- Email: [email protected]
Weighing environmental externalities: How to do it right
Weighing Environmental Externalities: How to Do It Right
Myrick Freeman is Professor, Department of Economics, Bowdoin College. Dallas Burtraw is a Fellow in the Quality of the Enmronment Division, Resources for the Future. Winston Harrington a n d Alan Krupnick are both Senior Fellows in the same division at RFF. The authors are grateful to the U.S. Department of Energy and the Electric Power Research Institute for supporting this research, and to Robin Cantor, Peter Fox-Penner, David Harrison, Robert D. Rowe, participants in seminars at the University of Laval, the John F. Kennedy School of Public Policy, and Project 88--Round II in Washington, D.C., for helpful comments. Some of the authors also have been involved with the State of New York's Externality Cost Study. The views expressed here are solely those of the authors and do not represent the views of any other organization.
18
Context makes all the difference in considering environmental externalities. Results may be surprising, depending on whether one is working within a setting of command-and-control regulation, a taxation approach, or an emissions trading regime. A. Myrick Freeman IlL Dallas Burtraw, Winston Harrington, and Alan J. Krupnick
ore than half the states have adopted methods for incorporating external costs in electric utility planning and decision making or are investigating the desirability of doing so. 1 These initiafives reflect the recognition by state utility regulators of the potential importance of divergence between private and social costs in choosing among electricity supply options. This problem is highlighted when public utility commissions oversee bidding processes in which utilities and independent power producers (IPPs) offer increments of supply at prices which presumably re-
M
flect their private costs and the utility regulators must oversee selection of the least-cost set of incremental supply options. PUCs have focused recently on environmental costs. While their current topicality led us in this paper to focus on environmental concerns, the examples below extend readily to other social concerns.2 Indeed, PUCs have sometimes been known to promote other social objectives, including equity and regional economic development. State PUCs have explored a variety of approaches to incorporating environmental considerations
The ElectricityJournal
in utility least-cost planning. These include allowing higher rates of return to low polluting sources and using environmental impacts as a "tie breaker'' in ranking alternatives, and various nonmonetary weighting and points systems. In this paper we focus on the procedure of incorporating monetary measures of external damages as "adders" to private bids and ranking alternatives on the basis of their social costs. We assume that it will be possible to develop methods and models to calculate environmental damages for specific electricity supply options. 3 We reject the use of marginal control cost as a proxy for marginal damages. 4 he question we address here is whether the marginal damage from an increase in electricity supply is the correct adjustment or adder to private cost for least-cost utility planning. 5 We will show that arriving at the correct adder depends on both the form that the existing environmental regulations take - - that is, whether pollution is controlled by taxation, tradable emissions permits, or direct regulation - - and, in the case of an emissions tax, whether the degree of control of pollution is optimal, too strict, or not strict enough according to the criterion of economic efficiency. We will make explicit the circumstances under which the correct adder will be equal to marginal damages, will be some other positive number, or zero, or even negative. If emissions reductions are achieved by direct regulation
T
August~September 1992
or command and control policies (CAC), then the correct adder is always just equal to marginal damages.
I. The Basic Theory of Environmental Policy An externality exists when one party's activity (in this case, say, an electric utility) has an effect (positive or negative) on the welfare or cost of others and the offending party insufficiently takes account of that impact in its private decision making. In this case, a private basis for decision-
The correct adder depends on the form that existing environmental regulations take. I [[111[ making is insufficient for the purpose of achieving economic efficiency. Since emissions cause impacts on third parties, they can be the source of negative externalities or external costs. For firms generating electricity, an externality exists if firms insufficiently take into account in their business decisions the damage imposed on others by their emissions. The result is termed a "market failure" because it is said that the market (here a regulated utility) fails to provide an efficient allocation of resources in the economy.
There are at least two ways a market failure can be remedied and economic efficiency can be achieved through environmental regulation. The first way is for the government to create a pseudo-price or a tax on emissions. The tax has the effect of f6rcmg firms to take account of the level of their emissions in private decision making. This is often referred to as "internalizing" the externality. If the tax is set equal to the marginal damage then firms will fully consider the external damage in their financial decisions. They will balance benefits of avoiding the tax and associated environmental damage with the cost of pollution control so as to limit emissions to the optimum level. he other alternative is some form of tradable emission permit system. In such a system, permits equal in number to the optimal level of pollution are either sold to firms or given away on some arbitrary basis. If permits may then be sold or traded, firms will exchange permits until the marginal cost of pollution abatement, which they can avoid by holding a permit, is just equal to the price of a permit in the market. Since the opportunity cost of an emission of one unit for a firm is the price of a permit, firms must take account of this cost in their private decision making, so again, the externality would be internalized in the firnxs' decision making. In both cases, whether with taxes or permits, third parties may still experience harm or damage which are sometimes •
.
.
T
19
known as residual damages, since such damages remain even after the implementation of a policy to restrict emissions. This illustrates an essential point: the presence of residual environmental damages does not necessarily mean that there is an externality or inefficiency. or the most part, U. S. environmental policy does not make use of either of the above approaches to internalizing externalities. Rather, the most common policy instrument is specific quantitative restrictions on the level of emissions from each firm through direct regulation, also known as command and control (CAC) regulation. In principle, it would be possible to establish a set of quantitative restrictions at an emissions level that is economically efficient - - that is, where the marginal benefit of additional pollution control just equals the marginal cost. But even if this level of total emissions was properly identified and could be achieved in a cost-effective way, it would not fully internalize the effects of emissions. This is because with a CAC pohcy, just as with taxes or permits, the cost of pollution control is reflected in the private financial decisions of firms. However, a tax also imposes a direct financial cost on the firm for residual emissions, and tradable permits impose an opportunity cost even if the firm is endowed originally with permits for free. Unlike taxes or permits, with a CAC policy the cost of residual damage is not taken into account. Consequently, with a CAC policy the presence of residual
F
20
damages has economic significance. This can be particularly important in u n i t y capacity planning decisions, wherein major new investments could be made without properly considering the full social costs of each alternative. All this is prelude to the issue at hand. In a first-best world, the utility regulators and the environmental agency would work together to implement efficient pollution control. However, making optimal adjustments to the environmental policy instruments is
Making optimal adjustments to the environmental policy instruments is beyond the power of the utility regulator.
beyond the power of the utility regulator. Rather, the regulator should evaluate the alternative supply options taking existing pollution control policy as given. Social costing has been proposed as a means to deal with decisionmaking when one anticipates an increase in the demand for electricity and must plan for an increase in capacity in light of existing environmental regulations and associated residual emissions that may occur. This is a problem of the "second best," to which w e n o w turn.
II. F i n d i n g t h e Correct A d d e r
We make the following assumptions about utility regulators. First, they have no control over environmental policy per se. They must take such things as emission standards, pollution tax rates, or numbers of tradable emission permits as given. We assume the regulator has the objective of minimizing the increment to social cost (inclusive of environmental concerns) associated with meeting a given increase in the demand for electricity. 6 They achieve this objective through their choice of the mix of new capacity and demand-side management. e assume that utility regulators have before them a set of proposals from a utility and (perhaps) IPPs. Each proposal is characterized by a quantity of electricity, a price or bid which we take to represent private cost, and a description of its emissions. In the presence of CAC regulation, private cost will include only the costs of controlling emissions to meet standards. In the presence of other approaches to regulation the private cost will also include the pollution taxes to be paid, the cost of permits, etc. We also assume that the PUC wishes to rank the proposals by social costs, so any external cost should be added to each proposal's private cost to determine its social cost. This external cost is the "adder." In the rest of this section we identify the correct adder under each of three alternative en-
W
The Electricity Journal
I vironmental regulatory systems: command and control, a tax on emissions, and tradable emissions permits. In each case, we also consider the possibility that environmental regulation may not be optimal. We will show that in some circumstances, the magnitude of the adder will depend on whether existing environmental policy is optimal, over-controlling, or under-controlling.
A. Command and Control Regulation The CAC case is the simplest of the three. With CAC, there is no mechanism for internalizing the marginal damages of residual emissions in the private cost or bid of each option. Thus, the adder should be equal to the marginal damages for each source. o see this, suppose that the PUC has two alternatives, a clean source (no environmental damages) and a dirty source in the sense that it has large environmental damages even with the presently required controls on emissions. For the clean source, the marginal social cost equals the marginal private cost. Let us assume this marginal cost is 9c/kWh. And for the dirty source, the marginal social cost is equal to the marginal private cost (say, 8c/kWh) plus marginal damage (say, 2c/kWh). The utility or regulator looking only at private costs would choose the dirty source. However, the marginal social cost of the dirty source is 10c/kWh. Since the regulator wishes to minimize the marginal social cost of the new capac-
T
August~September 1992
ity, it should choose the clean source. Including an adder equal to the marginal damage of the dirty source in this example reverses their ranking on the basis of cost. Suppose the CAC environmental policy is inefficient - - that is, suppose the marginal benefits of the environmental standard are not equal to the marginal cost of pollution control. It is sometimes asked whether the regulator can "correct" for the effects of bad environmental policy by using a dif-
Can the regulator "correct" for the effects of bad environmental policy by using a different adder? The answer is no.
ferent adder. 7 The answer is no. Any effort on the part of the regulator to adjust its ranking of supply options in an effort to "correct" for inappropriate environmental policy will necessarily make things even worse. Whether the CAC policy is efficient, over-controlling or undercontrolling is irrelevant. The regulator cannot do anything to make matters better than to consider marginal damage from residual emissions and to choose the source with the lowest social cost. For instance, suppose that the regulator believes that environ-
mental policy involves emissions standards that are too weak from an economic efficiency perspecfive. Could the regulator "correct" for this under-control by choosing the clean source even though its marginal social cost exceeds that of the dirty source? If tl~e regulator applies an adder in excess of marginal damage in order to tilt the scale in favor of the dean source, it winds up choosing a source that imposes additional private costs on society without correcting for the excessive abatement costs at existing sources. Given that a dollar of private cost has the same significance for economic efficiency and welfare as a dollar of external cost, the regulator should choose the source with the lowest social cost. he bottom line is that if the regulator is trying to minimize social cost it should treat the costs imposed on society by inappropriate environmental standards for existing sources as sunk costs and ignore them. However, the environmental consequences of an incremental investment in capacity (given existing environmental regulation) are not sunk but variable. If control costs are excessive, they are already built into the bid or private cost, so the regulator need only find the external cost and add it to the private cost. The appropriate adder is marginal damage.
T
B. An Emissions Tax8 With an emissions tax system, a potential new source's bid will be based on the sum of production 21
cost and the tax on the remaining emissions. The production cost will include any cost associated with controlling emissions in response to the incentives created by the tax. The regulator will wish to rank the supply options by marginal social cost, which will be the sum of the production costs (induding costs of control) and the marginal damages imposed by the remaining emissions. Marginal social costs will differ from bids to the extent that marginal damages differ from the tax on emissions. The regulator should adjust the bids by including an adder equal to the difference between marginal damage and the tax (MD - t).9 'f existing environmental pol.icy is too strict, the emissions tax is greater than marginal damages. The regulator should correct for this by incorporating a negative adder equal to MD - t. This negative adder corrects for the fact that part of the tax included in the private cost or bid is not reflective of a social cost, but rather it is just a transfer payment. Similarl3~ if the environmental policy is not strict enough, the regulator should adjust the bid upward by an amount equal to the excess of marginal damage over the emission tax.
l
C. Tradable E m i s s i o n s Permits
This case is of special interest because of the offset requirements for new sources in air pollution non-attainment areas and the SO2 allowance trading system being established under the Clean Air Act Amendments of 19907 o
22
Calculation of the correct adder in the case of tradable emissions permits depends on whether the marginal damage per unit of emission varies across sources, and if so, the extent to which the rules determining how permits are traded reflect any variation in marginal damages, u In general, the marginal damage per unit of emission depends both on the effect of emissions on ambient environmental quality and on the marginal damage associated with a decrease in ambient quality. Both of these relationships can vary significantly across sources of emissions at different locations. The origins of this variation include differences in the ability of the environment to absorb emissions, differences in the characteristics and numbers of agents experiencing the impacts of the emissions (e.g., urban vs. rural areas), and nonlinearities in the damage function. irst consider a simple case where the marginal damage per unit of emission is the same for all sources. An example would be a globally or regionally well-mixed pollutant with a uniform ambient concentration. The effect of a one unit increase in emissions on ambient concentration and damages is therefore independent of the location of the source of the emission. The key feature of this case is that any new source must purchase permits for every unit of emission, and since the total number of permits available is fixed, emissions from other sources must be reduced by an
F
equal amount as they sell permits to the new source. Given these assumptions, the marginal social cost of a n e w source is the sum of the four quantities in Table 1. If the pollutant is uniformly mixed, the two marginal damage terms are equal and cancel out. ]2 If the permit market works effectively, the price of a permit will equal the marginal control cost at the source selling permits. The new source must include the price of its permits in its bid, so the bid will equal the sum of marginal production cost and
Table 1: Marginal social costs of
production at a source in the case of tradable emissions permits.
1. Marginal production costs at a new source, including control costs, plus 2. Marginalcontrol costs at the source selling permits to the new source, plus 3. Marginal dama .ge of emissions at new source, mll'lUS
4. Marginal damages avoided at the source selling permits and reducing emissions.
the cost of permits. This will be equal to the marginal social cost of the new source. In other words, the price of permits internalizes the cost of reducing emissions at other sources. The appropriate adder is thus zero. This is true regardless of whether the number of permits available in the market is optimal, too large, or too small. N o w consider the case where the marginal damage per unit of emission varies across sources. If the environmental agency has
The Electricity Journal
adopted optimal trading rules to reflect this variation, these rules will require that any source that desires to increase its emissions by purchasing permits from another source must purchase a sufficient number of permits to offset the marginal damages of its new emissions, rather than just the quantity of emissions. 13 For instance, suppose that marginal emissions at the new source create three times the damage as marginal emissions at an existing source. Optimal trading rules would require that the new source purchase three permits from the existing source for every one additional unit of emissions. Hence the price of one unit of emissions at the new source is three times the price of a permit. The new source will minimize its total costs by purchasing enough permits so that its marginal control cost is equal to the marginal permit cost it must pa3a gain, the marginal social cost of the new source is the sum represented in Table 1. Since the optimal trading rules require that the new source purchase three permits for each unit of its emissions, the last two components are equal to each other and cancel out. And the marginal control cost at the source selling the permits is internalized in the bid of the new source. So the adder is zero in the case of a permit system with optimally designed trading rules. Note that this result is obtained without making any assumption about the optimal level of environ-
A
m
_
mental quality or the number of permits issued for trading. N o w let's see what happens when the trading rules are not optimal. Suppose that in the above example permits are traded one for one, even though marginal damage per unit of emissions at the new source was three times as great. Then, the last two components of the summation in Table 1 would not cancel out. Although the private bid would internalize the marginal control costs imposed on the selling source through the permit price, the difference between the marginal damages at the two sources would have to be added to the bid to reach the correct marginal social cost. In other words, the environmental adder would be the difference between the marginal damages at the new source and the selling source. If the marginal damages at the selling source were greater than the marginal damage at the new source, the adder would be negative. Alternatively, suppose that the marginal damages per unit of emissions are the same at both sources but that permits do not
trade on a one for one basis. For example, suppose that the new source must purchase more than one permit for each unit of its emissions so that there will be a reduction in total emissions24 Then, the correct adder is based on the product of marginal dam~ges and the net change in emissions. If total emissions are reduced, the adder is negative, is III. C o n c l u s i o n s
The correct environmental adder depends on the nature of the environmental policy instrument and the details of the implementation of that instrument. The results that demonstrate this are summarized in Table 2. In the case of CAC regulation, which is the typical case in the U.S., the adder equals marginal damage. However, the correct adder is not always equal to marginal damage. The adder will be zero when the true costs of the new source, broadly construed, are fully internalized in the private financial cost of the bidder. With a tradable emission permit system and optimal trading rules, the adder is always zero because
Table 2: Summary of AddersunderVarious Regulatory Regimes Level of EnvironmentalControl Policy Instrument
Under Control
Optimal Control
Over Control
Emissiontax
Adder=MD-t>O Adder--MD-t--O Adder--MD-t
Tradable EmissionPermits: OptimalTradingRules
Adder = 0
Tradable EmissionPermits: Non-op~malTradingRules
Adder-- MD(new) - MD(selling)
Command and Cont]'ol
Adder= MD
.
August/September 1992
23
the cost that the new source imposes on society is the additional control cost at the selling source and this cost is already internalized through the cost of permits. But if the trading rules are not optimal, the adder equals the difference in marginal damages across trading sources. here are a number of issues and problems involving social costing and electricity supply planning that we have not dealt with. One is that using externality costs only in capacity planning and not in dispatch decisions can lead to a bias against new sources. As a consequence, older dirty sources might be kept in service longer and are dispatched more heavily because the dispatch decisions do not take account of their external costs2 6 Of course, the answer to this problem is to use correctly calculated externality costs in dispatching decisions as well as in capacity decisions. Although the marginal damages associated with an increment to capacity may be different from
T
the damages associated with a change in the rate of the utilization of a facility, the principles governing the relationship between marginal damages and the correct adder will be the same. In the interim, if external costs are considered for resource supply planning only, it is important to evaluate investments in life extensions for existing facilities on an equal footing with new sources. The present procedures for social costing include external costs in supply decisions but not directly in setting electricity prices. Thus the demand for electricity is higher than it should be because consumers may not be getting the correct price signals. Of course this deviation from efficient pricing is an overlay to existing pricing practices that typically deviate from marginal cost pricing. Whether the net effect is to amplify or diminish departures from efficient pricing is a technical and empirical question best reserved for another discussionJ 7 However, other things being equal, one
A zero adder is correct only in certain circumstances.
24
can see that consumers are not being forced to recognize the environmental consequences of their electricity use. nother problem is that the use of social cost adders in the least-cost capacity planning process only for the utility system puts new plants at a competitive disadvantage. Some utility customers may find it privately advantageous to generate their own electricity, switch fuels, or even relocate their industrial facilities rather than purchase electricity from a utility that is making decisions on the basis of least social cost. This "by-pass" problem is a consequence of applying environmental costs only to utilities. If nonutility sources of energy have uninternalized environmental costs associated with them, social welfare could be reduced by requiring externality adders only in the utility system. These two problems of incorrect electricity price signals and bypass arise not because social costing is wrong, but rather because it is incomplete in its coverage. The solution to these problems is to develop a comprehensive system for internalizing environmental costs not only on utilities but on other sources of these emissions as well. This can be done either through a comprehensive system of taxes on emissions or a comprehensive marketable emissions permits system. This suggestion is not new. What is new is the alliance between proponents of market-based approaches to environmental policy and utility regulators who have taken leader-
A
The Electricity Journal
ship positions in the social costing movement. Together, they can push for an extension of externality pricing as the answer to the criticism that their present second best approaches might make matters worse. • Endnotes:
1. S. D. Cohen, J. H. Eto, J. Beldock, and G. Crandall, A Survey of State Regulatory C o m m i s s i o n Activities to Incorporate Environmental Externalities into Electric Utility Planning and Regulation, Lawrence Berkeley Laboratory (1990). 2. The distinction between these and other n o n e n v i r o n m e n t a l external costs (and benefits) that appears in some discussions is artificial. See D. Burtraw and A. J. Krupnick, The Social Costs o.~ Electricity: H o w Much of the Camel to Let into the Tent?, Discussion Paper QE92-15, Ressources for the Future, Wash., D.C. (1992). 3. There are presently u n d e r w a y two major research efforts and several smaller ones to d e v e l o p such methods and models. One major effort is jointly sponsored by the U.S. Department of Energy and the Commission of the European Community. The other is the N e w York State Environmental Externalities Cost Study. See also R.L. OTT1NGER, ET AL, ENVIRONMENTAl, COSTS OF ELECTRICITY (1990).
4. Several states, including Massachusetts and Nevada, have adopted this approach. The rationale for it is that environmental regulators have revealed a "willingness to pay" for reduced environmental impacts by imposing pollution control requirements with these costs. There are at least three sufficient reasons for rejecting this rationale. First, the typical uniform emissions or discharge standard results in very different marginal control costs across different sources of the same pollutant; so there is no unique marginal control cost curve to intersect with a marginal d a m a g e curve. Second, the same a m o u n t of
Augus t/September 1992
discharge of a pollutant is likely to result in very different marginal d a m ages across different locales. Third, most of the federal environmental laws do not require, and some preclude, balancing of benefits and costs in establishing control requirements. So there is no reason to expect any particular relationship between marginal damages and marginal abatement costs. 5. Several of the issues w e address were identified by Professor Paul Joskow in this journal. See P.L. Joskow, Weighing Environmental Externalities: Let's Do It Right!, ELEC. J., May 1992, 53-67. 6. For a different v i e w of the responsibilities and objectives of public utility
So the adder is calculated as (MD - t) times the level of emissions per kWh. 10. Proposed new sources of air emission in regions presently not attaining primary national air quality standards must certify offsetting reductions in emissions from other sources as a condition for obtaining an emissions license. They m a y obtain offsets by ipurchasing reductions in emissions from other sources. 11. For a rigorous demonstration of this point, see B.F. Hobbs, What Do SO2 Emissions Cost?, External Costs and Allowance Prices U n d e r the 1990 Clean Air Act A m e n d m e n t s , working paper, Case Western Reserve U. (1992). 12. There is a change in the incidence of damages. Some gain as their damages are reduced while others lose. But the net effect is zero. 13. For details concerning optimal trading rules, see, e.g., T.H. TIETENBERG, EMISSIONS TRADING: AN EXERCISE IN REFORMING POI.LUTION POI.-
IcY (Resources for the Future, Wash., D.C. 1985) and P. Bohm and C. S. Russell, Comparative Analysis of Alternative Policy Instruments, in HANDBOOKOF ENVIRONMENTALAND RESOURCE ECONOMICS, vol. 1 (A. V. Kneese and J. L. Sweeney, eds., 1985). 14. This requirement is a feature of the present federal offset program. regulation, see P.A. Agathen, Dealing with Environmental Externalities, PuB. UrIL. FORT., Feb. 15, 1992, at 23-24. 7. Joskow, supra note 5, alleges this is the agenda for some utility regulators. 8. Currently there are no examples of these in the U.S., although they are prevalent in Europe. ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT, ECONOMIC INSTRUMENTS FOR ENVIRONMENTAL PROTECTION
(1989). 9. The statement that the a d d e r is equal to the difference between marginal d a m a g e and the tax (MD - t) is an over-simplification. MD and t are usually expressed in units of dollars per unit of emissions, while the adder is expressed in units dollars per kWh.
15. Joskow, supra note 5, is reluctant to offer such a recommendation in this situation. However, we think such a policy is well founded and should be favored by the environmental c o m m u nity, because it gives the utility credit for i m p r o v e m e n t in environmental quality. 16. K. Palmer and H. Dowlatabadi, Implementing Environmental Costing in the Electric Utility Industry, Quality of the Environment Division Discussion Paper QE91-13, Resources for the Future, Wash., D.C. (1991). 17. See D. Burtraw, W. Harrington, A. M. Freeman llI and A. J. Krupnick, The Analytics of Social Costing, Res. for the Future, Wash., D.C., unpublished manuscript (1992).
25
Myrick Freeman is Professor, Department of Economics, Bowdoin College. Dallas Burtraw is a Fellow in the Quality of the Enmronment Division, Resources for the Future. Winston Harrington a n d Alan Krupnick are both Senior Fellows in the same division at RFF. The authors are grateful to the U.S. Department of Energy and the Electric Power Research Institute for supporting this research, and to Robin Cantor, Peter Fox-Penner, David Harrison, Robert D. Rowe, participants in seminars at the University of Laval, the John F. Kennedy School of Public Policy, and Project 88--Round II in Washington, D.C., for helpful comments. Some of the authors also have been involved with the State of New York's Externality Cost Study. The views expressed here are solely those of the authors and do not represent the views of any other organization.
18
Context makes all the difference in considering environmental externalities. Results may be surprising, depending on whether one is working within a setting of command-and-control regulation, a taxation approach, or an emissions trading regime. A. Myrick Freeman IlL Dallas Burtraw, Winston Harrington, and Alan J. Krupnick
ore than half the states have adopted methods for incorporating external costs in electric utility planning and decision making or are investigating the desirability of doing so. 1 These initiafives reflect the recognition by state utility regulators of the potential importance of divergence between private and social costs in choosing among electricity supply options. This problem is highlighted when public utility commissions oversee bidding processes in which utilities and independent power producers (IPPs) offer increments of supply at prices which presumably re-
M
flect their private costs and the utility regulators must oversee selection of the least-cost set of incremental supply options. PUCs have focused recently on environmental costs. While their current topicality led us in this paper to focus on environmental concerns, the examples below extend readily to other social concerns.2 Indeed, PUCs have sometimes been known to promote other social objectives, including equity and regional economic development. State PUCs have explored a variety of approaches to incorporating environmental considerations
The ElectricityJournal
in utility least-cost planning. These include allowing higher rates of return to low polluting sources and using environmental impacts as a "tie breaker'' in ranking alternatives, and various nonmonetary weighting and points systems. In this paper we focus on the procedure of incorporating monetary measures of external damages as "adders" to private bids and ranking alternatives on the basis of their social costs. We assume that it will be possible to develop methods and models to calculate environmental damages for specific electricity supply options. 3 We reject the use of marginal control cost as a proxy for marginal damages. 4 he question we address here is whether the marginal damage from an increase in electricity supply is the correct adjustment or adder to private cost for least-cost utility planning. 5 We will show that arriving at the correct adder depends on both the form that the existing environmental regulations take - - that is, whether pollution is controlled by taxation, tradable emissions permits, or direct regulation - - and, in the case of an emissions tax, whether the degree of control of pollution is optimal, too strict, or not strict enough according to the criterion of economic efficiency. We will make explicit the circumstances under which the correct adder will be equal to marginal damages, will be some other positive number, or zero, or even negative. If emissions reductions are achieved by direct regulation
T
August~September 1992
or command and control policies (CAC), then the correct adder is always just equal to marginal damages.
I. The Basic Theory of Environmental Policy An externality exists when one party's activity (in this case, say, an electric utility) has an effect (positive or negative) on the welfare or cost of others and the offending party insufficiently takes account of that impact in its private decision making. In this case, a private basis for decision-
The correct adder depends on the form that existing environmental regulations take. I [[111[ making is insufficient for the purpose of achieving economic efficiency. Since emissions cause impacts on third parties, they can be the source of negative externalities or external costs. For firms generating electricity, an externality exists if firms insufficiently take into account in their business decisions the damage imposed on others by their emissions. The result is termed a "market failure" because it is said that the market (here a regulated utility) fails to provide an efficient allocation of resources in the economy.
There are at least two ways a market failure can be remedied and economic efficiency can be achieved through environmental regulation. The first way is for the government to create a pseudo-price or a tax on emissions. The tax has the effect of f6rcmg firms to take account of the level of their emissions in private decision making. This is often referred to as "internalizing" the externality. If the tax is set equal to the marginal damage then firms will fully consider the external damage in their financial decisions. They will balance benefits of avoiding the tax and associated environmental damage with the cost of pollution control so as to limit emissions to the optimum level. he other alternative is some form of tradable emission permit system. In such a system, permits equal in number to the optimal level of pollution are either sold to firms or given away on some arbitrary basis. If permits may then be sold or traded, firms will exchange permits until the marginal cost of pollution abatement, which they can avoid by holding a permit, is just equal to the price of a permit in the market. Since the opportunity cost of an emission of one unit for a firm is the price of a permit, firms must take account of this cost in their private decision making, so again, the externality would be internalized in the firnxs' decision making. In both cases, whether with taxes or permits, third parties may still experience harm or damage which are sometimes •
.
.
T
19
known as residual damages, since such damages remain even after the implementation of a policy to restrict emissions. This illustrates an essential point: the presence of residual environmental damages does not necessarily mean that there is an externality or inefficiency. or the most part, U. S. environmental policy does not make use of either of the above approaches to internalizing externalities. Rather, the most common policy instrument is specific quantitative restrictions on the level of emissions from each firm through direct regulation, also known as command and control (CAC) regulation. In principle, it would be possible to establish a set of quantitative restrictions at an emissions level that is economically efficient - - that is, where the marginal benefit of additional pollution control just equals the marginal cost. But even if this level of total emissions was properly identified and could be achieved in a cost-effective way, it would not fully internalize the effects of emissions. This is because with a CAC pohcy, just as with taxes or permits, the cost of pollution control is reflected in the private financial decisions of firms. However, a tax also imposes a direct financial cost on the firm for residual emissions, and tradable permits impose an opportunity cost even if the firm is endowed originally with permits for free. Unlike taxes or permits, with a CAC policy the cost of residual damage is not taken into account. Consequently, with a CAC policy the presence of residual
F
20
damages has economic significance. This can be particularly important in u n i t y capacity planning decisions, wherein major new investments could be made without properly considering the full social costs of each alternative. All this is prelude to the issue at hand. In a first-best world, the utility regulators and the environmental agency would work together to implement efficient pollution control. However, making optimal adjustments to the environmental policy instruments is
Making optimal adjustments to the environmental policy instruments is beyond the power of the utility regulator.
beyond the power of the utility regulator. Rather, the regulator should evaluate the alternative supply options taking existing pollution control policy as given. Social costing has been proposed as a means to deal with decisionmaking when one anticipates an increase in the demand for electricity and must plan for an increase in capacity in light of existing environmental regulations and associated residual emissions that may occur. This is a problem of the "second best," to which w e n o w turn.
II. F i n d i n g t h e Correct A d d e r
We make the following assumptions about utility regulators. First, they have no control over environmental policy per se. They must take such things as emission standards, pollution tax rates, or numbers of tradable emission permits as given. We assume the regulator has the objective of minimizing the increment to social cost (inclusive of environmental concerns) associated with meeting a given increase in the demand for electricity. 6 They achieve this objective through their choice of the mix of new capacity and demand-side management. e assume that utility regulators have before them a set of proposals from a utility and (perhaps) IPPs. Each proposal is characterized by a quantity of electricity, a price or bid which we take to represent private cost, and a description of its emissions. In the presence of CAC regulation, private cost will include only the costs of controlling emissions to meet standards. In the presence of other approaches to regulation the private cost will also include the pollution taxes to be paid, the cost of permits, etc. We also assume that the PUC wishes to rank the proposals by social costs, so any external cost should be added to each proposal's private cost to determine its social cost. This external cost is the "adder." In the rest of this section we identify the correct adder under each of three alternative en-
W
The Electricity Journal
I vironmental regulatory systems: command and control, a tax on emissions, and tradable emissions permits. In each case, we also consider the possibility that environmental regulation may not be optimal. We will show that in some circumstances, the magnitude of the adder will depend on whether existing environmental policy is optimal, over-controlling, or under-controlling.
A. Command and Control Regulation The CAC case is the simplest of the three. With CAC, there is no mechanism for internalizing the marginal damages of residual emissions in the private cost or bid of each option. Thus, the adder should be equal to the marginal damages for each source. o see this, suppose that the PUC has two alternatives, a clean source (no environmental damages) and a dirty source in the sense that it has large environmental damages even with the presently required controls on emissions. For the clean source, the marginal social cost equals the marginal private cost. Let us assume this marginal cost is 9c/kWh. And for the dirty source, the marginal social cost is equal to the marginal private cost (say, 8c/kWh) plus marginal damage (say, 2c/kWh). The utility or regulator looking only at private costs would choose the dirty source. However, the marginal social cost of the dirty source is 10c/kWh. Since the regulator wishes to minimize the marginal social cost of the new capac-
T
August~September 1992
ity, it should choose the clean source. Including an adder equal to the marginal damage of the dirty source in this example reverses their ranking on the basis of cost. Suppose the CAC environmental policy is inefficient - - that is, suppose the marginal benefits of the environmental standard are not equal to the marginal cost of pollution control. It is sometimes asked whether the regulator can "correct" for the effects of bad environmental policy by using a dif-
Can the regulator "correct" for the effects of bad environmental policy by using a different adder? The answer is no.
ferent adder. 7 The answer is no. Any effort on the part of the regulator to adjust its ranking of supply options in an effort to "correct" for inappropriate environmental policy will necessarily make things even worse. Whether the CAC policy is efficient, over-controlling or undercontrolling is irrelevant. The regulator cannot do anything to make matters better than to consider marginal damage from residual emissions and to choose the source with the lowest social cost. For instance, suppose that the regulator believes that environ-
mental policy involves emissions standards that are too weak from an economic efficiency perspecfive. Could the regulator "correct" for this under-control by choosing the clean source even though its marginal social cost exceeds that of the dirty source? If tl~e regulator applies an adder in excess of marginal damage in order to tilt the scale in favor of the dean source, it winds up choosing a source that imposes additional private costs on society without correcting for the excessive abatement costs at existing sources. Given that a dollar of private cost has the same significance for economic efficiency and welfare as a dollar of external cost, the regulator should choose the source with the lowest social cost. he bottom line is that if the regulator is trying to minimize social cost it should treat the costs imposed on society by inappropriate environmental standards for existing sources as sunk costs and ignore them. However, the environmental consequences of an incremental investment in capacity (given existing environmental regulation) are not sunk but variable. If control costs are excessive, they are already built into the bid or private cost, so the regulator need only find the external cost and add it to the private cost. The appropriate adder is marginal damage.
T
B. An Emissions Tax8 With an emissions tax system, a potential new source's bid will be based on the sum of production 21
cost and the tax on the remaining emissions. The production cost will include any cost associated with controlling emissions in response to the incentives created by the tax. The regulator will wish to rank the supply options by marginal social cost, which will be the sum of the production costs (induding costs of control) and the marginal damages imposed by the remaining emissions. Marginal social costs will differ from bids to the extent that marginal damages differ from the tax on emissions. The regulator should adjust the bids by including an adder equal to the difference between marginal damage and the tax (MD - t).9 'f existing environmental pol.icy is too strict, the emissions tax is greater than marginal damages. The regulator should correct for this by incorporating a negative adder equal to MD - t. This negative adder corrects for the fact that part of the tax included in the private cost or bid is not reflective of a social cost, but rather it is just a transfer payment. Similarl3~ if the environmental policy is not strict enough, the regulator should adjust the bid upward by an amount equal to the excess of marginal damage over the emission tax.
l
C. Tradable E m i s s i o n s Permits
This case is of special interest because of the offset requirements for new sources in air pollution non-attainment areas and the SO2 allowance trading system being established under the Clean Air Act Amendments of 19907 o
22
Calculation of the correct adder in the case of tradable emissions permits depends on whether the marginal damage per unit of emission varies across sources, and if so, the extent to which the rules determining how permits are traded reflect any variation in marginal damages, u In general, the marginal damage per unit of emission depends both on the effect of emissions on ambient environmental quality and on the marginal damage associated with a decrease in ambient quality. Both of these relationships can vary significantly across sources of emissions at different locations. The origins of this variation include differences in the ability of the environment to absorb emissions, differences in the characteristics and numbers of agents experiencing the impacts of the emissions (e.g., urban vs. rural areas), and nonlinearities in the damage function. irst consider a simple case where the marginal damage per unit of emission is the same for all sources. An example would be a globally or regionally well-mixed pollutant with a uniform ambient concentration. The effect of a one unit increase in emissions on ambient concentration and damages is therefore independent of the location of the source of the emission. The key feature of this case is that any new source must purchase permits for every unit of emission, and since the total number of permits available is fixed, emissions from other sources must be reduced by an
F
equal amount as they sell permits to the new source. Given these assumptions, the marginal social cost of a n e w source is the sum of the four quantities in Table 1. If the pollutant is uniformly mixed, the two marginal damage terms are equal and cancel out. ]2 If the permit market works effectively, the price of a permit will equal the marginal control cost at the source selling permits. The new source must include the price of its permits in its bid, so the bid will equal the sum of marginal production cost and
Table 1: Marginal social costs of
production at a source in the case of tradable emissions permits.
1. Marginal production costs at a new source, including control costs, plus 2. Marginalcontrol costs at the source selling permits to the new source, plus 3. Marginal dama .ge of emissions at new source, mll'lUS
4. Marginal damages avoided at the source selling permits and reducing emissions.
the cost of permits. This will be equal to the marginal social cost of the new source. In other words, the price of permits internalizes the cost of reducing emissions at other sources. The appropriate adder is thus zero. This is true regardless of whether the number of permits available in the market is optimal, too large, or too small. N o w consider the case where the marginal damage per unit of emission varies across sources. If the environmental agency has
The Electricity Journal
adopted optimal trading rules to reflect this variation, these rules will require that any source that desires to increase its emissions by purchasing permits from another source must purchase a sufficient number of permits to offset the marginal damages of its new emissions, rather than just the quantity of emissions. 13 For instance, suppose that marginal emissions at the new source create three times the damage as marginal emissions at an existing source. Optimal trading rules would require that the new source purchase three permits from the existing source for every one additional unit of emissions. Hence the price of one unit of emissions at the new source is three times the price of a permit. The new source will minimize its total costs by purchasing enough permits so that its marginal control cost is equal to the marginal permit cost it must pa3a gain, the marginal social cost of the new source is the sum represented in Table 1. Since the optimal trading rules require that the new source purchase three permits for each unit of its emissions, the last two components are equal to each other and cancel out. And the marginal control cost at the source selling the permits is internalized in the bid of the new source. So the adder is zero in the case of a permit system with optimally designed trading rules. Note that this result is obtained without making any assumption about the optimal level of environ-
A
m
_
mental quality or the number of permits issued for trading. N o w let's see what happens when the trading rules are not optimal. Suppose that in the above example permits are traded one for one, even though marginal damage per unit of emissions at the new source was three times as great. Then, the last two components of the summation in Table 1 would not cancel out. Although the private bid would internalize the marginal control costs imposed on the selling source through the permit price, the difference between the marginal damages at the two sources would have to be added to the bid to reach the correct marginal social cost. In other words, the environmental adder would be the difference between the marginal damages at the new source and the selling source. If the marginal damages at the selling source were greater than the marginal damage at the new source, the adder would be negative. Alternatively, suppose that the marginal damages per unit of emissions are the same at both sources but that permits do not
trade on a one for one basis. For example, suppose that the new source must purchase more than one permit for each unit of its emissions so that there will be a reduction in total emissions24 Then, the correct adder is based on the product of marginal dam~ges and the net change in emissions. If total emissions are reduced, the adder is negative, is III. C o n c l u s i o n s
The correct environmental adder depends on the nature of the environmental policy instrument and the details of the implementation of that instrument. The results that demonstrate this are summarized in Table 2. In the case of CAC regulation, which is the typical case in the U.S., the adder equals marginal damage. However, the correct adder is not always equal to marginal damage. The adder will be zero when the true costs of the new source, broadly construed, are fully internalized in the private financial cost of the bidder. With a tradable emission permit system and optimal trading rules, the adder is always zero because
Table 2: Summary of AddersunderVarious Regulatory Regimes Level of EnvironmentalControl Policy Instrument
Under Control
Optimal Control
Over Control
Emissiontax
Adder=MD-t>O Adder--MD-t--O Adder--MD-t
Tradable EmissionPermits: OptimalTradingRules
Adder = 0
Tradable EmissionPermits: Non-op~malTradingRules
Adder-- MD(new) - MD(selling)
Command and Cont]'ol
Adder= MD
.
August/September 1992
23
the cost that the new source imposes on society is the additional control cost at the selling source and this cost is already internalized through the cost of permits. But if the trading rules are not optimal, the adder equals the difference in marginal damages across trading sources. here are a number of issues and problems involving social costing and electricity supply planning that we have not dealt with. One is that using externality costs only in capacity planning and not in dispatch decisions can lead to a bias against new sources. As a consequence, older dirty sources might be kept in service longer and are dispatched more heavily because the dispatch decisions do not take account of their external costs2 6 Of course, the answer to this problem is to use correctly calculated externality costs in dispatching decisions as well as in capacity decisions. Although the marginal damages associated with an increment to capacity may be different from
T
the damages associated with a change in the rate of the utilization of a facility, the principles governing the relationship between marginal damages and the correct adder will be the same. In the interim, if external costs are considered for resource supply planning only, it is important to evaluate investments in life extensions for existing facilities on an equal footing with new sources. The present procedures for social costing include external costs in supply decisions but not directly in setting electricity prices. Thus the demand for electricity is higher than it should be because consumers may not be getting the correct price signals. Of course this deviation from efficient pricing is an overlay to existing pricing practices that typically deviate from marginal cost pricing. Whether the net effect is to amplify or diminish departures from efficient pricing is a technical and empirical question best reserved for another discussionJ 7 However, other things being equal, one
A zero adder is correct only in certain circumstances.
24
can see that consumers are not being forced to recognize the environmental consequences of their electricity use. nother problem is that the use of social cost adders in the least-cost capacity planning process only for the utility system puts new plants at a competitive disadvantage. Some utility customers may find it privately advantageous to generate their own electricity, switch fuels, or even relocate their industrial facilities rather than purchase electricity from a utility that is making decisions on the basis of least social cost. This "by-pass" problem is a consequence of applying environmental costs only to utilities. If nonutility sources of energy have uninternalized environmental costs associated with them, social welfare could be reduced by requiring externality adders only in the utility system. These two problems of incorrect electricity price signals and bypass arise not because social costing is wrong, but rather because it is incomplete in its coverage. The solution to these problems is to develop a comprehensive system for internalizing environmental costs not only on utilities but on other sources of these emissions as well. This can be done either through a comprehensive system of taxes on emissions or a comprehensive marketable emissions permits system. This suggestion is not new. What is new is the alliance between proponents of market-based approaches to environmental policy and utility regulators who have taken leader-
A
The Electricity Journal
ship positions in the social costing movement. Together, they can push for an extension of externality pricing as the answer to the criticism that their present second best approaches might make matters worse. • Endnotes:
1. S. D. Cohen, J. H. Eto, J. Beldock, and G. Crandall, A Survey of State Regulatory C o m m i s s i o n Activities to Incorporate Environmental Externalities into Electric Utility Planning and Regulation, Lawrence Berkeley Laboratory (1990). 2. The distinction between these and other n o n e n v i r o n m e n t a l external costs (and benefits) that appears in some discussions is artificial. See D. Burtraw and A. J. Krupnick, The Social Costs o.~ Electricity: H o w Much of the Camel to Let into the Tent?, Discussion Paper QE92-15, Ressources for the Future, Wash., D.C. (1992). 3. There are presently u n d e r w a y two major research efforts and several smaller ones to d e v e l o p such methods and models. One major effort is jointly sponsored by the U.S. Department of Energy and the Commission of the European Community. The other is the N e w York State Environmental Externalities Cost Study. See also R.L. OTT1NGER, ET AL, ENVIRONMENTAl, COSTS OF ELECTRICITY (1990).
4. Several states, including Massachusetts and Nevada, have adopted this approach. The rationale for it is that environmental regulators have revealed a "willingness to pay" for reduced environmental impacts by imposing pollution control requirements with these costs. There are at least three sufficient reasons for rejecting this rationale. First, the typical uniform emissions or discharge standard results in very different marginal control costs across different sources of the same pollutant; so there is no unique marginal control cost curve to intersect with a marginal d a m a g e curve. Second, the same a m o u n t of
Augus t/September 1992
discharge of a pollutant is likely to result in very different marginal d a m ages across different locales. Third, most of the federal environmental laws do not require, and some preclude, balancing of benefits and costs in establishing control requirements. So there is no reason to expect any particular relationship between marginal damages and marginal abatement costs. 5. Several of the issues w e address were identified by Professor Paul Joskow in this journal. See P.L. Joskow, Weighing Environmental Externalities: Let's Do It Right!, ELEC. J., May 1992, 53-67. 6. For a different v i e w of the responsibilities and objectives of public utility
So the adder is calculated as (MD - t) times the level of emissions per kWh. 10. Proposed new sources of air emission in regions presently not attaining primary national air quality standards must certify offsetting reductions in emissions from other sources as a condition for obtaining an emissions license. They m a y obtain offsets by ipurchasing reductions in emissions from other sources. 11. For a rigorous demonstration of this point, see B.F. Hobbs, What Do SO2 Emissions Cost?, External Costs and Allowance Prices U n d e r the 1990 Clean Air Act A m e n d m e n t s , working paper, Case Western Reserve U. (1992). 12. There is a change in the incidence of damages. Some gain as their damages are reduced while others lose. But the net effect is zero. 13. For details concerning optimal trading rules, see, e.g., T.H. TIETENBERG, EMISSIONS TRADING: AN EXERCISE IN REFORMING POI.LUTION POI.-
IcY (Resources for the Future, Wash., D.C. 1985) and P. Bohm and C. S. Russell, Comparative Analysis of Alternative Policy Instruments, in HANDBOOKOF ENVIRONMENTALAND RESOURCE ECONOMICS, vol. 1 (A. V. Kneese and J. L. Sweeney, eds., 1985). 14. This requirement is a feature of the present federal offset program. regulation, see P.A. Agathen, Dealing with Environmental Externalities, PuB. UrIL. FORT., Feb. 15, 1992, at 23-24. 7. Joskow, supra note 5, alleges this is the agenda for some utility regulators. 8. Currently there are no examples of these in the U.S., although they are prevalent in Europe. ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT, ECONOMIC INSTRUMENTS FOR ENVIRONMENTAL PROTECTION
(1989). 9. The statement that the a d d e r is equal to the difference between marginal d a m a g e and the tax (MD - t) is an over-simplification. MD and t are usually expressed in units of dollars per unit of emissions, while the adder is expressed in units dollars per kWh.
15. Joskow, supra note 5, is reluctant to offer such a recommendation in this situation. However, we think such a policy is well founded and should be favored by the environmental c o m m u nity, because it gives the utility credit for i m p r o v e m e n t in environmental quality. 16. K. Palmer and H. Dowlatabadi, Implementing Environmental Costing in the Electric Utility Industry, Quality of the Environment Division Discussion Paper QE91-13, Resources for the Future, Wash., D.C. (1991). 17. See D. Burtraw, W. Harrington, A. M. Freeman llI and A. J. Krupnick, The Analytics of Social Costing, Res. for the Future, Wash., D.C., unpublished manuscript (1992).
25