OCTOBER 9, 1998

At the request of the House of Representatives' Committee on Science, the Energy Information Administration has completed a detailed study of the potential, carbon-related impacts of the Kyoto Protocol on the price of energy in the United States and on the overall economy. The uncertainties involved in such an analysis are many, not the least of which is ambiguity about how much carbon emissions from energy use would actually have to be reduced in the United States versus how much of the Kyoto Protocol requirement could be met by other factors, such as reduced emissions from other greenhouse gases and international emissions trading. For this reason, further studies will clearly be needed.

Despite the limitations of this study, we believe the general direction of our conclusions are likely to be reflected in the actual course of events, if the reductions in emissions are eventually agreed to and the Kyoto Protocol is adopted and implemented. Our study of the Kyoto Protocol impacts assumes no change to current policies, except for the Kyoto Protocol itself. As a result, it does not explore energy policy options that might mitigate or aggregate the magnitude of the impacts.

As in much of the work of EIA, proponents on all sides of the debate can find some support in this report, but our goal has been to independently develop the strongest possible analysis regardless of what the impacts might be on the debate. Whatever their weaknesses - and there are several -- models, such as the National Energy Modeling System (NEMS) used by EIA - provide disciplined methodologies for analyzing complex issues.

The National Energy Modeling System has several strengths and weaknesses. Unlike some other models, which are global in nature, it does not explicitly represent several important measures in the Protocol, such as international emissions trading, or greenhouse gases other than carbon dioxide from energy use. Changes to the other gases could either increase or decrease the difficulty for U.S. compliance with the Protocol. On the other hand, more than 80 percent of the human-originated greenhouse gas emissions are energy related, and we attempted to deal with the uncertainties about the extent of international trading and other potential offsets. NEMS deals with year-to-year changes in energy equipment and economic factors, so it captures short-term or transitional effects that might be missed by models without this level of detail. It also deals with energy sectors in greater detail than other energy models. It is widely used by outside groups, and its methodologies are relatively transparent to other energy and economic experts.

To deal with the many acknowledged uncertainties, particularly those concerning the extent of international trading, we have portrayed the Kyoto Protocol impacts on the domestic energy sector across a broad range of possible domestic emissions reduction requirements, in order to provide the maximum amount of relevant information to policy makers. Although emission opportunities in other countries are not explicitly identified in this analysis, the broad range of possible reduction scenarios analyzed implicitly includes a range of emission opportunities in other countries as they may relate to U.S. purchases of emissions permits. This range of reduction requirements is compared to a base case in which EIA projects that, if current policies continue, carbon emissions from energy use will reach 1791 million metric tons in 2010, an increase of 33 percent over 1990 levels.

To cover a broad range of possibilities, we have focused on a menu of six specific cases. These range from one where carbon emissions from the energy sector in 2008-2012 are cut 7 percent below 1990 levels to a case where carbon emissions are constrained to grow only 24 percent above 1990 levels. The former represents a situation with no offsets from forestry activities, other gases, or international activities to reduce the obligation to cut domestic carbon emissions. The latter is a case that requires high levels of reliance on reductions or emissions credits from outside of the domestic energy sector, through mechanisms such as international emissions trading or the Clean Development Mechanism in addition to offsets from forestry, sinks, and other gases. The growth in the least stringent case (the 24 percent above 1990 case) represents a cut of 122 million tons (or 7 percent) of carbon emissions from the base (no new policies) case, or about 22 percent of the total effort to comply with the Kyoto Protocol on greenhouse gas reductions without offsets from gases other than CO2 or international emissions trading. In all cases, end-use consumers in the U.S. start to take action to meet the Kyoto Protocol target beginning in 2005, while capital-intensive supply businesses (e.g., electric utilities) begin to take action as early as 1999.

For purposes of this study, the level of domestic energy-related carbon emissions was held at each of the six alternative target levels throughout the 2008 to 2020 period. In reality, future domestic emissions under a scenario in which the U.S. undertook to meet the Kyoto Protocol's emissions reduction commitment would depend on the levels of the emissions reduction obligation after 2012 (the Protocol now covers only the 2008 to 2012 period) and changes in other regions of the world that could affect opportunities for international emissions trading.


Energy Prices

One of our two primary conclusions is that, using only free market forces (e.g., no new policies), it will take a substantial increase in the price of energy produced from fossil fuels to provide sufficient incentives for the needed changes to technologies in some sectors and consumer

choices in all markets to reduce carbon emissions from domestic energy use to required levels for the cases analyzed. The price increase could come from a variety of approaches including a cap that restricts the use of carbon-based fuels, an auction of carbon permits, or a carbon tax.

Using prices based on the carbon content of each fuel as the mechanism for pricing carbon emissions, EIA estimates that the resulting carbon price would range between $67 and $348 a metric ton (1996 dollars) in the year 2010 in the six cases analyzed. Obviously, the price increase falls most heavily on the most carbon intensive fuels, particularly coal. These price signals generally would be sufficient to produce the fuel switching, efficiency gains, and reductions in the demand for energy services needed to reach the Kyoto target. The price range narrows to $99 to $305 by 2020, as the passage of time allows adoption of more cost effective measures in the more stringent cases, and continued economic growth with increasing energy demand making adaptation more difficult for the least stringent cases.

Estimates of the cost per ton to reduce carbon emissions have little meaning for many people. As a result, it is important to look at the impacts of the prices energy users actually pay. The price of a gallon of gasoline, for instance would increase an estimated 14 to 66 cents in 2010 (from a projected level of $1.25 per gallon). The price of electricity would be more heavily impacted, rising an estimated 20 to 86 percent over the baseline price in 2010 (a projected 5.9 cents per kilowatthour). This would equate to a 4 to 62 percent increase over current (1996) levels.

All economic studies have concluded that higher prices will be needed, but our estimates tend to be higher than most other analyses, particularly during the transition period of 2005-2012.

Embedded Capital Stock. Both the equipment that generates energy and the equipment that consumes it have long useful lives. The existing stock of this equipment serves to slow very rapid adoption of advanced energy technologies, since it is usually expensive to install new (usually more energy efficient) equipment before the old is scheduled for replacement. Much of the capital stock purchased in 1999, (for example, buildings), will likely still be in use in the year 2010. If the useful life of existing stock did not serve as a barrier to rapid penetration of more efficient technologies, the price signal needed to encourage rapid changes in the way Americans use energy would not have to be as high.

NEMS provides the most detailed portrayal of the capital stock that generates and consumes energy among the various models used to assess the cost of restricting fossil fuel use. While these differences may not be significant for longer term studies, the existing stock is a critical issue for studies focusing on the period 2008-2012. With its broader coverage of the relevant capital stock, NEMS provides, in our opinion, a more realistic picture of the opportunities and challenges to rapid, inexpensive gains in energy efficiency.

Consumer Responses to Change in Energy Prices. Two major issues in assessing the impact of restrictions on carbon emissions are

If small price signals are sufficient to change behavior, the response is considered elastic; if large price changes are required to generate a behavior change, the response is considered inelastic. In the real world, there is a whole range of elasticities, but whatever they net out to be, they have a big impact on the estimated price change needed to reduce carbon emissions from energy use.

If energy users change behavior in advance of actual changes in energy prices, they exhibit "foresight." If not, there is no immediate effect from the announcement (as opposed to the implementation) of new policies. Foresight allows more gradual adaptation and lowers the estimated cost of carbon emissions during the period of peak impacts.

Energy prices have been extremely volatile over the last twenty-five years. As a result, we have had many opportunities to see how energy consumers react to rising and falling prices. It is generally acknowledged that responses to changes in energy prices tend to be inelastic compared to other commodities. Energy is such a basic part of doing business and of the quality of life that energy consumers are reluctant to reduce sharply their need for energy services, even in the face of rapidly escalating prices. In addition, the slow turnover of energy capital stock makes it difficult to respond quickly to higher prices with fuel switching or the use of more efficient equipment. While EIA projections do not assume that the future will always be like the past, we have not seen any reason to believe that consumer response to price changes will become more elastic than those observed historically. While large price increases in the 1976-1986 period resulted in both efficiency improvements and shifts in the mix of goods produced, and approximately a 30-percent reduction in energy intensity, the period between 1986-1996 had stable energy prices and exhibited virtually no improvement to energy intensity.

We have also looked closely at the issue of foresight. We allow the impacts of a price signal to begin in 2005 for end-use markets and the electric, gas pipeline, and refinery industries to anticipate the price signal in decisions before 2005. Other sectors do not exercise foresight in our model. On balance, we believe this allows a reasonable treatment of foresight. In some other models, foresight occurs across all sectors and has already begun to influence energy decisions. Other phase-in periods (shorter, longer, earlier, later) could affect the transitional costs and the associated carbon prices, particularly the peak carbon prices.

Prices for Carbon Will Be Set by Marginal Costs. A number of reports tell of reductions in carbon emissions that can be achieved at very low cost (for instance, encouraging reforestation or improving energy efficiency in developing countries). In some cases, these reductions are already contained in the projections for the United States in the base (no new policies) projections in EIA's Annual Energy Outlook reference case. As a result, they represent no additional progress in meeting Kyoto Protocol requirements. In other cases, the reports do reflect efforts that go beyond the base case in cutting emissions and the costs are very low.

Even if options are available that are low cost and represent additional effort beyond the AEO base case, these options may have relatively little bearing on the price of carbon in a competitive environment. It is the cost of the last ton reduced, wherever in the world that occurs, that will set the price in a competitive market for carbon emissions reductions.

Comparing Models. Comparison of models can be a daunting task and is covered in some detail in Chapter 7 of our report. One simpler alternative is to compare them on the relationship between consumer behavior and the retail price of gasoline. In the EIA case where carbon emissions are restricted to 9 percent above 1990 levels in 2010, the projected average efficiency of new cars rises above the 30.6 miles per gallon in the base case to 33.6. However, because new cars make up only a small percentage of the entire fleet, the increase in overall average fuel efficiency is much lower. In addition, projected miles driven is 4.9 percent lower than in the base case. These changes are achieved with a price signal of a 30-cent increase in the cost of a gallon of gasoline. Someone who believes the price signal would not need to be as great probably thinks the EIA model is not elastic enough. Someone who thinks the signal would need to be greater probably thinks the EIA model is too elastic.


Impacts on the Macro Economy

Impacts on the macro economy are typically represented by effects on the Gross Domestic Product (GDP). We have portrayed these impacts two ways -- effects brought about through changes in the quantity of energy consumed within the economy, represented by changes in the Potential GDP; and transitional effects brought about when energy prices also change, represented by changes in the Actual GDP. To further illustrate how the transitional effects can be influenced by fiscal policy, we have analyzed each carbon constraint under two revenue recycling schemes. Therefore, for each of the six carbon reduction cases, there are three estimates of the effect on the Gross Domestic Product. Each of these estimates is portrayed as effects on the overall size of the GDP and effects on the annual percentage growth rate. Our major findings are:

EIA tends to show more severe economic impacts than other studies during the transition period of 2005-2010. Many of the reasons for this difference (e.g., equipment turnover) were discussed earlier. Most other models do not incorporate year-to-year analysis. In addition, many assume that the economy operates at full-employment with perfect knowledge of the future, and, therefore, do not incorporate all of the transitional effects in the EIA analysis. Over the longer period to 2020, many models seem to agree that the projected impacts on the overall economy should be small, and in some cases, even imperceptible.

Even in the EIA cases showing more severe economic impacts than other studies, the effects might be smaller than many would anticipate, given the magnitude of the increases in energy prices discussed earlier. In recent American history, big run ups in energy prices have been associated with economic slow downs. Each of our major recessions since 1970 has occurred during periods of extremely large increases in energy prices (i.e., 1974-75, 1980-82, 1990-91). There are two major factors, however, mitigating the impact of higher energy prices on the economy today.

First, energy is a smaller part of the economy now, muting the impact of that sector on the overall economy. In 1972, the year before a major Arab oil embargo against the United States, the U.S. consumed more than 19 thousand Btu for each dollar of GDP ($1992). Twenty five years later, the number had fallen to 13 thousand Btu per dollar, a drop of 32 percent. Shifts to less-energy intensive industries (e.g., information-based industries) and to higher levels of energy efficiency provide some protection to the overall economy from any possible energy price increase.

Second, the revenues from the higher energy prices projected in our analyses can be recycled back into the American economy. Earlier price increases often resulted in a surge of U.S. dollars being sent to foreign oil producers. In the Kyoto Protocol scenarios in our study, the majority of the money from higher energy prices is recycled domestically (only the revenue used to purchase permits internationally flows abroad). Most other models assume implicitly or explicitly that the money from higher energy prices is recycled domestically. Individuals and firms with high energy costs may turn out to be negatively affected, but some with low energy costs may turn out to be positively affected. The projected benefits of revenue recycling are time sensitive. Recycling is less able to cushion the impacts of higher energy prices during the transition period 2005 to 2010, when the energy using characteristics of the capital stock are relatively fixed, than over the longer term.


Impacts on the Energy Economy

Macroeconomic analysis (that is, analysis of the entire domestic economy) often blurs shifts that are occurring within the overall economy. Of all such shifts, the most dramatic is the projected impact of reduced carbon emissions on the coal industry. The projected production of coal drops sharply, as it is replaced by other fuels -- particularly natural gas -- and the overall demand for energy declines.

Increases in coal productivity in the base case are projected to reduce employment in the coal industry from 83,462 in 1996 to 68,519 in 2010 (a decrease of 18 percent) while allowing coal production to increase by almost 18 percent. Domestic gas production is projected to increase by 24 percent in the same period. In the least stringent case for carbon reduction, coal use drops 18 percent from the base case by 2010 and 40 percent by 2020. Production drops well below current levels and by 2020 has returned to the levels of the early 1980's.

In the most stringent case for carbon reduction, coal use drops 77 percent from base by 2010 and 92 percent by 2020. From another perspective, coal would be produced at about 16 percent of 1996 levels in 2020.

Natural gas and renewable energy (mainly wind and biomass) grow faster in the carbon reduction cases, and there are fewer retirements of nuclear plants. These fuels replace the energy share lost by coal. Overall energy consumption, however, drops below the base case in all the carbon reduction cases we analyze in our projections. Because of fuel switching, however, energy consumption is higher than current levels, even in the most stringent case.