity are constant, the production function states that output also will be constant- there will be no economic growth. 7  The fundamental determinants of long-run living standards (Solow model) 1. An increase in the saving rate, s, causes long run output to rise 2. An increase in the rate of population growth, n, causes long run output to fall 3. An increase in productivity causes long run output to rise In the long run, the rate of productivity improvement is the dominant factor determining how quickly living standards rise. 8 Question: Do economies converge? Three possible scenarios for the evolution of living standards throughout the world: unconditional convergence, conditional convergence, and no convergence. Unconditional convergence: poor countries eventually catch up to the rich countries so that in the long run living standards around the world become more or less the same Conditional converge: living standards will converge only within groups of countries having similar characteristics No convergence: poor countries don’t catch up over time = living standards may even diverge (the poor get poorer and the rich get richer). 9  For a variety of reasons (such as different cultures, political systems, and economic policies) countries do differ in characteristics such as saving rates, so conditional convergence seems to be the more likely outcome Please note: according to the Solow model, if economies are open and international borrowing and lending flow freely, some additional forces support unconditional convergence What is the evidence? A. Little empirical support for unconditional converge. B. The evidence for conditional converge seems much better (Barro and Sala-i-Martin(1992); Mankiw, Romer and Weil ((1992). 10  New growth theory The Solow model does have one serious shortcoming: it takes the rate of productivity growth as given The new growth theory includes explanations of productivity change within the Solow model and it has two principal strands: 1. Focuses on the role of human capital (the economist’s term for the knowledge, skills, and training of individuals) The links between human capital and growth flow two ways a. As economies become richer they become more likely to “invest in people,” through improved nutrition, schooling, and on-the-job training. b. A healthier and more highly skilled labor force is more productive, which leads to improved living standards.11 2. Emphasizes the importance of technological innovation by private firms as a source of productivity growth = firms improve product design and production efficiency in various ways (formal research and development programs, learning by doing, etc) New growth theorists have tried to incorporate the process of innovation by firms into the Solow growth model. The new growth theory is promising: Policy implication In the Solow model the ability of government policies to influence long-run living standards is limited 12 Government policies to raise long-run living standards Include: 1. raising the rate of saving 2. increasing productivity: a. Investing in public capital (infrastructure) b. Encouraging the formation of human capital c. Increasing research and development (R&D) 3. industrial policy: the government uses subsidies and other tools to influence the nation’s pattern of industrial development-in particular, to stimulate high-tech industries Critics contend that in practice the government cannot successfully pick and subsidize only “winning” technologies. 13  An augmented theory of economic growth Broadly speaking, one can write down a production function that relates total output to total inputs such as: Y = F (K, L, H, A, X) where K: capital L: labor H: human capital A: level of technology X: production inputs such as: geography/climate/ecology (“disease burden”), economic (property rights) institutions, war, 14 culture (Protestantism)  Production function with climate change Output = f (capital, labor, climate change) Climate change will influence economic output via its effect on the availability of commodities essential for economic growth, such as 1. Water: population pressure and water-intensive activities (irrigation) will strain water supply (Southern Europe, West Coast of USA, Australia) 2. Food: although agriculture makes up only a small fraction of GDP (2-5%), it still vulnerable to climate change (Canada, Russia, and Northern Europe will increase production; Southern Europe, Australia will decrease) 3. Energy: High latitude regions reduce energy for heating and will increase energy for cooling; low latitude regions increase energy for cooling; disruption of energy production 15 (in nuclear plants) So, The set of mechanisms through which climate may influence economic outcomes, positively or negatively, is extremely large and difficult to investigate comprehensively = even if the effect of climate on each relevant mechanism were known, one would will be faced with the challenge of how various mechanisms interact to shape macroeconomic outcomes. The traditional approach to estimating the overall impact of climate change is to use “Integrated Assessment Models” (IAM) = they take some subset of mechanisms, specify their effects, and then add them up 16 Note: implementations of the IAM approach require many assumptions about which effects to include, how each operates, and how they aggregate. Even users of these models acknowledge their substantial limitations “Making such estimates is a formidable task in many ways (discussed below). It is also a computationally demanding exercise, with the result that such models must make drastic, often heroic, simplifications along all stages of the climate-change chain. What is more, large uncertainties are associated with each element in the cycle. Nevertheless, the IAM remain the best tool available for estimating aggregate quantitative global costs and risks of climate change” (the Stern Report, 2007) 17 IPCC AR5: “Changes in many extreme weather and climate events have been observed since about 1950” Source: Mizutori/Guha-Sapir, Economic Losses, Poverty and Disasters 1998-2017 18  Numbers of disasters by type 1998-2017 Source: Mizutori/Guha-Sapir, Economic Losses, Poverty and Disasters 1998-2017 19 20  Breakdown of recorded economic losses (US$) per disaster type 1998-2017 Source: Mizutori/Guha-Sapir, Economic Losses, Poverty and Disasters 1998-2017 21  Recorded climate-related disaster losses per income group compared to GDP losses 1998-2017 Source: Mizutori/Guha-Sapir, Economic Losses, Poverty and Disasters 1998-2017 22  Relative human and economic costs of climate- related disasters on continents 1998-2017 Source: Mizutori/Guha-Sapir, Economic Losses, Poverty and Disasters 1998-2017 23 Climate-related disaster affected totals in absolute numbers and percentage of population potentially exposed (PPE) 2000-2017 Source: Mizutori/Guha-Sapir, Economic Losses, Poverty and Disasters 1998-2017 24 Climate-related disaster deaths in absolute numbers per million population potentially exposed (PPE) 2000-2017 Source: Mizutori/Guha-Sapir, Economic Losses, Poverty and Disasters 1998-2017 25 Remember: The effects of climate change on economic growth will vary greatly over the world. Vulnerability to climate change depends on 1. Exposure to changes in the climate 2. Sensitivity: the degree to which a system is affected by or responsive to climate stimuli = susceptibility: the degree to which a system is open, liable, or sensitive to climate change 3. Adaptive capacity: the ability to prepare for, respond to and tackle the effects of climate change 26  Climate change and economic growth in LDCs Climate change will have a disproportionately damaging impact on economic growth of less developing countries (LDCs) due to their 1. Geographic exposure: the geography of many LDCs leaves them especially vulnerable to climate change a. they are located in low latitudes (tropical areas) b. they endure climate extremes and intra- (rainfall is concentrated in a single season) and inter- (large differences in the annual total of rainfall because of El Nino) annual variability in rainfall c. they are “already too hot” 27 2. Sensitivity: Developing economies are very sensitive to direct impacts of climate change because a. they are heavily dependent on i. agriculture (61% of people in South Asia and 64% in Africa are employed in the rural sector) ii. ecosystems (>20C=>large area in Amazon will dry up=> more than 1 billion people will be affected) b. they are experiencing i. rapid population growth (a 2 billion population increase) will add to the existing strain on natural resources) ii. rapid urbanization (city planners and public services unable to keep up with the huge influx of people=>people live in slums and hazard prone areas such as floodplains and steep slopes) 28 iii. food insecurity (40% of the sub-Saharan population is undernourished) iv. malnutrition (54% of child deaths in LDCs is associated with malnutrition) v. health problems (malaria) 29 3. Adaptive capacity: people will adapt to changes in the climate as far as their resources and knowledge allow. But LDCs lack both. = They have a. poor water-related infrastructure and management (less than 5% of land is irrigated in Africa; inappropriate water pricing=> excessive use of groundwater pumping=> aquifers drain, water tables fall) b. low incomes ( climate change will act as a poverty multiplier by increasing the number of poor people and by making poor people even poorer (SDG Fourth US National Climate Assessment report: the US economy could lose hundreds of billions of dollars – or, in the worst-case scenario, more than 10% of its GDP – by the end of the century if global warming continues apace. 36  The poor (country and individuals) will suffer the bulk of damages from climate change, whereas the rich countries will likely benefit. Should low latitude rural areas be compensated for damages from climate change? Insurance, cash or development aid? 1. Compensation is needed (Global Environmental Facility (GEF)) 2. Insurance fund (difficult to administer) 3. Invest in their economic development to increase adaptive capacity and thus reduce vulnerability (World Bank to administer the development program) 37  Sustainable development “Development that meets the needs of the present without compromising the ability of future generations to meet their own needs“ Brundtland Commission (1987) (formally known as the World Commission on Environment and Development (WCED)) It contains within it two key concepts: a) the concept of needs, in particular the essential needs of the world's poor, to which overriding priority should be given; and b) the idea of limitations imposed by the state of technology and social organization on the environment's ability to meet present and future needs.“ All definitions of sustainable development require that we see the world as a system—a system that connects space; and a system that 38 connects time. 39  Economic growth and pollution The environmental Kuznets curve posits an inverted-U relationship between pollution and economic development (Kuznets’s inverted-U relationship between income inequality and development) 40 Basic Environmental Kuznets Curve (EKC) 41  Environmental Kuznets Curve: Income  environmental quality 42  Economic Growth and Environmental Degradation 43 Scale, composition, technique effects  Numerous empirical studies have tested the environmental Kuznets curve model 44  45 Empirical evidence is inconclusive so, 1. Need for better theory (for example, corruption plays a role) 2. Need for better data (Global Environmental Monitoring System (GEMS) focuses on a few criteria pollutants and doesn’t incorporate toxic pollutants) 3. Think about the curvature of the environmental Kuznets curve 4. Compare the cross-section results with time series ones when the data are available 46  Research suggests that the shape of the environmental Kuznets curve is not fixed (curve can be lower and flatter)  developing countries don’t have to grow rich before they start to care about their environment Reasons for optimism: 1. Environmental regulation: dominant factor in explaining the decline in pollution as countries grow beyond middle-income status 2. Economic liberalization: privatization and elimination of government subsidies reduce the scale of economic activity and change its composition => decrease pollution + firms adopt clear technologies (cheaper and more profitable) 47 3. Pervasive Informal regulation: communities and NGOs a. where formal regulators are present, use the political process to influence the strictness of environmental regulations b. where regulators are absent/ineffective, pursue informal regulation= polluting factories negotiate directly with local actors in response to threats of political, social or physical sanctions if they fail to compensate the community (Coase theorem) or reduce emissions 4. Pressure from market agents: consumers may boycott products of polluting firms; bankers may refuse credit (worries about environmental liability); investors may withhold investment (worries about potential financial losses from regulatory penalties and liability settlements) 48 5. Better methods of environmental regulation: a) Target regulatory monitoring and enforcement on polluting industries/firms b) Move away from command-and-control policies and towards market-oriented forms of regulation (such as pollution charges) 6. Better information: improved information about pollutants, pollution damages, local environmental quality and cost of pollution abatement enhances the ability of a. local communities to protect themselves b. regulators to enforce environmental standards c. market agents to reward clean firms and punish heavy 49 pollutants. But Possible concerns: 1. Will countries need to suffer lower environmental quality in the short and medium run? No, if they enact and enforce appropriate environmental regulation 2. Globalization and risk of a “race to the bottom”: because high environmental standards in high-income countries impose high costs on polluters, firms relocate to low-income countries where people are eager for jobs and their environmental regulations are weak or non- existent (the “pollution havens” hypothesis). The “race to the bottom” argument is not supported empirically (few cases that pollution havens can emerge in extreme cases) 50 3. Are other pollutants rising? Yes, emissions of toxic organic chemicals remain largely unregulated almost everywhere 4. Building effective regulatory capability: appropriate legal measures for regulation, effective monitoring and enforcement of regulatory compliance 5. International assistance: trade and aid sanctions are inappropriate and ineffective methods, the international community can support programs aiming at a. providing public, easily accessible information about polluters, pollution damages, local environmental quality and cost of pollution abatement b. supporting the development of stronger regulatory institutions and cost-effective measures to reduce 51 pollution Environment and Development Economics 11: 159–178
C 2006 Cambridge University Press doi:10.1017/S1355770X05002755 Printed in the United Kingdom The distributional impact of climate change on rich and poor countries ROBERT MENDELSOHN* Yale School of Forestry and Environmental Studies, 230 Prospect Street, New Haven CT 06511 ARIEL DINAR World Bank, 1818 H Street NW, Washington DC 20433 LARRY WILLIAMS Electric Power Research Institute, 3412 Hillview Ave, Palo Alto, CA 94303 ABSTRACT. This paper examines the impact of climate change on rich and poor countries across the world. We measure two indices of the relative impact of climate across countries, impact per capita, and impact per GDP. These measures sum market impacts across the climate-sensitive economic sectors of each country. Both indices reveal that climate change will have serious distributional impact across countries, grouped by income per capita. We predict that poor countries will suffer the bulk of the damages from climate change. Although adaptation, wealth, and technology may influence distributional consequences across countries, we argue that the primary reason that poor countries are so vulnerable is their location. Countries in the low latitudes start with very high temperatures. Further warming pushes these countries ever further away from optimal temperatures for climate- sensitive economic sectors. 1. Introduction There is a broad consensus among climate scientists that further emissions of greenhouse gases will cause temperatures to increase 1.5◦ C to 5.8◦ C and precipitation patterns to shift by 2100 (Houghton et al., 2001). These changes in temperature will in turn cause ecosystems to move poleward and seas to rise. All of these changes will have effects on the global economy and the quality of life around the globe. In this paper, we focus on the distributional impact of climate change on the economies of rich and poor countries. We provide empirical support for a hypothesis, first suggested by Schelling (1992), that the poor may bear the brunt of the economic damages from climate change. * Correspondence: Email: [email protected] We want to thank the reviewers and editor for their very helpful comments. The views in this paper are those of the authors and should not be attributed to the World Bank. 160 Robert Mendelsohn, Ariel Dinar, and Larry Williams experimental cross sectional temperate country market impacts tropical country climate change Figure 1. Generic hill-shaped impact response function We want to make clear that this paper does not explain why some coun- tries are poor and others are rich. The standard neoclassical economic growth framework asserts that growth depends primarily on basic eco- nomic inputs such as trained labor, capital, and technological development (Solow, 1956). These basic economic factors have been extended to in- corporate government policies, the accumulation of human capital, fertility decisions, and the diffusion of technology (Barro, 1997; Bloom and Sachs, 1999; Easterly and Devine, 1998; Barro and Sali-i-Martin, 2004). Our paper has nothing to add to this important debate on growth. The purpose of this paper is to examine whether there are important distributional consequences of climate change impacts. The early literature on greenhouse gases did not raise serious concerns about the distributional impact of climate change (Nordhaus, 1991; Tol, 1995; Fankhauser, 1995; Pearce et al., 1996). This early literature largely assumed that damages were a linear or quadratic function of the change in temperature. As a result, the early models predicted that every country would suffer damages from warming and that they would be roughly proportional to income. Developing countries were predicted to be slightly more vulnerable because so much of their economies were in climate-sensitive sectors such as agriculture and because low technology operations are expected to have less substitution (Fankhauser, 1995; Tol, 1995). The literature at this time, however, assumed that almost every region would be damaged by warming (Pearce et al., 1996). Further, this sentiment extended to other chapters in the Second Assessment Report of the IPCC where cross-country equity and compensation to low-income countries was overlooked (Arrow et al., 1996; Jepma et al., 1996). Subsequent empirical research on climate impact sensitivity has re- vealed new insights into how temperature affects climate-sensitive eco- nomic sectors (Mendelsohn and Neumann, 1999; McCarthy et al., 2001; Mendelsohn, 2001; Tol, 2002). The new research indicates that several climate-sensitive sectors have a hill-shaped relationship with absolute temperature. Figure 1 illustrates this relationship in general. For each sector,  Environment and Development Economics 161 there is an optimum temperature that maximizes welfare in that sector. For farmers in regions that are cooler than the optimum temperature, warming would cause net revenues to go up. For farmers in regions that are warmer than the optimum temperature, warming would cause net revenues to fall. These results imply that countries that happen to be in relatively cool regions of the world will likely benefit from warming and that countries that happen to be in relatively warm regions of the world will likely be harmed by warming. We quantify the market impacts of climate change on every country in the world by combining a range of future climate scenarios with a range of climate response functions and background information from each country. The next section describes the country-specific climate forecasts from the three Atmospheric Ocean General Circulation Models (AOGCM’s) used in the paper (Houghton et al., 2001). Section 3 discusses the two sets of climate response functions used to evaluate the climate forecasts (Mendelsohn and Schlesinger, 1999). One set of response functions has a high and the other set a low climate sensitivity. Within each set, there is a separate response function for each of the five major economic sectors that are expected to be affected by climate: agriculture, water, energy, timber, and coasts1 (Smith and Tirpak, 1990; Pearce et al., 1996; Mendelsohn and Neumann, 1999; McCarthy et al., 2001). Using country-specific background information on such variables as cropland, coastland, population, and GDP, we use previously calibrated response functions (Mendelsohn and Schlesinger, 1999) to develop quantitative forecasts of the impacts in each sector for each country for each climate scenario (Mendelsohn et al., 2000a). We then sum these sectoral impacts to get an aggregate impact for each country. These country-level market impacts are then summed to get multi- country/regional aggregate outcomes. Non-market impacts to health, the environment, and aesthetics are not included in the calculations. Although non-market effects will certainly add to expected damages, reliable estimates of the magnitude of the resulting welfare impacts do not yet exist. The reported market impacts thus underestimate the total impacts of warming. However, it is likely that the non-market impacts will not change the distributional consequences of warming. Health effects and aesthetic effects are likely to strike low latitude countries the hardest as well. Only ecological changes have an ambiguous distributional outcome. In section 4, we use these tools to evaluate the distributional impacts of climate change. First, the world’s population is divided into quartiles on the basis of per capita income in 2100. The predicted impacts of the three climate models and two response functions are displayed for each quartile. The results indicate that the poorest half of the world’s nations suffer the bulk of the damages from climate change, whereas the wealthiest quarter has almost no net impacts. 1 Technically, the early literature also assumed that there would be commercial fishery losses. They have not been included here because the ecological link between warming and fishery losses is still speculative and so it is not known how fisheries would change. Further, fishery impacts are likely to be small since this is a small sector, but that may not be true for some countries. 162 Robert Mendelsohn, Ariel Dinar, and Larry Williams There are many reasons why rich and poor countries are different and they are all included in the results across quartiles. We consequently engage in two tests to isolate whether climate change and initial climate play an important role. In order to test the role of climate change, we assume all countries face identical climate change, although everything else about the countries may differ. We do a similar test with respect to initial climate by assuming that every country has the identical initial climate. The tests reveal that forcing climate change to be the same has no effect on the distributional outcome of impacts. The poor still bear the brunt of the world’s damages. However, forcing the initial climate to be the same for all countries changes the distributional results. If all countries had the same initial climate, the absolute magnitude of climate damages would rise with income, because richer nations have larger climate-sensitive sectors. As a fraction of GDP, poorer nations would still suffer higher climate damages than richer nations, but the difference is small. These tests reveal that the poor nations of the world bear the brunt of climate change damages primarily because they are located in the low latitudes and are already too hot. The rich nations may well benefit from climate change because they are located in the mid to high latitudes and are currently cool. The proportion of GDP in agriculture, technology, wealth, and adaptation contribute to the distributional outcome, but play a smaller role. These strong distributional results across countries suggest that compensation needs to be a part of the greenhouse gas policy agenda along with mitigation and adaptation. The final section of the paper discusses some policy alternatives. 2. Climate scenarios We explore the results of three AOGCM’s to predict the impact of green- house gases. The Parallel Climate Model (PCM) comes from the National Center Atmospheric Research (Washington et al., 2000). The Center for Climate Research Studies (CCSR) model was developed at NIES (Emori et al., 1999). The Canadian General Circulation Model (CGCM1) was de- veloped at the Canadian Climate Centre (Boer et al., 2000). All three models are dynamic coupled ocean–atmosphere models that include greenhouse gases and sulfates. The PCM and CCSR models assume the IS92a path of greenhouse gases and the CGCM1 model assumes a 1 per cent exponential path of greenhouse gases. These two paths result in CO2 levels of 685ppmv for PCM and CCSR and 808ppmv in CGCM1 by 2100. These three models were selected to demonstrate the consequences of a full range of climate scenarios. Each model predicts changes in individual grid points across the earth. We use the grid points in each country to create a climate change scenario by country for 2100. The grid points are weighted by population and not by area. We prefer the population weighting method of evaluating climate change forecasts because most impacts occur near where people are living (Williams et al., 1998; Mendelsohn et al., 2000b). The population-weighted changes for each country are the inputs to the impact model. We also use population weights to generate regional average changes in temperature and precipitation. The three models make very different forecasts of global temperature change: PCM predicts 2.5C, CCSR predicts 4.0C, and CGCM1 predicts 5.2C  Environment and Development Economics 163 Table 1. Changes in temperature and precipitation predicted by each climate model in 2100 PCM CCSR CGCM1 Region T(C◦ ) P(%) T(C◦ ) P(%) T(C◦ ) P(%) L. Amer. 2.0 5.9 3.3 −10.8 4.9 −4.3 Africa 2.3 11.9 3.9 11.6 6.2 −10.3 S. Asia 2.4 21.5 3.6 16.1 4.5 −1.6 Pacific 1.8 7.3 2.6 14.4 4.2 −8.6 N. Amer. 2.4 7.9 5.5 −23.1 5.4 1.7 Europe 2.4 8.2 5.3 −6.0 3.9 −1.8 N. Asia 2.9 22.5 4.0 10.0 6.4 −12.6 FSU 3.5 10.2 5.7 9.6 7.1 9.9 Globe 2.5 15.5 4.0 7.7 5.2 −5.6 Note: Temperature changes measured in centigrade and precipitation in percentage changes. The climate measurements are weighted by population not area. by 2100. Global precipitation changes also vary by model. PCM predicts a 16 per cent increase, CCSR predicts an 8 per cent increase, and CGCM1 predicts a 6 per cent decrease in global precipitation by 2100. The distribution of population weighted temperature changes across continents varies as seen in table 1. Warming is expected to increase with latitude (Houghton et al., 2001). PCM and CCSR follow this accepted pattern and show more warming in the polar and temperate regions versus the tropical regions. CCSR predicts that this difference across latitudes will be extreme, whereas PCM shows more modest differences. CGCM1 shows a more random flux of temperatures, with Africa getting especially hot but Western Europe warming less than the rest of the planet. The AOGCM models also predict a wide range of changes in population- weighted precipitation. PCM predicts higher precipitation in every con- tinent, but especially in Africa, North Asia, and South Asia. CCSR predicts large losses of precipitation in Latin America and North America and small losses in Europe but large gains elsewhere. CGCM1 predicts losses of precipitation in every continent except North America and North Asia. There is clearly no consensus across the models about what will happen to local precipitation. However, the models do suggest that local precipitation might change significantly in very different ways across the planet. The climate scenarios provide four driving forces that can have an effect on economic sectors: changes in mean or seasonal temperature, changes in mean or seasonal precipitation, increases in carbon dioxide, and increases in sea level. In addition to these climate changes, global warming can also cause an increase in the variance of temperature and precipitation, a slowing of the thermohaline circulation (resulting in northern cooling), and the sudden loss of ice sheets (rapid sea level rise). These latter forces were not evaluated in this paper, partially because they are more speculative and partially because there is less known about their timing and magnitude. 164 Robert Mendelsohn, Ariel Dinar, and Larry Williams 3. Impact methodology We look at two different empirical approaches to determine the climate sensitivity of each economic sector: experimental and cross-sectional studies. The experimental studies have been done in controlled settings such as laboratories or greenhouses (see Reilly et al., 1996 for a good summary of experimental results in agriculture). These studies carefully control for unwanted variables but they struggle to include adaptation fully. In contrast, cross-sectional studies examine actual outcomes from place to place in order to measure climate impacts. The Ricardian method is a good example of this approach in agriculture where the values of farms in different climates are compared (Mendelsohn et al., 1994). The cross-sectional studies include efficient adaptation by design, but they struggle to control for unwanted influences. Comparing the experimental and cross-sectional method, the strength of each empirical methodology is the weakness of the other. The experimental method has the added advantage that it can measure the direct effect of carbon dioxide, which the cross-sectional method cannot.2 Both models predict that temperature has a hill-shaped relationship with agriculture, forestry, water, and energy; that increased precipitation is generally beneficial; and that coastal damages increase as sea level rises. Details about the shapes of these functions can be found in Mendelsohn and Schlesinger (1999). The results from experimental studies lead to steeper hill-shaped climate response functions compared with the cross-sectional results. The experimental model predicts that countries that are cooler than optimum will gain more from warming and countries that are warmer than optimum will lose more than the cross-sectional model predicts. As in the cross- sectional model, precipitation is predicted to have a beneficial impact on agriculture, forestry, and water but no effect on energy. The experimental model depends only on average annual climate, whereas the cross-sectional model captures a full array of seasonal temperatures and precipitation levels. Carbon dioxide, through fertilization, is strictly beneficial and helps forestry and especially agriculture in all regions (see Reilly et al., 1996 and Sohngen et al., 2002). It is assumed that carbon fertilization benefits increase with the log of CO2 and are the same in both models. In a complete general equilibrium model, global warming could change the supply and demand of all goods and services, leading to new global prices for everything. In practice, the climate changes expected over the next hundred years will not change overall economic conditions enough to affect most prices. For example, across a host of climate scenarios, market damages as a fraction of GDP were estimated to be less than 1 per cent (Mendelsohn et al., 2000b). Even the higher estimates found in the early literature sug- gested that damages would be just 2 per cent of GDP (Pearce et al., 1996). Such small changes in output do not warrant using a general equilibrium model. Most price changes that will occur because of warming will be limited to the sectors directly affected by climate change. In these sectors, warming would affect consumers and suppliers across the world through direct 2 Carbon dioxide effects from the experimental studies are used to predict carbon dioxide impacts in the cross-sectional results.  Environment and Development Economics 165 effects and prices could change. Models that take these price effects into account have been constructed to study climate impacts on timber (Sohngen et al., 2002). Reliable global models for agriculture and energy have not yet been developed.3 This paper relies on studies that assume climate has no effect on output prices in agriculture, timber, and energy. Global impact studies of these sectors have not been done but it is likely that climate-induced global price changes would be small. However, if this is not the case, assuming constant prices biases the welfare estimates. If climate change causes global scarcity and therefore increases prices, the presented results will underestimate total damages and miss consumer damages completely. If climate change increases abundance and reduces prices, the presented results will overestimate total benefits and miss consumer benefits completely. What will happen to supplier welfare is ambiguous. In contrast to the sectors with global markets, water is likely to have only a regional market because it is hard to transfer across basins. Water supply and demand in specific regions can change dramatically across climate scenarios and so there could be profound local price effects. These are captured by the model, which measures basin water prices using water supply and demand changes (Hurd et al., 1999). Two response functions to sea level rise are used in the model (Neumann and Livesay, 2001). In the cross-section model, we assume that landowners have foresight and so they depreciate buildings, anticipating they will be abandoned to sea level rise. In the experimental model, we assume that landowners have no foresight and that leads to slightly higher costs. The coastal study examines a series of decisions made each decade to either protect or abandon coastline in response to the rising seas. By stretching out responses across the century, costs are held to a relatively low level in each decade. At least in the US and Singapore, the model predicts that valuable coastlines will be protected (Neumann and Livesay, 2001; Ng and Mendelsohn, 2005). However, coastal protection is an adaptive response that generally requires government planning and coordination. It is not clear whether governments will make efficient decisions to protect coastlines. We find that the hill-shaped response functions are slightly different between developed and developing countries (Mendelsohn et al., 2001). The developing countries have lower crop net revenues per hectare and they are more temperature sensitive. The agriculture crop response functions to temperature in both developed and developing countries are hill shaped. But the developed country response function is both higher and flatter than the developing country response function, presumably because the high technology farmers have more capital and they can substitute capital for climate. The model predicts that agriculture in developing countries is more vulnerable to higher than optimal temperatures. We assume that this more vulnerable climate response function applies to countries whose 2100 per capita income is less than $7,000. Empirical research suggests that 3 There are some well-calibrated general equilibrium models in agriculture that examine country-specific impacts (see for example Adams et al., 1999), but these simply assume global price changes. 166 Robert Mendelsohn, Ariel Dinar, and Larry Williams Table 2. Aggregate market impacts in 2100 (USD Billions/yr) Climate predictions PCM CCSR CGCM1 CS Exp CS Exp CS Exp Total 63.8 217.1 −22.8 −93.5 −19.1 −273.3 % GDP 0.03% 0.10% −0.01% −0.04% −0.01% −0.13% Per capita 6.69 22.77 −2.39 −9.81 −2.00 −28.7 Note: We assume that sea level rise by 2100 is equal to 0.3 m by 2100 in PCM, 0.5 m in CCSR, and 0.9 m in CGCM1. agriculture in developing countries is more climate sensitive (Mendelsohn, et al., 2001), but this particular income cutoff value is arbitrary. The cutoff roughly separates out the poorer from the richer half of all nations. Both the cross-sectional models and the experimental-simulation models assume efficient adaptation. Users are assumed to maximize their net benefits: gross private benefits minus the costs of adaptation. The results reported in this paper include the most recent efforts to incorporate adaptation. The experimental-simulation results are consequently not as severe as some of the earlier analyses in the literature, which made less of an effort to include efficient adaptation (see Pearce et al., 1996). Each country has numerous characteristics such as land, length of coasts, population, and GDP that also play a role in determining country-specific impacts. Some of these factors are constant over time such as coastline and land. However, several of these factors will change over time and the changes could be profound by 2100. All the future scenarios use the same economic and demographic assumptions. Population growth is assumed to decline over time in every country. The average population growth over the next century for developed countries is 0.7 per cent a year, for China it is 0.5 per cent a year, and for all other developing countries it is 0.9 per cent a year. GDP is expected to grow by 1.9 per cent a year for developed coun- tries, by 0.9 per cent a year for Sub-Saharan Africa, and by 2.9 per cent a year for developing countries. Agriculture is expected to grow at 0.25 per cent a year for Africa and at 0.5 per cent a year for the rest of the world. Agriculture is consequently expected to be a smaller fraction of GDP over time. All of these assumptions are based on the IS92 scenario (Houghton et al., 1994), although they are consistent with predictions of other international agencies (World Bank Group, 2002). 4. The distributional impacts of climate change For each of the climate predictions from the three AOGCM’s and for both climate sensitivity functions, we calculate the global net market impacts by 2100 by aggregating individual country results. Table 2 presents these results for the world as a whole. Global impacts are positive or beneficial under the PCM climate, because it is a very mild climate change scenario. The experimental response function leads to larger benefits than the cross- sectional function. In contrast, the more severe climate change scenario  Environment and Development Economics 167 Table 3. Market impacts in 2100 by income (Billions USD/yr) Impacts by climate predictions PCM CCSR CGCM1 Income Cross Cross Cross Group section Experimental section Experimental section Experimental Poorest Impact −1.2 −8.0 −4.8 −69.4 −6.9 −140.7 Quartile %GDP −0.2 −1.4 −0.8 −11.8 −1.2 −23.8 Second Impact 4.5 19.7 −5.6 −30.2 −9.5 −92.0 Quartile %GDP −0.4 1.6 −0.5 −2.4 −0.8 −7.4 Third Impact 21.8 56.6 −0.7 −7.1 −4.5 −64.1 Quartile %GDP 0.8 2.1 −0.0 −0.3 −0.2 −2.4 Richest Impact 38.8 148.7 −11.7 13.2 1.8 23.5 Quartile %GDP 0.2 0.9 −0.1 0.1 0.0 0.1 in CCSR is predicted to lead to small global net damages in both cases with experimental results again being larger. Finally, under the severe climate change scenario in CGCM1, damages will be slightly smaller for the cross-sectional response function but much larger for the experimental response function. This range of global net impacts is consistent with the Third Assessment Report of the IPCC, although the Report focuses more on the potentially harmful end of this range (McCarthy et al., 2001). Annual market impacts as a fraction of GDP in 2100 range from slightly beneficial (+0.1 per cent of GDP) to slightly harmful (−0.12 per cent of GDP). These market impacts amount to an annual benefit of about $23 per person to a loss of about $27 per person. It is clear that the different climate change scenarios have a large impact on the overall results one sees for the world. Specifically future climate scenarios that predict larger temperature increases and precipitation losses, lead to larger overall net global damages. However, the focus of this paper is upon the distributional impacts of these global changes, not their overall magnitude. In order to understand how these climate impacts affect countries of different income levels, we order countries by per capita income in 2100. We then divide the country list into quartiles on the basis of their projected population in 2100. Each quartile represents one-fourth of the world’s population by 2100. A list of all countries and which quartile they fall in are shown in the Appendix. Table 3 shows the market impact results by quartile. The poorest quartile earns less than $4,380 per capita and includes 53 countries, mostly from Africa. The second quartile group earns from $4,380 to $5,785 per capita and includes only six countries, notably India and China. The third quartile earns between $5,785 and $25,000 and includes 65 countries from all over the world. Although the bulk of these countries are from warm latitudes, there are a few cooler countries in this group. The richest quartile of the world’s population includes 52 countries from North America, Europe, and the Middle East and a handful of countries from other continents. The richest quartile includes most of the countries in the mid–high latitudes and a scattering of countries 168 Robert Mendelsohn, Ariel Dinar, and Larry Williams Table 4. Market impacts assuming identical climate change in all countries (Billions USD/yr) Income +2C +3.5C +5C +3.5C +3.5C Quartile 0%P 0%P 0%P +10%P −10%P Poorest Exp −41.4 −102.0 −153.8 −78.7 −124.2 Crs −2.3 −4.6 −6.9 −4.5 −4.6 Second Exp −12.3 −50.3 −94.1 −39.3 −71.9 Crs −3.3 −5.4 −7.5 −5.4 −5.2 Third Exp 31.9 −1.9 −44.1 30.4 −35.7 Crs 19.2 10.3 −0.7 11.4 8.7 Richest Exp 96.7 65.9 12.4 126.6 2.8 Crs 43.6 22.5 −6.0 24.9 19.3 Note: Climate change is assumed to be uniform across the world. in the low latitudes. Most of the largest economies in the world are in this group. The richest quartile controls 78 per cent of the world’s GDP in 2100, the poorest quartile only 2.5 per cent. Table 3 shows what happens to aggregate market impacts in each quartile for each climate scenario and for each response function. Examining impacts across quartiles, the poorest quartile suffers damages across all six scenarios. The second poorest quartile suffers damages in all but the PCM scenario with the cross-sectional response function. The third richest quartile also suffers damages in all but the PCM scenario with the cross- sectional response, but these damages are smaller than what the two poorer quartiles suffer. In contract, the richest quartile suffers damages in only the CCSR scenario with the cross-sectional response. In all other cases, the richest quartile actually benefits from warming. The results provide strong evidence that the bulk of the damages from climate change will fall on the poor countries of the world. Table 3 also shows the impacts as a percentage of GDP. These calculations reveal that climate impacts are likely to be burdensome to the poorest countries. The lowest quartile would suffer damages from 12 per cent to 23 per cent of their GDP with the more severe climate scenarios and the experimental response function. In contrast, the range of impacts for the richest quartile is between a damage of 0.1 per cent to a benefit of 0.9 per cent of GDP. The results suggest there is a very large cross-country dis- tributional issue associated with climate change impacts. Why are poor countries so vulnerable? One hypothesis is that poor countries have more severe climate change scenarios than rich countries. We test this hypothesis by forcing climate change to be the same for all countries in the world. Table 4 presents the results of this experiment. The three levels of global temperature chosen roughly correspond to the global average predictions from the three AOGCM’s. The results provide a similar pattern as in table 3. Damages are greatest for the poorest quartile and they decline with income, eventually becoming beneficial for the richest group. Eliminating the difference in climate change predictions across countries does not change the distributional results.  Environment and Development Economics 169 Table 5. Current observed temperature and precipitation in each region Region Temperature (C) Precipitation Africa 29.1 7.2 South Asia 28.5 10.0 Latin America 25.9 11.9 Pacific 29.6 18.3 North Asia 19.7 7.4 North America 19.5 8.0 Europe 13.7 6.1 Former Soviet Union 12.0 4.8 Note: Observed measurements are population weighted averages not area weighted averages as usually shown. Table 6. Market impacts assuming identical climates and climate change in all countries (Billions USD/yr) Income +2C +3.5C +5C +3.5C +3.5C Quartile 0%P 0%P 0%P +10%P −10%P Poorest Exp 1.8 −37.4 −82.0 −8.5 −65.3 Crs −2.2 −5.0 −7.7 −5.0 −5.0 Second Exp −0.8 −29.7 −64.6 −9.2 −50.2 Crs −3.4 −6.1 −8.8 −6.0 −5.9 Third Exp −2.1 −60.9 −131.7 −19.6 −102.3 Crs 17.8 5.7 −8.6 7.0 4.6 Richest Exp −30.8 −156.2 −304.4 −80.0 −232.4 Crs 21.9 −14.9 −59.3 −11.6 −17.0 Note: All countries are assumed to have the identical global average climate. Climate change is assumed to be uniform across the world. An alternative hypothesis is that poor countries are more vulnerable because they are located in the low latitudes and have higher current observed temperatures. Table 5 shows the current variation in temperature and precipitation by region. These starting climates can be very important because they determine whether a sector is already too hot or too cool compared with the optimum for that sector. Table 5 shows that the low latitude regions are currently hot. These temperatures are actually beyond the optimum for most climate-sensitive economic sectors. In contrast, the mid latitude regions enjoy a range of current temperatures near the optimum. The former Soviet Union and northern Europe have cool current temperatures that make warming good for their economy. In tables 6 and 7, we assume that all countries have the same climate both now and in the future. This assumption places every country under the same climate experiment, although it allows countries to be different in other ways. As shown in table 6, if both present and future climates are the same in every country, it would no longer be true that the poorest countries would suffer the brunt of the damages from climate change. In fact, damages 170 Robert Mendelsohn, Ariel Dinar, and Larry Williams Table 7. Market impacts assuming identical climates and climate change in all countries (% GDP) Income +2C +3.5C +5C +3.5C +3.5C Quartile 0%P 0%P 0%P +10%P −10%P Poorest Exp 0.3 −6.3 −13.9 −1.4 −11.1 Crs −0.4 −0.9 −1.3 −0.8 −0.9 Second Exp −0.1 −2.4 −5.2 −0.7 −4.0 Crs −0.3 −0.5 −0.7 −0.5 −0.5 Third Exp −0.1 −2.3 −4.9 −0.7 −3.8 Crs 0.7 0.2 −0.3 0.3 0.2 Richest Exp −0.2 −1.0 −1.9 −0.5 −1.4 Crs 0.1 −0.1 −0.4 −0.1 −0.1 Note: All countries are assumed to have the identical global average climate. Climate change is assumed to be uniform across the world. would rise with income, because the climate-sensitive economic sectors in the richer quartiles are larger. Although agriculture plays a much larger role in developing countries today, it is expected to play a much smaller role by 2100. Further, energy and water are much larger sectors in the richer countries and their role does not shrink over time. Differences in current climates do explain why poor countries are predicted to suffer the net global impacts of climate change. Because most poor countries happen to be in the low latitudes, they begin with temperatures that are already too warm. Table 7 displays the results for this same experiment using impacts per GDP as the measure. Here, the results are not as dramatic as in table 6. Even if climates were the same across all countries, poor countries would still have higher impacts per GDP. The difference between rich and poor shrinks but does not disappear. Poor countries still have larger proportions of their economy in climate sensitive sectors (namely agriculture) and the absence of capital and technology still gives them fewer adaptation options. Location is not the only reason why poor countries are likely to have higher impacts per GDP than rich countries, but it is a very important reason. Figure 2 maps annual market impacts in 2100 for each country in the world, using the cross-sectional impact model. The figure illustrates the geographical distribution of impacts across the world. Three very different future climates provide a range of climate changes. The relatively flat response function of the cross-sectional model produces a subdued impact pattern across these future climates. Mid latitude regions do well and even tropical parts of the western hemisphere see small benefits under all three projected climates in 2100. Russia, Mongolia, Kazakhstan, and Eastern Europe do better than the rest of the world. However, across all the scenarios, the poor countries of Africa and southeastern Asia experience noticeable adverse impacts. Figure 3 maps 2100 impacts using the more steeply hill-shaped experi- mental response functions. The benefits accruing to Russia, Mongolia, Kazakhstan, and Eastern Europe are larger in figure 3 than with the cross- sectional model. Africa and the poor tropical countries are worse off in this  Environment and Development Economics 171 PCM cs 2100 (%GDP) no data to -5.0 to -2.0 to -0.5 to -0.1 to 0.0 to 0.1 to 0.5 to 1.0 over CCSR cs 2100 (%GDP) no data to -5.0 to -2.0 to -0.5 to -0.1 to 0.0 to 0.1 to 0.5 to 1.0 over CGCM1 cs 2100 (%GDP) no data to -5.0 to -2.0 to -0.5 to -0.1 to 0.0 to 0.1 to 0.5 to 1.0 over Figure 2. Annual market impacts (percent GDP) estimated for each country in 2100 using the cross-sectional impact model. The climatologies are based on the PCM, CCSR, and CGCM1 AOGCMs figure. Two of the three climate forecasts show that the experimental model leads to damages in tropical South America as well. Even the United States shows mild losses under the warming predicted by CCSR. 172 Robert Mendelsohn, Ariel Dinar, and Larry Williams PCM exp 2100 (%GDP) no data to -5.0 to -2.0 to -0.5 to -0.1 to 0.0 to 0.1 to 0.5 to 1.0 over CCSR exp 2100 (%GDP) no data to -5.0 to -2.0 to -0.5 to -0.1 to 0.0 to 0.1 to 0.5 to 1.0 over CGCM1 exp 2100 (%GDP) no data to -5.0 to -2.0 to -0.5 to -0.1 to 0.0 to 0.1 to 0.5 to 1.0 over Figure 3. Annual market impacts (percent GDP) estimated for each country in 2100 using the experimental impact model. The climatologies are based on the PCM, CCSR, and CGCM1 AOGCMs The maps of figures 2 and 3 (color maps are available from the authors upon request) reveal that under all climate forecasts and with both impact models, the poor countries of Africa and Southeast Asia are harmed by projected climate change in 2100. These results demonstrate that the low lati- tude regions will be hard hit by climate change. Almost all of the poor  Environment and Development Economics 173 countries of the world lie in the low latitudes. The maps consequently sup- port the analysis and suggest that warming will be hard on poor countries. The simulation model does not assume that damages increase with income (the model actually predicts the opposite). The key assumption in the model is that impacts have a hill-shaped relationship with temperature. Countries that are already hotter than optimal (the top of the hill) will suffer damages with warming. The distributional impacts predicted in this paper are caused because poor countries just happen to be in the low latitudes (which are already hot), whereas most rich countries happen to be in the mid to high latitudes (which are currently cool). 5. Conclusion The paper investigates whether the impacts from climate change have distributional consequences across countries. This analysis uses predictions about future climate change and calibrated climate response functions to calculate market impacts in 2100 for each country in the world. In order to capture the range of outcomes likely from climate change, we present the results of six scenarios (three climate scenarios times two response functions). For each scenario, we aggregate the impacts across countries by income per capita. Specifically, we divide the world’s population into quartiles on the basis of their GDP per capita. We then calculate the market impacts for each quartile. The results indicate that the poorest quartile will suffer damages in all scenarios. The next poorest quartile will suffer damages in all but the mildest climate change scenario. Although the third richest quartile also suffers damages in all but the mildest climate change scenario, the damages are quite small compared with the poorer half of the world. The richest quartile, in contrast, benefits in all but one case. Overall, the poor will suffer the bulk of the damages from climate change, whereas the richest countries will likely benefit. The analysis then tests whether poor countries face more devastating climate change scenarios than more wealthy countries. We test how impacts would change if every country faced identical climate changes. The results in this experiment are almost identical to the findings with the AOGCM predictions. Poor countries continue to bear the burden of climate change damages, whereas rich countries likely benefit. Finally, we test whether poor countries bear a larger burden of climate damages because they are already hot. In this test, we assume every country has the identical current climate as well as climate change. Countries still differ from one another because of economic, demographic, and geographic reasons. In this test, damages rise with income. Damages are concentrated in poor countries specifically because of their current climate. Because they happen to be located in low latitude regions, poor countries are currently much hotter than optimal, whereas more wealthy countries located in the mid to high latitudes are currently cool. Increases in temperature consequently cause more damages to poor countries compared with more wealthy countries. The fact that damages increase with income, once current climate is controlled, deserves some additional explanation. More wealthy countries bear larger damages because the climate-sensitive economic sectors in these 174 Robert Mendelsohn, Ariel Dinar, and Larry Williams countries are larger. By 2100, the importance of agriculture in GDP has shrunk, whereas other climate-sensitive sectors, namely water and energy, have maintained their relative share. The lower climate sensitivity implied in more advanced economies apparently has little effect on aggregate impacts. However, as a fraction of GDP, even controlling for climate, poor countries have higher impacts. The size of the climate impacts pales in comparison with the size of the economies of the richer nations. Poor countries consequently bear a larger burden as a fraction of their GDP than rich countries because of the many reasons raised in the literature, including lower capital, technology, and adaptation options. Although these national results are insightful, they do not necessarily predict what will happen to individual poor people. That is, many countries are large enough so that different regions will have different effects within national borders. Further, what happens to some countries in aggregate does not necessarily indicate what will happen to the poor residents of a country. The approach used in this paper cannot identify within-country effects. However, alternative studies such as rural income analysis (Mendelsohn et al., 2003) can identify how effects are distributed within a nation. There are several reasons to expect that individual poor will be burdened even more than the aggregate national numbers suggest. In most countries, there is a wide disparity of agricultural productivity across regions. In the low latitudes, the rural poor tend to live in the hotter and drier regions of each country. Warming is likely to damage these regions more harshly than the more temperate zones of each country. The poor are also likely to suffer larger damages than country averages because the poor do not have access to capital. Without capital, the poor will find it harder to adapt to warming. The poor may have more difficulty moving away from changes in climate, as their assets may be closely tied to specific pieces of property that may be of low value once climate changes. Finally, the poor cannot purchase their way out of reductions in crop productivity; they may not have the resources to buy food. An important limitation of this paper is that most of the empirical impact studies that support these results have been done in the US (Smith and Tirpak, 1990; Mendelsohn and Neumann, 1999; Mendelsohn, 2001). Only a few studies have attempted a more global reach (Rosenzweig and Parry, 1994; Sohngen et al., 2002) or have measured welfare impacts in other countries. Very few studies have been done in developing countries (Mendelsohn et al., 2001; Kurukulasuriya and Rosenthal, 2003). Most of what we assume will happen in the low and high latitudes is inferred from a few empirical studies. We consequently have less confidence in our results for the low and high latitudes. This is important to remember because the largest predicted impacts from climate change are in the low and high latitudes. This paper has shown that climate impacts have large distributional consequences. The bulk of the damages from climate change are likely to fall on the poor countries of the world. These results have bearing on climate change policy. If one applies equity weighting, the damages from climate change will be greater (Fankhauser, Tol, and Pearce, 1997; Azar, 1999; Tol, 2001) and the urgency to apply mitigation and adaptation will increase. However, we believe the most important policy change that  Environment and Development Economics 175 is required is to consider cross-national compensation. The distributional results found in this paper suggest that climate change negotiators must talk about compensating poor countries from the low latitudes. These countries will bear the brunt of the damages from climate change even though they made only a small contribution to cumulative emissions. If compensation is considered, how will a compensation program be designed? One idea that has been circulated in the UNFCCC (United Nations Framework Convention on Climate Change) negotiations is to provide some compensation to help poor countries mitigate emissions (see Marrakesh accords at www.unfccc.int). The Marrakesh accords also recommend holding a workshop to help developing countries insure themselves against the adverse impacts of climate change. The specifics of this idea are not yet developed. One possibility is that an international fund such as GEF (Global Environmental Facility) could subsidize adaptation. For example, the GEF could provide poor countries with financial and technical support for joint-public adaptations such as water projects, coastal protection, or endangered species protection. Efficient programs that support mitigation or adaptation are definitely possible compensation schemes. Another alternative is to create a climate impacts insurance fund for low latitude countries. Countries could apply for relief from the fund whenever they suffer a climate impact. In practice, this is likely to be difficult to administer because countries will claim harm with every weather event whether or not it is related to greenhouse gases. Unlike severe events such as hurricanes and floods, the gradual nature of global warming will make it very difficult to measure damages as they occur. Finally, paying victims compensation may create deleterious incentives that encourage people to put themselves in harm’s way. A final alternative is to compensate low latitude countries by investing in their economic development. Rapid development could help low latitude countries adapt to future climate change by reducing vulnerability, although it would increase emissions. As countries develop, they move away from agriculture, making their economies more resilient to climate change. An effective development program would also provide the needed technological progress that would make even climate-sensitive sectors less sensitive to future climate change. But, most importantly, an economic development program could address the imbalance between those who currently benefit from emissions and those who are likely to pay the consequences of climate change. A well-designed economic development program would bring large benefits directly to the people of poor countries. International development institutions such as the World Bank could administer such development programs. They could be an effective intermediary between the countries that generate greenhouse gas emissions and the countries likely to be harmed. For example, the Bank could collect a modest carbon tax on all countries in order to fund a development program for low latitude countries. They could help design an effective development program and then use the carbon revenues to fund it. Rather than focusing strictly on mitigation, the carbon program would modernize developing countries, making them more capable of taking care of themselves. The development program could address the fundamental inequity of greenhouse gases and provide the poor nations of the world with immediate benefits. American Economic Association Confronting the Environmental Kuznets Curve Author(s): Susmita Dasgupta, Benoit Laplante, Hua Wang, David Wheeler Source: The Journal of Economic Perspectives, Vol. 16, No. 1 (Winter, 2002), pp. 147-168 Published by: American Economic Association Stable URL: http://www.jstor.org/stable/2696580 Accessed: 23/02/2009 03:00 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aea. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected]. American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Economic Perspectives. http://www.jstor.org Journal of EconomicPerspectives-Volume16, Number1-Winter 2002-Pages 147-168 Confronting the Environmental Kuznets Curve Susmita Dasgupta, Benoit Laplante, Hua Wang and David Wheeler T he environmentalKuznets curve posits an inverted-U relationship be- tween pollution and economic development. Kuznets's name was appar- ently attached to the curve by Grossman and Krueger (1993), who noted its resemblance to Kuznets's inverted-U relationship between income inequality and development. In the first stage of industrialization, pollution in the environ- mental Kuznets curve world grows rapidly because people are more interested in jobs and income than clean air and water, communities are too poor to pay for abatement, and environmental regulation is correspondingly weak. The balance shifts as income rises. Leading industrial sectors become cleaner, people value the environment more highly, and regulatory institutions become more effective. Along the curve, pollution levels off in the middle-income range and then falls toward pre-industrial levels in wealthy societies. The environmental Kuznets curve model has elicited conflicting reactions from researchers and policymakers. Applied econometricians have generally ac- cepted the basic tenets of the model and focused on measuring its parameters. Their regressions, typically fitted to cross-sectional observations across countries or regions, suggest that air and water pollution increase with development until per capita income reaches a range of $5000 to $8000. When income rises beyond that level, pollution starts to decline, as shown in the "conventional EKC" line in Figure 1. In developing countries, some policymakers have interpreted such results as conveying a message about priorities: Grow first, then clean up. Numerous critics have challenged the conventional environmental Kuznets curve, both as a representation of what actually happens in the development process and as a policy prescription. Some pessimistic critics argue that cross- sectional evidence for the environmental Kuznets curve is nothing more than a Susmita Dasgupta, Benoit Laplante, Hua Wang, and David Wheelerare Economists, v DevelopmentResearchGroup,WorldBank, Washington,D. C. 148 Journal of EconomicPerspectives Figure 1 Environmental Kuznets Curve: Different Scenarios Pollution New Toxics Race to the Bottom Conventional EKC Revised EKC $ 5000 $ 8000 Income Per Capita snapshot of a dynamic process. Over time, they claim, the curve will rise to a horizontal line at maximum existing pollution levels, as globalization promotes a "race to the bottom" in environmental standards, as shown in Figure 1. Other pessimists hold that, even if certain pollutants are reduced as income increases, industrial society continuously creates new, unregulated and potentially toxic pol- lutants. In their view, the overall environmental risks from these new pollutants may continue to grow even if some sources of pollution are reduced, as shown by the "new toxics" line in Figure 1. Although both pessimistic schools make plausible claims, neither has bolstered them with much empirical research. In contrast, recent empirical work has fostered an optimistic critique of the conventional environmental Kuznets curve. The new results suggest that the level of the curve is actually dropping and shifting to the left, as growth generates less pollution in the early stages of industrialization and pollution begins falling at lower income levels, as shown by the "revised EKC"in Figure 1. The stakes in the environmental Kuznets curve debate are high. Per capita GDP in 1998 (in purchasing power parity dollars) was $1440 in the nations of sub-SaharanAfrica, $2060 in India, $2407 in Indonesia, and $3051 in China (World Bank, 2000). Since these societies are nowhere near the income range associated with maximum pollution on the conventional environmental Kuznets curve, a literal interpretation of the curve would imply substantial increases in pollution during the next few decades. Moreover, empirical research suggests that pollution costs are already quite high in these countries. For example, recent World Bank estimates of mortality and morbidity from urban air pollution in India and China suggest annual losses in the range of 2-3 percent of GDP (Bolt, Hamilton, Pandey and Wheeler, 2001). The stakes are not trivial for industrial societies, either. Those who believe in the "race to the bottom" model repeatedly advocate trade and investment restric-  SusmitaDasgupta, BenoitLaplante, Hua Wang and David Wheeler 149 tions that will eliminate the putative cost advantage of "pollution havens" in the developing world. If their assessment of the situation is correct, then industrial society faces two unpalatable options: Protect environmental gains by moving back toward autarky, but reducing global income in the process, or accept much higher global pollution under unrestrained globalization. Moreover, industrialized coun- tries surely must consider the daunting possibility that they are not actually making progress against pollution as their incomes rise, but instead are reducing only a few measured and well-known pollutants while facing new and potentially greater environmental concerns. In this paper, we review the arguments and the evidence on the position, shape and mutability of the environmental Kuznets curve. We ultimately side with the optimists-but with some reservations. Theory and Measurement of the Relationship between Economic Development and Environmental Quality Numerous theoretical and empirical papers have considered the broad rela- tionship between economic development and environmental quality. The focus of the theoretical papers has mainly been to derive transition paths for pollution, abatement effort and development under alternative assumptions about social welfare functions, pollution damage, the cost of abatement, and the productivity of capital. This theoretical work has shown that an environmental Kuznets curve can result if a few plausible conditions are satisfied as income increases in a society: specifically, the marginal utility of consumption is constant or falling; the disutility of pollution is rising; the marginal damage of pollution is rising; and the marginal cost of abating pollution is rising. Most theoretical models implicitly assume the existence of public agencies that regulate pollution with full information about the benefits and costs of pollution control. In addition, they assume that the pollution externality is local, not cross-border. In the latter case, there would be little local incentive to internalize the externality. L6pez (1994) uses a fairly general theoretical model to show that if producers pay the social marginal cost of pollution, then the relationship between emissions and income depends on the properties of technology and preferences. If prefer- ences are homothetic, so that percentage increases in income lead to identical percentage increases in what is consumed, then an increase in output will result in an increase in pollution. But if preferences are nonhomothetic, so that the pro- portion of household spending on different items changes as income rises, then the response of pollution to growth will depend on the degree of relative risk-aversion and the elasticity of substitution in production between pollution and conventional inputs. Selden and Song (1995) derive an inverted-U curve for the relationship between optimal pollution and the capital stock, assuming that optimal abatement is zero until a given capital stock is achieved, and that it rises thereafter at an 150 Journal of EconomicPerspectives increasing rate. John and Pecchenino (1994), John, Pecchenio, Schimmelpfennig and Schreft (1995), and McConnell (1997) derive similar inverted-U curves by using overlapping generations models. Recent analytical work by L6pez and Mitra (2000) suggests that corruption may also account for part of the observed relation- ship between development and environmental quality. Their results show that for any level of per capita income, the pollution level corresponding to corrupt behavior is always above the socially optimal level. Further, they show that the turning point of the environmental Kuznets curve takes place at income and pollution levels above those corresponding to the social optimum. Numerous empirical studies have tested the environmental Kuznets curve model. The typical approach has been to regress cross-country measures of ambient air and water quality on various specifications of income per capita. For their data on pollution, these studies often rely on information from the Global Environmen- tal Monitoring System (GEMS), an effort sponsored by the United Nations that has gathered pollution data from developed and developing countries. The GEMS database includes information on contamination from commonly regulated air and water pollutants. Stern, Auld, Common and Sanyal (1998) have supplemented the GEMS data with a more detailed accounting of airborne sulfur emissions. Although greenhouse gases have not been included in the GEMS database, carbon dioxide emissions estimates for most developed and developing countries are available from the U.S. Oak Ridge National Laboratories (Marland, Boden and Andres, 2001). Empirical researchers are far from agreement that the environmental Kuznets curve provides a good fit to the available data, even for conventional pollutants. In one of the most comprehensive reviews of the empirical literature, Stern (1998) argues that the evidence for the inverted-U relationship applies only to a subset of environmental measures; for example, air pollutants such as suspended particulates and sulfur dioxide. Since Grossman and Krueger (1993) find that suspended particulates decline monotonically with income, even Stern's subset is open to contest. In related work, Stern, Auld, Common and Sanyal (1998) find that sulfur emissions increase through the existing income range. Results for water pollution are similarly mixed. Empirical work in this area is proceeding in a number of directions. First, international organizations such as the United Nations Environment Programme and the World Bank are sponsoring collection of more data on environmental quality in developing countries. As more data is collected, new opportunities will open up for studying the relationship between economic development and envi- ronmental quality. In the meantime, it is useful to think about how to compensate for incomplete monitoring information. For example, Selden and Song (1994) develop estimates of air emissions based on national fuel-use data and fuel-specific pollution parameters that are roughly adjusted for conditions in countries at varying income levels. A second issue is that for many pollutants data is scarce everywhere, not just in developing countries. The GEMS effort has focused on a few "criteria"pollutants, so-designated because legal statutes have required regulators to specify their dam-  ConfrontingtheEnvironmentalKuznets Curve 151 aging characteristics. Criteria air pollutants, for example, have generally included ozone, carbon monoxide, suspended particulates, sulfur dioxide, lead and nitrogen oxide. A far broader class of emissions, known as toxic pollutants, includes mate- rials that cause death, disease or birth defects in exposed organisms. Among the hundreds of unregulated toxic pollutants that have been subjected to laboratory analysis, the quantities and exposures necessary to produce damaging effects have been shown to vary widely. Literally thousands of potentially toxic materials remain untested and unregulated. Data gathering in this area has started, as some countries have mandated public reports of toxic emissions by industrial facilities. For example, the United States has a Toxic Release Inventory; Canada has a National Pollutant Release Inventory; the United Kingdom has a Pollutant Inventory; and Australia has a National Pollutant Inventory. Using sectoral estimates of toxic emissions relative to level of output, developed from U.S. data by Hettige, Martin, Singh and Wheeler (1995), researchers have estimated toxic emissions in eastern Europe (Laplante and Smits, 1998) and Latin America (Hettige and Wheeler, 1996; Dasgupta, Laplante and Meisner, 2001). However, the underlying scarcity of data has as yet made it impossible to do more than speculate about the shape of an environmental Kuznets curve for toxics. A third empirical issue involves thinking about the curvature of the environ- mental Kuznets curve. In most cases, the implied relationship between income growth and pollution is sensitive to inclusion of higher-order polynomial terms in per capita income whose significance varies widely. Fourth, it is useful to compare the results of time series studies where the environmental evidence is available. De Bruyn, van den Bergh and Opschoor (1998) estimate time series models individually for Netherlands, Germany, the United Kingdom and the United States and show that economic growth has had a positive effect on emissions of carbon dioxide, nitrogen oxides, and sulfur dioxide. They argue that conventional cross-section estimation techniques have generated spurious estimates of the environmental Kuznets curve because they do not ade- quately capture the dynamic process involved. Given the data limitations, concerns over appropriate functional forms, and choices between cross-section and time series analysis, structural interpretations of the environmental Kuznets curve have remained largely ad hoc. In view of these uncertainties, few researchers have taken the next step and begun to study the sources of change in the marginal relationship between economic development and pollution. How the Environmental Kuznets Curve Can Become Lower and Hatter Research on the environmental Kuznets curve has suggested that its shape is not likely to be fixed. Instead, the relationship between growth in per capita 152 Journal of EconomicPerspectives income and environmental quality will be determined by how many parties react to economic growth and its side effects-including citizens, businesses, policymakers, regulators, nongovernmental organizations, and other market participants. A body of recent research has investigated these connections. The theme that emerges from this research is that it is quite plausible for developing societies to have improvements in environmental quality. It also seems likely that because of growing public concern and research knowledge about environmental quality and regula- tion, countries may be able to experience an environmental Kuznets curve that is lower and flatter than the conventional measures would suggest. That is, they may be able to develop from low levels of per capita income with little or no degradation in environmental quality, and then at some point to experience improvements in both income and environmental quality. The Primary Role of Environmental Regulation In principle, observed changes in pollution as per capita income rises could come from several different sources: shifts in the scale and sectoral composition of output, changes in technology within sectors, or the impact of regulation on pollution abatement (Grossman and Krueger, 1993). The absence of appropriate microdata across countries has precluded a systematic empirical approach to this decomposition. However, the available evidence suggests that regulation is the dominant factor in explaining the decline in pollution as countries grow beyond middle-income status. For instance, Panayotou (1997) estimates a decomposition equation for a sample of 30 developed and developing countries for the period 1982-1994. He incorporates policy considerations into the income-environment relationship while decomposing it into scale, sectoral composition and pollution intensity (or pollu- tion per unit of output) effects. His main finding, at least for ambient sulfur dioxide levels, is that effective policies and institutions can significantly reduce environ- mental degradation at low income levels and speed up improvements at higher income levels, thereby lowering the environmental Kuznets curve and reducing the environmental cost of growth. However, the estimated equation is not derived from any formal structural equation. In addition, in the absence of actual measures of environmental regulation, Panayotou uses indices of contract enforcement and bureaucratic efficiency as proxies. De Bruyn (1997) decomposes the growth- environment relationship in a sample of OECD and former socialist economies, using a divisia index methodology. Analyzing changes in sulfur dioxide pollution, he finds a significant role for environmental policy, but not for structural change in the economy. In a cross-country study of water pollution abatement, Mani, Hettige and Wheeler (2000) find that while some of the improvement in water quality with increases in per capita income is attributable to sectoral composition and technology effects, the main factor is stricter environmental regulation. There appear to be three main reasons that richer countries regulate pollution more strictly. First, pollution damage gets higher priority after society has com- pleted basic investments in health and education. Second, higher-income societies have more plentiful technical personnel and budgets for monitoring and enforce-  Susmita Dasgupta, Benoit Laplante, Hua Wang and David Wheeler 153 Figure 2 Air Pollution Regulation and Income Per Capita in 31 Countries 2.6 - $ 2.4- 2.2 - 2- 1.8- O 1.6- k 1.4- 1.2 - 1 I I I I 2 2.5 3 3.5 4 4.5 Log (Income Per Capita) Source:Dasgupta, Mody, Roy and Wheeler (2001). ment activities. Third, higher income and education empower local communities to enforce higher environmental standards, whatever stance is taken by the national government (Dasgupta and Wheeler, 1997; Pargal and Wheeler, 1996; Dean, 1999). The result of these mutually reinforcing factors, as shown in Figure 2, is a very close relationship between national pollution regulation and income per capita (Das- gupta, Mody, Roy and Wheeler, 2001). Economic Liberalization During the past two decades, many countries have liberalized their economies by reducing government subsidies, dismantling price controls, privatizing state enterprises and removing barriers to trade and investment. Easterly (2001) pro- vides strong evidence that measures of financial depth and price distortion have improved significantly for developing countries since 1980. The result has been an adjustment toward economic activities that reflect comparative advantage at undis- torted factor and product prices, which in turn can affect the level of pollution in an economy by shifting the sectoral composition. One result has been growth of labor-intensive assembly activities such as garment production. These activities are seldom pollution-intensive, although there are some notable exceptions such as electronics assembly that employs toxic cleaning solvents and fabric production that generates organic water pollution and toxic pollution from chemical dyes (Hettige, Martin, Singh and Wheeler, 1995). Another likely area of comparative advantage is information services with relatively low skill requirements, such as records maintenance for internationally distributed information-processing services. Such activities are typically not very polluting. More environmentally sensitive areas of comparative advantage include large-scale agriculture and production that exploits local natural resources such as forest 154 Journal of EconomicPerspectives products, basic metals and chemicals (Lee and Roland-Holst, 1997). These indus- tries are often heavy polluters, because they produce large volumes of waste residuals and frequently employ toxic chemicals. Elimination of government subsidies often has an environmentally beneficial effect in this context. The heaviest polluters often receive subsidies, because they operate in sectors such as steel and petrochemicals where state intervention has been common. Privatization and reduction of subsidies tend to reduce the scale of such activities, while expanding production in the assembly and service sectors that emit fewer pollutants (Dasgupta, Wang and Wheeler, 1997; Lucas, Hettige and Wheeler, 1992; Jha, Markandya and Vossenaar, 1999; Birdsall and Wheeler, 1993). Elimination of energy subsidies increases energy efficiency, shifts industry away from energy-intensive sectors, and reduces demand for pollution-intensive power (Vukina, Beghin and Solakoglu, 1999; World Bank, 1999). However, higher energy prices also induce shifts from capital- and energy-intensive production techniques to labor- and materials-intensive techniques, which are often more pollution- intensive in other ways (Mani, Hettige and Wheeler, 2000). Economic liberalization also has a common effect, at least in pollution- intensive sectors, of enlarging the market share of larger plants that operate at more efficient scale (Wheeler, 2000; Hettige, Dasgupta and Wheeler, 2000). This change often involves a shift toward publicly held firms at the expense of family firms. The improvement in efficiency means less pollution per unit of production, although larger plants may also concentrate pollution in a certain locality (Lucas, Dasgupta and Wheeler, 2001). In China, state-owned enterprises have much higher costs for reducing air pollution because they are operated less efficiently. Figure 3 displays recent econometric estimates of control costs for sulfur dioxide air pollu- tion in large Chinese factories (Dasgupta, Wang and Wheeler, 1997).1 The level of polluting emissions also reflects managers' technology decisions. In the OECD countries, innovations have generated significantly cleaner technol- ogies that are available at incremental cost to producers in developing countries. Even in weakly regulated economies, many firms have adopted these cleaner technologies because they are more profitable. Increased openness to trade also tends to lower the price of cleaner imported technologies, while increasing the competitive pressure to adopt them if they are also more efficient (Reppelin-Hill, 1999; Huq, Martin and Wheeler, 1993; Martin and Wheeler, 1992). Thus, firms in relatively open developing economies adopt cleaner technologies more quickly (Birdsall and Wheeler, 1993; Huq, Martin and Wheeler, 1993). While liberalization can certainly improve environmental conditions, it is no panacea. The evidence suggests that in a rapidly growing economy, the effect of lower pollution per unit of output as a result of greater efficiency is generally 1 Xu, Gau, Dockery and Chen (1994) have shown that atmospheric sulfur dioxide concentrations are highly correlated with damage from respiratory disease in China. Sulfur dioxide and other oxides of sulfur combine with oxygen to form sulfates and with water vapor to form aerosols of sulfurous and sulfuric acid. Much of the health damage from sulfur dioxide seems to come from fine particulates in the form of sulfates.  ConfrontingtheEnvironmentalKuznetsCurve 155 Figure3 Sulfur Dioxide Marginal Abatement Costs: Large Chinese Factories 300 250 z 200 -_/ E Non-state-owned enterprises 150 --- ;- State-owned enterprises f 100 50 0 t I It 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Rate of Abatement Source:World Bank (1999). overwhelmed by the rise in overall pollution as a result of rising output (Beghin, Roland-Holst and van der Mensbrugghe, 1997; Dessus and Bussolo, 1998; Lee and Roland-Holst, 1997). Thus, total pollution will grow unless environmental regula- tion is strengthened (Mani, Hettige and Wheeler, 2000). Pervasive Informal Regulation Low-income communities frequently penalize dangerous polluters, even when formal regulation is weak or absent. Abundant evidence from Asia and Latin America shows that neighboring communities can strongly influence factories' environmental performance (Pargal and Wheeler, 1996; Hettige, Huq, Pargal and Wheeler, 1996; Huq and Wheeler, 1992; Hartman, Huq and Wheeler, 1997). Where formal regulators are present, communities use the political process to influence the strictness of enforcement. Where regulators are absent or ineffective, nongovernmental organizations and community groups-including religious insti- tutions, social organizations, citizens' movements, and politicians-pursue infor- mal regulation. Although these pressures vary from region to region, the pattern everywhere is similar: Factories negotiate directly with local actors in response to threats of social, political or physical sanctions if they fail to compensate the community or to reduce emissions. The response of factories can take many forms. Cribb (1990) cites the case of a cement factory in Jakarta that-without admitting liability for the dust it generates-"compensates local people with an ex gratia payment of Rp. 5000 and a tin of evaporated milk every month." Agarwal, Chopra and Sharma (1982) describe a situation where, confronted by commun