Pollution and the Price of Power Donald N. Dewees* This study analyses the un-priced environmental harm caused by generating electricity from fossil fuels in the ECAR control region south of the Great Lakes in 2004 and again in 2015 when the recent Clean Air Interstate Rule will have its full effect. Using existing damage values, we estimate wholesale electricity under-pricing for coal-fired plants at about $40 per MWh in 2004, almost as much again as the $45/MWh actual price. Averaging across all fuels, the price of electricity was more than $30/MWh too low. The under-pricing will still be $18/MWh for coal plants and $15 for all generation sources in 2015, a decade after CAIR was adopted. Recognizing this environmental price now could reduce pollution levels, increase energy conservation and lead to wiser choices of new generation technology. 1. Introduction Air pollution regulation in the US and elsewhere generally proceeds from modest initial regulation to increasingly strict emission limits as our knowledge about harm caused and control cost improves. (Davies and Mazurek, 1998, chs. 2, 4, 5; Freeman, 2002.) Legislative and regulatory procedures delay the promulgation of regulations, along with industry lobbying and administrative and court challenges. Industry must be allowed time to purchase and install pollution control equipment, or to replace inherently dirty production equipment. A decade or more may pass between a decision to regulate and the achievement of the intended emission rate. Even then, if harm is underestimated, costs are overestimated or there is insufficient political will to equate marginal costs and benefits, the degree of control is inefficiently low.
The Energy Journal, Vol. 29, No. 2. Copyright ©2008 by the IAEE. All rights reserved. *
Department of Economics, University of Toronto, Toronto, ON, Canada M5S 3G7; email:
[email protected] .
I would like to thank my research assistant Noureen Shah for her work and the Social Sciences and Humanities Research Council Canada for funding. I would also like to thank Ross McKitrick and three anonymous referees for their comments and suggestions on earlier drafts of this paper.
81
82 / The Energy Journal Even when the control is optimal, if it is imposed by regulation, the product of the polluting firms will be under-priced because the polluter does not pay for the harm caused by the remaining emissions. Too much of the product will be produced and consumed. If control is sub-optimal the remaining harm will be larger and the under-pricing greater. This pattern of insufficient control and under-pricing is still playing out with respect to air emissions from fossil-fuelled electricity generating units, particularly coal units. In 2004, after a half century of state and local regulation of air emissions and over three decades of federal regulation, studies showed that reducing air pollution emissions from fossil-fuelled electricity generation units would still give rise to large net benefits, mostly from improved public health. Banzhaf, Burtraw and Palmer (2004, p. 318) estimated optimal US national average effluent charges for sulfur dioxide (SO2) and nitrogen oxide (NOx) emissions at $3,500/ton and $1,100/ ton respectively, far more than the price of emission allowances in 2004. See Table 1. They estimated that such charges would lead to an 89% reduction of SO2, a 70% reduction of NOx, and a 5% reduction in carbon dioxide (CO2) nationwide in 2010 compared to a baseline without the charges. These damage values mean that the average US coal plant caused marginal environmental harm worth about $26 per megawatt-hour (MWh). The next year, the US EPA (2005a, p. 2-4) reported that the benefits of its proposed Clean Air Interstate Rule, hereafter CAIR, which will use a cap-and-trade program to reduce utility emissions in 2015 of SO2 and NOx by 70% and 60% respectively in 28 eastern states, greatly exceed the total costs. It appears that after decades of regulation emissions will still exceed the optimum. But if the net benefits of abatement are great, the damage per MWh of electricity generated today must also be great, implying that electricity is significantly under-priced. This will lead to inadequate investment in clean generation and over-consumption of electricity. These are serious concerns when some jurisdictions are struggling to find ways to supply increasing electrical demand and when
Table 1. Marginal External Harm and Allowance Prices ($/ton)
2004
2015
SO2 Harm 3500 Allowance Price 455
3500 1000
NOx Harm (annual average) 1100 Allowance Price (annual average) 751 Harm (ozone season) 2640 Allowance Price (ozone season) 2253
1100 667 2640 1600
CO2 Harm 10 Price 0
10 0
See text for sources.
Pollution and the Price of Power / 83 the new generation plants will last for decades. The under-pricing of electricity will be greatest in regions where population density is high and coal is the dominant power source, such as the states south of the Great Lakes. US studies generally ignore harm caused in Canada by US utility emissions. Neither do they ask how much wholesale electricity is under-priced as a result of the shortfall in regulation. This paper examines the control of air emissions from fossil-fuelled generating stations in the Great Lakes region of the United States in 2004 and in 2015 after the full effects of the 2005 Clean Air Interstate Rule are felt. We use existing estimates of the marginal harm from air emissions and the emission rates from existing and projected generation facilities to estimate the under-pricing of electricity in five states in both years. We consider the traditional pollutants, sulfur dioxide and nitrogen oxides, plus carbon dioxide. While CO2 emissions have yet to be regulated by the United States Congress or the EPA they are increasingly recognized as causing significant future harm and regulation at the state and federal level seems inevitable. We suggest that in our study region wholesale electricity prices fell short of covering full private plus external costs by more than $30/MWh in 2004, two-thirds of the wholesale price in that year. Despite great reductions in forecast NOx and SO2 emissions by 2015 the wholesale price of power will still fall short of full costs by over $15/MWh in 2015, or 30%. This enormous under-pricing of electricity must result in over-consumption and inadequate conservation of electricity, a distortion of generation investment incentives and excessive environmental harm. The reasons are several. First, emission rates for SO2 and NOx exceed the rate at which marginal cost equals marginal benefit in this region, even with CAIR. Second, as a direct consequence, electricity prices today include only a fraction of the marginal damage costs from discharging conventional pollutants because pollution allowance prices are far less than marginal damage costs in this region. Third, most studies ignore global warming, yet analyses have suggested benefits from near-term CO2 reductions of as much as $10/MWh for coal-fired generation. Fourth, the US analysis ignores any benefits that might accrue in Canada, yet half of the air pollution in southern Ontario blows in from the US. (Yap et al. 2005.) Finally, regulated utility prices do not fully include the opportunity cost of allowances because they are distributed at no cost. We study states in the ECAR electricity control region, excluding those only fractionally in ECAR.1 We call these states “ECAR Lite.” We answer two questions for coal and gas-fired power plants: • By how much did the 2004 wholesale price fall short of private plus social cost because of the marginal external harm from criteria pollutants, the Canada adjustment and CO2? • By how much does the EPA’s forecast 2015 wholesale price with the CAIR program fall short of marginal social cost because of the marginal external 1. ECAR includes Michigan, Indiana, Ohio, Kentucky, West Virginia, part of western Pennsylvania, and the western end of Virginia. We exclude Virginia and Pennsylvania.
84 / The Energy Journal harm from criteria pollutants, the Canada adjustment and CO2, given forecast 2015 emission rates? We make two comparisons: one for a competitive plant and another for a plant subject to traditional rate regulation. The difference is that the former will fully include allowance prices in its price while the latter will only include them to the extent that their emissions exceed the free distribution of allowances. We use plant-specific generation activity and emissions from actual 2004 data and from the EPA’s simulated 2015 CAIR generation and emissions to estimate average emissions by state for coal and gas-fired generation. We then apply damage estimates to calculate the harm caused per MWh of generation by fuel and state. We use a pricing model to calculate the under-pricing of coal and gas-fired electricity by state and the state average under-pricing for both regulated and competitive utilities. 2. Efficient Pricing of Electricity Basic price theory says that when price equals marginal cost (MC), consumer surplus and producer surplus are maximized if there are no externalities. (Varian, 1990, ch. 28; Joskow and Schmalensee, 1983, p. 81.) Where production causes harm to another party, the efficient level of production is achieved if the marginal external cost is added to the marginal private cost to set the price to which the consumer will then equate to her value. (Varian, 1990, ch. 30.) If consumers pay less than this marginal cost, there is a welfare loss arising from excess production. More recently, general equilibrium analyses have shown that existing distorting taxes such as income and sales taxes cause the optimal price of a polluting good to exceed private costs by less than the marginal external harm caused, in some cases much less. (Bovenberg and Goulder, 1996.) General equilibrium analysis is beyond the scope of this paper, so the under-pricing estimated here is overstated to the extent of this general equilibrium effect. Joskow and Schmalensee (1983, p. 88) note that wholesale and retail power prices “are currently not generally based on marginal cost pricing principles.” Regulated rates for most consumers are designed to cover average total costs, not to represent marginal private costs. But total costs for regulatory purposes do not include the cost of environmental harm arising from generation. While environmental regulations will force utilities to control some of their air emissions and the costs of those controls will be paid for by consumers, the utility will not pay for the harm caused by the remaining emissions. Where this un-priced harm is substantial and the unit is marginal, electricity is substantially under-priced. This study focuses on the wholesale price. The retail price includes charges for transmission and distribution which in many jurisdictions are regulated to cover the average total costs of those operations. Much of the time, the transmission grid and distribution networks are uncongested so the marginal cost of using them is below average total cost; when they are congested, marginal cost may be very high. (Joskow, 2006, p. 9.) Thus transmission and distribution charges may be generally greater than short-run marginal costs. On the other hand
Pollution and the Price of Power / 85 one could argue that the high cost of expanding this capacity to meet relentlessly rising demand would justify much higher unit prices based on long run marginal cost pricing. The exploration of deviations of transmission and distribution pricing from the efficient ideal is beyond the scope of this paper. We will focus on wholesale electricity prices. Utility prices would be corrected for environmental harm if every utility were required to pay for most of the damage that its emissions cause. Nowhere in North America is this required. However in the US and Ontario some air emissions are controlled by cap-and-trade programs which distribute free allowances to polluters and require them to surrender one allowance for each ton of pollution discharged.2 In equilibrium the allowance price should represent the marginal cost of control. The discharge of a ton of pollution thus imposes an opportunity cost on the utility. One might expect that these cap-and-trade programs would reflect the marginal pollution damage in the price of power, but in practice the cap is too high so only a fraction of the external cost is internalized. Furthermore the cap-and-trade programs distribute allowances at no cost to the utility. A substantial literature has shown that the welfare effects of this free distribution are different from an auctioned distribution or the imposition of an effluent charge in part because of distorting taxes elsewhere. In general the free allowances yield smaller welfare gains because there is no revenue to reduce the other distorting taxes. (Parry, 2003.) The implication is that if we use free allowance distribution we should control less, and thus aim for a lower allowance price, than if we were to auction the allowances. Analysis of this tax interaction effect is beyond the scope of this paper, but we note that when free allowances are used the under-pricing is again less than that computed here. A regulated utility may set rates that recover reasonable and necessary costs, which includes pollution control cost and the net cost of allowances: the cost of allowances purchased less the revenue from allowances sold. Some utilities may have to surrender more allowances than they receive, while other utilities may surrender fewer than they receive. Since allowances are initially distributed at no cost in the relevant US and Ontario trading programs, the net cost (revenue) of allowances is the excess (shortfall) of allowances retired less free allowances received multiplied by the market price of allowances. Thus emissions trading increases the product price of regulated utilities by a fraction of the increase that would result from an equivalent effluent charge. A generator in a competitive jurisdiction will include the opportunity cost of necessary allowances in its calculation of the marginal cost of generating a MWh when bidding its electricity, since the allowance may be bought or sold at the market price. The market price of allowances is therefore automatically imbedded in the competitive electricity price. If the cap has been set optimally, the allowance price should equal the marginal damage cost. If the cap is too lenient, the allowance price falls short of the damage cost and the electricity price does 2. The Ontario emissions trading program is discussed in “Emissions Trading and NOx and SO2 Emissions Limits for Ontario’s Electricity Sector,” http://www.ene.gov.on.ca/envision/env_reg/er/ documents/2001/RA01E0020-C.pdf .
86 / The Energy Journal not cover full social costs. Imperfect competition will increase prices compared to the perfectly competitive result, which will reduce the under-pricing found here. While the exercise of market power has been much discussed in electricity markets modeling that phenomenon is beyond the scope of this paper. Joskow (2006, pp. 6, 7, 23) concludes that there is effective wholesale competition in much of the northeast US and retail competition for all customers from Michigan and Ohio eastward although only a small fraction of retail customers had chosen competitive suppliers in 2004. Palmer and Burtraw (2005, p. 877) assume that electricity prices are set competitively in five control regions including ECAR, and they further assume that large consumers in those states face the market price. However even those states that have a competitive wholesale market often charge regulated rates to small and medium size consumers rather than the competitive price. We will calculate both regulated and competitive under-pricing in each state. Some public utility commissions have considered “adders” to represent environmental harm, but most of these would have affected the choice of new generating units to build or the dispatch of units, not the price of power. (Burtraw, Palmer and Krupnik, 1997.) We calculate the environmental under-pricing of wholesale electricity as follows. Assume a jurisdiction in which there are G types of generation unit, with all units of a type being identical. Let: Xi = marginal harm from the discharge of one ton of pollutant i, for up to N pollutants; Eij = the rate of discharge of pollutant i from a source of type j, for up to G types, in tons/MWh; Pi = price of an allowance to discharge one ton of pollutant i; Hj = harm caused by generating one MWh of electricity from source type j; Uj = the extent to which electricity generated by a source of type j is under-priced. The external harm caused by generating one more MWh of electricity from source type j is: N
Hj = Σ Eij * Xi $/MWh i=1
(1)
In a competitive jurisdiction, under-pricing equals the sum of external costs for each pollutant emitted by generating one more MWh from a particular fuel less the market price of allowances surrendered: N
UCj = Σ Eij * (Xi – Pi ) $/MWh i=1
(2)
If the price of allowances Pi equaled the external harm, the under-pricing would be zero.
Pollution and the Price of Power / 87 In a regulated jurisdiction with only type j generators, the utility can pass on the net cost of allowances purchased (or sold). Suppose that proportion δi of the allowances surrendered for pollutant i were purchased (sold) or taken from (added to) inventory. The effective average cost of allowances per ton of pollution would be δi*Pi. The under-pricing for the regulated firm equals the sum of the external costs of each pollutant emitted by generating one more MWh from a particular fuel, Xj, less the effective average cost of allowances: N
URj = Σ Eij * (Xi – δi * Pi ) $/MWh i=1
(3)
Equations 2 and 3 do not account for the general equilibrium effects of an emissions trading system in an economy with other distorting taxes, so the under-pricing calculated here will be somewhat overstated. Suppose that the jurisdiction has several types of generation unit. In a regulated market, where each type of unit generates a share αj of total MWh of electricity, the extent to which the regulated market price UR falls short of the efficient price is the weighted average of the under-pricing of power from each type of unit: G UR = Σ aj URj $/MWh (4) j=1
In a competitive market, where each type of generation unit is the price-setting type for share βj of the total MWh of electricity generated, the average competitive market price falls short of the efficient price in proportion to the fraction of power sold when each type of unit is the price-setter: G
UC = Σ bj * UCj $/MWh j=1
(5)
These calculations assume that within each type of generation unit all units are identical, which is sufficiently realistic for the approximate calculations undertaken here. Each type represents a fuel (coal, gas, etc). Will our assumption that units are identical bias the estimate of under-pricing if actual emission rates and marginal costs differ within a type? It will not for regulated firms, because in the simple linear world of equation 4 the under-pricing represents the weighted average emission rate for all units, whether identical or not. However in a competitive market, the extent of under-pricing depends on which units set the price when a variety of units are running. Suppose that cleaner units have higher marginal costs. The cleaner units will under-price less than dirty units, yet dirty units, having lower costs, will always run when clean units run. When clean units set the price, the under-pricing will be greater than if all plants of that type were equally clean. Equation 5 will under (over) -estimate the extent of under-pricing if units of a type are not identical and if marginal cost is inversely (directly) correlated with the emission rate.
88 / The Energy Journal Finally, consider a pollutant such as NOx that causes more harm in one season than in another. The NOx cap applies only during the ozone season, May through September, when most of the ozone formation and damage occurs. In theory we should disaggregate the formulas above by season. However the benefit (harm) data for NOx that we use are available only on an annual average basis, so we cannot disaggregate the damage function. Furthermore it appears that the controls chosen for NOx generally have high capital costs and relatively low operating costs so that they are generally operated all the time, not just in the ozone season.3 (US EPA 2005a, p. 7-9.) This provides two analytical choices. We can follow the benefit data and do the analysis on an annual basis but recognize that the NOx allowance price applies only to a fraction of the year and thus apply an annual average NOx price equal to that fraction of the market NOx price in equations 2 and 3. Alternatively we can attribute all NOx benefits to the ozone season, accept that controls are the same in both seasons, recognize that the NOx allowances are only applicable during the ozone season, and compute under-pricing separately for the ozone season and the non-ozone season. Our principal calculations will focus on the annual average, but we will present the seasonal results as well. 3. The Marginal Harm from Electricity Generation The goal of this study is not to derive new estimates of the relationship between emissions and harm, but rather to use existing data on this relationship to estimate the under-pricing of fossil-fuel generated electricity. We rely primarily on a study by Banzhaf, Burtraw and Palmer (2004, p. 323-4) which in turn uses concentration/response functions based on the same studies used by the US EPA to perform its benefit-cost analysis of the Clean Air Act Amendments. The EPA’s concentration/response functions have been subjected to peer review and considerable examination and some criticism, summarized in the EPA’s regulatory impact analysis for CAIR. (US EPA, 2005a, section 4.1.5.) One line of criticism is that the many studies of health effects relied on by the EPA yield only weak positive associations between mortality and exposure to moderate concentrations of fine particulates; the concentration/ response coefficients vary widely among the individual studies; and toxicological studies have failed to find effects at these concentrations. (Green and Armstrong, 2003.) Still, the studies underlying the EPA models have been subjected to thorough examination and criticism, and they seem to be the best available models for this analysis. The critics have not suggested an alternative data set that has been as thoroughly reviewed and could be regarded as more widely accepted. We accept the EPA’s health effects models as used by Banzhaf, Burtraw and Palmer (2004) as an appropriate basis for both setting policy and criticizing it. Banzhaf, Burtraw and Palmer (2004, p. 318) find that the marginal benefits of reducing SO2 emissions in the US are worth between $1,800 and $4,700 per 3. The EPA’s data for NOx emissions show a similar emission rate during the ozone season and the rest of the year for both the 2004 and 2015 simulations.
Pollution and the Price of Power / 89 short ton while NOx emissions are worth from $700 to $1,200 per short ton.4 The most likely values are $3,500 and $1,100 respectively. These are the values that we use in this study. Effluent charges in this amount would lead to an 89% reduction of SO2, a 70% reduction of NOx, and a 5% incidental reduction in CO2 nationwide. They compute pollution reduction benefits on a regional basis as well, finding benefits of $3,500/ton of SO2 reduction in Indiana, Illinois, Ohio, Pennsylvania, and New York and still higher benefits in another six states, while benefits are low in western states other than California. (p. 329.) They do not differentiate benefits by season. Palmer, Burtraw and Shih (2005, p. 76) use the same model to find a lower bound estimate of the benefits of SO2 reduction of $2,900 to $3,100 per ton. The EPA’s Regulatory Impact Analysis (RIA) for CAIR finds health and environmental benefits of CAIR valued at more than 25 times the cost of compliance. (EPA, 2005a, p. 2-4.) A statistical life is valued at $5.5 million in 1999 $US, more than twice the value used by Banzhaf, Burtraw and Palmer (2004). Benefits are not estimated separately by region, by state, nor by season. In Ontario, a study by DSS/RWDI (2005) analyses the costs and benefits in Ontario (ignoring US benefits) of reducing Ontario air emissions. It finds that the average Ontario coal-fired generating station causes health-related damages of US $96/MWh generated. (DSS/RWDI, 2005, p. 29.) These values are several times greater than previous health effects estimates because they are based on recent long-term cohort studies of health effects which embody significantly higher exposure-response relationships than the previous literature. (DSS/RWDI, 2005, p. 19.) The assumed value of a statistical life is $3.55 million US, higher than the RFF studies, but lower than the EPA’s value. The total damages from coal-fired generation including environmental effects are about US $113/MWh. (DSS/RWDI, 2005, p. ii.) This suggests that the US studies that we rely on are not over-estimating damages from coal-fired power plants. The omission of Canada from the US benefit estimates is not easy to correct without the full model simulations. However a study of transborder air pollution flows concluded that 55% of air pollution health and environmental harm in Ontario was caused by pollution emitted from United States sources. (Yap et al., 2005, p. 54.) Since that analysis focuses on ultimate health and environmental effects, it considers transport from sources to Ontario receptors, and reflects population density at those receptors. Still it is difficult to disentangle the air modeling from the health effects models used by Yap et al. Some idea of the implications of Canadian benefits may arise from looking at Banzhaf’s map of benefits per ton of 4. Banzhaf, Burtraw and Palmer (2004) and Palmer, Burtraw and Shih (2005) model the electric utility sector in considerable detail, using plant-level data on heat rates, emission rates and costs to determine the mix of generation that would be used by cost-minimizing utilities under varying regulatory assumptions. Costs include the cost of pollution control and the cost of effluent charges or the opportunity cost of emission allowances. They model 13 regions, four daily time periods and three seasons, determining the electricity market equilibrium supply and demand in each. (Banzhaf, Burtraw and Palmer, 2004, pp. 321-323.) The electricity model produces emissions which an air dispersion model distributes and a damage model values. Both studies use a value of statistical life equal to $2.25 million ($US 1999). Both studies ignore benefits in Canada and benefits from controlling CO2.
90 / The Energy Journal Table 2. Share of Electricity Generated by Fuel: 2004 (%)
Coal
Gases
Oil
Nuclear
Hydro*
Michigan
57.9
12.78
0.75
25.8
3.69
Indiana
94.4
4.35
0.35
0
0.88
Ohio
86.4
1.13
0.94
10.8
0.74
Kentucky
91.1
0.61
3.83
0
4.45
West Virginia
97.6
0.44
0.30
0
1.67
ECAR Lite Avg.
84.9
4.03
1.14
8.03
2.12
US DOE EIA (2005, EIA 906). * Includes renewable energy and “other” generation.
SO2 (Banzhaf et al., 2004, p. 329). Kentucky, North Carolina, Tennessee, Virginia, and New Jersey all fall in the highest benefit range: $3,829 to $6,062 per ton. The next tier of states to the north, Illinois, Indiana, Ohio, Pennsylvania, and New York fall in a lower benefit range: $3,338 to $3,688 per ton of SO2. Michigan and West Virginia have benefits of $2,795-3,245/ton. The more southerly state benefits are 15% to 64% greater. The population density of the more northerly states is not significantly less than the first group, so the lower benefits seem likely to be caused by the frequent winds blowing to the northeast and the minimal populations to their north if Canada is ignored. Since the population density of southern Ontario and southwestern Quebec5 is similar to that of the adjoining Great Lakes states, including Canada in the benefits model would raise the benefits per ton for the ECAR Lite states, perhaps to the level of the more southerly states. To quantify the Canadian supplement, we assume a perfectly mixed airshed among the ECAR Lite states and southern Canada. If the airshed is mixed, omitting the Canadian portion omits damages proportional to the southern Canadian population divided by the population of ECAR plus southern Canada. In 2004 the ECAR Lite population was 33.77 million; the southern Canadian population was 14.868 million. Adding southern Canada would add 30.57% to the US damages. To be conservative we assume a Canadian supplement equal to 25% of US damages. The benefits of reducing greenhouse gas emissions have been estimated in several studies in the last decade. Gillingham, Newell and Palmer (2004, pp. 67, 85) reviewed the major empirical studies of the environmental benefits of reduced GHG emissions. These include the IPCC Working Group III contribution to the Second Assessment Report (Pearce et al., 1995), the Third Assessment Report of the IPCC and additional reports including Tol (1999). These estimates depend significantly on assumed discount rates; Tol (1999, p. 69) estimated the benefits at $9 to $23 per metric tonne of carbon for real discount rates of 5% and 3% respectively. Gillingham, Newell and Palmer find a mean damage estimate of 5. We include the Ontario population south of Sudbury and the population of southwestern Quebec: Montreal, Sherbrooke, and Trois Rivieres, a total of 14.868 million in 2004.
Pollution and the Price of Power / 91 $30 per metric tonne of carbon discharged, in 2003 US dollars. Since one tonne of carbon implies 3.67 tonnes of CO2, this is equal to $8.17 per tonne of CO2. The US National Commission on Energy Policy (NCEP, 2004, p. 23) surveyed the literature and found benefit values ranging from $3/tonne to $19/tonne of CO2. We will use US $10 per metric tonne of CO2 to represent the benefits for CO2 reduction, a value that is also within the range of the literature discussed above. Since coalfired generating stations emit almost one tonne of CO2 per MWh, $10 CO2 implies climate change damage of almost $10 per MWh for coal-fired generation. The US Energy Information Administration publishes annual emissions by pollutant, fuel type and state and annual generation by state and fuel type. (US DOE EIA, 2005.) We have used year 2004 state average emission rates for coal and for natural gas and other gases for the ECAR Lite states. A plant-level EPA data set reveals that a significant fraction of the natural gas burned in our states is burned in coal plants where it represents less than 1% of the total fuel, which means that it is burned in peaking turbines, which have low efficiency and thus high carbon emissions/MWh. For 2015 we use the EPA’s simulations of electricity generation and pollution emissions by state and fuel type, reported in the Regulatory Impact Analysis and its supporting documents. (US EPA, 2005a, b.) Table 3 summarizes the marginal harm caused by pollution discharged from a set of electric generating stations using 2004 discharge rates, assuming an annual average value of $3,500/ton for SO2 discharge, $1,100 per ton for NOx, and a 25% Canada supplement for emissions from the ECAR Lite states that pass into Canada as discussed above. The calculation is based on equation 1. CO2 is valued at $10 per metric tonne. The total external harm from coal plants in ECAR Lite ranges from over $34/MWh in Michigan to over $51/MWh in Ohio, with an average of $42/MWh, over half of which is caused by conventional pollutants. These are huge costs, since the average industrial price of electricity, which is close to the wholesale price, was under $45 in ECAR Lite. In Indiana, Ohio and Kentucky the total external cost exceeds the industrial price of electricity. The marginal external costs in Table 3 represent an annual average cost for NOx emissions. If we attribute all of the NOx harm to the ozone season, then the NOx harm during the ozone season is 12/5 that shown in the table and zero during the rest of the year. This seasonality makes relatively little difference because NOx is the smallest component of the total damage. On average in ECAR Lite, it would cause ozone season damages to exceed those in the rest of the year by 16%. Natural gas emissions and external costs are much lower than those of coal, with the external costs of a typical combined cycle gas generator falling below $0.50/MWh for conventional pollutants, but totaling about $4 if CO2 is included. The average emissions from Indiana gas-fired power plants in 2004 caused harm valued at $0.69 per MWh for conventional pollutants and $7.66 including CO2. The ECAR gas average total cost was $6.35. While gas is a clean fuel compared to coal, the externality is still 15% of the industrial price of power.
92 / The Energy Journal Table 3. Generation Plant Marginal External Costs: 2004 ($US/MWh) Plant SO2 NOx
CDN CO2 Total Supp
Coal US Average 22.75 3.30 NA 10.20 36.25 Clean 2.5 .36 NA 9.07 12.66 Michigan 18.08 1.91 5.00 9.79 34.77 Indiana 25.45 2.09 6.88 9.66 44.07 Ohio 31.41 2.37 8.44 9.46 51.68 Kentucky 20.63 2.05 5.67 9.60 37.95 West Va. 19.64 2.20 5.46 9.29 36.59 ECAR Lite Avg.
24.10
2.15
6.56
9.55
42.35
Ozone season avg. Rest of year avg.
24.10 24.10
5.16 0
7.32 6.03
9.55 9.55
46.11 39.67
Electricity Price* 52.70 NA 49.20 41.30 48.90 33.40 38.30
