Investigating where environmental efficiency and good public policy intersect.

More than a decade after adopting the first national cap-and-trade approach to regulating pollution from electricity generators, Congress is considering another round of cap-and-trade regulations on a number of gases emitted by electricity generators. Three bills were introduced into the 107th Congress-and are expected to be reintroduced in a similar form in next year's 108th Congress-that propose dramatic reductions in emissions of sulfur dioxide (SO₂), nitrogen oxides (NO_x), and mercury from the electricity sector. In addition, several states already have adopted limits on emissions of multiple pollutants from electricity generators, and others are considering doing so.

A central question in designing multi-pollutant regulation is at what level to set the emission caps. Economists define the efficient level of an emission cap as that at which the incremental benefit of another ton of emission reduction equals the incremental cost of obtaining that reduction. In this article, we describe the results of our analysis of the efficient levels for NO_x and SO₂ based on health science and economics. We discuss the geographic distribution of the likely air quality improvements, and likewise the consequences for electricity producers and customers. We also use insights from this analysis to evaluate, indirectly, all three legislative proposals introduced in the 107th Congress.

In brief, we find the emission caps for SO₂ and NO_x for all three legislative proposals are well within the likely range for the efficient level of reductions. However, a number of qualitative issues regarding how the regulation takes shape affect whether the emission cap levels we find to be efficient actually would be good public policy.

Efficient Emission Levels for SO₂ and NO_x

To investigate the efficient emission levels of SO₂ and NO_x from electricity generators, we used two simulation models to calculate the incremental costs and incremental environmental benefits associated with emission fees imposed in the year 2010. We focus on 2010 because it is a reasonable date by which additional emission reductions might be achieved. The effects on fuel choice, investment in generation technologies and emission controls, costs, prices, and emissions at different locations were simulated using Resources for the Future's (RFF) Haiku electricity market model.¹ The impact of the resulting emissions changes on atmospheric concentrations of particulate matter and other pollution, and subsequent effects on the environment and human health, were simulated using the Tracking and Analysis Framework (TAF) model.² These benefits are expressed in dollar terms and compared to costs to find the efficient pollution levels.

We identified the efficient point in our models by searching for the level of a "pollution tax" that would induce the marginal cost of reducing pollution to equal the marginal benefits. Our analysis suggests that the efficient level of an SO₂ emissions fee in 2010 ranges from $1,800 to $4,700 per ton, corresponding to annual emissions between 0.9 and 3.1 million tons (in 1999 dollars). The most likely value is $3,500 per ton, corresponding to annual emissions of about 1.1 million tons per year. For comparison, we forecast 2010 emissions to be 9.2 million tons under current policies. For NO_x emissions, the efficient fee level is between $700 and $1,200 per ton, which corresponds to annual emissions of 1 million to 2.8 million tons. The most likely value of the fee is $1,100, which achieves annual emissions of 1.4 million tons. In comparison, we forecast 4.6 million tons for 2010 under the status quo. Note that these fees are equivalent to , not the average cost per ton reduced. (We discuss the costs and technologies associated with these reductions below.)

The emission fees approach that we analyze is analogous to an emission cap-and-trade program that achieves the same level of total emissions, with allowance prices equal to the emission fees under the following conditions. First, emission trading must occur without interference from state utility regulators or less flexible regulation (such as new technology requirements), which would increase costs.

Second, the program must use an auction to distribute emission allowances. A different approach to distributing emission allowances, such as grandfathering or output-based allocation, would incur greater cost to the economy than an auction, and therefore raise the efficient level of emissions.³ The allowance auction that we model represents a substantial change from the granting of allowances to existing emitters, the predominant method used to allocate SO₂ allowances under the Clean Air Act. Generally, an allowance auction does impose higher costs on fossil-fired generators than under grandfathering, due to the need to purchase allowances. However, an auction typically leads to higher electricity prices than grandfathering, which helps to enhance profitability. An auction also provides a source of revenue that could be used to compensate those who suffer losses as a result of the policy. Past research has shown that the government would only need to give away a fraction of the allowances to compensate generators for the losses they would incur as a result of stricter emission caps. If an approach different than an auction were used, the efficient target would be less stringent than the level we identify.

On the other hand, some important omitted considerations could lead to the identification of an efficient emission cap that is lower than the level we identify. Our analysis of benefits accounts for health effects from SO₂, NO_x and secondary particulates. However, it does not account for the effects of ozone or acidification on human health or the environment. Including these effects would raise marginal benefits and lead to a lower emission level. Uncertainty analysis in the modeling accounts for the parameter estimates from the health literature for estimating and valuing the effects of air pollution that are used in the TAF model, and for other key judgments.

The Geographic Scope of Benefits

A simple cap-and-trade program treats all emissions equally, but it is important to recognize that there are significant regional differences in the effects of pollution. Emissions from California and states in the mid-Atlantic area cause the greatest economic damages because they lead to changes in exposure for a large population. In contrast, emissions in other western states have the lowest impacts. Overall, damages from SO₂ are about 43 percent higher east of the Mississippi than they are to the west. Likewise, damages from NO_x emissions are about 22 percent higher east of the Mississippi. These large differences suggest there would be advantages to differentiating the programs by origin of emissions.

Fortunately, most of the emission reductions likely would come from the higher-damage areas. Our model suggests that under the efficient policy, the largest reductions would come from a cluster of Midwestern states from Pennsylvania to Missouri, a cluster of Southeastern states from North Carolina to Alabama, and Texas. Accordingly, reductions of ambient pollution levels would be greatest in these areas as well, and in states to the east and north. Western states would see the smallest improvements in air quality. Significantly, however, our analysis suggests that no part of the country would see worse air quality, even with free trading among utilities. Consequently, pollution "hot spots," a concern that has partly motivated technology standards on individual plants, are not likely to be a danger. Figure 1 displays the predicted reductions in particulate matter by state under the efficient policy.

Effects of Efficient Emission Levels on the Electricity Sector

There are three ways to reduce emissions from electricity generators: installing post-combustion controls, switching fuels, or reducing generation. All three methods are used to achieve the efficient level of 1.1 million tons of SO₂ emissions. Scrubbers will be required at over 270 GW of coal-fired capacity (including the capacity that is currently scrubbed under Title IV of the Clean Air Act), with an incremental annual cost for additional scrubbers of roughly $7.5 billion.

Fuel switching, both from high-sulfur coal to low-sulfur coal, and from coal to natural gas and nuclear, is also important. Under the efficient policy, low-sulfur coal use increases by 50 percent to about 1 billion tons.⁴ Fuel switching from coal to gas and nuclear also contributes to achieving the efficient emission cap. At the efficient level of emissions, coal-fired generation is 7 percent below its level in the baseline, while gas-fired generation is 15 percent above its baseline level and nuclear generation is 4 percent higher. Even under the efficient policy, coal-fired generation accounts for roughly half of total generation.

The third way to achieve emission reductions is through a reduction in total generation, which results from an increase in electricity price. The efficient policy leads to a 4 percent increase in electricity price but, because electricity demand is fairly insensitive to changes in price, leads to only a 1 percent drop in electricity generated.

Figure 2 illustrates how reductions of SO₂ are achieved to move from the baseline to the efficient level of emissions. We find the largest piece of the pie (55 percent) represents an increase in scrubbing of coal-fired generation. Switching from high- to low-sulfur coal at unscrubbed units accounts for 32 percent of the reductions. Fuel switching occurs at scrubbed units as well, accounting for 6 percent of the emission reductions. Fuel switching from coal to other fuels, including nuclear and natural gas, accounts for 5 percent, and the decrease in total generation accounts for about 2 percent of the emission reductions.

Reductions in NO_x emissions are achieved primarily through post-combustion controls at coal-fired plants that will incur incremental annual costs of $4.4 billion. The baseline policy already has a substantial amount of post-combustion control in place due to the requirements of the NO_X SIP Call, part of the Clean Air Act implementation. Currently run during the summer ozone season only, such controls would be utilized year-round under the efficient policy. Although gas-fired generation has no SO₂ emissions, it has only slightly lower NO_x emissions than the controlled rate for coal-fired generation. Consequently, switching from coal to gas accounts for only 5 percent of the NO_x emission reductions. Also, we find little change in total generation in response to tightening of the NO_x cap.

Despite the lack of regulation of CO₂ in our analysis, capping emissions of SO₂ and NO_x leads to ancillary reductions in CO₂ emissions. Compared to the baseline, CO₂ emissions fall by 4.9 percent or 152 million short tons in 2010 under the efficient policy. The stringent cap on SO₂ emissions is the primary reason for these reductions. The reductions in CO₂ are due primarily to the substitution of natural gas and nuclear for coal in electricity generation, and also to the reduction in overall electricity demand. Changing the cap on NO_x has very little effect on CO₂ emissions.

The cost estimates we provide are not forecasts of the cost of any of the specific multi-pollutant proposals, which differ in stringency and design from the scenario we model in a number of ways. But our estimate does provide a measure of the cost for emission restrictions that can be supported by cost-benefit analysis.

Summary of the Multi-Pollutant Bills

Having analyzed an efficient policy, we now turn to a discussion of the three multi-pollutant bills introduced in the 107th Congress. All three of the bills adopt a cap-and-trade approach to controlling emissions of SO₂ and NO_x, under which a ton of emissions of each pollutant must be accompanied by the surrender of an emission allowance to EPA. In Table 1 the annual cap on the allocation of emission allowances, the timing, and the scope of these bills are summarized and compared to the range of emission levels that we find to include the efficient level and to recent experience.

The bill (S.556) sponsored by Sen. James Jeffords (I-Vt.) is the most aggressive of the three. Voted out of the Senate Environment and Public Works Committee in June, the bill by 2008 caps annual allocations of NO_x emission allowances at 25 percent of their 1997 levels (about 1.5 million tons); annual allocations of SO₂ emission allowances at 25 percent of the 8.9 million tons allocated annually under Title IV (2.25 million tons); and annual emissions of mercury at 10 percent of 1999 levels (5 tons). The total allocation of allowances for CO₂ emissions is capped at 1990 levels, or 2.05 billion tons. The bill allows for emissions trading and some degree of banking for all gases except mercury.

The most moderate of the three plans is the Bush administration's "Clear Skies" proposal, sponsored by Sen. Bob Smith (R-N.H.) as S.2815 and Rep. Joe Barton (R-Texas) as H.R.5266. This proposal caps annual emissions of SO₂ at 4.5 million tons in 2010 and three million tons in 2018; annual emissions of NO_x at 2.1 million tons in 2008 and 1.7 million tons in 2018; and annual emissions of mercury at 26 tons in 2010 and 15 tons in 2018. It permits the trading of emission allowances for all three pollutants. The proposal does not include a cap on CO₂ emissions.

A compromise between these two, by Sen. Thomas Carper (D-Del.) (S.3135), was introduced in October 2002. The proposal caps annual emissions of SO₂ at 4.5 million tons in 2008, phasing down to 2.25 million tons in 2015; annual emissions of NO_x at 1.87 million tons in 2008 and 1.7 million tons in 2012; and annual emissions of mercury at 24 tons in 2008 and between five and 16 tons in 2012. It also includes a cap of approximately 2.56 billion tons on CO₂ emissions in 2008, falling to roughly 2.39 billion tons in 2012. This proposal permits the trading of allowances for all four gases and includes trading with non-electricity sources of carbon.

While the caps on NO_x and SO₂ proposed in the bills differ importantly from the specific efficient emission levels that we analyze in detail above, especially with regard to SO₂, all three bills propose caps that are within the range of efficient emission caps suggested by our analysis. The Jeffords and Carper bills ultimately limit SO₂ emissions to about 2.25 million tons, while Clear Skies sets an ultimate cap of three million tons, both of which lie within the efficient range of 0.9 million to 3.1 million tons identified in our analysis. Thus, the aggressive targets in these proposals appear to be well justified from the perspective of economic efficiency. Likewise, for NO_x emissions, both the Jeffords target of 1.5 million tons and the Carper/Clear Skies target of 1.7 million tons are within the one million to 2.8 million ton range for efficient emission levels supported by current knowledge.

According to our forecasts, the level of SO₂ emissions that is most likely to be efficient would yield an allowance price of $3,500 per ton. The multi-pollutant bills all would yield allowance prices under $2,000 if the caps were implemented under a perfect trading system without additional regulations including mandatory caps on CO₂ or mercury.⁵ The efficient level of NO_x emissions yields an allowance price of $1,100 per ton, whereas the multi-pollutant bills all would yield allowance prices between $1,000 and $1,100 under the assumptions in our model. However, more important than our specific estimates is our finding that the NO_x and SO₂ caps for all of the legislative proposals appear justified when comparing benefits and costs.

Further, it is important to stress that there are qualitative differences between the efficient policy that we model and the current proposals, and among the proposals as well. These qualitative differences have important implications for the efficiency of the policy, and the distribution of benefits and costs. For example, the bills differ in their treatment of technology-based standards in the Clean Air Act known as new source review (NSR). By requiring certain pollution control equipment on new facilities and old facilities that receive substantial investment, NSR creates a perverse incentive to maintain older, dirtier power plants. The Jeffords bill eventually would eliminate this distinction by requiring that, when they are 40 years old, all existing plants must come into compliance with new source performance standards. However, this would not reduce aggregate pollution (which is already set by the cap), but instead merely reduce the efficiency gains from permit trading. Over time, this technology-based standard may even become more binding than the emissions caps, effectively eliminating the trading program. In contrast, the Clear Skies proposal has been linked to a phase-out of the technology-based standards already present in NSR. The Carper bill also would weaken NSR, and would repeal emission offset requirements for new electricity generators in non-attainment areas.

Another area where the bills differ substantially is in their approaches to allocating emission allowances. Under the Jeffords bill, over 60 percent of the allowances are allocated to households in the first year, effectively giving them the proceeds of an allowance auction, with this fraction eventually growing to nearly 100 percent. In contrast, the Clear Skies proposal allocates the vast majority of the allowances to existing firms, and gradually phases in an auction of allowances over 50 years. The Carper bill continues the approach used in allocating allowances in Title IV for SO₂, but it adopts an output-based allocation approach for NO_x, mercury, and CO₂. Under this approach, allowances are allocated to generators based on their share of total generation from covered facilities in recent years, thereby providing an incentive for low or non-emitting facilities to increase their generation.

As noted previously, our model allocates permits through an auction. This allocation is substantially more cost-effective from a societal perspective than grandfathering or an output-based approach. However, allocation of allowances does have important distributional consequences. In past research with carbon allowances, we found that virtually all existing firms prefer the grandfathering approach to the other two approaches, while consumers feel the opposite. However, existing firms prefer the auction, in turn, to output-based allocations, as it generally does a better job of preserving their asset values. As noted previously, a compromise approach of grandfathering a fraction of the allowances would be sufficient to protect the valuation of existing firms.

Finally, our analysis suggested that a national cap-and-trade program for both NO_x and SO₂ might be more efficient if it differentiated emissions according to where they originate. While none of the bills propose the type of non-uniform trading ratios we suggest, the Jeffords bill proposes separate trading areas for SO₂, and Clear Skies proposes separate areas for NO_x.

When tradable permits were introduced for SO₂ in 1990, air pollution policy took a turn toward a more efficient strategy. Regardless of the level of pollution reductions attempted, tradable permits provide a way to achieve that level at less cost than traditional regulation. Our research suggests that-if the qualitative aspects of the policies remain efficient-lowering the pollution cap for SO₂ and extending it to NO_x would be an additional improvement in economic efficiency. The three bills introduced in the last Congress all fall within a reasonable range of emission levels. Yet the importance of the qualitative aspects of air pollution policy should not be underestimated. How emission allowances are distributed is expected to have a large effect on the cost of the program, and therefore on the efficient level of emission reductions. In addition, the power of a cap-and-trade strategy is strongest in the absence of superfluous technology-based regulations.

---

1. For more information about the Haiku model, see Anthony Paul and Dallas Burtraw, "The RFF Haiku Electricity Market Model," Washington, DC: Resources for the Future, June 2002. Available at http://www.rff.org/reports/PDF_files/haiku.pdf (accessed 10/28/02).
2. TAF was developed to support the National Acid Precipitation Assessment Program (NAPAP). An earlier version of the model, along with documentation, is available at www.lumina.com aflist.
3. Dallas Burtraw, Karen Palmer, Ranjit Bharvirkar, and Anthony Paul, "The Effect of Allowance Allocation on the Cost of Carbon Emission Trading," Resources for the Future Discussion Paper, Jan. 30, 2001.
4. As a baseline, we assume implementation of the summertime NO_x SIP Call trading program in 2004 in 19 eastern states, and the continuation of the Title IV SO₂ cap-and-trade program. We assume no additional regulations affecting mercury or CO₂ and no additional enforcement of new source review beyond settlements already announced. We assume limited restructuring of the electricity sector with retail competition in about half of the country.
5. The addition of such caps would promote fuel switching and other abatement activities that would be likely to lower the cost of SO₂ and NO_x allowances.

Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.