economic growth sensitivity case, the annual decline increases to 1.9 percent. In the 1990+9% low economic growth case, the decline in energy intensity slows to 1.3 percent per year.

Technological Progress

The assumed rate of development and penetration of energy-using technology has a significant impact on projected energy consumption and energy-related carbon emissions. Faster development of more energyefficient or lower carbon-emitting technologies than assumed in the reference case could reduce both consumption and emissions; however, because the AEO98 reference case already assumes continued improvement in both energy consumption and production technologies, slower technological development is also possible. To examine the influence of technology improvement, two sensitivity cases were analyzed relative to the 1990+9% case. The high technology case includes more optimistic assumptions on the costs, efficiencies, market potential, and year of availability for the more advanced generating and end-use technologies, assuming increased research and development activity. This sensitivity case also assumes a carbon sequestration technology for coal- and natural-gas-fired electricity generation, which would capture the carbon dioxide emitted during fuel combustion and store it in underground aquifers; however, use of the technology is not projected to be economical relative to other technologies within the time frame of this sensitivity case because of high operating costs and storage difficulties. The low technology case assumes that all future equipment choices are made from the end-use and generation equipment available in 1998, with building shell and industrial plant efficiencies frozen at 1998 levels.

Because faster technology development makes advanced energy-efficient and low-carbon technologies more economically attractive, the carbon prices required to meet carbon reduction levels are reduced. Conversely, slower technology improvement requires higher carbon prices (Figure 24). In the 1990+9% case with high technology assumptions, the carbon price in 2010 is $121 per metric ton-$42 per metric ton lower than the price of $163 per metric ton in the 1990+9% case with reference technology assumptions. With the low technology assumptions, the projected carbon price is $243 per metric ton in 2010.

Total energy consumption in 2010 is lower by 2.1 quadrillion Btu in the high technology case, about 2 percent below the projection in the 1990+9% case, and average energy prices, including carbon prices, are 10 percent lower. As a result, direct expenditures on energy are 13 percent lower in the high technology case. Demand in both the industrial and transportation sectors is lower as efficiency improvements in industrial processes and

most transportation modes outweigh the countervailing effects of lower energy prices. In the residential and commercial sectors, the effect of lower energy prices balances the effect of advanced technology, and consump tion levels are at or near those in the 1990+9% case. With the high technology assumptions in the generation sector, coupled with the lower carbon permit price, coal use for generation is 3.8 quadrillion Btu higher than the 9.7 quadrillion Btu level associated with reference technology assumptions.

In the low technology case, the converse trends prevail. In 2010, total consumption is higher by 1.5 quadrillion Btu with the low technology assumptions, and energy expenditures are 17 percent higher. Industrial and transportation demand is higher, and residential and commercial demand lower, suggesting that industry and transportation are more sensitive to technology changes than to price changes, and that the residential and commercial sectors are more sensitive to price changes. With the higher carbon prices in the low technology case, coal use is further reduced in the generation sector, with more natural gas, nuclear power, and renewables used to meet the carbon reduction targets.

Nuclear Power

In the AEO98 reference case, nuclear generation declines significantly, because 52 percent of the total nuclear capacity available in 1996 is expected to be retired by 2020. A number of units are retired before the end of their 40-year operating licenses, based on industry announcements and analysis of the age and operating costs of the units. In the carbon reduction cases, life extension of the plants can occur, if economical, and there is an increasing incentive to invest in nuclear plant refurbishment with higher carbon prices; however, no construction of new nuclear power plants is assumed, given continuing high capital investment costs and institutional constraints associated with nuclear power.

A nuclear power sensitivity case was developed to examine the potential contribution of new nuclear plant construction to carbon emissions reductions, assuming that new nuclear capacity would be built when it was economically competitive with other generating technologies. In the nuclear power sensitivity case, electricity generators were assumed to add nuclear power plants when it became economical to do so. In addition, the reference case assumptions about higher costs incurred for the first few advanced nuclear plants were relaxed by reducing the premium in costs for the first phase of new nuclear plant additions.

In the 1990+9% case, even with the nuclear power sensitivity assumptions, nuclear plants are not competitive with fossil and renewable plants. In the 1990-3% case. however, when the new nuclear assumptions are used, I gigawatt of new nuclear capacity is added by 2010, and

41 gigawatts, representing about 68 new plants of 600 megawatts each, are added by 2020. (In a trial case in which first-generation cost premiums were left unchanged, only 3 gigawatts of nuclear capacity was added.) The availability of this no-carbon capacity offsets about 25 million metric tons of carbon emissions from additional natural gas plants in 2020; on the other hand, more coal is used, because the projected carbon prices are lower. Most of the impact from the new nuclear plants comes after the commitment period of 2008 through 2012. As a result, there is little impact on carbon prices in 2010. By 2020, however, carbon prices

are $199 per metric ton with the assumption of new nuclear plants, as compared with $240 per metric ton in the 1990-3% case with the reference nuclear assumptions.

In the 1990-3% case, total energy consumption is about the same in 2010 with new nuclear plants allowed and higher by about 1.8 quadrillion Btu in 2020. Somewhat lower energy prices induce higher consumption in all sectors, and the greater availability of carbon-free nuclear generation allows the carbon reduction target to be met with higher end-use consumption.

Background

This chapter provides in-depth analyses of the carbon emissions reduction cases for the four end-use demand sectors-residential, commercial, industrial, and transportation. Additional analyses are included for a number of alternative cases, including low and high technology sensitivity cases, which have the most direct impacts on energy end use.

Primary and Delivered Energy
Consumption

In each of the reduction cases, carbon emissions are reduced through a combination of switching to carbonfree or lower-carbon fuels, reductions in energy services, and increased energy efficiency. The latter two options lower total energy consumption (Figure 25).

Electricity generation typically consumes about three times as much energy, on the basis of British thermal units (Btu), as is contained in the electricity delivered to final consumers. In AEO98, total delivered energy consumption in 1996 is estimated at 70.4 quadrillion Btu, compared with total primary energy consumption of 94.0 quadrillion Btu (Table 3). The difference comes from electricity-related generation and transmission losses and, consequently, is relatively small for the transportation sector, where little electricity is consumed. Although the delivered price of electricity per Btu generally is more than three times the delivered price of other energy sources, the convenience and efficiency of electricity use outweigh the price difference for many applications.

2005 2010 2015 2020 Sources: History: Energy Information Administration, Annual Energy Review 1997, DOE/ELA-0384(97) (Washington, DC, July 1996). Projections: Office of Integrated Analysis and Forecasting, National Energy Modeling Syalem runs KYBASE D080308A, FD24ABY D0803988, FD1998 D0803088, FD09ABV D0803988, FD1990.0060398B, FD03BLW.D0803988, FD07BLW.00803988.

energy-consuming equipment or to use a particular fuel, their decisions are based on the cost and performance characteristics of the technology, mandated efficiency standards, and energy prices. End-use energy prices include all the direct costs of providing energy to the point of use.

The distinction between end-use and primary energy consumption is an important one for the evaluation of efficiency standards and other energy policies. Reducing electricity demand through the use of more efficient technologies reduces primary energy consumption by a factor of three. In addition, although electricity at its point of use produces no carbon emissions, reductions in electricity use produce savings in emissions from the fuels used for its generation.

Integrated Energy Market Analysis

The analysis in this report is a fully integrated analysis of U.S. energy markets, representing the interactions of energy supply, demand, and prices across all fuels and sectors. For example, initiatives to lower energy consumption may lower the prices of the energy supplied, causing some offsetting increase in energy consumption. An integrated market analysis can capture such feedback effects, which may be missed in an analysis that focuses on end-use demand for energy without accounting for impacts on energy prices.

The Energy Information Administration's Annual Energy Outlook 1998 (AEO98), includes results from a number of alternative sensitivity cases in addition to its reference case projections. Sensitivity cases generally are designed by varying key assumptions in one of the demand, conversion, or supply modules of the National Energy Modeling System (NEMS), in order to isolate the impacts of the revised assumptions. For example, the high technology sensitivity cases for the end-use demand sectors in AEO98 do not include any feedback effects from energy prices, and energy consumption in each sector is lower than in the reference case solely due to the revised assumptions about technology costs and efficiencies. The sensitivity cases described in this report, in contrast, were combined into an integrated analysis. As a result, lower energy consumption in the high technology case leads to lower energy prices, which in turn produce some offsetting increases in consumption.

Carbon emission reduction targets and carbon prices further complicate the integrated market analysis. In the high technology sensitivity cases presented in this chapter, the carbon reduction targets are the same as those in the comparable cases that use the AEO98 reference case technology assumptions. For example, the 9-percentabove-1990 (1990+9%) case and the 1990+9% high technology sensitivity case have the same carbon emissions target. The effect of the high technology assumptions is to lower the projected carbon price that would be required to achieve the same level of carbon emissions, which also reduces the delivered price of fuel. With lower carbon prices, adverse impacts on the macroeconomy and on energy markets are moderated. Assuming that the technological advances posited in the high technology cases for the various end-use sectors could in fact be achieved, energy consumption levels would not necessarily be lower in each sector. Rather, the carbon

price would be lower, and it would be less costly to achieve a given emissions reduction target.

Residential Demand

Background

As the largest electricity-consuming sector in the United States, households were responsible for 20 percent of all carbon emissions produced in 1996, of which 63 percent was directly attributable to the fuels used to generate electricity for the sector. Electricity is a necessity for all households, and with electricity use per household growing at 1.5 percent per year since 1990, the projected increase in residential sector electricity consumption has become a central issue in the debate over carbon stabilization and meeting the goals of the Kyoto Protocol.

The number of occupied households is the most important factor in determining the amount of energy consumed in the residential sector. All else being equal, more households mean more total use of energy-related services. From 1980 to 1996, the number of U.S. households grew at a rate of 1.4 percent per year, and residential electricity consumption grew by 2.6 percent per year. In the reference case, the number of households is projected to grow by 1.1 percent per year through 2010, and residential electricity consumption is projected to grow by 1.6 percent per year. Strong growth in the South. which features all-electric homes more prominently than do other areas of the country, and the advent of many new electrical devices for the home have significantly contributed to high electricity growth since 1980. Although these trends are projected to continue through 2010, efficiency improvements-due in part to recent Federal appliance standards, utility demand-side management programs, building codes, and nonregulatory programs (e.g., Energy Star)-should dampen electricity growth somewhat as residential appliances are replaced with newer, more efficient models.

Within the residential sector, all of the major end-uses (heating, cooling, lighting, etc.) are represented by a variety of technologies that provide necessary services. Technologies are characterized by their cost, efficiency, dates of availability, minimum and maximum life expectancies, and the relative weights of the choice criteria-installed cost and operating cost. The ratio of the weight of installed cost to that of operation cost gives an estimate of the "hurdle rate" used to evaluate the energy

efficiency choice. When more emphasis is placed on installed cost, the hurdle rate is higher. The hurdle rates for residential equipment range from 15 percent for space heating technologies to more than 100 percent for some water heating applications. The range in part reflects differences in the way consumers purchase the two technologies. In the case of water heaters, for example, purchases tend to occur at the time of equipment failure, which tends to restrict the choice to equipment readily available from the plumber. Space conditioning equipment, on the other hand, is not used all year round, allowing some latitude in terms of timing the replacement of an older unit. It is assumed that residential consumers expect future energy prices to remain at the current level at the time of purchase when calculating the future operating cost of a particular technology.

Technological advances and availability play a large role in determining future energy savings and carbon emission reductions. Even in today's marketplace, there exist many efficient technologies that could substantially reduce energy consumption and carbon emissions, however the relatively high initial cost of these technologies restricts their widespread penetration. Over time, the costs of more advanced technologies are assumed to fall as the technology matures, one example being natural gas condensing water heaters. In addition, technologies that are not available today but are nearing commercialization are assumed to become available in the future. Three technology menus are used in the analysis below: a reference technology menu, a high technology menu (reflecting more aggressive research and development). and a "frozen" menu limited to equipment available today. In all cases, the menu options and characteristics are fixed. In the high technology sensitivity case, for example, the cost of a condensing natural gas water heater is assumed to fall by almost 75 percent by 2005, relative to the reference case, and a natural gas heat pump water heater becomes available for purchase, by

2005.

In response to energy price changes, residential elasticities, defined as the percent change in energy consumed with a 1-percent change in price, range from -0.24 to 0.28 in the short run, depending on the fuel type, to -0.33 to -0.51 in the longer term. The elasticities reported here are derived from NEMS by a series of simulations with only one energy price varying at a time, beginning in 2000.25 These price elasticities reflect changes in both the

demand for energy services and the penetration rate of more efficient technologies. In the absence of energy price changes, energy intensity, as defined as delivered energy consumption per household, declines at an average rate of 0.5 percent per year through 2010. This nonprice-induced intensity improvement reflects the efficiency gain brought about by ongoing stock turnover, equipment standards, new housing stock, and the future availability of new technologies.

Energy consumption, including the combustion of various fossil fuels, is the major source of U.S. carbon emissions. Energy use in the residential sector is greatly affected by year-to-year variations in seasonal temperatures, particularly in the winter, as illustrated by the decline in delivered energy use in 1990 (Figure 26). which was one of the warmest winters on record. The projections in this analysis assume normal seasonal temperatures over the 1996-2020 forecast period.

In the 3-percent-below-1990 (1990-3%) carbon reduction case, which assumes an emissions target of 3 percent below 1990 levels for the United States, a sharp drop in residential energy use is projected between 2005, when