The Cost of Reducing SO2 (It?s Higher Than You Think)

Fortnightly Magazine - May 15 1998

LAST YEAR, IN JUSTIFYING THE PROPOSED NEW NATIONAL AMBIENT Air Quality Standards (NAAQS) for particulate matter and ozone, Environmental Protection Agency Administrator Carol Browner testified that: "During the 1990 debates on the Clean Air Act's acid rain program, industry initially projected the costs of an emission allowance¼ to be approximately $1,500¼ Today those allowances are selling for less than $100." %n1%n

Later in 1997, at the White House briefing announcing President Clinton's Global Climate Change Plan, Katie McGinty, chairwoman of the Council on Environmental Quality, said of the plan to reduce greenhouse gas emissions in the U.S.: "We've reduced the emissions that cause acid rain by more than 40 percent of what was required¼ for less than a tenth of the price that was predicted¼ We will put [the same] market forces to work to help us take on this [climate change] objective." %n2%n

Statements like these attempt to justify some of the most ambitious air quality initiatives ever considered. However, the initial cost projections never ran as high as those cited today by the White House or the EPA. Initial cost estimates for achieving the sulfur-dioxide emissions reductions envisioned under the fully implemented Phase II cap in the acid rain provisions of Title IV of the Clean Air Act ranged much lower (em from $225 to $500 per ton. Further, costs were projected even lower during the period preceding the date of full compliance with the Phase II cap, and we are still in that period.

These discrepancies stem from inappropriate comparisons. It is no longer possible with market-based regulations to directly compare costs just because they are all expressed in "dollars per ton." It is now essential to understand whether a cost is an average or a marginal cost, short-run or long-run, a capital investment, current expenditure or a market price, or even ex ante or ex post. Sound confusing? It is. The quotes above are evidence of just how far off-base policy prescriptions can go if one doesn't take care to avoid comparing numerical values that are like apples and oranges.

Title IV capped annual emissions at 9 million tons, to be achieved in two phases. Phase I began in 1995, and required only 110 power plants with 263 generating units to balance their emissions with allowances. About 182 additional units opted in. Phase II will begin in the year 2000, when essentially all major fossil units must obtain allowances to operate; only about 9 million allowances will be distributed each year.

Utilities were given the option in Phase I of "banking" unused allowances. Allowances not used in Phase I could be banked for use in Phase II, smoothing the transition to the ultimate cap of 9 million tons.

Banking strategies encouraged early compliance with emissions reductions targets, whereby the units covered by Phase I reduced emissions more than required. This overcompliance is what the CEQ's McGinty was referring to when she cited reductions of 40 percent more than what was required by law. However, that 40-percent overcompliance reflects only the early years of a multi-year phase-in (see Figure 1).

Starting in 2000, the beginning of Phase II will usher in a lower cap, marking the start of an era of "late compliance," as plant owners draw down the bank, allowing them to delay the full force of the Phase II cap. Current estimates say the bank will supplement compliance strategies until sometime between 2005 and 2012, depending on a variety of factors. %n3%n

Most of the historical cost estimates alluded to here apply to a fully implemented SO2 cap with no remaining bank. They do not necessarily show what Title IV may have cost to date. However, the experience gleaned from the allowance market during the past several years can prove useful in analyzing the actual costs of Title IV.

Today, the average cost actually experienced in Phase I is about $200 per ton. This figure falls within the range of the initial projections for Phase I. Today's most up-to-date estimates for Phase II average costs run about $185 to $220 per ton. This interval lies at the low end of the initial range of estimates for Phase II. Actual allowances trade much lower, but we will show that current prices nevertheless remain consistent with actual average costs of about $200 per ton.

Early Forecasts: Accounting for

Flexibility in Control Measures

The current lore is that initial cost estimates for Title IV exceeded $1,000 per ton. This perception appears to have been falsely created by confusion regarding the distinction between marginal and average costs (see sidebar, "Program Terminology").

Estimates in the range of $1,000 per ton or more have always been for the marginal costs, i.e., costs associated with the most difficult-to-control sources. That narrow focus overlooks the flexibility made possible through emissions trading.

For example, a paper from as long ago as 1985 clearly shows that control costs would exceed $1,000 per ton only for scrubbing of units that are already using lower sulfur fuels. %n4%n Barring emissions trading, many units had been estimated to face such high costs, yet it was readily acknowledged that a well-functioning allowance market would reach equilibrium at far less cost because the few units at the high-cost end of the range would have the flexibility to purchase allowances from lower-cost sources, who in turn would control more than would be required under the less flexible regulation.

Many of the proposals for legislation prior to 1987 did not envision much flexibility regarding what control measures might be selected by individual sources. Some of the proposals would have entailed widespread use of flue-gas desulfurization (FGD), which can be very costly for some power units. By the late 1980s, the idea of emissions trading had started to emerge from academic discussions as a political reality and proposals for legislation of SO2 became increasingly flexible in terms of implementation.

As flexibility became an increasingly important feature of regulatory proposals, it was viewed as less and less likely that any units might be forced into these more costly measures. In fact, when the Title IV legislation was being written, $1,500 per ton was viewed as such an unlikely cost that it became the price set for a reserve supply of allowances that the government guaranteed to make available to new companies as a last resort in the event of hoarding. The punitive charge associated with failure to comply was set at $2,000 per ton, "several times more than the estimated average cost per ton of reducing SO2 emissions." %n5%n This penalty was selected because it was viewed as being so much higher than any expected costs of obtaining allowances that it would serve as a deterrent to non-compliance.

Policymakers generally rely on total annual cost to measure total control costs. They may also refer to average annual cost, particularly where there are variations in estimates of the total tons reduced. Table 1 lists a number of such cost estimates for Title IV, starting at the time that its final form was emerging in legislative bills in 1989. The estimates from the original studies have all been converted to a common year (1995 constant dollars). Since the cap is different for Phase I and Phase II, the costs will be different between the two time periods and are shown separately. The estimates for Phase II are all for the year 2010, by which time the bank can reasonably be expected to have been used up. Thus, 2010 marks the first period in the different studies that exhibits a comparable degree of stringency and therefore comparable cost estimates.

Table 1 shows that, even after inflating early cost estimates to 1995 dollars, the estimates for average cost per ton generally vary in a range of $150 to $300 per ton for Phase I and $225 to $500 per ton for full implementation in Phase II. TBS noted that its cost estimates would be 20 to 25 percent lower if emissions trading were included in the estimate. %n6%n This adjustment is shown in the table. All other estimates in Table 1 did incorporate fully flexible emissions trading.

Table 1 also provides estimates of marginal costs, which, though not particularly useful for understanding a program's total control costs, are nevertheless useful for forecasting allowance prices. (With the advent of trading, they have often been the values cited in trade press summaries of new studies.) Analyses by EPRI have focused almost entirely on marginal costs. Estimates of all three cost indicators are presented in the same table to help eliminate confusions from the past.

Phase I Costs:

Recognizing Realities in Fuel Markets

Table 1 shows how the pre-implementation estimates for Phase I ranged from about $150 to $300 in terms of average cost per ton. Yet, anyone who has given passing attention to the press is likely to be aware that allowance prices have always appeared lower than expected, with a particularly notable drop to approximately $70 per ton in March 1996. Why the discrepancy? Had Title IV actually incurred lower control costs than estimated?

The most recent in-depth assessment of actual costs incurred for Phase I comes from the Massachusetts Institute of Technology, in work funded by the National Acid Precipitation Assessment Program. Their results are shown in Table 2.

MIT researchers find that the actual total costs of SO2 control measures in Phase I were $0.7 billion per year in 1995, for a reduction of 3.5 to 3.9 million tons of SO2 relative to what would have occurred without Title IV. Thus, the actual long-run average costs of Phase I appear to be about $187-210 per ton.

MIT's analysis indicates there have been errors in expectations during the market start-up that have raised costs above the minimum achievable. A key cause appears to have been inability to anticipate economic displacement of midwestern high-sulfur coal by western low-sulfur coal, combined with the fact that many of the control decisions involved irreversible capital investments with 3- to 4-year lead times, or fuel contract rigidities. Scrubber-related bonuses and outright political pressure to use flue-gas desulfurization added to the market-information gaps to bias compliance strategies towards FGD. The result was aggregate "overcompliance" with the cap was greater than had been expected when companies were first making their decisions on (and financial commitments to) compliance strategies.

At the same time, costs have fallen for individual control measures. For example, FGD now appears to cost about half what it cost in 1990. %n7%n Low-sulfur coals are also substantially cheaper, particularly delivered to parts of the Midwest that have access to coals from Wyoming (Powder River Basin, or "PRB" coal), made cheaper due to railroad productivity improvements and heightened competition that has occurred since the mid-1980s. However, the flexibility built into Title IV allowed owners to take advantage of the suddenly cheaper low-sulfur coals as a compliance option, increasing incentives for FGD manufacturers to reduce costs to retain what they could of their expected market. Further, the flexibility of Title IV increased the number of ways in which technology costs could be reduced: FGD could now be installed without costly backup systems that would have been essential if 95-percent control levels were mandated. Instead, a much less costly version of FGD has been made possible (e.g., single large vessels), where any failures of the control equipment could be paid for via additional allowance consumption rather than insured against through more costly capital investment.

Rethinking Phase II:

What Demand for Coal Generation?

How much do we now think Title IV ultimately will cost in light of new information about load growth patterns, market performance, and technological improvement?

The ultimate costs of Phase II are still unknown, but

pre-implementation estimates ranged between $1.5 billion to $6.5 billion per year, with average costs between $225 and $500 per ton (Table 1). The lower ends of the ranges were associated with lower levels of coal-fired generation. Using these assumptions, the lower end of the range was $225 to $350 per ton.

Because of the unexpectedly large allowance bank (whose final size is uncertain), full implementation of Phase II remains almost as far in the future as it was at the time that original Title IV cost estimates were being made. That is, in 1990, Phase II was expected to be fully implemented by about 2002, or about 12 years in the future. Eight years later, we find full implementation of the Phase II cap still to be about nine to 12 years away. Thus, forecasts of the ultimate total costs of Title IV remain a long-run projection (em not something that anyone can state with confidence even today.

Overall the range has narrowed among the most up-to-date estimates of ultimate costs for Phase II. This narrowing can be illustrated by work at EPRI and Resources for the Future. However, the range still overlaps with the low end of earlier estimates.

EPRI's updated estimates of total Title IV costs factor in lower FGD, lower sulfur coal prices and a range of uncertainties (see Table 2). As with pre-implementation estimates, the key uncertainty comes from load growth (em or, more specifically, the demand for coal-fired generation, which are shown in separate rows in the table. Looking into the future, arguments can be made for either higher or lower levels of fossil plant use. Higher growth could stem from higher economic growth generally, and from electrification, as well as possibly from lower prices from deregulation. Lower demand growth could develop from conservation and end-use efficiencies. Furthermore, future regulatory initiatives, such as controls for particulates or carbon dioxide, could dramatically alter demand for coal-fired generation. That prospect supplies the primary impetus for EPRI's "low-demand" scenario (bottom row, Table 2).

The middle row of Table 2 shows EPRI's long-run average and marginal cost estimates using an updated base-case demand for coal-fired generation. The base-case range depends mainly on whether control investments are amortized over 10 or 20 years. With lower assumptions for coal plant utilization (which EPRI does not consider a likely outcome unless there are significant new regulatory initiatives), the marginal costs could be as low as $275 per ton. These estimates take into account all of the various cost reductions that have been observed, including lower costs for low-sulfur coals and reduced costs for FGD, reflecting reduced backup requirements and cheaper technology.

EPRI's best judgment of the additional cost to achieve Phase II is $0.8 billion to $1.4 billion per year. (If coal-fired generation demand growth rates were reduced by half, that figure would drop to $0.4 billion per year.) Adding these estimates to the $0.7 billion per year cost to achieve Phase I, EPRI now figures the total cost for Title IV compliance will be $1.5 billion to $2.1 billion per year (or as low as $1.1 billion per year with lower coal plant use, which EPRI does not consider a likely outcome).

Researchers affiliated with Resources for the Future are developing an econometric model of SO2 reduction costs that differs from the engineering approach seen in all the studies cited in Tables 1 and 2, particularly in that it estimates endogenous rates of change in cost parameters and demand. The RFF model is particularly interesting because it fairly closely reproduces the early engineering cost estimates of Table 1 when applied using the assumptions about demand, prices, and technologies that were accepted around the time of those studies. It also tracks fairly closely the actual costs of Phase I when actual compliance actions are applied. At the same time, the model suggests that $0.2 billion to $0.3 billion per year of the actual costs in Phase I could be avoided (while still meeting the actual 1995 emissions levels) by further optimizing fuel choices. If so, this work suggests that possibly as much as 25 percent of the actual Phase I costs already incurred may not be permanently fixed, leaving room for some further downward revisions in the updated cost estimates discussed above. Although their work is still in progress, RFF researchers offer their own "best judgment" of $1.3 to $1.4 billion per year for total costs of Title IV SO2 reductions.

Current Allowance Prices:

Depressed by Excess Capacity

In contrast to control measure costs, allowance prices have proved much lower than expected. In a well-functioning market, allowance prices would roughly coincide with long-run marginal costs. Yet, while marginal costs for Phase I sometimes exceeded $500 per ton, allowance prices have remained around $100 per ton. How can these lower-than-anticipated allowance prices be explained?

One possible answer is unintended overinvestment in compliance. This resulted from the challenge that individual companies faced prior to Phase I in estimating full-market control costs in conditions of uncertainty to devise their control strategies, cognizant of both immediate and coming requirements.

For example, if participants overestimate marginal costs, they may invest too heavily in control measures, creating more allowances for sale than are needed to achieve the cap in any given year. Allowance prices fall. Moreover, because SO2 control measures exhibit a large degree of "irreversibility," such excess supply conditions can well persist. Thus, one benchmark for how low allowance prices might fall before companies stop creating more of them lies in the short-run cost of running a scrubber. That cost reflects daily operating costs, such as reagent, power consumption and labor, but not those associated with recovering capital investment. In fact, for many potential sellers of allowances, the cost of continuing to generate excess allowances is zero.

Consider companies that achieved compliance earlier than Phase I because of increasingly lower costs of delivering western low-sulfur coal to mid-western power plants (due to falling rail rates), or because of the need to comply with more stringent state regulations. For these sellers, the price question is not what the allowances cost, but what they are worth to others, or to themselves for future internal use.

In 1995, it became apparent that overcompliance had been substantial: emissions of the 445 units affected by Phase I of Title IV totaled only 5.3 million tons of SO2 relative to a cap in 1995 of 8.7 million tons. %n8%n The actual amount of banking, 3.4 million tons, exceeded earlier estimates by as much as 1.5 million tons. (The number will decline by about 1 million tons in 1997 due to the drop in bonus allowances awarded for the first two years of scrubbing.) The fact that the excess allowances could be saved for a future day when the cap will be more binding is what gives these allowances financial value.

A second consideration is the ability to bank allowances for future years, which strongly influences today's allowance trading prices. In fact, if Phase II control costs turn out to be high enough, Phase I excess controls may yet generate a return despite the low current price for allowances. How high must prices rise to make banking a profitable strategy? If we take an average price in 1996 of about $100 per ton, and assume that the average real rate of return of utilities is 8 percent, then the 1996 price appears consistent with expected allowance prices in 2008 to 2010 of $233 to $272 per ton (in constant dollars). With equally plausible, higher discount rates, today's low prices appear consistent with even higher future marginal costs. This sort of comparison should not be overdone, however; it compounds uncertainties (coal generation levels, choices of discount rates, many cost assumptions and other regulatory and business risks not explicitly accounted for).

Thus, today's allowance prices do tell us something, but it isn't what total control costs have been so far. First, they indicate the present value of today's expectation of future compliance costs, using a rate that embodies a probably substantial risk premium for future regulatory changes. Second, the current allowance price offers a picture of today's short-run marginal costs.

In the short-run, the considerable FGD capacity now installed creates an effective floor to allowance prices at FGD's operating cost of about $50 to $65 per ton. %n9%n This cost appears consistent with the very lowest allowance price experienced of about $70 per ton. Today's higher allowance prices are thus consistent with the short-run cost of switching to low-sulfur coal, which is based on the price premium for coals with lower-sulfur contents, and is $100 to $120 per ton of SO2 reduced. Thus, the current short-run marginal cost of switching is the current cost most related to today's allowance price.

Future Price Trends:

Dependent on Ozone, Particulates and CO2?

What might happen to future allowance prices if additional control costs were imposed through new regulations for ozone, particulate matter or carbon dioxide? Answer: It all depends on how the new regulations might be implemented.

PARTICULATES. The new NAAQS for fine particles (particulates less than 2.5 microns, or PM2.5) is one regulatory change that substantially affects expectations of the SO2 allowance market. (A standard for particulates would target SO2 emissions, which contribute to fine particle formation.)

EPRI has estimated this new target would increase projected control costs by $3 billion to $5 billion per year greater than those of Phase II. It finds marginal costs could be as high as $1,350 per ton, %n10%n and the benefits of emissions trading relative to a straight technological requirement are reduced.

In the SO2 market, if the Title IV cap-and-trade program was retained and allowance allocations were cut in half through new legislation, then allowance prices would increase substantially. The stricter cap would greatly reduce the range of options, since the sulfur content of coal simply does not go low enough. If, however, any additional SO2 reductions were to come from local requirements laid on top of the present Title IV cap (as might be expected for regulations directed at ambient air quality), then this greater stringency would render the Title IV cap meaningless. The new NAAQS requirements would create a flood of allowances and would shrink the pool of potential buyers. Although allowance prices would plummet, costs to reduce SO2 would increase. Current allowance prices (which depend in part on future prices) would also fall.

CARBON DIOXIDE. Though less imminent, any cap imposed on carbon-dioxide emissions to address climate change would also likely affect SO2 allowance prices. The primary way to achieve CO2 cuts is to replace coal and oil with natural gas or renewable fuels or to improve energy efficiency. Both methods will reduce demand for coal-fired power and cause the baseline of SO2 emissions (and the demand for allowances) to drop. Overall, this result would dampen the growth in allowance prices but these lower prices would not be attributable to the benefits of allowance trading.

TRADING PROGRAMS. What if new NAAQS requirements were imposed through an entirely new emissions trading market, such as allowances for NOx or for CO2? Many lessons might be drawn from SO2 trading.

One key feature of the SO2 market would need to be present again. The SO2 cap was set at a level that left a wide range of options open to each individual source: an approximate 50-percent reduction was required in a situation where there was a technologically proven option that could achieve reductions of 95 percent. Combined with that was the presence of a wide continuum of lower percentage reductions possible via fuels with many different sulfur content levels. Thus, there was room for applying the most costly control measures on only a small fraction of the regulated units, where they would be truly cost-effective. This situation also laid the groundwork for substantial price competition among very different types of control options. A more stringent cap would have reduced the degree of flexibility to manage costs, and probably also the degree of price competition among suppliers of control options. The ability to take advantage of cheaper low-sulfur coals from the West would have been greatly diminished if aggregate SO2 reduction requirements were substantially greater than 50 percent.

Market-based regulatory approaches are here to stay. Emissions trading has proven an effective tool, but will not make the cost of controlling emissions disappear. It is not fair to say that the Title IV experience shows that initial cost estimates are not to be trusted for guiding policy decisions. As always, the main determinant of cost is the stringency of the measure. F

Anne E. Smith is vice president with Charles River Associates. Jeremy Platt is manager, fuel supply target, EPRI. A. Denny Ellerman is executive director of the Center for Energy and Environmental Policy Research, MIT. The authors drew on data and analyses performed independently at MIT and EPRI to provide an updated perspective on Title IV's costs and to contrast these current data appropriately with initial cost estimates.

Program Terminology

Benchmarks for Cost Comparisons

TOTAL CONTROL COST: Sum of control costs (both capital and operating) for all individual sources covered in program. Often stated in terms of total annual costs, to reflect actual expenditure flows. (Note: All definitions listed exclude ancillary costs such as government administration and monitoring.)

AVERAGE COST: Total control cost (see above) divided by estimated tons of emissions reduced, allows comparability by policymakers for programs with different control objectives. Often confused with marginal cost (below), since each is expressed in dollars per ton.

MARGINAL COST: The highest per-ton cost incurred by any individual source or source category included in calculating average cost (see above). Always equals or exceeds average cost. Difference between marginal and average cost can vary significantly between programs. Often considered a good predictor of long-run allowance prices for market-based emissions programs.

ALLOWANCE PRICE: The cost of buying an allowance. Also represents price received for selling an allowance created and earned by undertaking physical control measures, but does not necessarily represent the cost incurred thereby.

LONG-RUN COSTS: Includes cost of building or installing a control measure (the capital cost). Capital costs are always included in regulatory cost projections, and for the most part, both average and marginal cost projections represent long-run estimates.

SHORT-RUN COSTS: The cost of operating the control equipment. Is most relevant for explaining disequilibrium market price behavior. Short-run costs, and hence midterm market outcomes, may vary significantly from the actual total costs of the control program, though each is expressed in dollars per ton.

TIME FRAME: Is the cost estimate of a Phase I cost, an early Phase II cost, or a later, "fully implemented" Phase II cost?

1 Before the Committee on Environment and Public Works, U.S. Senate, Feb. 12, 1997.

2 Washington D.C., Oct. 22, 1997.

3 EPRI, SO2 Compliance and Allowance Trading: Developments and Outlook, prepared by K. D. White, Palo Alto, California, April 1997; tr-107897 4129; p. 1-12.

4 Crocker T. D.; J. L. Regens. "Acid Ceposition Control," Env. Sci. & Tech., 1985. 19(2). pp. 112-116; (Table 2).

5 U. S. General Accounting Office. Allowance Trading Offers an Opportunity to Reduce Emissions at Less Cost, Washington, D.C., December 1994; gao/rced-95-30; p 18.

6 Temple, Barker & Sloane, Incorporated, Economic Evaluation of H.R.3030/S.1490 "Clean Air Act Amendments of 1989": Title V, The Acid Rain Control Program, slides prepared for the Edison Electric Institute, Aug. 30, 1989, p. 4.

7 Ellerman A. D.; Schmalensee R.; et al., Emissions Trading Under the U.S. Acid Rain Program: Evaluation of Compliance Costs and Allowance Market Performance, MIT Center for Energy and Environmental Policy Research, Massachusetts Institute of Technology: Cambridge, Massachusetts, 1997; mit e40-279, p 43.

8 Ellerman A. D.; Schmalensee R.; et al., op. cit. p 15.

9 Ibid, Table 8 and p 49.

10 EPRI, The New Environmental Drivers: Challenges to Fossil Generation Planning and Investment, prepared by K. D. White, Palo Alto, Calif., March 1998 (in press), epri tr-110261, p. 4-23.


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.