5
Climate Change

OVERVIEW OF QUANTIFYING AND VALUING CLIMATE-CHANGE IMPACTS

Burning fossil fuels creates externalities through its impact on the stock of greenhouse gases (GHGs) in the atmosphere and the subsequent effects of GHG concentrations on climate. This chapter provides a general overview of these effects and various attempts that have been made to quantify and monetize the damages associated with GHG emissions. The chapter begins by summarizing information on trends in Earth’s temperature over the past century, the relationship between GHG concentrations and climate, and predictions of future changes in climate associated with various emissions trajectories. That summary is followed by an overview of the approach that economists have taken to quantifying the damages associated with GHG emissions, including a discussion of three integrated assessment models (IAMs), which provide estimates of the monetary impacts of GHG emissions. Given its resource constraints, it was not feasible for the committee to conduct a detailed critical review of the IAMs.

Estimates of the damages associated with GHG emissions in IAMs rest on estimates of the physical and monetary impacts of temperature changes in various market and nonmarket sectors. The next section of the chapter describes the physical impacts of climate change on weather, snow and ice formations, and water systems. That is followed by estimates of the physical and monetary impacts of climate change on individual market and nonmarket sectors, including water, agriculture, coastal infrastructure, health, and ecosystems. The next section discusses how monetary impacts



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 248
5 Climate Change OVERVIEW OF QUANTIFYING AND VALUING CLIMATE-CHANGE IMPACTS Burning fossil fuels creates externalities through its impact on the stock of greenhouse gases (GHGs) in the atmosphere and the subsequent effects of GHG concentrations on climate. This chapter provides a general over- view of these effects and various attempts that have been made to quantify and monetize the damages associated with GHG emissions. The chapter begins by summarizing information on trends in Earth’s temperature over the past century, the relationship between GHG concentrations and cli- mate, and predictions of future changes in climate associated with various emissions trajectories. That summary is followed by an overview of the ap- proach that economists have taken to quantifying the damages associated with GHG emissions, including a discussion of three integrated assessment models (IAMs), which provide estimates of the monetary impacts of GHG emissions. Given its resource constraints, it was not feasible for the com- mittee to conduct a detailed critical review of the IAMs. Estimates of the damages associated with GHG emissions in IAMs rest on estimates of the physical and monetary impacts of temperature changes in various market and nonmarket sectors. The next section of the chapter describes the physical impacts of climate change on weather, snow and ice formations, and water systems. That is followed by estimates of the physical and monetary impacts of climate change on individual market and nonmarket sectors, including water, agriculture, coastal infrastructure, health, and ecosystems. The next section discusses how monetary impacts 248

OCR for page 248
24 CLIMATE CHANGE reported in the literature are aggregated across sectors and countries and presents estimates of the marginal damage of a ton of carbon dioxide equivalent1 (CO2-eq) from various IAMs. The committee did not conduct its own modeling analyses of damages related to climate change. We deter- mined that attempting to estimate single values would be inconsistent with the rapidly changing nature of knowledge about climate change and the extremely large uncertainties associated with estimation of climate-change effects and damages. Climate-Change Observations, Drivers, and Future Projections According to the Intergovernmental Panel on Climate Change (IPCC), scientists have documented that Earth’s climate system is warming, the last decade was the warmest on record, global average temperatures have increased about 1.3°F since 1990, and sea levels at the end of the 20th cen- tury were rising almost twice as fast as over the century as a whole (IPCC 2007a,b).2 Arctic sea ice and glaciers are rapidly shrinking. Economic losses from extreme weather events, such as tropical cyclones, heavy rain storms, flooding, severe heat waves, and droughts, are increasing rapidly (CCSP 2008). The IPCC states that “most of the observed increase in global average temperatures since the mid-twentieth century is ery likely due to the ob- served increase in anthropogenic GHG concentrations” (IPCC 2007a, p.5). With high and increasing confidence, a range of “fingerprinting” techniques attribute a substantial fraction of recent warming to anthropogenic causes (IPCC 2007a). Although the greenhouse effect is a natural process necessary for life on Earth, humans have inadvertently intervened in this process so that the greenhouse effect is now trapping additional heat in Earth’s atmosphere, which is driving climate change. Specifically, human activities have led to a significant increase in the amount of CO2 and methane (CH4) in the atmo- sphere. These additional GHGs absorb more energy and let less heat escape to space. Therefore, Earth’s climate is warming.3 GHG emissions have steadily grown since the Industrial Revolution, with a 70% increase between 1970 and 2004. Burning fossil fuels, agri- 1 CO 2-eq expresses the global warming potential of a GHG, such as methane, in terms of CO2 quantities. 2 The IPCC is an intergovernmental scientific body given to the assessment of climate change. It does not conduct research. IPCC estimates are derived from literature reviews and assess- ments, not from its independent predictions or projections. 3 Airborne particles may have either a warming or cooling effect. Sulfate particles reflect incoming sunlight and cause a cooling effect at the surface. Other types of particles, referred to as carbon black, absorb incoming sunlight and trap heat in the atmosphere.

OCR for page 248
20 HIDDEN COSTS OF ENERGY culture, and deforestation are the primary anthropogenic sources of these GHG emissions. In 2004, the burning of fossil fuels accounted for 56.6% of the GHGs emitted. Of the total anthropogenic emissions released in 2004, energy supply produced 25.9%, transportation produced 13.1%, and industry produced 19.4% (Figure 5-1) (IPCC 2007a). Future Projections Using global climate models, scientists predict that, in the absence of concerted action to reduce GHG emissions, climate will warm substantially over the next century. The IPCC has developed scenarios that characterize a wide range of internally consistent, feasible alternative futures, charac- terized by trajectories in population, industrialization, governance, gross domestic product (GDP), and GHG emissions (IPCC 2000). By inputting these emission scenarios into global climate models, scientists have devel- oped sophisticated estimates of what atmospheric temperatures could look like in 2100 (Figure 5-2). If carbon concentrations were kept constant at the level produced in 2000, these models predict that Earth’s climate would continue to warm (see Figure 5-2). Scenario A2 describes a heterogeneous world with a focus on self-reliance and regional identity and having relatively slow economic and technology growth. This scenario ends the 21st century with very high emissions and dramatic warming. Scenario A1B describes a future with rapid economic growth and human population that peaks around 2050 and then starts to decline. This scenario assumes significant interregional cooperation and a balanced portfolio of energy sources. A1B predicts continued warming that starts to slow by 2100. Scenario B1 describes the same population and economic trends as in scenario A1B. However, B1 incorporates a rapid shift toward a service and information economy, reduced material intensity, and the widespread adoption of efficient low- carbon energy technologies. B1 predicts a less dramatic increase in global average temperatures. Since 2000, industrial carbon emissions have increased more rapidly than in any of the scenarios (Raupach et al. 2007). Moreover, natural feed- back processes, such as melting permafrost and more extensive wild fires, are releasing carbon into the atmosphere more quickly than anticipated (IPCC 2007b). On the other hand, as of mid-2009, the carbon budget data have not yet been updated to reflect changes resulting from the global eco- nomic crisis of 2008-2009. The U.S. Global Change Research Program (GCRP) concluded that climate-related changes are already under way in the United States and sur- rounding coastal waters, and the quantity and growth rate of these changes are dependent upon human choices in the present day (Karl et al. 2009).

OCR for page 248
49.0 44.7 39.4 35.6 28.7 GtCO2-eq / yr CO2 from fossil fuel use and other sources CO2 from deforestation, decay and peat F-gases CH4 from agriculture, waste and energy N2O from agriculture and others FIGURE 5-1 Global anthropogenic greenhouse gas (GHG) emissions. (a) Global annual emissions of anthropogenic GHGs from 1970 to 2004; (b) share of different anthropogenic GHGs in total emissions in 2004 in terms of CO2-equivalent; and (c) share of different sectors in total anthropogenic GHG emissions in terms of CO2-equivalent (forestry includes deforestation.). SOURCE: IPCC 2007a, p. 5, Fig. SPM.3. Reprinted with permission; copyright 2007, Intergovernmental Panel on Climate Change. 21

OCR for page 248
22 FIGURE 5-2 Atmosphere-Ocean General Circulation Model (AOGCM) projections of surface warming. Left panel: Solid lines are multimodel global averages of surface warming (relative to 1980-1999) in the IPCC Special Report on Emission Scenarios (SRES) A2, A1B, and A1, shown as continuations of the 20th century simulations (IPCC 2000). The orange line is for the experiment where concentrations were held constant at year 2000 values. The bars in the middle of the figure indicate the best estimate (solid line within each bar) and the probable range assessed for the six SRES marker scenarios at 2090-2099 relative to 1980-1999. The assessment of the best estimate and probable ranges in the bars includes the AOGCMs in the left part of the figure, as well as results from hierarchy of independent models and observational constraints. Right panels: Projected temperature changes for the early and late 21st century relative to the period 1980-1999. The panels show the multi-AOGCM average projections for the A2 (top), A1B (middle), and B1 (bottom) SRES averaged over decades 2020-2029 (left) and 2090-2099 (right). SOURCE: IPCC 2007b, p. 14, Figure SPM.5; Right Panel: IPCC 2007b, p. 15, SPM.6. Reprinted with permission; copyright 2007, Intergovernmental Panel on Climate Change.

OCR for page 248
23 CLIMATE CHANGE IPCC Mitigation Findings The IPCC concluded that the impacts of climate change can be reduced, delayed, or avoided through mitigation strategies designed to stabilize at- mospheric carbon concentrations. These concentrations can be stabilized primarily by reducing anthropogenic carbon emissions and, secondarily, by increasing carbon sinks (see Table 5-1). Figure 5-3 depicts the future carbon emission profiles needed to achieve the various stabilization concentrations and the global mean temperature associated with each stabilization concen- tration. The IPCC strongly suggests that the technology needed to achieve the needed stabilization levels is already or will very soon be available. They also claim that 60-80% of the needed emission reductions would have to come from the energy sector, via a shift to noncarbon-based energy sources and energy efficiency (IPCC 2007a,d) In response to a request from Congress, the National Research Council (NRC) has undertaken America’s Climate Choices (ACC), a suite of studies designed to inform and guide responses to climate change across the nation. A final ACC report, addressing strategies to reduce or adapt to the impacts of climate change, is expected to be complete in 2010. Overview of Quantification Methods, Key Uncertainties, and Sensitivities Defining the Marginal Damage of GHG Emissions The combustion of fossil fuels is a major source of GHG emissions, which create externalities through their impact on the stock of GHGs in the atmosphere and the subsequent effects of GHG concentrations on climate. Evaluating the external costs of energy due to climate change is a daunting task. The principal difficulty is the complexity arising from the fundamental dimensionality of the climate problem. The relevant dimensions are time (indexed by t), location (indexed by l), the set of relevant climatic variables (indicated by the index c), the categories of physical impacts as a result of climate changes (indexed by i), and the categories of damage incurred by these impacts (indexed by d). The external cost of an additional ton of GHGs, E0, emitted at time t = 0 depends on the following: a. The effect of emissions on Earth-system processes and, in turn, climatic variables in candidate locations over future time periods, Cc,l,t. b. The contemporaneous effect of climate changes in each location on various categories of physical impacts, Ii,l,t. c. The contemporaneous effect of impacts in each location on various categories of damage, Dd,l,t.

OCR for page 248
TABLE 5-1 Characteristics of Post-Third Assessment Report (TAR) Stabilization Scenarios and Resulting Long- 24 Term Equilibrium Global Average Temperature and the Sea-Level Rise Component from Thermal Expansion Only Change in Global Average Global Average Sea- CO2-Equivalent Concentration Temperature Increase Level Rise above CO2 Global CO2 Concentration at Stabilization Emissions in above Preindustrial Preindustrial Level at Stabilization Including GHGs Peaking Year 2050 (Percent Level at Equilibrium, at Equilibrium from Number of (2005 = 379 and Aerosols of 2000 Using “Best Estimate” Thermal Expansion Assessed for CO2 Category ppm)b (2005 = 375 ppm)b Emissionsa,c Emissions)a,c Climate Sensitivityd,e Onlyf Senarios ppm ppm Year Percent °C Meters I 350-400 445-490 2000-2015 −85 to −50 2.0-2.4 0.4-1.4 6 II 400-440 490-535 2000-2020 −60 to −30 2.4-2.8 0.5-1.7 18 III 440-485 535-590 2010-2030 −30 to +5 2.8-3.2 0.6-1.9 21 IV 485-570 590-710 2020-2060 +10 to +60 3.2-4.0 0.6-2.4 118 V 570-660 710-855 2050-2080 +25 to +85 4.0-4.9 0.8-2.9 9 VI 660-790 855-1130 2060-2090 +90 to +140 4.9-6.1 1.0-3.7 5 aThe emission reductions to meet a particular stabilization level reported in the mitigation studies assessed here might be underestimated because of missing carbon-cycle feedbacks. bAtmosphere CO concentrations were 379 parts per million (ppm) in 2005. The estimate of total CO -equivalent concentration in 2005 for all 2 2 long-lived greenhouse gases (GHG) is about 455 ppm, and the corresponding value including the net effect of all anthropogenic forcing agents is 375 ppm CO2-eq. cRanges correspond to the 15th to 85th percentile of the post-TAR scenarios distribution. CO emissions are shown so multigas scenarios can be 2 compared with CO2-only scenarios. dThe best estimate of climate sensitivity is 3°C. eNote that global average temperature at equilibrium is different from expected global average temperature at the time of stabilization of GHG concentrations due to the inertia of the climate system. For the majority of scenarios assessed, stabilization of GHG concentrations occurs between 2100 and 2150. fEquilibrium sea-level rise is for the contribution from ocean thermal expansion only and does not reach equilibrium for at least many centuries. These values have been estimated using relatively simple climate models (one low-resolution AOGCM [Atmosphere-Ocean General Circulation Model] and several EMICs [Earth System Models of Intermediate Complexity] based on the best estimate of 3°C climate sensitivity) and do not include contributions from melting ice sheets, glaciers, and ice caps. Long-term thermal expansion is projected to result in 0.2 to 0.6 m per degree Celsius of global average warming above preindustrial levels. SOURCE: IPCC 2007a, p. 20, Table SPM.6. Reprinted with permission; copyright 2007, Intergovernmental Panel on Climate Change.

OCR for page 248
FIGURE 5-3 Global CO2 emissions for 1940 to 2000 and emission ranges for categories of stabilization scenarios from 2000 to 2100 (left panel); and the corresponding relationship between the stabilization target and the probable equilibrium global average temperature increase above preindustrial levels (right panel). Approaching equilibrium can take several centuries, especially for scenarios with higher levels of stabilization. Colored shadings show stabilization scenarios grouped according to different targets (stabilization categories I to VI). The right-hand panel shows ranges of global average temperature change above preindustrial levels, using (i) “best-estimate” climate sensitivity of 3°C (black line in middle of shaded area), (ii) upper bound of probable range of cli- mate sensitivity of 4.5°C (red line at top of shaded area), and (iii) lower bound of probable range of climate sensitivity of 2°C (blue line at bottom of shaded area). Black dashed lines in the left panel give the emissions range of recent baseline scenarios published since IPCC (2000). Emission ranges of the stabilization scenarios comprise CO2-only and multigas scenarios and correspond to the 10th to 90th percentile of the full scenario distribution. NOTE: CO2 emissions in most models do not include emissions from decay of above-ground biomass that remains after logging and deforestation and from peat fires and drained peat soils. SOURCE: IPCC 2 2007a, p. 21, Figure SPM.11. Reprinted with permission; copyright 2007, Intergovernmental Panel on Climate Change.

OCR for page 248
26 HIDDEN COSTS OF ENERGY The dependence can be summarized in Equation 5-1, which is the analogue of the impact-pathway approach used in the analysis in Chapters 2 and 3: , Equation 5-1 where δt = (1 + r)−t is a factor that discounts damages in future year t back to the present, and r is the rate of discount. Effect b is captured by the term ΔCI, which summarizes the results of physical impact models. These models suggest how changes in temperature and precipitation may affect agricul- tural yields or how changes in climate will affect biodiversity. Effect c is captured by the term ΔID, which captures the monetary damages associated with changes in agricultural yields or loss of species diversity. Attempts to measure these damages are reviewed briefly later in this section. In many cases, the relationships between climate impacts and damages are based on judgment, assumptions, or analogy because data are lacking. This last point highlights the second difficulty facing any assessment of the costs of climate change: lack of information and uncertainty regarding effects a-c. The terms represented by ΔEC, which include the extent of ice sheet melting and shifts in regional distribution of precipitation are still subject to considerable uncertainty. A vast amount of effort is actively be- ing dedicated to elaborating the elements of ΔCI. However, while natural scientists mostly agree on the climatic variables whose impacts should be studied, there is less consensus on what impacts are significant and should be examined. Moreover, even those impacts thought to be significant (for example, species loss) respond to changes in climate in ways that are poorly understood. The difficulties in estimating monetary damages are described in more detail below. The climate equivalent of the Air Pollution Emission Experiments and Policy (APEEP) model would embody estimates of ΔEC, ΔCI, and ΔID and combine them according to Equation 5-1 to produce a summary measure of marginal damage. The committee did not have access to an integrated assessment model (IAM) of this kind. Indeed, such a model does not ex- ist—the IAMs used for climate studies are designed not to produce de- scriptively realistic, spatially disaggregate responses of climatic impact and damage variables but rather to bring together key stylized facts about these responses within the framework of Eq. 5-1 as a means of elucidating their joint implications. The benefits of such integration are often gained at the expense of introducing substantial theoretical and empirical weaknesses into IAMs. Although each element of ΔEC, ΔCI, and ΔID can be thought of as a model in its own right, IAMs adopt reduced-form approaches, which

OCR for page 248
2 CLIMATE CHANGE reduce these complex relationships into simplified response surfaces, in the process oversimplifying the complexities of the underlying science. In ad- dition, IAMs typically cope with the curse of dimensionality by consider- ing only relatively narrow sets of impacts or types of damage. IAMs also tend to trade off coarse regional coverage in favor of broad global scope, so relationships validated for restricted geographic domains are implicitly scaled up to broader spatial scales. The remainder of this section describes in general terms how IAMs, such as the Regional Integrated Model of Climate and the Economy (RICE) and the Dynamic Integrated Model of Climate and the Economy (DICE), the Climate Framework for Uncertainty, Negotiation, and Distribution (FUND) model, and the Policy Analysis of the Greenhouse Effect (PAGE) model, evaluate marginal damages. In a typi- cal IAM, the marginal damages from a ton of CO2-eq emissions (E) emitted today (year 0) can be expressed as , Equation 5-2 where = year of impact, t MD0 = marginal damage from GHG emissions in year 0 ($/ton CO2-eq), = mean global temperature in year t relative to preindustrial levels Tt (°C), E0 = GHG emissions in year 0 (ton CO2-eq), Dt = total climate damages in year t ($), δt = discount factor from year t to 0 = (1 + r)–t, where r is the discount rate, and = final year for which climate damages are included. tf The expression indicates that the marginal damages from GHG emissions (MD) depend on how much temperatures increase in response to a unit increase in emissions (dT/dE), how much additional climate damage results from this temperature increase (dD/dT), how one values future damages rel- ative to the present (δ), and how far into the future one aggregates impacts (tf). In terms of the preceding discussion, climate effects have been reduced to temperature, and the link between climate and impacts and impacts and damages has been condensed into a single step. The relationship between a ton of CO2-eq emitted today and future temperature depends on the effects of GHG emissions on the concentration of GHGs in the atmosphere, and on the effects of GHG concentrations on temperature. Spatially detailed predictions of the impact of GHG concentra- tions on temperature and precipitation are provided by general circulation

OCR for page 248
28 HIDDEN COSTS OF ENERGY models. IAMs typically simplify these relationships and describe changes in mean global temperature corresponding to a ton of CO2-eq emissions. In the most disaggregated IAMs, the monetary damages associated with a change in temperature are calculated by estimating damages by sector (for example, energy, health, and agriculture) and geographic region. Dam- ages are expressed as a percentage of GDP using methods described in the next subsection: Approach to Measuring Marginal Damages in Integrated Assessment Models. dDt/dTt represents the aggregation of impacts across sectors and regions. How change in GDP are aggregated across regions— whether using equity weights or by summing the monetary changes in GDP—is discussed below. Future monetary damages are discounted either at the market rate of interest (the revealed preference approach), or using the Ramsey formula (the prescriptive approach), which describes how the discount rate, r, var- ies along an optimal growth path (see Pearce et al. 2003). These two ap- proaches are discussed in detail later in the chapter. Marginal damages are extremely sensitive to the choice of discount rate, given the fact that the climate impacts of a ton of CO2-eq emissions will be felt for centuries (tf is typically 100 to 300 years). The marginal-damage formula in Equation 5-2 assumes that the effect of a ton of CO2-eq emissions on temperature and the effects of temperature on the economy are certain—which are clearly not the case. Indeed, a major difference between quantifying the local air pollution effects of fossil fuels and the impacts of GHG emissions is that the two differ significantly in their time dimension, their spatial scale, the variety of impacts, and, hence, in the certainty with which they can be estimated. In contrast to SO2 or NOx, CO2 is a pollutant that resides in the atmosphere for centuries.4 This factor implies that the effects of a ton emitted today must be estimated on a time scale (centuries) in which the state of the world is inherently more uncertain than the period during which effects of local air pollutants are estimated (months or years). Key sensitivities in Equation 5-2 include the impact of a change in atmospheric concentrations of CO2 on temperature (termed climate sensitivity) and how dD/dT varies with T. There is, in reality, a distribution of damages associated with any given temperature change. IAMs typically handle this uncertainty in two ways. One is to calcu- late marginal damages using Monte Carlo analysis: Parameters used for dT/dE and dD/dT are drawn from probability distributions and used to calculate the corresponding distribution of marginal damages. A second approach is to acknowledge that, corresponding to each change in mean global temperature from pre-industrial levels, there is a probability of 4 Theatmospheric lifetime of CO2 is complex. About half disappears in 40 years, but about 20% remains in the atmosphere for many centuries, essentially indefinitely.

OCR for page 248
28 HIDDEN COSTS OF ENERGY FIGURE 5-9 Dependence of GHG damage on the amount of temperature change. The lines show the PAGE 2002 damages for damage exponents between 1 and 3. The damage function of the DICE model is also shown for comparison. In this fig- ure, positive values indicate economic losses, and negative values indicate benefits from warming. SOURCE: Stern 2007, Technical Appendix. associated range of global impact that varies by a factor of almost 6 for a 6°C temperature increase (see Figure 5-9, where the PAGE damage function for 6°C is about 2.4 times the level at 2.5°C for linear damages, while it is about 14 times as high assuming cubic damages). Yet, in the absence of substantial mitigation action, projections of baseline GHG emissions tend to imply estimates of likely temperature in- crease that are significantly greater than that associated with a doubling of GHG concentrations. For example, the IPCC (2007a, p. 180, Figure 5.1) referenced plausible projections of GHG concentrations that go near to and beyond 1,000 ppm by 2100, with an associated best estimate global mean temperature increase above preindustrial levels of about 5-6°C and a likely range from just under 4°C to over 8°C. However, little is known about the precise shape of the temperature-damage relationship at such high tem- peratures.17 Figure 5-10 illustrates the dependence of GHG damage, as a percentage of global GDP, on the amount of temperature change. 17 See discussion in Stern (2007, pp. 659-662).

OCR for page 248
color R01631 uneditable bitmapped image scaled for portrait above, landscape below FIGURE 5-10 Dependence of greenhouse gas (GHG) damage, as a percentage of global gross domestic product (GDP), on the amount of temperature change. (a) Damage estimates represented as a percentage of global GDP, as a function of increases in global mean temperature. (b) Damage estimates, as a percentage of global GDP per capita, are correlated with increases in global mean temperature. In this figure, positive values indicate benefits from warming. SOURCE: IPCC 2001, p. 1032 and Stern 2007 (as cited in Yohe et al. (2007, p. 822). 2

OCR for page 248
300 HIDDEN COSTS OF ENERGY MARGINAL IMPACTS OF GREENHOUSE GAS EMISSIONS Given an estimate of the monetized global impact of a particular climate-change scenario at a particular future point in time, this total- damage estimate can then be translated into a marginal damage per ton estimate (often called the “social cost of carbon”) by evaluating the linkage between current GHG emissions and future climate-change impacts (see Equation 5-2). It is usually estimated as the net present value of the impact over the next 100 years (or longer) of 1 additional ton of CO2-eq emitted into the atmosphere. It is this marginal damage per ton of emissions that is normally used as a measure of the global climate externality. This measure requires assumptions about the emissions-temperature and temperature- damages linkages over time, as well as the rate at which future damages are discounted back to the present to account for differing valuation of monetary impacts felt at different points in time. Finally, uncertainties at each step of the analysis imply that different possible future conditions may yield widely differing impacts. The expected value of damages may be more sensitive to the possibility of low-probability catastrophic events than to the most likely or best-estimate values. There have been many previous reviews of existing estimates of the marginal damages from GHGs, including Pearce et al. (1996), Tol (1999, 2005b, 2008), Clarkson and Deyes (2002), and Yohe et al. (2007). Tol (2008) identifies 211 marginal-damage estimates from 50 studies, although this number does not imply we know more about marginal than total damages (there are only 12 global total-damage estimates, as shown in Ta- ble 5-7). The explanation for how so many marginal costs can be generated from so few total-damage estimates lies in the variety of additional model- ing assumptions that must be incorporated to translate total into marginal damages. As alluded to above, in addition to the benchmark estimate of total damages, important other assumptions include the change in damages with increased warming and with growth and changes in the composition of economic activity over time, the assumed emissions scenario, the climate sensitivity to GHG concentrations, the rate used to discount future impacts to the present, the timeframe over which impacts are considered, and the treatment of uncertainty and risk aversion. Box 5-2 discusses approaches used to determine a discount rate. Pearce et al. (1996, p. 215, Table 6.11) summarized early estimates of marginal GHG damages, which ranged from $3 to $62 per ton of CO2-eq for emissions occurring in the 2001-2010 decade.18 As part of a United Kingdom effort to assess the social cost of carbon, Clarkson and Deyes (2002) suggested a pragmatic approach could be to use a central 18 IAM results usually include CO as the only GHG. Tol (2005b) refers to a cost per tonne 2 of carbon.

OCR for page 248
301 CLIMATE CHANGE estimate of $35 per ton of CO2-eq, along with a sensitivity range of half and double this amount ($17-70 per ton of CO2-eq). Tol (2005b) identified 103 marginal-climate-damage estimates from 28 published studies, finding a median estimate of $4 per ton CO2, a mean of $25 per ton CO2, and a 95th percentile of $96 per ton CO2-eq across the estimates. Tol (2005b) also found that the subset of studies published in peer-reviewed journals reported lower estimates on average, with a mean of $12 per ton CO2-eq. (The Tol [2005b, 2008] values are not adjusted for inflation.) Summariz- ing 211 estimates identified in Tol (2008) yields a median estimate of $8 per ton CO2-eq, a mean of $29 per ton CO2-eq, a 5th and 95th percentile of $0 and $105 per ton CO2-eq, respectively, and a peer-reviewed mean of $14 per ton CO2-eq (University of Hamburg 2009). In cases where a single study generated multiple estimates, Tol (2008) included a relative weight for each estimate that was provided by the author of each study. Using these weights, one can construct a single weighted estimate for each of the 50 studies. Summarizing these 50 estimates from individual studies yields a median estimate of $10 per ton CO2eq, a mean of $30 per ton CO2-eq, and a 5th and 95th percentile of $1 and $85 per ton CO2-eq, respectively. Note, however, that due to the lack of necessary information, Tol did not adjust individual estimates for inflation, nor did he account for the timing of emissions (that is, the year they occur) or the GHG concentration and temperature scenario onto which those emissions are added. Adjusting for inflation from the study year to current dollars would make these figures higher. The underlying estimates also differ in terms of their assumed dis- count rates and how they aggregate regional impacts (using output, equity, or population weighting), among other factors. To provide a more consistent comparison of marginal-damage esti- mates, it is helpful to focus on estimates using the most widely used impact assessment models, DICE, FUND, and PAGE, as shown in Table 5-9. The estimates represent the marginal damages from current emissions against an assumed reference case climate scenario without GHG mitigation. This subset of estimates spans approximately the same range as discussed above, from roughly 0 to $100 per ton of CO2-eq. The table demonstrates that virtually all of the variation can be understood as a function of differences across the studies in what is assumed about the discount rate and the mag- nitude of GDP losses expected from uncontrolled warming. Nordhaus and Yang (1996) made the same point, noting that “the two crucial parameters are the discount rate (which indicates the relative importance of the future compared to the present) and the damages from climate change (which measure the willingness to pay to prevent or slow climate change). It is interesting to note that both major uncertainties inole human preferences rather than pure questions of ‘fact’ about the natural sciences” (emphasis in original).

OCR for page 248
302 HIDDEN COSTS OF ENERGY BOX 5-2 Discounting and Equity Weighting Quantifying the damages from GHG emissions requires aggregation of dam- ages that occur at different times extending centuries into the future and to dif- ferent populations across the globe at each point in time. The method chosen for aggregation has implications for how effects on different people are weighed. Two methods for aggregating effects on different people are common: using monetary and utility measures. The monetary measure assumes that $1 of benefit to one person is equally as good as $1 of benefit to another. The utility measure assumes that the gain in utility (or well-being) from receiving $1 is larger for a poor person than a rich person because the poor person is likely to have more pressing needs. Aggregating across people using the monetary measure is straightforward: One simply sums the monetary values of benefits and harms across the relevant population. To implement the utility-based approach, one needs to make some assumptions about how individual utility varies with income (or wealth). Often, it is assumed that utility is proportional to the logarithm of income or to a power function of income, where the power is less than 1. These functions have the property that utility increases with income but at a diminishing rate. After choos- ing a function, one can weight the monetary value of benefits and harms to each person by the incremental utility of income and sum these values. This “equity weighting” gives more weight to the same monetary value of damages when they are suffered by a poor person rather than a rich person. For aggregating effects across time, it is conventional to discount the mon- etary value of future effects by a factor of [1/(1 + r)]t that depends on the discount rate r and number of years in the future t at which the effect occurs. The present value of a stream of effects occurring at various times in the future is calculated by summing the discounted monetary values of the effects. In determining the appropriate discount rate to use for aggregating effects on the current and some future generation, one can distinguish between descriptive and prescriptive ap- TABLE 5-9 Indicative Marginal Global Damages from Current GHG Emissions ($/Ton CO2-eq) Damages from Benchmark Warming Discount Rate Relatively Low Higher 1.5% 10 100 3.0% 3 30 4.5% 1 10 NOTE: Only order-of-magnitude estimates appear warranted.

OCR for page 248
303 CLIMATE CHANGE proaches. The descriptive approach infers the rate at which society chooses between consumption at different times from market interest rates. In contrast, the prescriptive approach derives the appropriate discount rate on monetary values (the consumption discount rate) as the sum of utility and growth discount rates. The utility discount rate is the rate at which the future generation’s utility is discounted relative to the present generation’s. Many scholars have suggested that it is inappropriate to value other people’s well-being less simply because they come later in time and so argueed for a utility discount rate of zero. The growth discount rate accounts for differences in income between the current and future generation. If the future generation will have greater income than the current generation, it will lose less utility from $1 of damages than the current generation will. To aggregate the effects on utility, it is necessary to down-weight the monetary value of the damages to the future generation, just as one would down-weight the monetary value of effects on rich people at the same point in time in accordance with equity weighting. The extent of this growth discounting effect depends on the economic growth rate (that determines the difference in income between the two generations) and the utility function (that determines how much the incremental effect of income on utility falls). If the future generation is poorer than the present, the growth discounting effect will apply in the opposite direction and will give greater weight to the monetary value of damages suffered by the future generation. Following the prescriptive approach, Stern (2007) adopt a near-zero utility discount rate of 0.1% per year, a relatively small value of the rate at which the incremental effect of income on utility falls to 1 (corresponding to a logarithmic utility function), and a low rate of economic growth, 1.3% per year. Together, these yield a consumption discount rate of 1.4%. In contrast, Weitzman (2007a,b) suggests that more plausible values are roughly 2%, 2, and 2%, yielding a much larger consumption discount rate of 6%. Nordhaus (2008) uses the descriptive approach; he calibrates his model parameters so that the consumption discount rate is consistent with market interest rates, yielding a discount rate of 4.5%. Perhaps the clearest illustration of the influence of the discount rate is a comparison of the “no-control” relatively high (4.5%) discount rate scenario of Hope (2006) and the low discount rate (1.4%) of Stern (2007), which yield marginal damage estimates of $6 and $102 per ton, respec- tively: a 17-fold difference. Both studies used the same version of the PAGE model, so the only significant difference in assumptions is the discount rate. When Hope and Newbery (2008) applied approximately the same discount rate as Stern (2007) to the PAGE model, they found a similar marginal- damage estimate of $108 per ton CO2. Similarly, the Nordhaus (2008)

OCR for page 248
304 HIDDEN COSTS OF ENERGY estimate of $8 per ton CO2, which also used a relatively high discount rate of about 4.5%, is quite close to the estimate of Hope (2006) using a 4.5% discount rate. Finally, when Nordhaus (2008) applied low discount rates similar to Stern’s to the DICE model, he found a marginal-damage estimate similar in magnitude to Stern’s ($88 per ton CO2-eq). The rate at which future damages from current emissions are converted to present values in Stern (2007) is only slightly greater than the rate at which damages for an incremental increase in global temperature are pro- jected to grow over time. In Stern (2007), damages are a fraction of world GDP per capita that depends on climate change. The rate at which GDP per capita is assumed to grow (1.3% per year) is nearly as large as the dis- count rate (1.4% per year). With impacts rising almost as fast as they are being discounted, it is primarily the limited time horizon (2200 in the PAGE model) that constrains the marginal-damage estimate from becoming virtu- ally unbounded, given that the effects of current emissions on climate will persist for centuries. In contrast, the discount rates assumed in Nordhaus (2008) and Hope (2006) are high enough that even after accounting for these additional growth effects the present value of damages in the distant future is low. The growth in incremental damages over time underpins the rationale for a marginal (per ton) GHG damage that rises over time. For example, in the PAGE model, marginal damages rise by about 2.4% per year (Hope and Newbery 2008), and in the DICE model, marginal damages rise by about 2.0% per year (Nordhaus 2008). Over a 20-year period (for example, from 2010 to 2030), marginal damages rising at a rate of 2-3% per year would increase in total by a factor of 50-80%. This estimate is due to a combination of a larger economy being affected and increasing proportion- ate impacts of increasing temperatures (that is, nonlinearity of the damage function). The marginal damages from current emissions do not decrease ap- preciably for alternative scenarios with significantly lower GHG emissions and temperature increases in Nordhaus (2008) or in Hope and Newbery (2008). According to Hope and Newbery (2008), this finding is due to convexity of the damage function being roughly offset by concavity in the concentration-temperature relationship, which is logarithmic. Given that Stern also uses the PAGE model, it is surprising that Stern (2007) found that marginal damages fall dramatically from $102 per ton under a no-control scenario to $36 per ton CO2-eq under a 550-ppm stabilization scenario. Although Stern (2007) does not provide an explanation for the derivation of these results, it appears to be a result of the much lower discount rate assumed in Stern, which gives higher weight to future damages and which are much lower in a stabilization than no-control scenario. In contrast, Nordhaus (2008) and Hope (2006) used significantly higher discount rates

OCR for page 248
30 CLIMATE CHANGE where these future damages (or lack thereof) matter less. One implication is that even low discount rate scenarios that give rise to high marginal dam- ages with no climate mitigation may be consistent with substantially lower marginal-damage estimates (and corresponding Pigouvian emission prices) if, in fact, controls are undertaken. Put differently, even if one accepts marginal-damage estimates on the order of $100 per ton, the implication is not that emission prices at this level would be efficient. At the other end of the range in Table 5-8 are the estimates from the FUND model. These estimates also demonstrate the importance of the dis- count rate for present value marginal GHG damages, implying that GHG emissions move from having negligible effects (and in some scenarios posi- tive benefits) with relatively high discounting of 5%, to a larger impact of $6 per ton CO2-eq with relatively low discounting of 2%. The generally lower estimates of FUND are clearly due to the assumed damage function, which specifies benefits to global GDP up until about a 2-2.5°C of warming. Even after this point, damages do not go much beyond about 1% of lost GDP even for large temperature increases, in contrast to the other models where damages increase nonlinearly. The marginal damages of GHG emissions may be highly sensitive to the possibility of catastrophic events. Although a number of potentially catastrophic outcomes have been identified (for example, release of methane from permafrost that could rapidly accelerate warming, collapse of the West Antarctic or Greenland ice sheets raising sea level by several meters, and changes in North Atlantic currents that would dramatically alter European climate), the damages associated with these events and their probabilities are very poorly understood. Nordhaus and Boyer (2000) and Stern (2007) included some provision for catastrophic outcomes that could result in the loss of perhaps one-quarter of world GDP. Weitzman (2009) raised the even more sobering possibility that the probabilities of extreme outcomes are much larger than currently estimated. If taken into account, low-probability extreme outcomes, such as the possibility of a 10° or even 20°C increase in global mean temperature that could virtually destroy civilization as we know it, could dominate the expected value of damages, making it much greater than the values described above.19 Given the uncertainties and the still preliminary nature of the climate- damage literature, the committee finds that only rough order-of-magnitude estimates of marginal climate damages are possible at this time. Depending on the extent of future damages and the discount rate used for weighting future damages, the range of estimates of marginal global damages can vary by two orders of magnitude, from a negligible value of about $1 per ton to $100 per ton of CO2-eq. Roughly an order of magnitude in difference 19 For further discussion and alternative view, see Aldy et al. (2009).

OCR for page 248
306 HIDDEN COSTS OF ENERGY can be attributed to discounting assumptions, and another to assumptions about future damages from current emissions. Table 5-9 summarizes these findings for discount rates of 1.5%, 3.0%, and 4.5%, respectively, and for relatively low and higher climate-damage assumptions (corresponding roughly to FUND-level damages versus DICE- or PAGE-level damages). For a discount rate of about 3%—a typical rate for use in long-term en- vironmental analysis in the United States and elsewhere—the comparable marginal-damage estimates could be on the order of about $3 per ton to $30 per ton CO2-eq for relatively low versus higher damage assumptions.20 As discussed earlier, however, the damage estimates at the higher end of the range are associated only with emission paths without significant GHG controls. Therefore, care must be taken in translating these estimates for use in policies for decreasing GHG emissions. In Stern (2007), for example, the marginal-damage estimate is $36 per ton CO2-eq for a stabilization trajec- tory associated with stabilization at about 550 ppm CO2-eq, not the $102 per ton Stern found associated with uncontrolled emissions. As described above, marginal-damage estimates for emissions in 2030 could be as much as 50-80% larger than those estimates. Estimates of the damages specifically to the United States would be a fraction of those levels because the United States is only about one-quarter of the world’s economy, and the proportionate impacts on the United States are generally thought to be lower than for the world as a whole (see Table 5-7). Table 5-10 presents three different estimates of external global damages from GHG emissions on a per unit basis. The damages were calculated by multiplying GHG emission rates from Chapters 2 (electricity), 3 (transpor- tation), and 4 (heat) by each of the committee’s assumed low, middle, and high marginal damages of $10, $30, and $100 per ton CO2-eq. In conclusion, the committee finds that the relative weight placed on potential impacts occurring decades to centuries in the future is absolutely central to the determination of a present value measure of the damages from current GHG emissions. Over these time horizons, the discount rate carries with it implications for intergenerational distribution. As with any social analysis involving significant distributional impacts, it is therefore 20 To gain a rough sense for how marginal damages change as a function of growth and discounting, it is useful to consider the relative magnitude of the present value of a growing stream of damages discounted at different rates. As mentioned above, a typical climate eco- nomic model might imply marginal damages growing over time at about 2% per year due to economic growth and a convex damage function. Accumulated over several hundred years, the present value of a stream of damages growing at 2% per year increases by a factor of 2.5 using a discount rate of 3% rather than 4.5%. Using a discount rate of 1.5%, the cumulative value of a stream of damages growing at 2% per year is only bounded by the time horizon of the sum. As another point of reference, studies cited in reviews by Tol (2005b, 2008) using dis- counts rates of 3% also show a mean marginal damage in the range of $30 per ton CO2-eq.

OCR for page 248
30 CLIMATE CHANGE TABLE 5-10 Illustration of Ranges of Climate-Related Damages for Selected Categories of Energy Use in the United States, 2005 Climate-Related Damages for $10-30-100/Ton CO2-eqa Sector Fuel and Technology Electricity Coal plants (biomass)b 1-3.0-10 ¢/kWh Natural gas plants 0.5-1.5-5 ¢/kWh Nuclear, wind, solar Much lower than natural gas Transportation Cellulosic E85/car 0.02-(0.15-0.25)-2 ¢/VMT CNG 0.04-0.4-4.0 ¢/VMT Gasoline hybrid 0.04-0.4-4.0 ¢/VMT Gasoline/car 0.06-0.6-6.0 ¢/VMT E10 0.06-0.6-6.0 ¢/VMT H2(g) 0.03-0.3-3.0 ¢/VMT Diesel/car 0.05-0.5-5.0 ¢/VMT E85 corn/car 0.05-0.5-5.0 ¢/VMT Grid-dependent HEV or EVc 0.05-0.5-5.0 ¢/VMT Building and Industrial for Heating Natural gas combustiond 0.07-0.7-7.0 $/MCF aRounded to one digit, 2007 USD. bBiomass can be co-fired with coal in quantities up to about 20%. cRanges based on use of the fuel in a representative group of vehicles. Grid-electric cars are usually smaller than fleet average cars, so their better performance per vehicle mile traveled (VMT) is also dependent on use of smaller cars with lesser driving ranges. dFuture additions to supplies may include imported liquified natural gas, which will include nonclimate damages outside the United States at the source and will have increased climate damages in the range of 30% or more depending on the gas field and the liquefaction plant details. ABBREVIATIONS: CNG = compressed natural gas; HEV = hybrid electric vehicle: EV = electric vehicle; MCF = thousand cubic feet. crucial for decision makers not only to look at singular summary statistics (such as present value marginal damages) but also to understand the mag- nitude of impacts as individuals will bear them, both across time and at dif- ferent points in time across regions. This concern is not particular to climate change, but the very long time frames associated with GHG residence in the atmosphere and with thermal inertia of the oceans raise the issue of dis- counting to a level that is present in few other problems. Nonetheless, the committee also finds that a consistent framework for discounting impacts occurring over similar time frames across all potential policy investments is essential for reasoned policy analysis.

OCR for page 248
308 HIDDEN COSTS OF ENERGY RESEARCH RECOMMENDATIONS The committee makes the following recommendations to improve the understanding of physical, biological, and human impacts, as well as eco- nomic valuation aspects related to climate change. • More research on climate damages is needed, as current valuation literature relies heavily on climate-change impact data from the year 2000 and earlier (see Tol [2008] for a number of fruitful areas). • Marginal damages of GHG emissions may be highly sensitive to the possibility of catastrophic events. More research is needed on their impacts, the magnitude of the damage in economic terms, and the probabilities as- sociated with various types of catastrophic events and impacts. • Estimates of the marginal damage of a ton of CO2 include aggre- gate damages across countries according to GDP, thus giving less weight to the damages borne by low-income countries. These aggregate estimates should be supplemented by distributional measures that describe how the burden of climate change varies among countries.