Source: Energy Information Administration, Annual Energy Review1996, pp. 9 and 339, and Energy Infor- The trend in aggregate carbon emissions depicted in Figure 1 is the sum of emissions from three fossil fuels: petroleum, coal and natural gas. The carbon content of the fossil fuels could, in principle, affect trends in carbon emissions. The EIA (EIA, 1995) has developed carbon emission coefficients for each fuel on an annual basis for the 1984-1994 period. A carbon coefficient is defined as the million metric tons of carbon per quadrillion Btu of energy at full combustion. For crude oil, coal (used by electric utilities) and natural gas, the EIA estimates of the carbon coefficients are 20.21, 25.71 and 14.47 respectively for 1994 (see Table 2). A review of the EIA data indicates these coefficients are stable over time. Carbon emissions per unit of fuel are relatively constant for each fuel. Changes in these coefficients cannot explain the time trends in C/E depicted in Figure 2. Changes in the composition of fossil energy, such as shifts in fuel shares, could perhaps account for trends in the C/E ratio. Table 3 depicts the share of each fossil fuel in total fossil energy. During the historical period (1980 to 1995), the share of coal increased relative to the other fossil fuels, which would produce an increase in the C/E ratio. Nevertheless, the C/E ratio declined during the historical period, as shown in Figure 2. The EIA projects the share of natural gas to increase at the expense of coal during the forecast period (1995-2015). As seen in Figure 2, carbon emissions will increase relative to energy use during the forecast period. A shift from coal to natural gas would produce a decrease in these emissions. The shift in fossil fuel shares (away from coal and into natural gas) does not account for the projected increase in carbon emissions during the forecast period. Fuel substitutions between coal and natural gas are not the main force in historical or projected trends in the C/E ratio. Changes in the fossil energy share in total energy consumption are the third possible explanation of trends in the C/E ratio. As noted in Table 2, nuclear energy and renewable energy do not produce carbon emissions, hence changes in the share of fossil and non-fossil energy would affect the C/E ratio. As shown in Table 4, the fossil energy share of total primary energy declined significantly from 1980 to 1995, which would account for a declining C/E ratio during this period. The EIA projects the share of fossil energy to increase through the year 2015, which would account for the increasing C/E ratio during the forecast period. Non-fossil energy includes nuclear power, hydroelectric power, geothermal, biomass, and solar technologies. The conTable 3 Percent Of Fossil Fuel In Total Fossil Energy Use Source: Data for 1980 through 1995 obtained from Annual Energy Review 1996, p. 5. Data for the forecast year Source: Data for 1995 and 2015 obtained from Annual Energy Outlook 1997, p. 99. Data for 1980 and 1987 ob- tribution of hydropower to total electricity genera- The main component of non-fossil fuels that changes is the share of nuclear energy. Figure 3 depicts the share of total energy consumption accounted for by fossil energy and nuclear energy. During the historical period (1980 to 95), the share of fossil fuel declined from 92.1 percent to 84.6 percent. During this period, the share of nuclear power more than doubled. The shift from fossil energy to nuclear energy accounts for much of the decline in the C/E ratio. During the forecast period, the share of fossil energy increases from 84.6 percent to 88.4 percent. The share of nuclear power declines as nuclear units are retired with no new units being constructed. The rise and fall of nuclear power appears to be the critical factor in accounting for changes in the C/E ratio. down. Since 1996, six plants have been decommissioned, or are expected to be decommissioned in the near future. These plants are part of the EIA Reference Case with 40 year lifetimes. The six plants are: Connecticut Yankee, Maine Yankee, Lyon 1 and Lyon 2, Oyster Creek and Big Rock Point. Plants are typically retired when they require a large capital investment to continue operation. For instance, a crack in the reactor vessel, or reactor embrittlement, would be a sufficient cause for shutdown. The most reasonable current estimate for the average lifetime of nuclear plants is less than 40 years and the EIA is currently revising its forecast downward. In AEO98 (available on internet), the EIA assumes that 24 nuclear units will be retired before the end of their 40year operating licenses. The early retirement of nuclear capacity implies that carbon emissions will be even higher than indicated in the EIA's Reference Case. This result is counter to an assumption of extended lifetimes made in a recent more optimistic study, authored by five national laboratories." B. Trends in Energy Use The Kaya identity can account for trends in carbon emissions in an accounting sense. The form of the Kaya identity used here is:7 "Energy Information Administration, Nuclear Power Generation and Fuel Cycle Report 1997, Washington, DC, September 1997, DOE/EIA-0436(97), p. 69. • Interlaboratory Working Group, Scenarios of U.S. Carbon Reductions: Potential Impacts of Energy Technologies by 2010 and Beyond, Office of Energy Efficiency and Renewable Technologies, U.S. Department of Energy, Washington DC, September, 1997. 7 Jones, Russell and Barbara Tierney, "Carbon Emissions A Kaya Identity Perspective on Historic Emissions and Proposed Emission Reduction Targets and Timetables "International Energy Markets: Competition and Policy, Conference Proceedings of the 18th Annual Conference of the 10 94 API Discussion Paper #089 Figure 3 Percent Of Total Energy Consumption From Fossil Fuels And Nuclear Energy which accounts for carbon emissions using a carbon/energy ratio, the energy/GDP ratio and GDP. Table 5 depicts the annual average percentage change in each variable in Equation (2) during the historical and forecast periods. As indicated in Table 5, GDP growth is positively associated with the growth of carbon emissions; although emissions increase more slowly than GDP because of the declining energy/GDP ratio. The variables in this equation provide more of an accounting than economic explanation of carbon trends, because the equation does not explain changes in the terms, including the E/GDP ratio. An economic component is added to the above ratios by relating aggregate energy demand to an intercept, energy prices and real GDP growth. The model states that percentage changes in energy consumption (E) result from an autonomous trend (A) • The derivative of the (natural) logarithm Eq. (2) with respect to time expresses the terms in percentage rates of in energy use, percentage changes in GDP, and the percentage changes in the price of energy (PE). The suggested equation, In E In A+ ẞ, In GDP - B2 In PE, is expressed in logarithms to reflect percentage rates of change, where ẞ, is the income elasticity of demand and ẞ2 is the price elasticity of demand. Applying the model requires knowing the coefficients. The income elasticity implicit in the ELA's NEMS model is 0.6. With GDP increasing at 1.9 percent per year in the EIA Reference Case, economic growth produces an income effect on energy use of 1.14 percent per year (0.6 X 1.9 = 1.14). In the EIA model, GDP grows at 0.5 percent per year faster or slower than in the Reference Case in the high growth and low growth cases respectively. The NEMS model projects that primary energy consumption grows faster or slower than in the Reference Case at 0.3 percent per year in the high growth and low growth cases. In each case, the change in energy growth is 60% of the %% Nuclear Energy 7 A review of several models indicates an appropriate price elasticity. The Energy Modeling Forum 13 (EMF 13) reviewed 9 major energy models and reported the implicit aggregate energy price elasticities. In the review of these models, EMF 13 concludes that "The models indicate that energy demand would fall by only 3 to 8% after 20 years with a 25% increase in delivered energy prices" (EMF 13, p. x). The energy price elasticity implicit in these estimates ranges from -0.12 to 0.32. These elasticities are lower than those estimated for earlier decades and the EMF view is that energy price elasticities have declined. According to EMF 13 (p. 13), these low price elasticities occur because the marginal costs of energy conservation increase substantially. Assuming these modeling analyses are approximately correct, energy consumption has become less responsive to changes in energy prices. However, the low estimated price elasticity does not affect the EIA's projected growth rate of energy use because the EIA projects stable energy prices over the forecast years. The EIA energy projections therefore depend on other factors. The intercept in the above equation indicates the composite autonomous influence of all non-random variables, other than energy prices and GDP, on energy use. The intercept is likely to be negative, indicating an autonomous decline in such energy use. For instance, with energy prices unchanged, a 2.0 percent growth rate for GDP and a 0.6 income elasticity would predict a 1.2 percent growth rate for energy use. If energy use increases at a rate of only 1.0 percent, the autonomous decline in energy use must be the remainder, or -0.2 percent. This rate of decline in autonomous energy use (of 0.2%) is not the same as the decline in energy intensity, which is also declining. Autonomous energy use, measured as aggregate Btu of energy, is assumed independent of GDP and energy prices. Energy intensity, measured as the ratio of aggregate energy to GDP, is influenced by GDP trends and by 11 energy prices. 10 The Energy Modeling Forum notes that declines in energy intensity, unrelated to energy prices, are in the range of -0.6% to -1.0% per year. The income elasticity in the demand equation of 0.6 implies that energy consumption will increase at only 0.6 the rate of growth of GDP. The declines in the energy intensity reported by EMF include the effect of an intercept and an income elasticity less than one. The autonomous decline in energy use, rising energy prices, and increasing GDP each contribute to explaining the decline in E/GDP. With an income elasticity less than one, economic growth produces a decline in energy intensity. During the historical period, declines in E/GDP were affected by the drastically rising energy prices of the 1970s. If energy prices remain stable during the forecast period, E/GDP would be expected to decline at a lower rate than in the past. The autonomous decline in energy use accounts for part of the decline in energy intensity during both the historical and forecast periods. issions grew at a an total primary share of nuclear arket was increas This historical behavior and future projections provide a good explanation of carbon emissions. During the historical period, carbor much slower rate (0.6 percent) energy (1.2 percent) because th power in the electric generation 1 ing. During the forecast period, carbon emissions grow faster (1.2 percent) than pri nary energy consumption because fossil fuels displace nuclear power in the generation market. Energy consumption also affects carbon emissions. An aggregate energy demand equation explains energy use with GDP growth, energy price trends and autonomous declines in energy use. Primary energy prices are projected to remain constant and therefore not to contribute to changes in future energy consumption trends. The projected growth rate in energy consumption of 1.0 percent per year can be explained by the effect of GDP growth of 1.14 percent and an autonomous decline in energy use of -0.14 percent per year." Overall, the growth in carbon emissions 10 In the energy demand Equation (3), the intercept, A, is autonomous energy use, but energy intensity is defined in log form as In E/GDP - In A/GDP + (ẞ1-1)In GDP + B2 InPE/GDP. The equation indicates that energy intensity declines with GDP if ẞ,<1. 11 The effect of GDP growth is estimated as the income elasticity (0.6) times the projected the projected GDP growth rate. The autonomous energy trend of -0.14 off is positively affected by the increasing use of fossil energy in the generation market; negatively by the assumed continuous decline in energy use, and positively by increasing GDP. This review of energy and carbon emission trends suggests several factors that bear on climate change policy commitments. First, carbon emissions will likely increase in the future as natural gas and coal displace nuclear power in the market for electricity generation. This Reference Case result shows an increased growth rate of carbon emissions compared with historical trends. Further, the EIA projection is for electricity to become more carbon intensive even as it obtains an increased share of the residential, commercial and industrial markets. The EIA proj ects constant prices for primary energy and declining prices for electricity. The effect of the declining electricity prices is to encourage its increased use. Finally, in the EIA Reference Case, real GDP is projected to grow at only 1.9 percent per year, slower than during recent history. Primary energy consumption could increase faster if real GDP increases at a faster rate than projected. energy demand growth (1.0 percent) with constant energy |