Essays in Renewable Energy and Emissions Trading

Material Information

Essays in Renewable Energy and Emissions Trading
Kneifel, Joshua
Place of Publication:
[Gainesville, Fla.]
University of Florida
Publication Date:
Physical Description:
1 online resource (216 p.)

Thesis/Dissertation Information

Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Committee Chair:
Kenny, Lawrence W.
Committee Members:
Hamilton, Jonathan H.
Berg, Sanford V.
Luzar, E. Jane
Graduation Date:


Subjects / Keywords:
Coal ( jstor )
Compliance costs ( jstor )
Excess demand ( jstor )
Market prices ( jstor )
Prices ( jstor )
Scrubbers ( jstor )
Simulations ( jstor )
Sulfur ( jstor )
Total costs ( jstor )
Unit costs ( jstor )
Economics -- Dissertations, Academic -- UF
electricity, emissions, energy, environment, environmental, pollution, renewable, utility
Electronic Thesis or Dissertation
born-digital ( sobekcm )
Economics thesis, Ph.D.


Environmental issues have become a key political issue over the past forty years and has resulted in the enactment of many different environmental policies. The three essays in this dissertation add to the literature of renewable energy policies and sulfur dioxide emissions trading. The first essay ascertains which state policies are accelerating deployment of non-hydropower renewable electricity generation capacity into a states electric power industry. As would be expected, policies that lead to significant increases in actual renewable capacity in that state either set a Renewables Portfolio Standard with a certain level of required renewable capacity or use Clean Energy Funds to directly fund utility-scale renewable capacity construction. A surprising result is that Required Green Power Options, a policy that merely requires all utilities in a state to offer the option for consumers to purchase renewable energy at a premium rate, has a sizable impact on non-hydro renewable capacity in that state. The second essay studies the theoretical impacts fuel contract constraints have on a electricity generating unit's compliance costs of meeting the emissions compliance restrictions set by Phase I of the Title IV SO2 Emissions Trading Program. Fuel contract constraints restrict a utility?s degrees of freedom in coal purchasing options, which can lead to the use of a more expensive compliance option and higher compliance costs. The third essay analytically and empirically shows how fuel contract constraints impact the emissions allowance market and total electric power industry compliance costs. This paper uses generating unit-level simulations to replicate results from previous studies and show that fuel contracts appear to explain a large portion (65%) of the previously unexplained compliance cost simulations. Also, my study considers a more appropriate plant-level decisions for compliance choices by analytically analyzing the plant-level decision-making process to show how cost-minimization at the more complex plant-level may deviate from cost-minimization at the generating unit level. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis (Ph.D.)--University of Florida, 2008.
Adviser: Kenny, Lawrence W.
Statement of Responsibility:
by Joshua Kneifel.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Kneifel, Joshua. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
LD1780 2008 ( lcc )


This item has the following downloads:



























































































































































































































Full Text

significant, which brings up concerns about the variable coefficients and any possible

biases due to the missing data. The most important result from this specification is that

the additional variables have no effect on the policy variable coefficients, which remain

relatively unchanged relative to the original model.

The coefficient on SUGARCANE PROD CHANGE is insignificant. The missing fuel

cost dummy variable interacted with generation (M\ISSING FUEL COST*GEN) controls

for any measurement error caused by the extrapolation of the 38 missing data points and

is insignificant as well.

1.5.2 Regulatory Policy Variables: Ri

Table 4 estimates the statistically significant effects from both the control variables

and the policy variables based on a state with median generation levels. Clean Energy

Funds, Renewables Portfolio Standards with Capacity Requirements, and Required Green

Power Options have statistically significant effects on renewable capacity in the electric

power industry. Renewables Portfolio Standards with Generation/Sales Requirements and

State Green Power Purchasing Programs are marginally significant in Specification 1, but

lose their significance once control variables are introduced into the model.

CEF: CAP FUNDED, which measures the amount of capacity that the fund has

agreed to help finance, has a marginally statistically significant coefficient. This includes

capacity that has been agreed upon, but has not yet been built, either because the project

has not been finished or the project is later canceled. For each megawatt of capacity that

the Clean Energy Fund has funded or agreed to fund in the near future, approximately

0.206 MW has been constructed. This is not significantly different than the fraction of

capacity that has actually been constructed as of 2003, which was 0.33 MW per 1 MW.

Even though actual renewable capacity is probably not constructed linearly over the

lifetime of the policy, the estimates from the linear interpolation seem to be representative

of actual capacity construction due to the Clean Energy Funds.

If a generating unit faces PA > MCA"'8, it will use the maximum amount of low

sulfur spot market coal, which can be found from (3-41). Assuming only low sulfur spot

market coal use to meet demand, the maximum amount of low sulfur coal is expressed in

S- (3-59)

Replacing C i in (3-41) with the expression in (3-59) for C~,M gives an expression
for the minimum allowance excess demand in (3-60). A generating unit's minimum

excess demand must cover the difference between its initial allowance allocation (Ar)

and the amount of allowances needed to cover the unit's minimum actual emissions

[EJMIN ziri) (m)(Sl) ( )]. If a unit's initial allocation can cover its minimum

possible emissions, then it will have a negative net allowance position and be a net buyer

seller of allowances.
AMIN EMIN A (3-60)

If PA = MCA"'8, a generating unit may use any combination of high sulfur spot

market coal and low sulfur spot market coal and lead to any level of excess demand in the

range (E MIN A', EMAX A). The allowance excess demand can be represented by

A = (pEMAX (1 p)EIN A) where the constant p [0,1]. A unit that is indifferent

between fuel switching and allowances purchasing could be either a net buyer or a net


Combining the excess demands for each of the three cases creates the excess demand

correspondence shown below.

A AX if PA > MCAj',
Ai = pAAX (1 p)AMIN if PA MCA' Vp e [0,1]

A^IN if PA < MCA7',

A generating unit's excess demand correspondence can be seen graphically in Chapter 2.

High sulfur coal use corresponds to the right-hand vertical line where PA < MCA"'8. Low

(iii) For the range of allowance prices PA < MCA"'8, a high sulfur coal contract will
weakly increase excess demand if S > .h
Hfh H Tih
Proof of Proposition 2(i):

For allowance prices PA > MCA"', a unit prefers to use all low sulfur coal, which

leads to the minimum allowance excess demand (AMIN). A high sulfur coal contract

forces some high sulfur coal use and decreases low sulfur coal use from CSMAX to C8,MAX

which is the maximum amount of low sulfur coal a unit will use given its high sulfur coal

sMAX .sMAX Di (20)
CIS,, > JihH (240)

The decrease in low sulfur spot market coal use increases emissions from EMIN to E IN

which is the minimum emissions given the high sulfur coal contract constraint:

MN MIN (1 zr)( m) (S : s,MAX + Sicj (-41)
Si < r+ 2^i

Higher emissions must be covered by additional allowances, which results in the minimum

excess demand with a high sulfur coal contract to be greater than the minimum excess

demand with no high sulfur coal contract:

= AMIN < Ai, v -MIN A (2-42)

Therefore, a high sulfur coal contract will increase excess demand for allowance prices

PA > MCA8'".E

Proof of Proposition 2(ii) and 2(iii):

For allowance prices PA < MCA"'8, a unit prefers to use all high sulfur coal, which

leads to the maximum allowance excess demand (AjMAX). A high sulfur coal contract

forces a unit to decrease its high sulfur spot market coal use from CMAX to CMAX

Cis,MAX ^s,MAX Di -c H (24
ih >ih CH (2-43) i
H^sh i

The minimum unconstrained total industry compliance costs to the industry is

derived from the following problem. Each generating unit chooses its optimal A Ci >

0, C[* > 0, z* E {0, 1} based on the market equilibrium allowance price, P1. The

total compliance costs as shown in the objective function in (3-26) are the sum of the

compliance costs for each generating unit. Similar to the generating unit's problem in

C'!i lpter 2, there are two constraints that must be met. First, the sum of allowance

allocations must cover the sum of emissions that are emitted by all affected generating

units. Second, each generating unit must produce enough electricity to cover its electricity

demand requirement.

mmin z Pi + max{(P ClI + P Ch PCMAx), 0} (3-26)
zi,Ai',Ch'C i 1

s.t. (t zr) (Ci Si + CSi) (m) < Ae
i= 1 i= 1
(C$H C + CfHf) >D, Vie {1,...,n}

ze {0, 1} Vie {1,...,n}

If all generating units' cost-minimizing decisions result in an equilibrium in the

allowance market, then total industry compliance costs are minimized. We can prove this

by showing the first-order conditions for the minimized total industry compliance costs

match up with the first-order conditions from the generating unit's problem in ('! Ilpter 2,

and the optimal choices will be the same for both problems.

Solve for the first-order conditions given zi. For high sulfur coal:

Pih + A i(l zriTi)()(Sih) Ai2Hfh > 0, 0 if Cih > 0 (3-27)

For low sulfur coal:

Pai + A1(1 z r)(m)(Sjt) Ai2H > 0, 0 if Ca > 0 (3-28)

reductions can lead to significant differences between coal contract prices and spot market

coal prices.9

Ellerman and Montero (1998) found that investment and innovation in coal

production and delivery as well as greater competition between railroads due to the

Si .-.-. i Rail Act of 1980 created lower coal prices during the first year of Phase I,

especially for low sulfur coal from the Powder River Basin. These lower coal prices led to

lower marginal costs of abating SO2 emissions through fuel switching, which is reflected in

the lower than expected allowance price in 1995 (Burtraw et al., 2005). Considering the

downward rigidity of contract coal prices, the same coal price reductions also resulted in

lower spot market prices for both high sulfur and low sulfur coal relative to the coal prices

under contract.

2.4 Inefficiencies Resulting from Coal Contract Constraints

There are three plausible scenarios where binding contract constraints results

in sub-optimal compliance choices. First, during Phase I a utility with high sulfur

coal contracts may be unable to switch to low sulfur coal for compliance when it is

9 A utility cannot sell contracted coal on the spot market because of the transaction
costs involved in selling to another utility from both its contract for coal purchases and
its contract for coal transportation. A coal contract sets a given type and amount of coal
for an agreed upon price from a particular coal source. A transportation contract sets a
given price for delivery of coal purchases from a coal source. The combination of these
two contracts results in the delivered cost of a coal purchase. For one utility to sell coal
to another utility, it would need to either buy out its contract with a provision to deliver
the coal to the other utility or it would need to p ,li for the shipment of the coal from its
facility to the other utility. There are large additional costs associated with either of these
actions. As shown in Joskow, few contracts were bought out, breached, or renegotiated.
Also, there does not appear to be any sales of coal from one utility to another.

Consider a simple example where a plant operates two generating units, one that

is affected (Unit "i") and another that is not affected (Unit "j") by Phase I of Title IV.

Assume P < which is normally the case under Phase I.13 Furthermore, the plant
h I
prefers to switch from high sulfur coal to low sulfur coal to meet its emissions requirement

for Unit i because PA > MCA"'8. But the plant also has a high sulfur coal contract for

a small amount of high sulfur coal (Cc < ) that has identical characteristics to high

sulfur spot market coal. The plant must choose at which generating unit to use the high

sulfur contract coal. Since plant prefers to use low sulfur coal to lower its emissions at

Unit "i" and prefers to use high sulfur coal at Unit "j" to minimize its coal costs, then

the plant's contract will not bind and will not increase a plant's total costs. Alternatively,

if Unit "j" was an affected unit and PA > MCA"'7, a high sulfur contract would bind

and result in higher total costs for the plant. A non-affected unit will weakly decrease a

plant's total costs relative to if the unit WAS affected. High sulfur coal can be shifted to

the unaffected unit to relax the emissions constraint.

3.7.8 Scrubber Installation Choice

Up to this point, we have assumed a plant's scrubber choice for a generating unit

as given. Now consider a plant's scrubber installation choice for a given generating unit.

A plant chooses to use contract coal in different generating units depending on whether

it has installed a scrubber at the particular generating unit. The choice is based on the

scrubber choices because it is the only variable that can change the marginal cost of

abatement across a plant's generating units. Marginal cost of abatement with and without a scrubber

Finding the allowance price at which a generating unit will install a scrubber is not as

simple as in the generating unit model because a plant must take into account the costs of

13 Over i .'-. of generating units had higher delivered prices for low sulfur coal than high
sulfur coal on the spot market. Case 3: Install two scrubbers

For a plant to install a scrubber at both generating units, two conditions must hold.
First, the total costs of installing two generating units must be lower than the total costs
of installing one scrubber.

Pi, l PAA1+ PCi + P1 + C1 + PC(h + P CT1 + P,2

+PAA2 + PhC + PC1 + Pl C2 + Pl 1C

< + PAAI + P T + Pih + C + PhC h + P

+PAA2 + Ph + PIfCS + PC h + PC C02

From this equation, we can solve for PA for this condition to hold in (3-85).

D I psi cs _rs \sj Is s1s Cs S
P2 + rPhs(h ~h) + P1 W(2l C1) < PA (3-85)
(A2 A2)
Second, the total costs of installing a scrubber at both generating units must be lower
than the total costs of installing no scrubbers.

P, + PAAI + P l + PPCI + P Cih + P C1 + P2,

+PAA, + P"' P PCh 1 + PhC( + Pl FC

< PAA + PI'C;h + PfCl1 + PhCoh + PcC11

+PAA2 + P Ch + P1C1c + PC h + P C21

From this equation, we can solve for PA for this condition to hold in (3-86).

P1 + P (C'h CO) + P(Ci Ci) P2 + P(2h Ch) + P T Cs) Ps 86)
(A1 A) + (A2 A2)

The allowance price must be greater than both of these two indifference prices for two
scrubbers to be installed.

A binding high (low) sulfur coal contract constraint leads to two possible cost

inefficiencies, but only one of which increases compliance costs. First, a contract could lead

to a generating unit using more than the optimal amount of high (low) sulfur coal and

less than the optimal low (high) sulfur coal. In this case, actual compliance decisions are

altered and may lead to a sub-optimal coal mix. Second, a contract may force a unit to

use some high (low) sulfur coal that is "more exp' .i-,i. than the high (low) sulfur spot

market coal price, which increases total costs while leaving compliance costs unchanged. In

previous studies, both of these cost inefficiencies were identified as excess compliance costs

because contract restrictions were ignored in the baseline costs without the program (c).

( z +P +P +P +PC c +PC- c) > (zP P A* +PC,* +P, C* c)


The minimum compliance cost choices with no contract constraint for previous studies

are denoted by "*" while the contract constrained minimum compliance cost choices are

denoted by "A" in (3-20). The assumed "c" is the total costs of meeting demand assuming

no contract constraint or emissions constraint. The cost difference between these two sets

of choices will be the combination of changes in compliance decisions and fuel costs. Some

of these :; -- compliance < n-I may not be compliance costs, which makes the derived

compliance costs too high. So it is important to include the contract constrained coal use

in the baseline "c" to get the appropriate compliance costs. Assuming that contracted coal

is at least as expensive as spot market coal, the compliance costs will be greater for the

contract constrained case as shown in (3-20) and as expressed in (3-21).

Sz( z4) + P(A, Af+P(Csf ( Csf) + P(C PCi*P) C + P c > 0 (3-21)

Consider a simple example where high sulfur contract coal is more expensive than

high sulfur spot market coal, low sulfur coal is preferred over high sulfur coal, and no

scrubber would be installed if a generating unit had total freedom in its coal use choices.

Now compare a generating unit's compliance costs to the compliance costs a unit would

impacts by contract constraints. As can be seen in Table 3-4, contract constraints increase

scrubber installations in Ohio, Alabama, Florida, Indiana, Mississippi, and Missouri and

decrease scrubber installations in West Virginia, New York, Pennsylvania, Wisconsin,

Georgia, Kentucky, and New Jersey.

An unexpected result in Simulation 6 is that scrubber installations actually increase

relative to Simulation 5. These are likely a result of the assumed scrubber cost estimates

because the historical scrubber cost data used in the simulations is lower than the

engineering cost estimates. Scrubbers appear cheaper to install than the engineering

estimates state. Of the 404 generating units with at least a 911'. removal rate, 31 historical

capital cost estimates are higher, 84 are the same, and 289 are lower than the estimated

engineering capital costs.8 The higher cost estimates are not much higher than the

engineering costs with a difference of $11/kW. However, the historical capital cost

estimates that are lower than the engineering capital cost estimates range from $50/kW

to $216/kW higher, which could have some significant impacts on scrubber installation

choices. For example, historical capital cost estimates for generating units in Mizzouri are

assumed to be $50/kW while the engineering cost estimates are assumed to be $266/kW,

or a 43 difference.

As can be seen in Table 3-5, there is a significant difference in scrubber installations.

The use of engineering cost estimates for scrubber installation results in a decrease in

scrubber installations from 44 to 25 in Simulation 4 and from 61 to 38 in Simulation 6

because of the higher costs involved. Notice that introducing contract constraints results

in additional scrubber installations from 25 to 38, which is close to the actual installations

of 46 scrubbers. Compliance cost interpretations remain similar to simulations using

8 27 of the generating units are not directly comparable because historical data (50'".
removal rate) assumes a different scrubber technology than the engineering data (9, '.
removal rate).

occurred during Phase I because some generating units purchased 1C(i' high sulfur coal

through contracts in 1996.20

High sulfur coal is assumed to be the preferred coal to use prior to Title IV as units

would wish to use the cheapest coal regardless of sulfur content, which was usually high

sulfur coal. In Figure 2.8, a unit prefers to use low sulfur coal because PA > MCA"'8, and

the compliance costs from switching fuels to abate emissions can be seen in area (a + b).

Area (b + c) is the revenue gained (negative cost) from selling the remaining allowances

that are available due to abating emissions below a unit's initial allowance allocation. The

net compliance costs for a unit will be the costs of switching fuels minus the revenues from

allowance sales, or (a + b) (b + c) = (a c). If (a c) < 0, then a unit will have negative

compliance costs.

Now consider how a high sulfur coal contract will impact the excess demand

correspondence, compliance costs, and total costs. As has already been shown above

and can be seen in Figure 2.8, a high sulfur coal contract will increase the minimum excess

demand from A"IN to A~If and may increase or decrease the maximum excess demand

depending on the relative sulfur to heat content ratio of contract to spot market coal.

These shifts in a unit's excess demand may have three distinct effects on a unit's

costs. The first cost impact is the additional compliance costs to a unit from lost

allowance sales, which is represented by area (d) in Figure 2.8. A unit must use some

high sulfur coal, which results in a unit covering additional emissions through more

expensive allowances instead of switching to low sulfur coal.

The second impact results from the difference in the sulfur to heat content ratio
between high sulfur contract and spot market coal. If h < then the maximum

emissions a unit can create will decrease and AjAX < AMAX. This will decrease the

compliance costs by (f) in Figure 2.8(i) for reducing emissions to the allowance allocation

20 Available from FERC-423 Data.

more unit of low sulfur spot market coal is weakly less than or equal to the additional

costs of using one more unit of high sulfur spot market coal inclusive of emissions and

demand requirements.

P i + A,(l ziri)(m)(S) Ai2H > P8 + Ai(l zir)(m)(S) AX2H (3-55)

Once again we can derive the relation of PA to MCA'"8. By comparing these two

expressions and solving for Ai2, you get an inequality comparing Ai = PA and MCA'".

Since the generating unit uses low sulfur spot market coal, the allowance price is weakly

greater than the marginal cost of abatement (PA > MCA"'8) as shown in (3-56).
ps ,ps

= PA > MCA's -Hf (3-56)

3.7.5 Excess Demand Correspondence

When only considering the use of spot market coal (no contract constraints), a unit's

excess demand correspondence will look nearly identical to the correspondence in ('!i plter

2. The minimum and maximum excess demand for allowances can be derived for each of

the three cases described above in the same manner as in C'!i plter 2.

First, if a generating unit faces PA < MCA"'j it will use the maximum amount of

high sulfur spot market coal. The maximum amount of high sulfur spot market coal is

expressed in (3-57).

C8 MAX (3 57)
h ih

Replacing CQ in (3-41) with the expression in (3-57) for Ck',MAX gives an expression

for the maximum allowance excess demand in (3-58). A generating unit's maximum

excess demand must cover the difference between its initial allowance allocation (Ar)

and the amount of allowances needed to cover the unit's maximum actual emissions
[EMAX (1 X)(m)(SnZ)S)( )].

A AX = EAX Ae (358)


[1] T. Arimura. An empirical study of the SO2 allowance market: effects of PUC
regulations, J. Environ. Econ. Manage., 44 (2) (2002) 271-289.

[2] R. E. Baldwin, C.S. Magee, Is trade policy for sale? Congressional voting on recent
trade bills, Public C'! i... 105 (2000) 79-101.

[3] L. Bird, M. Bolinger, T. Gagliano, R. Wiser, M. Brown, B. Parsons. Policies and
market factors driving wind power development in the United States, Energy Pol. 33
(11) (2005), 1397-1407.

[4] D.R. Bohi. Utilities and state regulators are failing to take advantage of emission
allowance trading, Electricity J., 7 (2) (1994) 20-27.

[5] D.R. Bohi, D. Burtraw. Utility investment behavior and the emission trading market,
Resources and Energy, 14 (1992) 129-153.

[6] D.R. Bohi, D. Burtraw. SO2 allowance trading: How experience and expectations
measure up, Discussion Paper 97-24, Resources for the Future, Washington, D.C.,

[7] M. Bolinger, R. Wiser. 2004. The impact of state clean energy fund support for
utility-scale renewable energy projects, Lawrence Berkeley National Laboratory and
Clean Energy States Alliance.

[8] M. Bolinger, R. Wiser, G. Fitzgerald. 2006. The impact of state clean energy fund
support for iul iii ---I i1.- renewable energy projects, Lawrence Berkeley National
Laboratory and Clean Energy States Alliance.

[9] M. Bolinger, R. Wiser, G. Fitzgerald. An overview of investments by state renewable
energy funds in large-scale renewable generation projects, Electricity J. 18 (1) (2005)

[10] M. Bolinger, R. Wiser, L. Milford, M. Stoddard, K. Porter. Clean energy funds: An
overview of state support for renewable energy, Energy Efficiency and Renewable
Energy, Office of Power Technologies, Department of Energy, 2001.

[11] D. Burtraw. Cost savings sans allowance trades? Evaluating the SO2 emission trading
program to date, Discussion Paper 95-30, Resources for the Future, Washington,
D.C., 1995.

[12] D. Burtraw, Evans, A. Krupnick, K. Palmer, R. Toth. Economics of pollution trading
for SO2 and NOx, Discussion Paper 05-05, Resources for the Future, Washington,
D.C., 2005.

[13] C. Carlson, D. Burtraw, M. Cropper, K. Palmer. Sulfur dioxide control by electric
utilities: What are the gains from trade?, J. of Political Economy, 108 (6) (2000)

3.5 Compliance Costs

Total compliance costs for a generating unit are the additional costs due to the

li. -' ,ri including costs from switching fuels, the costs from its net allowance position,

and scrubber installation costs (seen in (3-18)).

zPi, + PjA, + max{(PCfi, + PFCf Pi MAX), 0} (3-18)

Compliance costs may be positive or negative depending on its compliance decisions

and its initial allowance allocation. The scrubber installation costs are represented by

P,, and will only attribute to a unit's compliance costs if a scrubber is installed in

response to the program (zi = 1). The costs of a unit's net allowance position is the

difference between a generating unit's initial allowance allocation and its actual emissions

multiplied by the allowance price (PAAj). The costs of switching fuels is the larger of

two values: (1) total costs of actual coal purchases (PCihh + Pji C + PFhC + PFCi)

minus the costs of purchasing only high sulfur spot market coal given any contracted
-s s ,M A X '. c cc
coal (P CA M + PFhC + PFC?), or (2) zero. The latter will only occur if it is weakly

cheaper for the generating unit to use low sulfur coal without the emissions restrictions

( > H7). It is important to consider that the contracted coal will be used regardless
ih ii
of the program and will have no direct impact on compliance costs. However, a contract

could have an indirect impact by altering compliance decisions.

3.5.1 Compliance Costs with Coal Contracts Relative to Compliance Costs
from Previous Studies

Previous studies have assumed no restrictions on coal use, which results in a different

estimation of compliance costs. The optimal compliance choices in this case will not

account for any coal use restrictions. The constant "c" is the minimum costs for a

generating unit to meet its electricity demand if emissions and coal use are not restricted.

Pi + PA + PC + P C c (3-19)

There are two cases for which the value of c must be derived to determine the

contracts impact on the allowance indifference price, the first of which will have two

subcases. As will be shown for each of the cases, if ( > ), then c will be positive and

may decrease the allowance price at which a unit will prefer to install a scrubber.

In the first case, without a high sulfur coal contract, a unit prefers to use low sulfur

spot market coal without a scrubber and high sulfur coal with a scrubber. The two

subcases will be determined by the generating unit's characteristics and the size of the

contract constraint.

In the first subcase, both with or without a high sulfur coal contract, a unit prefers

to use low sulfur spot market coal if it does not install a scrubber and high sulfur spot

market coal it does install a scrubber. For this to hold, (MCA"'8 < PA < MCA7'8) and

(MCA'<" < PA < Pf).

If a unit that has no high sulfur contract coal (C =- 0), a unit uses the maximum

amount of high sulfur coal with a scrubber and the maximum amount of low sulfur coal

without a scrubber, and is indifferent to installing a scrubber at allowance price PS:

Pi + PASMAX I ps i,MAX S MIN + Pss,MAX (2 61)
4iz + P Ai + ii A i + il

(2-61) can be rearranged to find an expression for Pi:

SF = PS(A MIN ASMAX pS cs,MAX ps s,MAX (2 62)

A high sulfur coal contract will change the optimal values of the other parameters, which

will change the indifference allowance price by some value e:

S+ (PA e MAX + PM i PS ii} V psAY P iMh (2 ps MAX -63)
P. + (PjS )+ (P c)A;i p1 picc (2 63

(2-63) can be rearranged to find an expression for Pi:

(ptS (A' I ASMAX) + pssMAX s issMAX (2 64)
,^ ( ) \ A ~ ~) + i 'ilil ;ih ih.

type, C- for high sulfur coal and CU for low sulfur coal. The amount of contract coal at

a generating unit is a choice variable because a plant is able to choose which generating

unit to use its allotted contract coal. The Lagrange multiplier of each contract coal type is

represented by Ph for high sulfur coal and p, for low sulfur coal.

3.7.3 First-Order Conditions

The partial derivative with respect to Ai yields the impact of a one unit change in its

net allowance position on the unit's total costs.

PA Ail 0 (3-46)

Since Ai can be either positive or negative based on the net allowance position, (3-46) will

hold with equality. The additional cost to the firm of emitting one more ton of emissions is

equivalent to the allowance price, Ai = PA, which is the same result as from the previous


The partial derivative with respect to C[ represents the impact a one unit change in

high sulfur spot market coal has on the unit's total costs.

Ph + Ai(li zri)(m)(S)() Ai2H, > 0, 0 if Ch > 0 (3-47)

If the generating unit uses some amount of high sulfur spot market coal (C[ > 0), then

(3-47) holds with equality.

The partial derivative with respect to C1 represents the impact a one unit change in

low sulfur spot market coal has on the unit's total costs.

PI + Ail(l zzri)(m)(SS ) Ai2 H > 0, 0 if Ci > 0 (3-48)

Similar to high sulfur spot market coal, if the generating unit uses some low sulfur spot

market coal (Cu > 0), then (3-48) holds with equality.


MCA' -

-AM7N 0

Figure 2-12. Compliance Costs: PA

p $
P =MC ^S




4MN 4MMA 0 40 'N 0 ^M4X

Figure 2-13. High Sulfur Coal Contract




0 \ AM


MCA5,s --


A^~ jMW


0 AMr MA4X
4 (ii

Figure 2-14. Low Sulfur Coal Contract

MC4sc -
PA = MC4'-

3.6 Simulation Model

3.6.1 Introduction

It has been shown analytically that contracts may alter compliance choices by altering

coal use and scrubber installation and lead to greater compliance costs. These results may

be able to explain a large portion of the excess compliance costs found in previous studies.

This section will show how much of these excess compliance costs can be explained by

contract constraints.

The first portion of this section will look at the data used to parameterize the

simulation model, which includes a description of where the data was obtained, the

techniques used to create the parameters, and some issues regarding the data. The second

portion will look at the model design and approach. The last portion will summarize

the simulation results in terms of the total industry, individual states, and individual

generating units.

3.6.2 Data

All data used in the simulation, with an exception for the coal contract data,

were obtained from Dr. Paul Sotkiewicz. Using the same data allows for direct result

comparisons to determine the coal contract impacts. Dr. Sotkiewicz originally hand-collected

the the data from the EIA's "Cost and Quality of Fuels 1996," the EIA's "Electric Power

Annual 1997," and the EPA's "1996 Compliance Report."

Up to this point all generating units are assumed to be coal-fired units because it is

the primary fuel options for electricity generation in the U.S. However, there are 24 units

of the 431 units that used fuel oil or natural gas for electricity generation. The treatment

of these units is described in Section 6.2.1. Fuel data

Fuel data on heat content, sulfur content, and delivered price were obtained from the

EIA's "Cost and Quality of Fuels 1996", which compiled information from FERC Form

423. The coal contract data was gathered directly from the FERC Form 423 database for

industry choices are shifted, but that individual generating units are not able to make

their cost-minimizing choices. Generating unit scrubber installation choices

There are three issues to consider regarding generating units' scrubber choices: (1)

impacts from the endogenous scrubber choice without contract constraints, (2) impacts

from the endogenous scrubber choice with contract constraints, and (3) impact of contract

constraints on scrubber choices. First, comparing scrubber installations in Simulation 4 to

Simulation 3 will show how scrubber decisions would be altered if the scrubber decision

is made endogenous and contract constraints are excluded from the model. Allowing the

scrubber choice to be endogenous results in 48 generating units altering their scrubber

choice, including 25 scrubbers to be removed and 23 to be added for a total of 2 fewer

installed scrubbers. 21 generating units maintain the same scrubber choices, but 17 of

those were installed for NSPS. So only 4 scrubber choices remained the same.

Second, comparing scrubber installations in Simulation 6 to Simulation 5 will show

how scrubber decisions would be altered if the scrubber decision is made endogenous

and contract constraints are included in the model. Allowing the scrubber choice to be

endogenous results in 53 generating units altering their scrubber choice, including 19

scrubbers to be removed and 34 to be added for a total of 15 more installed scrubbers. 27

generating units maintain the same scrubber choices, but 17 of those were installed for

NSPS. So 10 scrubber installations remained the same.

Finally, it is important to consider the impacts contract constraints have on a

unit's endogenous scrubber installation decision, which can be determined by comparing

Simulation 6 to Simulation 4. 33 scrubber choices are altered as a result of contract

constraints, including 25 units that will now install a scrubber and 8 units that no longer

install a scrubber. Of the 25 units that chose to install a scrubber, 20 of them had a high

sulfur coal contract. Of the 8 units that chose to not install a scrubber, all 8 of them

had a low sulfur coal contract. So 28 of 33 scrubber choices appear to have been directly

Consider the following example for a plant with three generating units each requiring

200,000 mmBtu of heat input to cover their electricity demand and one unit has a

scrubber. Contracts account for 101' i. of coal purchasing agreements where there are

contracts for 300,000 mmBtu for both high sulfur contract coal and low sulfur contract

coal. Based on the distribution approach described above, Unit 1 will use as much of the

high sulfur coal as possible (200,000 mmBtu) and the remaining high sulfur coal will be

distributed equally among the remaining two units (50,000 mmBtu each for Unit 2 and

Unit 3). The low sulfur coal will be distributed to Unit 2 and Unit 3 equally because

neither has a scrubber installed (150,000 mmBtu each). Since contracts account for 101'.

of coal use, there is no need to purchase any coal on the spot market. If there had been

any excess coal, such as an extra 10,000 mmBtu of low sulfur contact coal, it would not

change the allocation in Table 3-1 and would be considered coal stored at the plant for use

the following year.

Sulfur content is the percentage of each ton of coal, barrel of oil, or 1,000 cubic

feet of natural gas that is sulfur. The data must be manipulated to create the desired

variable, which is pounds of sulfur dioxide per million Btus of heat. Phase I of Title IV

distributes allowance allocations based on 2.5 pounds of SO2 per million Btus of heat. So

fuel is considered "high sulfur" if contains greater then 2.5 lbs. of SO2 per mmBtu and

"low sulfur" if it contains less than 2.5 lbs. of SO2 per mmBtu. Any coal use that has a

higher (lower) sulfur content will increase (decrease) emissions above (below) its allocation


Getting the data in terms of pounds of sulfur dioxide per million Btus of heat requires

an emissions factor, which is the amount of sulfur dioxide emissions that will result from

a unit of sulfur. Emissions factors were found in the EIA's "Electric Power Annual 1997".

By taking the sulfur content multiplied by the emissions factor divided by the millions of

2 Coal sulfur content can vary significantly within each category (high or low sulfur).

2.5 Model and Parameters

The model is a static production cost model that draws heavily from Sotkiewicz

(2003) and Fullerton et al. (1997), which simulates production costs at the generating unit

level with constraints on demand for electricity and emissions levels, and introducing high

sulfur coal and low sulfur coal contract constraints. It would seem that adding contract

constraints to the model would not cause any in iir, disruptions. However, the model

results in rather complex interpretations due to how the contract constraints interact

with the non-convexities of a unit's scrubber choice. Let "i" be the index of units. The

parameters in the model are described below.

Technology Parameters:

* zi E {0, 1} represents a generating unit's discrete scrubber choice where zi = 1 if a
unit installs a scrubber and zi = 0 if a unit does not install a scrubber.

Pi represents the levelized yearly cost of a scrubber, which are the average annual
costs from depreciation and use of capital plus the operation and maintenance costs
of installing and operating a scrubber.15

ri E [0, 1] represents the scrubber emissions capture rate or emissions removal
efficiency rate, which is the fraction of emissions that the scrubber removes from the
exhaust stream. The removal rate is independent of the sulfur content of the coal
used by a utility because it removes some percentage of emissions after production.
Depending on the scrubber technology and vintage, it can remove 25-9' -. of SO2

Demand Parameter:

Di represents electricity demand, in million Btus of heat input, for a given
generating unit. Demand is derived by taking the total kilowatt-hours of electricity

15 The capital costs are assumed to be $260/kW under Phase I and $141.34/kW under
CAIR. Capital costs are discounted at a 10' rate based on a 20 year equipment lifespan
( (+di). The operation and maintenance costs are assumed to be 2.0 mills/kWh under
Phase I and 1.23 mills/kWh under CAIR.
16 Table 30: "Flue Gas Desulfurization (FGD) Capacity in Operation at U.S. Electric
Utility Plants as of December 1996" from the 1996 Electric Power Annual Vol. II

for installing one scrubber will be weakly less than the allowance at which a plant would

install scrubbers at both it's generating units. Case 2: Install one scrubber

We know that a plant will install a scrubber at the generating unit with the lowest

ACAi, which in this case is assumed to be Unit 1. For a plant to install one scrubber, two

conditions must hold. First, the total costs of installing one generating unit must be lower

than the total costs of installing no scrubbers.

Pi1 + PAA, + PCr + PC + PS + P + CI + AA2 + PSCs + PI C + PhIC

From this equation, we can solve for Ps for this condition to hold in (3-83).

Piz + P (Csh Ch) + P1 (C1o Ci)
A < A1 (383)

Second, the total costs of installing one generating unit must be lower than the total

costs of installing two scrubbers.

Piz + PAA1 + PhC' + PlC1 Ph + P/cCt

+PAA2 + P + P1 1 + P h + PI 1

SP + + PAAl + PhC + PC11 + PCh PC( + P2z

+PAA2 P + P1S + PhC C P'C

From this equation, we can solve for Pc for this condition to hold in (3-84).

D PTss PAs I ns( nds s_ Csl) S A r

(A2 A2)
The allowance price must be between these two indifference prices, which results in

PAS < PA for Unit 1 and PAS > PA for Unit 2.

Table 2-2. High Sulfur Coal Contract: Assumptions
Coal Type Price/mmBtu Hf Price/Ton
cQ $1.60 24 $38.40
Ch $1.30 24 $31.20
Ch (Ex. 1) $1.50 24 -:1, 10
Ch (Ex. 2) $1.20 24 -.- 0O

PA $300.00
A' 20,000 tons
Di 24,000,000 mmBtu

Table 2-3. High Sulfur Coal Contract: Results
Costs Compliance Total Costs
Costs (Ex. 1) (Ex. 1)
Unconstrained $4.62 million -2 million
Constrained $5.01 million $38.61 million
CI' ii,';, $390,000 $2.79 million

Allowance Use (Tons)
Minimum 11,400 tons
Maximum 38,000 tons
Constrained Min. 24,700 tons

Costs (Ex. 2)
$4.62 million
$5.01 million

MCAf" (Ex. 1)
MCAf'" (Ex. 2)

Total Costs
(Ex. 2)
-< ". million
S;". ill million

- 1 ,1g.90

Table 2-4. Low Sulfur Coal Contract Examples: Assumptions
Coal Type Price/mmBtu HI Price/Ton SY
CQ $1.60 24 $38.40 0..',
Ch $1.30 24 $31.20 2.0'C
cQ (Ex. 1) $1.80 24 ;.20 0..'
CQ (Ex. 2) $1.50 24 ., 1)0 0.,' ,

20,000 tons
24,000,000 mmBtu

0.i .' .
2.0' ,
2.0' ,
2.0' ,

A state's annual weighted average real fuel cost (in 2002 dollars) per million Btus

(FUEL COST) measures the impact of both a state's composition of fossil fuel mix and

a state's average costs for each fossil fuel type: coal, natural gas, and fuel oil.13 FUEL

COST captures the effects of all these variables, which may have offsetting effects on

renewable capacity. FUEL COST is used instead of creating separate variables for the

cost and capacity of each fossil fuel for several reasons. First, using one variable instead of

five variables simplifies the model. Second, data on specific fossil fuel costs are missing for

many states.14

Levelized cost of each renewable source is the estimated real cost of production

per kilowatt-hour of electricity over the lifetime of the equipment, including all federal

production incentives.15 It captures the economic competitiveness of each renewable

13 Fuel cost data can be found on the EIA website in the electricity databases section
under Monthly Cost and Quality of Fuels for Electric Plants Database (FERC Form-423).
The cost per unit, Btus per unit, and number of units purchased for every fuel purchase
made by all public utilities are used to obtain a nominal average fuel cost measure. The
data are ., i'--regated and deflated using the Consumer Price Index for all goods from the
Federal Reserve Bank of St. Louis to get the state's annual average real fuel cost per
million Btus in January 2002 dollars. FUEL COST has 30 missing observations for 8
different states. Idaho is the only state without any fossil fuel purchases. Estimates of
the fuel costs are used to fill in the missing data. The non-Idaho missing observations
are extrapolated from the existing data for a state from 1990-2003. Idaho's observations
are generated by using the average fuel costs of the states bordering Idaho. A missing
data dummy variable is included in the model to capture any bias created through the
extrapolation and approximation.
14 There are missing fuel cost observations for coal (69), natural gas (65), and fuel oil
(58). This might be due to no deliveries of a particular fuel to a state, or it could be the
missing observations are due to changes in data reporting requirements during the sample
15 Levelized cost is calculated by a model that accounts for the initial capital costs of
constructing the capacity, expected lifetime of the equipment, interest rates on debt,
inflation rate, fuel costs, operational and maintenance costs, capacity factor of the
equipment, and federal production incentives. Read McVeigh (1999) for a more detailed
description of levelized cost used in this paper.

positive excess demand (x > 0), p(x) =PA. If there is a negative excess demand (x < 0),

or excess supply, p(x) = 0. Since [0, PA] is compact, the graph of p(x) is closed, which

implies p(x) is upper semi-continuous.

Now define F(x, PA) i(x) x Am(PA). Since p(x) and Am(PA) satisfy all properties

needed to apply Kakutani's Fixed Point Theorem, there exists a fixed point (x*, Pj) such

that PA E [0, PA] and x* c X such that PA E p(x*) and x* e Ar(PA).E

In English...There is a market excess demand correspondence for which each excess

demand value can only result from only one allowance price while each allowance price will

result in at least one market excess demand value.

Once the scrubber choice is introduced into the decision-making process, the

correspondence becomes more complex, as seen in Figure B.2. An equilibrium may in

fact exist, but there is no way to guarantee an equilibrium because the excess demand

correspondence is no longer a convex set. The average of the two excess demand values

(ASMAX and A"IN) is not in the excess demand correspondence.

B.2 Technical Details of Simulation Model Design

The equilibrium allowance market price is solved by using a bisection iterative

process. An upper limit ($1,000) and lower limit ($0) for the allowance price are chosen.

The initial allowance price is set to the upper limit and the simulation solves for each

generating unit's cost-minimizing choices. Then it checks if the allowance market is in an


If market excess demand is positive, the allowance price is too low and the allowance

price is increased by one-half the difference between the upper and lower limits. The old

price now becomes the new lower limit while the upper limit remains the same. If the

market excess demand is negative, the allowance price is too high and the allowance price

is decreased by one-half the difference between the upper and lower limits. The old price

becomes the new upper limit and the lower limit remains the same. In this case, the upper

S2',Sj are the sulfur content for high sulfur and low sulfur contract coal for a given
unit, respectively. The sulfur content will differ across regions of the U.S. due to the
heat content of coal from different coal mines across the country.

represents the rate at which sulfur is transformed into SO2, which is assumed to
be a constant (1.9) for simplicity.18

C0, Ci represent the contract constraints for a given unit, which requires the use of
a minimum amount of each coal type.

Allowance Parameters:

* Ei represents a generating unit's tons of SO2 emissions.

* Ae represents a generating unit's initial allowance allocation in tons of SO2

Ai represents a generating unit's net allowance position in tons of SO2 emissions,
which is the difference between the actual allowances used and a unit's initial
allowance allocation. A unit is a net buyer of allowances (positive excess demand)
if it uses more allowances than its initial allocation (Ai > 0), a net seller (negative
excess demand) if it uses fewer allowances than its initial allocation (Ai < 0), and
neither if it uses exactly the same amount of allowances as its initial allocation
(A =- 0).

PA is the allowance price, which is endogenously determined in the model by
the decisions of the utilities. Each allowance that is bought (sold) will increase
(decrease) the utility's production costs by PA. Each generating unit takes PA as

2.6 Generating Unit Level Decision-Making Process

The model is a static model with decisions made at the generating unit level where

each generating unit chooses its coal use, net allowance position, and scrubber choice to

is Sulfur content is the tons of sulfur per ton of coal. In this paper, any coal that results
in emissions greater than 2.5 lbs. SO2/l\ I 1tu is considered high sulfur coal. Under Phase
II of Title IV and CAIR, the high sulfur-low sulfur cut-off value is reduced from 2.5 to
1.2 lbs./\ i\l Ibtu. Under CAIR the new allowance allocation is based on 0.6 lbs./\l\I Ibtu,
which cannot be met by fuel switching alone because low sulfur coal normally ranges from
0.7-1.2 lbs./\ lil Itu with few shipments of low sulfur coal resulting in emissions of 0.6
lbs./\ l ktu. m = 1.9 for bituminous and anthracite coal, m = 1.75 for subbituminous
coal, m = 1.5 for lignite coal. These differ due to each coal types composition. Cost savings of fuel switching versus allowance purchases when
PA > MCAf'

Assuming that high sulfur spot market coal is cheaper than low sulfur spot market

coal ( < ), a generating unit that does not have an emissions constraint prefers to

use high sulfur spot market coal to meet electricity demand.19 Allowing generating units

to have a choice in their compliance options can lead to costs savings in several cases.

First, consider the case that PA > MCA"'j. Using low sulfur spot market coal leads to

the minimum number of allowances used by the unit (A"IN). Assuming that the initial

allocation by the firm is larger than the minimum possible allowance use, the unit will sell

its remaining allowances after meeting its allowance requirement. The excess demand for

such a unit can be seen in (2-39), where the excess demand will actually be negative.

The unit's cost savings from switching fuels over purchasing allowances is (PA -

MCAs's)(A AX AF"), which is the area "a+b" seen in Figure 2.8(ii). The dark-shaded

area (a) is the cost savings for the generating unit from abating emissions through fuel

switching instead of purchasing additional allowances. The light-shaded area (b) is the

cost savings from abating more than its initial allocation and selling the extra allowances. Effects of high sulfur coal contracts on excess demand and costs

Now consider how a high sulfur coal contract will impact a generating unit's allowance

excess demand correspondence in Figure 2.8, which is summarized in Proposition 2.

Proposition 2: Given the scrubber choice,

(i) For the range of allowance prices PA > MCA"'8, a high sulfur coal contract will
weakly increase a unit's allowance excess demand.

(ii) For the range of allowance prices PA < MCA"'8, a high sulfur coal contract will
Sc S
weakly decrease excess demand if Sh < .h

19 The assumption that high sulfur coal is cheaper than low sulfur coal is supported by
actual coal prices.

and will cancel out. Fill in for known values and combine like terms:

AMIN Di -I Ae I r 'V Di /17 r~ ,Y c A-
Ai Sm A il A Hiiil i

P(A1i1 AMIN) PCiH l( ) (A-13)
i ( Ai A i I HI H i H isi l
The change in compliance costs will be the increase in costs from the increase in a unit's

net allowance position. If > S a unit's compliance costs will increase.E

Proof of Proposition 1(iv):

If PA < MCA"'8, a unit prefers to use all high sulfur coal and purchase allowances

instead of switching fuels to meet its emissions requirement because it is the least-cost

compliance option. Without a low sulfur coal contract, a unit will use all high sulfur coal

(C7 = CMAX and C"* 0) and require the largest net allowance position to cover the
maximum emissions level (AMAX). With a low sulfur coal contract, a unit will use less

high sulfur coal (CiA < sih A), which will decrease the emissions level and requires

fewer allowances (AAX > AMAX). Since a unit prefers to use high sulfur coal with and

without the emissions constraint, coal use will remain the same. Fill in for known values

and combine like terms:

D/ iI'V AMIN)Ps\ CS,MAX _s,MAX s ,MAX s s,MAX it i, V AMIN) (A 1)
SA~i~i Aih\h ~ih hi (iC ) C C ~ A A (A

The change in compliance costs is the change in costs from the change in net allowance

position. Now fill in for the net allowance position:

AMAX SDi 8 e I ; _i (HD -s ,+ CSc c Ae
S- m Ai Ai + C,-Sm Aii
Hih ih

CcC HP [i ihA5
HzPm sh < 0 (A15)
Since ~ < compliance costs will decrease.E

Another way of looking at the impacts of coal contract constraints on compliance

costs is to find the change in total costs for a unit facing an emissions constraint with

Policy dummy variable values are determined by a policy's enactment date, zero

before enactment and one after enactment. The enactment date is the year that the policy

is passed by the state legislator, created through an executive order, or announced as a

mandate under new PUC guidelines. Some of these policies allow a grace period for power

producers to meet the new regulations. The effective date is the year that the policy

requirements must be met. The average lag from the enactment to effective date is a little

over one year, but can be longer for Renewables Portfolio Standards. The enactment year

is a better choice to determine when the policy begins to impact the power producers.

Once a power producer becomes aware of a future requirement, it may begin to construct

any necessary renewable capacity. These actions could lead to large amounts of renewable

capacity being constructed between the enactment date and effective date.

Regulatory policies described below include a Renewables Portfolio Standard with

a Capacity Requirement, Renewables Portfolio Standard with a Generation/Sales

Requirement, Clean Energy Fund, Net A ii.- Interconnection Standards, State

Government Green Power Purchasing, and Required Green Power Options. Table 3-2

summarizes the data for the policy variables.

The first policy that will be discussed is a Renewables Portfolio Standard, which

specifies an amount of a state's electricity production, sales, or capacity that must be

renewable-based. Renewables Portfolio Standards can be differentiated into three main

structural forms, policies that set (1) mandatory renewable generation or sales levels,

(2) voluntary renewable generation or sales goals, and (3) mandatory renewable energy

capacity requirements.

The first type of Renewables Portfolio Standard sets a percentage of total generation

or sales for each power producer/retailer that must originate from renewable sources,

usually increasing every year or every few years. For example, Arizona's tiered renewable

levels that have to be met began at 0.' in 2001 and increased by 0.' each year,

Inserting C,,)MAX in (2 37) for C, in (3-3) gives an expression for the minimum allowance

excess demand, which is the difference between the minimum emissions level (EMIN) and

the initial allowance allocation (Ar):

AMIN MN AE (- z ()(S Ae (2-39)
11 Hli

If a unit's initial allocation can cover its minimum possible emissions, then it will have a

negative net allowance position and be a net seller of allowances.

If PA = MCA'8, a unit may use any combination of high sulfur spot market

coal and low sulfur spot market coal, which leads to any level of excess demand in the

range (E,1N A', EAX Ar). The allowance excess demand can be represented by

Ai (OEfAX (1 O)EMIN A) where the constant 0 [0, 1]. A unit that is indifferent

between fuel switching and allowances purchases could be either a net buyer or a net


Combining the excess demands for each of the three cases creates the Excess Demand


A AX if PA > MCA'
A, = A AX (1 )AMIN if PA MCA'8 VO e [0, 1]

AIN if PA < MCA'"

A generating unit's excess demand correspondence can be seen graphically in Figure

2.8(i). High sulfur spot market coal use corresponds to the right-hand vertical line where

PA < MCA"'8. Low sulfur spot market coal use corresponds to the left-hand vertical line

where PA > MCA"'>. The case where a generating unit uses some combination of low

sulfur spot market coal and high sulfur spot market coal is represented by the horizontal

line at which PA = MCA"'.

Although actual industry compliance costs are higher than the least-cost results at $716

million, these compliance costs are lower than the compliance costs found in Sotkiewicz

(2003) and Sotkiewicz and Holt (2005) at $990 million and Carlson et al. (2000) at $910

million. Contract constraints appear to explain some of the excess compliance costs found

in previous studies, which implies that generating units' decisions were more cost-effective

than previously stated in the literature. Industry and generating unit coal use

By comparing coal use in Simulation 5 to that in Simulation 3, the impact of contract

constraints assuming scrubbers as given can be derived. Simulation 3 uses a total of

2,607,025,040 mmBtu of high sulfur coal (40.7' ) and 3,794,557,150 mmBtu of low sulfur

coal (59.;:'. ). Simulation 5 uses a total of 2,041,933,640 mmBtu of high sulfur coal (31.9' .)

and 4,359,648,550 mmBtu of low sulfur coal (68.1 .). Introducing contract constraints into

the model results in a" '. decrease in high sulfur coal relative to Simulation 3. Contract

constraints results in less high sulfur coal than would have otherwise been preferred.

Contract constraints led to 16 units (3.7'. of affected units) using a suboptimal coal

combination. 15 of the 16 units had 1(1C' of its coal use altered. Only 2 of these units had

an increase in high sulfur coal.

Simulation 4 uses a total of 2,651,410,732 mmBtu of high sulfur coal (41. !',) and

3,750,168,009 mmBtu of low sulfur coal (58..', ). Simulation 6 uses a total of 2,614,773,732

mmBtu of high sulfur coal (411 -',) and 3,786,804,109 mmBtu of low sulfur coal (59.2'.).

Including contract constraints into the model results in a 0..' decrease in high sulfur

coal relative to Simulation 4. Overall coal use does not appear to have been significantly

altered. However, this does not tell the whole story. Contract constraints led to 27 units

(6.;:'. of affected units) using a suboptimal coal combination. 5 of the 27 units had a

change of at least '- of coal use and 15 of the 27 had at least a 50', change in coal use.

14 of the 27 units had an increase in high sulfur coal use while 13 units had an decrease

of high sulfur coal use. The concern with the contracts is not necessarily that the entire

2.6.3 ('!, ii :terizing a Unit's Spot Market Fuel C('!i... and Marginal
Cost of Abatement from Fuel Switching . . . 65 Necessary conditions for using both high sulfur and low
sulfur coal ...... ........ ... .. ...... 65 Only high sulfur coal use: Necessary conditions . 65 Only low sulfur coal use: Necessary conditions ...... ..66
2.6.4 Coal Use Under a High Sulfur Coal Contract Constraint ...... 66
2.6.5 Coal Use under a Low Sulfur Coal Contract Constraint ...... ..68
2.6.6 Generating Unit-Level Compliance Costs . . ..... 70
2.6.7 Generating Unit's Net Allowance Position: Excess Demand
Correspondence .................. .. 72 Cost savings of fuel switching versus allowance purchases
when PA > MCA'.. .................. .. 74 Effects of high sulfur coal contracts on excess demand and
costs .................. .......... .. 74 Cost savings of allowance purchases versus fuel switching
when PA < MCA'.. ............... .. .. 80 Effects of low sulfur coal contracts . . ... 80 Fuel switching versus allowance purchases when PA
M CAf'" ...... . .. ... ..... 85
2.6.8 Generating Unit's Scrubber Installation C('!i i. . . 86 When will a generating unit install a scrubber? . 86 Different marginal costs of abatement . . .... 87 Excess demand correspondence . . ..... 87
2.6.9 Impact of Coal Contracts on Excess Demand Correspondence . 89 Impact of a binding high sulfur coal contract . ... 90 Impact of a binding low sulfur coal contract . ... 100
2.7 Possible Implications on the Allowance Market and Industry Compliance
C costs . . . . . . . . .. 109
2.8 Conclusions .................. ................ .. 110


3.1 Introduction ............. .. ........ ..... 127
3.2 Review of Generating Unit Model ................ ... 128
3.2.1 Generating Unit Problem .................. ...... 128
3.2.2 Optimal Compliance C('!. ~ .. ............... 129
3.3 Allowance Market Equilibrium ..... . . .... 132
3.4 Comparative Statics: Effects on the Allowance Market . .... 134
3.4.1 Comparative Statics: Effect of Relative Fuel Cost on the Allowance
Market ................... ............... 134
3.4.2 Comparative Statics: Effect of Coal Contracts on the Allowance
Market Given the Scrubber C('! i... .................. 135 Impact of high sulfur coal contract on allowance market 135








Impacts of Low Sulfur Coal Contract . .........

Given Scrubber C('!. Shift from High Sulfur Contract .

Given Scrubber C('!... Shift from Low Sulfur Contract .

With Scrubber C('!. .. Shift from High Sulfur Contract .

With Scrubber C('!. Shift from Low Sulfur Contract .

Upper Semi-Continuous Correspondence . .......

Correspondence with Scrubber C('!i.... .........

.. 195

. . 96

. . 96

. . 97

. . 98

.. 210

.. 210

considered a consumption tax on electricity to fund deployment of renewable capacity in

the industry. In Minnesota, a settlement with the electric utility Xcel Energy created a

similar fund that is p .iing for renewable energy research and deployment. Maine created

a voluntary fund similar to a Clean Energy Fund for the state's customers to donate

money. 28

Similar to Renewables Portfolio Standards, Clean Energy Funds must be differentiated

to understand how effective these policies are at increasing renewable deployment in a

state. The variable used in this model is a variable that measures the amount of capacity

that is being funded for iuil' ii--, i1.- projects from Clean Energy Funds (CEF: CAP


Some customers may prefer to build generating capacity to provide their home with

some of their own electricity. Net Metering (NET METERING) allows customers that

are able to produce more electricity than they consume in a given month to sell any

excess to the utility to offset the charges for electricity in months the customer is a net

purchaser. The effect of net metering is expected to be negative because if renewable

energy demanders produce their own renewable electricity through a solar PV system

or small wind turbine, they will demand less renewable capacity from power producers.

From a utility perspective, if it is required to reach a renewable capacity or sales target,

these customer-owned facilities may serve to offset a utility's needs to build renewable

28 Database of State Incentives for Renewable Energy (DSIRE) does not include New
Mexico as having a Clean Energy Fund, while Bolinger et al. (2001) verifies that New
Mexico does have a Clean Energy Fund.
29 The capacity obligations as of 2003 are interpolated backwards linearly to the
enactment year so that an equal amount of additional capacity obligations are made each
year and total the overall obligations as of 2003. The data for is variable originated from
the Database of Utility-Scale Renewable Energy Projects from the Clean Energy States
Alliance (CESA).

of one specific type of renewable energy because using only one type would preclude

any interesting cross-state comparison of policy effects of states with different available

renewable energy resources.7 For example, comparing the effects of a policy on Maine

and Texas using only wind power capacity excludes the policy effects on biomass capacity,

which is a more likely renewable choice for Maine. Both types of renewable resources must

be included to directly compare the effectiveness of policies across states.

The effects of state renewable energy policies are best estimated using total state

non-hydro renewable capacity as the dependent variable because several policies mandate

or fund a specific amount of renewable capacity. Policies that do not set specific renewable

capacity requirements can be measured in capacity terms by controlling for each state's

market size, which will be discussed in more detail in Section 4.

A large amount of renewable capacity created before 1996 originated from the Public

Utilities Regulatory Policy Act (PURPA), a federal policy passed in 1978 requiring

utilities to purchase electricity from Qualifying Facilities (QFs), which are IPPs that meet

specific requirements and include renewable-based facilities. For a variety of reasons, the

effects of PURPA varied from state to state. State dummy variables (Si) measure these

effects and other unchanging state factors, such as renewable resource availability.8

7 Hydropower is not included in the renewable energy capacity because most
hydropower was created well before the mid-1990s, with few changes in capacity or costs
over the time period being analyzed. These aspects allow hydropower to be considered
a type of current generating technology, which includes steam or gas turbines fired by
natural gas, coal, petroleum, or nuclear power. For hydropower to be a viable power
option there must be an available river or stream as well as a significant change in
elevation. Most of these sites in the U.S. already have hydropower capacity in place.
Removing hydropower from the dependent variable allows the focus of the paper to be
on the policy effects on the emerging technologies of wind, solar, biomass, and geothermal

8 (\ !, i, 2003). There is some concern that expiration and buyouts of PURPA
contracts during the 1990s have led to decreases in renewable capacity, especially in
California where deregulation in the early to mid-1990s created competition based on price

all units when choosing to install a scrubber. The marginal cost of abatement for all coal

types has already been derived. Now the MCAi for each coal combination both with and

without a scrubber can be derived. The only way the marginal cost of abatement can vary

across generating units is through the scrubber choice (zi and ri).

First, consider the marginal cost of abatement of switching from high sulfur spot

market coal to low sulfur spot market coal (MCA7'") with a scrubber in (3-68), and

compare it to the marginal cost of abatement without a scrubber in (3-69). As in '!i lpter

2, a scrubber decreases the savings from switching spot market fuels because a scrubber

causes the emissions reduction to be smaller than without a scrubber.
P Ph'
h I
Ail MCAs- I h (3-68)
(tl- r )(T )( S1
P18 Ph,

AiR MCAs ? H h (3-69)
(mn)() )

Second, consider the marginal cost of abatement of switching from high sulfur

contract coal to low sulfur spot market coal (MCAA'") with a scrubber in (3-70), and

compare it to the marginal cost of abatement without a scrubber in (3-71). As in the case

of only spot market coal, installing a scrubber decreases the savings from switching fuels.

In the case of a high sulfur contract, it is less costly to a plant to use high sulfur coal at a

particular generating unit if it has a scrubber.
= Ail MCAS + H s (3 70)
(1- rt)(m)(' ITT

= Al = MCAs + sh (3-71)
h I
Third, consider the marginal cost of abatement of switching from high sulfur spot

market coal to low sulfur contract coal (MCA'c) with a scrubber in (3-72), and compare

it to the marginal cost of abatement without a scrubber in (3-73). As in the two cases

above, installing a scrubber decreases the savings from switching fuels. In the case of a low

1.6 Conclusions

States have enacted many policies to increase the deployment of non-hydro renewable

capacity into the electric power industry in that state. The literature evaluating the

effectiveness of these programs consists of case studies and one statistical study, which

explains the use of wind power. My statistical study utilizes a larger panel, more policies,

and more control variables to explain the deployment of total renewable capacity in a


Three regulatory policies appear to be effective at increasing renewable capacity

deployment in a state. The significant results from these regulatory policies confirm many

of the findings from prior case studies, which find Renewables Portfolio Standards with

Capacity Requirements and Clean Energy Funds have increased renewable capacity. An

additional policy, Mandatory Green Power Options, is also found to increase capacity

deployment in a state as well.

The previous empirical study found Public Benefits Funds, which include any Clean

Energy Fund in a state, to be insignificant in their model. My paper finds that Clean

Energy Funds with ul il li--i 1 .J projects increase the deployment of renewable capacity in

a state. By using System Benefits C!i irges (SBCs) a state can effectively make consumers

p ,i for cleaner energy without creating a different market for renewable energy demand.

Similar to the case study findings by Bolinger et al. (2001, 2004, 2005), larger utility-scale

projects make Clean Energy Funds more effective at increasing renewable capacity

deployment in a state.

This paper finds that different types of Renewables Portfolio Standards have different

effects on renewable capacity. Each megawatt of capacity mandated by Renewables

Portfolio Standards with Capacity Requirements results in the deployment of one

megawatt of additional renewable capacity in a state. But recent Renewables Portfolio

Standards that mandate generation or sales levels appear not to have statistically

significant effects. These results mirror Petersik's case study in that only Renewables

allowance allocations similar to substitution and compensation units. There were seven

units that entered the program through this opt-in provision (Ellerman et al., 2000).6

Allowance prices were low compared to initial marginal abatement cost estimates,

and relatively stable throughout Phase I and the beginning of Phase II. Initial marginal

cost estimates used by the EPA ranged from $199-$226 (Smith and Ellerman, 1998). The

market opened in 1995 at a price of $150, soon hit a low of $70 in early 1996, and then

slowly rose back to around $150. Other than a slight spike in 1999 as utilities positioned

themselves for the start of Phase II, the allowance price remained relatively stable around

$150 (Burtraw et al., 2005). Phase II of Title IV

Phase II, which began in 2000 and will continue until the implementation of CAIR in

2010, includes all units over 25 MW in generating capacity. The more than 2,000 affected

generating units throughout the U.S. were allocated allowances based on an emissions rate

of 1.2 lbs. SO2/mmBtu of heat input, multiplied by the unit's baseline heat input during

1985-1987. New generating units were given no allowances and were required to purchase

any necessary allowances in the allowance market. Phase II allocations were capped at

10.0 million tons annually in 2000, have decreased to 9.5 in 2002 where it will remain until

2010, when it drops to 8.95 million tons.

The banking provision has allowed utilities to trade intertemporally with utilities

using the substantial allowance bank accumulated through Phase I for compliance in

Phase II leading to annual emission levels in excess of 10 million tons in each year from

2000 to 2005.

6 There was an incentive to opt-in generating units voluntarily if it is beneficial to
the utility. Opting Phase II units into Phase I give utilities additional v--i to decrease
emissions and sell allowances and, apparently more importantly, bank allowances for
future use in Phase II. Actual SO2 emissions by Phase I units were much lower during
Phase I than the total allowance allocation during Phase I, which allowed utilities to bank
additional allowances for use during Phase II (Ellerman et al., 1997).

A plant with a binding low sulfur coal contract has a smaller incentive to install a

scrubber at each of its generating units because it must use some low sulfur coal, even if

it would prefer to use high sulfur coal at all generating units. A low sulfur coal contract

increases the indifference price at which a unit will install a scrubber from PA to (PA + c)

for each generating unit. A plant is indifferent to using low sulfur contract coal at any of

its generating units if no scrubbers are installed because all units have the same MCA"'.

A binding low sulfur coal contract results in an increase in PS.

At Pf where...

P, + PAA, + PQ +P + Pl PAA + PC + PlC (3-79)

If low sulfur contract coal is at least as expensive as low sulfur spot market coal, we know


Pi + PAA + PhCh + PC4 + P1C > PAA, + Ph CJ + P8C4+/c V (3-80)

This is the same result as in C'i plter 2. A low sulfur coal contract results in an

inefficient coal use combination of high and low sulfur coal. The indifference price at which

a plant will install a scrubber at a given generating unit will weakly increase when a plant

chooses to use low sulfur contract coal at that unit.

3.7.10 Scrubber Installation Example: Plant with Two Affected Generating

For a simple example, consider the scrubber installation choices for a plant with only

two generating units. Assume that a plant will install a scrubber at Unit 1 before it will

install a scrubber at Unit 2 because (ACA1 < ACA2). There are three cases that may

result, each of which is described below with their own indifferent price for installing a


be seen in Figure 2.8, where MCA"'8 < MCA'"8 and there is actually a cost savings to the

unit from using low sulfur contract coal over low sulfur spot market coal. The compliance

costs remain the same as in the previous example, area (a + b + c). However, the total

costs to the unit will decrease relative to not having coal under the contracted price. Area

(c + d) is additional costs to the unit for not being able to switch to low sulfur coal. Area

(d) is the cost savings of using lower priced high sulfur contract coal over high sulfur spot

market coal.

By making the same assumptions as in Example 1 except changing the price of low

sulfur contract coal to $1.50/mmBtu, we can solve for compliance costs and total costs to

a unit. MCA'c is $180.45. The change in compliance costs remain the same at $940,000

while the costs due to the lower priced coal actually decrease by $1.2 million.22 The

unit actually gains by lowering its total costs by $260,000 through the coal contract even

though it must increase its compliance costs. Fuel switching versus allowance purchases when PA = MCA'8

In the knife-edge case a generating unit has no strict preference between purchasing

allowances and abating emissions because PA = MCA"'. The unit's excess demand may

be any value in the range [A -IN Ae, A AX- A(]. The generating unit has no preference

in compliance options because any combination of abatement and allowance purchases

result in the same compliance costs of area (a) in Figure 2.8.

A high sulfur coal contract will have the same impacts on the excess demand

correspondence when PA = MCA"' as in Section 5.7.2 where PA > MCA"'j. However,

the impacts on compliance costs and total costs will be different. The reasoning for this

is that the contract does not force a unit to use a more expensive compliance option. By

comparing Figure 2.8 to Figure 2.8, these differences can be derived. The shift in a unit's

22 For simplicity, the sulfur to heat content ratio is assumed to be equal for low sulfur
contract and spot market coal.








The SO2 Allowance Price . ..........

Excess Demand Correspondence and Compliance Cost
Over Allowance Purchasing .. ...........

High Sulfur Contract: Shift in Minimum Excess Dema

No Contract: Compliance Costs . ......

High Sulfur Contract: Compliance and Total Costs .

High Sulfur Contract: Relative Savings from Contract

Cost Savings from Using Allowances Over Fuel Switch

2-8 Low Sulfur Contract

No Contract: Compliance Costs . .......

Low Sulfur Coal Contract: MCA8'c . .

Low Sulfur Coal Contract: MCA8'c . .

Compliance Costs: PA = MCA8' . .

High Sulfur Coal Contract . ..........

Low Sulfur Coal Contract . ...........

Excess Demand Correspondence: MCA"'" < P ...

Excess Demand Correspondence: MCA"'" > PA .

Impact of a High Sulfur Coal Contract: MCA"' < PA

Impact of a High Sulfur Coal Contract: MCA"'" > PS

Impact of a Low Sulfur Coal Contract: MCA"'j > P .

Impact of a low sulfur Coal Contract: MCA^s < P .

Excess Demand Correspondence . .......

Impact of High Sulfur Coal Contract . .....

Impact of Low Sulfur Coal Contract . .....

Excess Demand Correspondence with Scrubber ('C!I .,

Impacts of High Sulfur Coal Contract . ....

from Fuel

. 119





d . . 120

. .. . 20

.. . 20

Coal . . ... 121

ig . . . 121

. .. . 121

. .. . 22

. .. . 22

. .. . 22

. .. . 23

. .. . 23

. .. . 23

. ... . 24

. ... . 24

. . . 125

. . . 125

. . . 125

. . . 26

. .. . 93

. .. . 93

. .. . 94

. . . 94

. .. . 95


















unit allows greater freedom in coal use and may greatly alter a plant's preferred

compliance options.11

Plants must consider the total combined costs over all generating units, which

may lead to plant level decisions that are contrary to a specific generating unit's

cost-minimizing choice. A plant may be able to lower total costs by increasing the costs at

one generating unit to save money at other units.

3.7.2 Plant-Level Problem

A plant solves the following problem. The subscript "i" represents a specific

generating unit for each operating plant while "n" represents the number of generating

units owned by the plant.

mm n
zACmin ,PCCC
subject to...A + Ai > (1

i= 1

'I > C
i= 1
zi E {0, 1}

Ch, ,'i > 0

i G {1,..., n}

The Lagrange multiplier on a generating unit's emissions constraint is represented by Ail.

The Lagrange multiplier on a generating unit's demand constraint is represented by Ai2.

Coal contract constraints require a plant to use a minimum amount of each contract coal

11 Due to the design of the policy with "compensation" provision, a plant can shift
coal use, but not electricity production to a non-affected unit without the unit being
incorporated into the program.

PS is the allowance price at which the unit is indifferent between installing a scrubber or

not. The amount of high sulfur and low sulfur contract coal will be the same both with

and without a scrubber and will cancel out, but the contract coal still affects the allowance


A generating unit's decisions will hinge on this allowance price. (2-56) can be used

to solve for PS, the minimum allowance price at which a generating unit will install a

pS > r. Zr Pih(vih ih) + Pil l ( Q81 Q8
PA + C )~8(C 0) (2 -57)
(A Ai)
A unit will prefer to install a scrubber at PA if the average costs of abatement from using

a scrubber is weakly less than the costs of purchasing an allowance. Different marginal costs of abatement

The installation of a scrubber leads to an increase in the unit's marginal abatement

cost of switching from high sulfur spot market to low sulfur spot market coal relative to

the marginal abatement cost without a scrubber installed:
ps ps ps ps
il ih il ih
MCA '" iH ih H MCA'8 il ih (2-58)
(l- rj)(Tn)( ) (T)(|__- )
"Th 7T Hh 7T

Scrubber installation decreases the size of the denominator of MCA7'" by (ri)(m)(

$), which is due to the fact that only a fraction (based on the scrubber's reduction rate)

of the emissions reduction from switching fuels is realized. Excess demand correspondence

A generating unit's excess demand correspondence becomes significantly more

complicated when a unit's scrubber installation choice is introduced into its decision-making

process. The excess demand correspondence is a combination of a unit's excess demand

23 See Appendix A for the derivation of this equation.

hold at least enough allowances to cover their yearly emissions to be in compliance. The

program allows generating units several degrees of freedom in choosing how to best meet

its compliance obligations: switch from high sulfur to low sulfur fuels; install scrubbers;

and buy or sell allowances; or any combination thereof.5 Phase I of Title IV

Phase I of Title IV, which ran from 1995-1999, capped the initial level of emissions

at 8.7 million tons of SO2 per year for the 110 largest polluting plants, which included

263 generating units. The EPA allocated allowances gratis to these affected units

based on average heat input during 1985-1987 multiplied by an emissions rate of 2.5

lbs. S02/mmBtu.

An additional 168 units participated in Phase I in 1996 based on the rules established

by EPA allowing a plant to "opt-in" units (7 units), designate substitution units (160

units), or designate compensating units (1 unit) as part of their Phase I compliance plans.

The voluntary participation of these additional units resulted in a total of 431 affected

generating units under Phase I. A "substitution unit" is a unit that would eventually

be affected in Phase II that voluntarily enrolled into Phase I to meet some or all of the

required emissions reductions for a Phase I unit (Sotkiewicz and Holt, 2005). Substitution

units receive an allowance allocation based on its historical heat input. A utility may

decide to reduce its electricity production at a Phase I affected unit. To do so, the utility

must have a "compensation unit" from the Phase II units the utility operates to cover the

necessary additional electricity. This compensation unit is then brought into Phase I and

given an allowance allocation based on its historical heat input. Industrial sources of S02

emissions could use the opt-in provision and voluntarily enroll into Phase I and receive

5 There are additional compliance options, including shutting down the affected unit
and shifting dispatch away from the affected unit (Energy Information Association, 1997) Impact of low sulfur coal contract on allowance market

A generating unit's excess demand for allowances may be altered by a binding

low sulfur coal contract constraint. From C'! Ilpter 2, a binding low sulfur coal contract

will result in more low sulfur coal use than would be optimal for a generating unit and

decrease a generating unit's excess demand. A decrease in a generating unit's excess

demand may decrease the allowance market price.

Proposition 3: Assuming the scrubber choice as given and H, a low sulfur

contract results in a weakly lower allowance price.

Proof of Proposition 3: As has already been shown in C'!i lpter 2, given the

scrubber choice and H H a low sulfur contract results in a weak decrease in a
H: H77T
generating unit's excess demand.

If PA > MCA"', a generating unit prefers to switch fuels from high to low sulfur coal,

create its minimum possible emissions, and have an excess demand of AIN. Assuming

the same relative sulfur content for spot and contract coal, a low sulfur coal contract will

not change total emissions or excess demand (AIN = AFIN) and PA remains unchanged.

If PA < MCA"', a generating unit prefers to use high sulfur coal, create its maximum

possible emissions, and have an excess demand of AMAX. However, the low sulfur coal

contract would force less than the maximum amount of emissions and weakly decrease

excess demand to AAX. A lower excess demand will lead to an increase in the allowance

market supply or both an increase in allowance market supply and a decrease in the

allowance market demand. In both cases PA weakly decreases.

3.4.3 Comparative Statics: Effect of Coal Contracts on the Allowance Market
with Endogenous Scrubber Choice

Previously when the scrubber choice was taken as given, the impact of coal contracts

on the allowance price was unambiguously non-negative or non-positive. High sulfur

contracts may only lead to an increase in the allowance price while low sulfur contracts

Table 1-3. Policy Variables

States with Policy Non-Zero Observations
8 51

The third E-iv analytically and empirically shows how fuel contract constraints

impact the emissions allowance market and total electric power industry compliance costs.

This paper uses generating unit-level simulations to replicate results from previous studies

and show that fuel contracts appear to explain a large portion (i.' .) of the previously

unexplained compliance cost simulations. Also, my study considers a more appropriate

plant-level decisions for compliance choices by analytically analyzing the plant level

decision-making process to show how cost-minimization at the more complex plant level

may deviate from cost-minimization at the generating unit level.

Also, this paper considers a more appropriate level for compliance decisions. The

literature has only considered compliance costs at the generating unit level. However,

actual compliance decisions would be made at the plant level because of the economic

relations and the physical proximity between generating units. Coal is delivered at

the plant level, where multiple generating units may be located. Since a plant is

only concerned about minimizing its total costs at the plant, the minimization at a

particular generating unit is not necessary for a plant to make optimal choices. This paper

analytically analyzes the plant level decision-making process to show why it is appropriate

to consider decisions as the plant level instead of the generating unit level, the interaction

of choices for economically related generating units, and how cost-minimization at the

plant level may deviate from cost-minimization at the generating unit level.

The paper will be structured in the following manner: Section 2 will look at

the conditions for an allowance market equilibrium while Section 3 will look at the

comparative statics of the allowance market. Section 4 will look at compliance costs,

both for an individual generating unit and the entire industry. Section 5 and Section

6 will explain the data used in the generating unit level simulations and the results

from the actual simulations, respectively. Section 7 will analyze the plant level decision

making process, how it differs from the generating unit level model, and what additional

information the model will add to the literature.

3.2 Review of Generating Unit Model

3.2.1 Generating Unit Problem

mm zA- + P A. + PC + PAA + + + P + P zi,Ai,Cg,C ,C',C
subject to...A7 + A, > (1 z)(m)(CS + CS + CS + CS>) Ai (3-2)

(CSH + C H + CQH8 + CHI) > D Ai2 (3 3)

C> /4 (3-4)

p P5 + Ai1(1 z+ri) (m) (S2) PA + Ai (l zri) (m)(S) (2
(2 23)
He He H
ih ih ih
If the additional costs, both from fuel costs and emissions, of using high sulfur

contract coal is more expensive than using high sulfur spot market coal, then tih > 0. If

the additional costs, both from fuel costs and emissions, of using high sulfur contract coal

is less than using high sulfur spot market coal, then Pih < 0. If the additional costs, both

from fuel costs and emissions, of using high sulfur contract coal is is the same as using

high sulfur spot market coal, then ih = 0.

Now assume a unit has a high sulfur coal contract and uses both high sulfur and low

sulfur spot market coal. So (2-10) and (2-11) hold with equality. Also, it has been shown

in (2-15) that when a unit uses both high and low sulfur spot market coal, PA = MCA"'8:
PA= MCA' = MCA<' + ih (2-24)
(l- z)r(m)( Tn))

By solving for Ph in (2-24), the sign of Pih can be determined. If MCA'" MCA' = 0,

then Pih = 0 and high sulfur contract coal has the same cost as high sulfur spot market

coal. If MCA'" MCA'"8 > 0, then Pih > 0 and contract coal is more expensive to use

than spot market coal. If MCA'" MCA'"8 < 0, then pih < 0 and contract coal is cheaper

to use than spot market coal.
P ih
(h Sc = MCA'8 MCAI" (2-25)

2.6.5 Coal Use under a Low Sulfur Coal Contract Constraint

pI- is the shadow price of the low sulfur contract constraint, which includes both the

change in costs due to fuel costs and emissions. If contract coal is more expensive than

spot market coal, then pil > 0 and it increases fuel costs. If contract coal is cheaper than

spot market coal, then pil < 0 and it decreases fuel costs.

Assume a unit has a low sulfur coal contract and uses only high sulfur spot market

coal. It has already been shown in (2-16) that when a unit uses only high sulfur spot

trying to minimize allowance sales to states whose emissions will eventually reach New

York/Wisconsin and result in hotspots, which are are areas with extreme emissions levels

that result in greater damages in a particular area relative to damages throughout the

rest of the region. Preventing utilities in New York and Wisconsin from selling allowances

to utilities outside their state would have resulted in more than double the compliance

costs in both states, and increased nationwide compliance costs. Some of the additional

costs from these restrictions would have been offset by lower costs for utilities not in New

York or Wisconsin that would have been able to sell more allowances due to the additional

demand no longer being met by allowances sales from New York and Wisconsin utilities.

Some studies further explain the inefficiencies by examining the actual lost cost

savings that are specifically a result of state PUC regulation under Phase I. Carlson et al.

(2000) find that the actual compliance costs were $339 million (59'-.) greater in 1996 than

the least-cost solution. The study concludes the difference between actual compliance costs

and the least-cost compliance may be attributable to ,Illustment costs associated with

changing fuel contracts and capital expenditures as well as regulatory policies." Sotkiewicz

and Holt (2005) find that PUC regulations resulted in $131 million of the additional

compliance costs relative to the least cost solution. However, there is a significant amount

of compliance costs that remained unexplained.8 My study conjectures long-term coal

contracts may be responsible for what appears to be inefficient behavior resulting in

additional compliance costs.

8 Sotkiewicz and Holt model the possibility of ex post "prudence", which assumes that
there is some cost, such as future state PUC cost disallowance, to the generating unit for
choosing a less cost-effective option. If PUC regulations allow for total pass-through of
costs without threat of ex-post prudence, then a generating unit is indifferent to costs and
may not make the lowest cost compliance option.

demand multiplied by the heat input required to generate one kilowatt-hour
of electricity. Demand at the unit level is assumed to be fixed. Modeling each
generating unit's hourly dispatch and hourly costs in the context of varying loads
and dispatch are not easily modeled, and would require arbitrary assumptions about
how units would be utilized. For these reasons, it is assumed in this paper that
utilities do not have the option to shift electricity production across generating units
to meet demand.

Coal Parameters:

* Ci,Ci are the quantities, in tons, of high sulfur and low sulfur spot market coal use
for a given unit, respectively.

C2,Cj are the quantities, in tons, of high sulfur and low sulfur contract coal use for
a given unit, respectively.

Pi,Pi are the delivered prices, in dollars/ton, of high sulfur and low sulfur spot
market coal for a given unit, respectively.

P,,P, are the delivered prices, in dollars/ton, of high sulfur and low sulfur contract
coal for a given unit, respectively. Delivered coal prices will differ across regions of
the U.S. due to the location of coal mines across the country.17 It is assumed that
generating units are price takers in purchasing coal.

H[,,H, are the heat content for high sulfur and low sulfur spot market coal for a
given unit, respectively. Heat content is the average amount of heat, in million Btus,
in one ton of coal. The delivered price is the dollars/mmBtu paid for coal.

H,,H, are the heat content for high sulfur and low sulfur contract coal for a given
unit, respectively. The heat content will differ across regions of the U.S. due to the
heat content of coal from different coal mines across the country.

Si,Si are the sulfur content for high sulfur and low sulfur spot market coal for a
given unit, respectively. Sulfur content is the percentage of a ton of coal comprised
of sulfur.

17 For example, a generating unit in Wisconsin will have different delivered coal prices
for a particular coal type relative to a unit in Georgia. Wisconsin is closer to the low
sulfur coal mines in the Powder River Basin in the Western U.S., which results in a much
lower delivered price to Wisconsin than to Georgia.


Joshua David Kneifel was born in 1981 in North Platte, N. ,i i-1: i. He grew up

in North Platte, graduating salutatorian from Hershey High School in 1999. Joshua

received his bachelor's degrees in economics and mathematics in 2003 from Doane College

in Crete, N. i i-1;: i. He received his Master of the Arts in economics in 2005 from the

University of Florida, where he specialized in industrial organization, public economics,

and econometrics. From Fall 2006 through Spring 2008, Joshua instructed four semesters

of a course in environmental economics. His classwork and research allowed him to obtain

his PhD in economics from the University of Florida.

Upon completion of his PhD program, he will take an economist position at the

National Institute of Standards and Technology (NIST) in Gaithersburg, Maryland. His

responsibilities at NIST will include research on the life-cycle cost and environmental

impacts of individual products used in the construction industry.

The partial derivative with respect to C represents the impact a one unit change in

high sulfur contract coal has on the unit's total costs.

c + A(il ziri)(m)(S) Ai2h- > 0, 0 if C > 0 (3-49)

Unlike with high sulfur spot market coal, high sulfur contract coal will be impacted by the

high sulfur coal contract constraint (ph). If the generating unit uses some amount of high

sulfur contract coal (Cc > 0), then (3-49) holds with equality.

The partial derivative with respect to Cf represents the impact a one unit change in

low sulfur contract coal has on the unit's total costs.

Pf + A1(1 ziri)(rm)(S) Ai2fHi- 1 > 0, 0 if Cf, > 0 (3-50)

Unlike with low sulfur spot market coal, low sulfur contract coal will be impacted by

the low sulfur coal contract constraint (pi). Similar to high sulfur contract coal, if the

generating unit uses some low sulfur contract coal (C, > 0), then (3-50) holds with


3.7.4 Characterizing a Unit's Spot Market Fuel Choices

A plant's choice of fuel type for a given generating unit is not only based on the

plant's marginal cost of abatement relative to the allowance price, but also the plant's

scrubber installation choice for each individual unit. For this section, we assume that the

scrubber choice is given and focus solely on a generating unit's marginal cost of abatement

and the allowance price, excluding the use of contract coal. The interaction of choices

for the use of contract coal, the consideration of multiple generating units, and scrubber

choice will be discussed later. As in previous models, three cases must be considered:

a generating unit uses both high and low sulfur spot market coal, only high sulfur spot

market coal, and only low sulfur spot market coal.

compliance options, Bohi and Burtraw recommend the more expensive be treated less

favorably. Fullerton et al. (1997) uses a numerical model to determine the impact state

regulation will have on a utility's compliance costs by modeling the cost-minimizing

utility compliance choices and a utility's compliance choices under its Public Utility

Commission rules. The study finds that .i-,iiiii', Ilical cost recovery rules can lead to

utility compliance costs much higher than the least-cost solution, and possibly higher than

a command-and-control approach.

2.3.3 Long-Term Coal Contracts

Joskow (1985, 1988) states that coal contracts decrease transaction costs in coal

purchasing that result from uncertainty and complexity in future coal markets. A utility

may be willing to lp .i more than the current spot market price for coal to protect itself

from unexpected higher rates in the future.

Joskow (1988, 1990) finds that during periods in which the spot market coal prices

were lower than the contracted prices, the contract prices failed to adjust downward. This

downward rigidity of coal prices can lead to utility coal costs being higher than is optimal

in the short run. Some renegotiation, breach of contract, and litigation has occurred,

but nearly all contracts appear to have continued unchanged. The main reason for the

constraints in altering these coal contracts is that less than 15' of coal consumed by

utilities is supplied by a coal company owned by the same utility (Joskow, 1987). Firms

have high legal or negotiation costs of breaking a coal contract when the agreement is

made with a firm that has no financial ties to the utility. Coal contracts may also be

a result of regulations protecting the local coal industry (Arimura 2002). Due to the

inability of contracted coal prices to decrease with spot market prices, large coal price

The two expressions for PiF in (2-99) and (2-101) can be set equal to solve for the

value of e:
( e = '- i i' (2-t10 2)

Filling in for allowances and coal use, it is possible to determine the sign of e.

AMAX D ) (Sih) () AD (DA-- eHs Sih + Sii) () A~
Hih ih

SMAX = )()(1-)-A MAX in the size of the coal contract () (-increases
iih iih
By filling into (2 102), the expression for e can be simplified to parameters for coal price,

sulfur content, heat content, and scrubber capture rate.

C e(A4" AiSMAX) = STnihHih ) (2103)

Since ( < t e < 0 and an increase in the size of the coal contract (i) increases

the magnitude of e.

The third example in Table 2-8 uses data reflective of delivered costs and coal

characteristics for a unit in Florida. A unit initially prefers to use high sulfur coal both

with and without a scrubber because (PA = 'i111.00) < (MCA"' = $3, 887.56). A unit will

prefer to install a scrubber in this example because the indifferent price is PS 1

A low sulfur coal contract for 50'. of coal use will increase the indifference price to

(Pjs e) = ".1 12.42, which will result in a unit not installing a scrubber.
Based on the above three cases, the sign of epsilon can be summarized in Proposition


Proposition 6: Given a low sulfur coal contract, e < 0.

Given Proposition 3 and Proposition 6, the impact of a low sulfur coal contract is

derived in Proposition 7(a) and 7(b).
Proposition 7(a): Assuming ) and allowing for the scrubber choice...S
Proposition 7(a): Assuming (-> >-) and allowing for the scrubber choice...

spot market coal given any contracted coal (Pi'M + PCA + P5C'), or (2) zero

,MAX Di- C H C Ht (2 32)
The latter will only occur if it is weakly cheaper for the generating unit to use low sulfur
ps ps
coal without the emissions restrictions (Q > -l ). Notice that the contracted coal will be

used regardless and will cancel out.

ziPi, + PAA, + max{(P Cf1 + P, Pzsh MA), 0} (2 33)

Combining each of the three cost components results in a unit's total net compliance costs.

Even though the contract coal has no direct affect, the contracts will indirectly affect

compliance costs through a unit's allowance position and scrubber choice. Proposition 1

shows the sufficient conditions under which a coal contract will either increase or decrease

compliance costs.

Proposition 1: Given the scrubber choice

(i) If PA > MCA"'? and the sulfur to heat content ratio of high sulfur contract coal
(h) is greater than the sulfur to heat content ratio of high sulfur spot market coal
("), then a high sulfur coal contract increases compliance costs.

(ii) If PA < MCA"'j and the sulfur to heat content ratio of high sulfur contract coal
(h) is greater than the sulfur to heat content ratio of high sulfur spot market coal
( ), then a high sulfur coal contract increases compliance costs.

(iii) If PA > MCA"'. and the sulfur to heat content ratio of low sulfur contract coal ( )
is greater than the sulfur to heat content ratio of low sulfur spot market coal (>),
then a low sulfur coal contract increases compliance costs.

(iv) If PA < MCA"'S, then a low sulfur coal contract decreases compliance costs.

See the Appendix A for detailed proofs of Proposition 1. Proposition l(iv) may seem

counter-intuitive, but it shows the importance of being careful about defining a unit's

Given the scrubber choice, a generating unit that does not face any coal contract

constraints will make its optimal choices based on the relationship of the allowance

price (PA) and the marginal cost of abatement from switching fuels from high sulfur

spot market coal to low sulfur spot market coal (MCAj'"). If PA > MCA"'S, then

it is cheaper to meet the emissions constraint by decreasing emissions by switching

fuels than purchasing allowances. A unit will use all low sulfur coal and purchase the

minimum amount of allowances, which will result in the following compliance choices:

ch = Cf1 = s,MAX Di, and A, AMI. If PA < MCA'", then it is cheaper

to meet the emissions constraint by purchasing allowances than decreasing emissions by

switching fuels. A unit will use all high sulfur coal and purchase the maximum amount

of allowances, which will result in the following compliance choices: Ch= C, Di,

C, 0, and Ai = AAX.

Introducing a high sulfur coal contract constraint will restrict spot market coal use by

Cc and will alter excess demand (Ai). The maximum amount of high sulfur spot market

coal use decreases from Di = MAX to Di C = ,AX The maximum amount

of low sulfur spot market coal decreases from Di CI,MAX to Di Cc = ,MAX

If PA > MCA"', a unit will use it's contract constrained maximum amount of low

sulfur spot market coal (Cs,MAX) and excess demand will increase from AMIN to AMIN.

If PA < MCA"'8, a unit will use it's contract constrained maximum amount of high

sulfur spot market coal (,MAX) and excess demand will shift from AMAX to MAX. If
Sh < S then AAX < MAX. If i> i then AMAX > MAX
H H S s H H i
ih i- h ih ih
Introducing a low sulfur coal contract constraint will restrict spot market coal use by

Ci and will alter excess demand (Ai). The maximum amount of high sulfur spot market

coal use decreases from Di = CsMAX to Di Cc = sMAX The maximum amount
of low sulfur spot market coal decreases from Di C"SM to Di Ci= ",' f

PA > MCA8', a unit will use it's contract constrained maximum amount of low sulfur

spot market coal (iMAX) and excess demand will shift from AMIN to AI. If < H

decreases the maximum amount of low sulfur spot market coal use from Cs MAXt to

Cis,MAX > s,MAX Di c H (251)

If < :, the sulfur content per unit of heat content is lower for low sulfur contract

coal than low sulfur spot market coal and will decrease the minimum emissions from


SEMIN > EMIN ( )Qs (s,/1MAX +ScC) (2 52)
-^ ^i > 'i ( ,- ( ~ )i) ,., + (Sil) pc s( )

Lower emissions result in a decrease in a unit's minimum excess demand from AyAX to

AMAX in Figure 2.8(i):

SAIN > A' J EIN A7 (253)

Therefore, a low sulfur coal contract for coal with a lower sulfur to heat content ratio will

decrease excess demand for allowance prices PA > MCAj'".U
If >HZ the sulfur content per unit of heat content is greater for low sulfur

contract coal than low sulfur spot market coal and will increase the minimum emissions
from EMIN to EMIN

= EMIN < EMIN (1- r)(m)( ,MAX + SC) (2-54)

Greater emissions result in an increase in a unit's minimum excess demand from AIN to

AIIN in Figure 2.8(ii):

SAmIN > A'''~ EN At (2-55)

Therefore, a high sulfur coal contract for coal with a higher sulfur to heat content ratio

will increase excess demand for allowance prices PA > MCAj'".E

A binding low sulfur coal contract that restricts a unit's ability to use allowance can

force a net buyer of allowances to decrease their allowance purchases from AMAX to AMAX

and abate more emissions than the generating unit would prefer as shown in Figure 2.8. If

an increase in fuel costs will have a larger impact on renewable capacity in California

than in Rhode Island. Larger states should have more funding to 1 iv for projects to

increase renewable capacity. Renewables Portfolio Standards with Sales Requirements set

requirements on the percent of generation that must originate from renewable sources.

States with more generation will have more total generation that is required to originate

from renewable resources, which should lead to more renewable capacity in those states.

The following three variables are included in the model to control for market

structure. Two of these variables are hydropower capacity (PCT HYDROPOWER)

and nuclear power capacity (PCT NUCLEAR) as a percentage of total capacity excluding

non-hydro renewables. Hydropower should lead to less non-hydro renewable capacity

because hydropower has low marginal production costs, and the capacity typically was

constructed many years ago. With lower marginal costs and sunk capital costs associated

with hydropower, hydropower will be the first renewable energy to be implemented

because it is more economically competitive than most non-hydropower renewables

available to the electric power industry. Consumer and/or policy driven demand for

renewable-based electricity may not differentiate between hydropower and other renewable

sources, which allows hydropower to be a substitute of non-hydro renewables.

Similar to hydropower, nuclear power has low marginal costs of producing base load

electricity, has sunk capital costs, and has no emissions. If non-hydro renewable capacity

is deploy, ,1 based on economic factors, given similar emissions profiles, greater nuclear or

hydropower capacity should decrease the amount of non-hydro renewable capacity.

An alternative possibility is that regulators in states with large amounts of nuclear

power encourage power producers to use other resource types to meet new demand.

Renewable energy may be used by utilities to alleviate pressure from environmentalists

over nuclear power, thus leading to greater deployment of renewable energy capacity in

states with large amounts of nuclear capacity. The sign of PCT NUCLEAR will depend on

which of these two factors has the larger effect on power producers.

Proposition 7(b) is shown graphically in Figure 2.8(ii). As with a high sulfur contract,

some of the price ranges may not exist for a particular case. However, the remaining parts

of the propositions hold.

2.7 Possible Implications on the Allowance Market and Industry Compliance

Under Phase I of Title IV, there were many generating units facing high sulfur

coal contracts for at least a fraction of their total coal use. If a binding high sulfur coal

contract leads a unit to choose a suboptimal compliance choice, such as purchasing

additional permits or installing a scrubber instead of switching fuels, and results in

weakly higher compliance costs. A unit's suboptimal choices not only increases a unit's

compliance costs, but should also increase compliance costs for the industry as a whole.

As has been show in several examples, high sulfur coal contracts for a large fraction

(50-101i i.) of coal use can greatly reduce a unit's "indifference price" to installing a

scrubber. Some units under Phase I initially appear to have installed a scrubber when

it was not a unit's optimal compliance option, increasing a unit's compliance costs.

Additional scrubber installations should have resulted in greater emissions reduction,

which should simultaneously lower demand and increase supply of allowances as a unit

switches from a net demander to a net seller. In doing so, the equilibrium allowance

market price should be driven lower, which may explain the lower than expected allowance

prices realized during Phase I. Even though the allowance market price was lower than

expected, the inefficient unit compliance choices resulted in higher than expected total

industry compliance costs.

Under future CAIR regulation, a unit's compliance options may be restricted by

low sulfur coal contracts agreed upon during the 1990s to meet Title IV emissions

requirements. A unit may find installing a scrubber and using high sulfur coal to be it

best compliance option. However, low sulfur coal contracts may lead a unit to choose a

suboptimal compliance choice, such as switching fuels or installing a scrubber while using

Following along the same thought process, if a plant chooses to install a scrubber at two

generating units, the scrubbers will be installed at the generating units with the two

lowest ACAi. The above condition will only hold for affected generating units. A plant

will not install a scrubber at a non-affected generating unit because reducing emissions at

a non-affected unit does not relax any emissions constraints.14 At which generating units will a plant install a scrubber?

The order a plant will install scrubbers at its generating units and a plant's marginal

costs of abatement for both with and without a scrubber are known. Now it must be

defined when a plant will install a scrubber at a given generating unit by finding the

allowance price at which a plant is indifferent to installing a scrubber. This is tough to

analytically show because a plant has multiple choices to minimize its total costs through

scrubber installation.

Assume that all generating units operated by a plant are affected by Phase I, each

unit uses its cost minimizing combination of coal and allowances based on its scrubber

choice, and the units are sorted by ACA, from smallest to largest (ACA1 < ACA2 < ... <

ACA,_1 < ACA,). A plant's scrubber choices will be based on the relative total costs of

each possible combination of scrubber installation. A plant will not install a scrubber at

any generating units if its total costs are lower with no scrubbers installed than installing

a scrubber at the generating unit with the lowest ACAi (C(zi = 0, z2 = 0, ..., z = 0) <

C(zi = 1, 2 = 0,..., z = 0)) where C(*) is the total costs for a given scrubber choices

and optimal coal and allowance choices. We already know that any other combination of

scrubber installation must result in higher costs.

14 Some plants voluntary enrolled units into Phase I, which are labeled as substitution
units or compensation units. Although these units were not initially affected, they were
enrolled into the program and face the same types of requirements as the original units.
For this reason, they are considered affected units.

(i) For the range of allowance prices PA > MCA'"8, a low sulfur coal contract will
,, ., /l.; increase excess demand.

(ii) For the range of allowance prices (Pf e) < PA < MCA'8 a low sulfur coal contract
will ;, '.it l,; decrease excess demand.

(iii) For the range of allowance prices PA' < PA < (PF e), a low sulfur coal contract will
,, .. I; increase excess demand.

(iv) For the range of allowance prices MCA"'8 < PA < PFA, a low sulfur coal contract will
.i,;, increase excess demand.

(v) For the range of allowance prices 0 < PA < MCA"', a low sulfur coal contract will
., l.; decrease excess demand.

Proof of Proposition 7(a):

(i) When a unit faces PA > MCA'A", a unit prefers to install a scrubber and use all
low sulfur coal. From Proposition 3(iii), given S > S a low sulfur coal contract
increases the minimum emissions level, which will weakly increase a unit's allowance
excess demand.

(ii) When a unit faces (Pf c) < PA < MCA' ", a unit prefers to install a scrubber and
use all high sulfur coal. From Proposition 3(i), a low sulfur coal contract decreases
the maximum emissions level, which will weakly decrease a unit's allowance excess

(iii) From Proposition 6, when a unit faces PA < PA < (PA e), a low sulfur coal
contract increases a unit's indifference allowance price of installing a scrubber above
the allowance price, which leads a unit to not install a scrubber where it initially
would have done so and increases a unit's emissions and a unit's excess demand.
From Proposition 3(iii), given the scrubber choice and S > 7 a low sulfur coal
contract will weakly increase a unit's emissions and excess demand. The combined
net effect is weakly positive and weakly increases excess demand.

(iv) When a unit faces MCAA'j < PA < Pf, a unit does not install a scrubber and prefers
to use low sulfur coal. From Proposition 3(iii), given > a low sulfur coal
contract increases a unit's minimum emissions level and excess demand.

(v) When a unit faces 0 < PA < MCA"'8, a unit does not install a scrubber and prefers
to use high sulfur coal. From Proposition 3(i), a low sulfur coal contract decreases a
unit's maximum emissions level and excess demand.E

Proposition 7(a) is shown graphically in Figure 2.8(i).
Proposition 7(b): Assuming that ( < and allowing for the scrubber choice...
Proposition 7(b): Assuming that (- < H) and allowing for the scrubber choice...

2.3 Literature Review

2.3.1 Title IV: Phase I

There has been significant research done on the Title IV S02 Cap-and-Trade

Program, both for Phase I and Phase II (Ellerman et al. (2000); Burtraw et al. (2005)).

From Table 2-1, it can be seen that, in general, the compliance cost estimates before

Phase I took effect were higher than the estimates made after Phase I became effective

and actual data could be used in the estimates. The pre-policy estimates range as high

as $1.34 billion/year with most estimates at least 1.11 million/year. The actual cost

estimates are towards the lower end of this range between $730-$990 million/year.

There are several reasons for the differences between initial estimates and actual

.,.- -regate industry compliance costs. The most important factor was the decrease in

delivered low sulfur coal prices. At the unit level, lower low sulfur coal prices decreased

the marginal cost of reducing emissions through fuel switching, which was the compliance

option chosen by 5"' of all affected units, while 3:'. of affected units chose to purchase

allowances, 10'-. installed a scrubber, :'-. shut down, and :'-. chose other methods.7

Several of these studies have estimated the cost savings resulting from the allowance

trading system. Carlson, et. al. (2000) used an econometric-based simulation model to

estimate the potential cost savings from trading in the program compared to a uniform

emissions rate standard. The potential savings was estimated at -'>iI million, Of of

which is a result of switching from high to low sulfur coal and 211' from technical change,

such as improved scrubber technology (Burtraw et al., 2005).

Keohane (2002) simulates which generating units would have installed scrubbers

under a uniform emissions-rate standard and finds that the total number of scrubbers

would have been one-third higher than the actual number of installed scrubbers under the

7 Energy Information Association, The Effects of Title IV of the Clean Air Act
Amendments of 1990 on Electric Utilities: An Update"

increase the emissions level and require additional allowances (AMIN > AMIN).

P~iIV ( is, MAX\ ,s( s,MAX 0)] [ *AMIN+ ps / -s,MAX\ srs,MAX)
[PI -i T1 Pish (0- C"A +pis Cil i^ I ~{J \

The change in compliance costs will be:

J/IV- AMIN) S (C MAX is,MAX r s s,MAX s,MAX) (A 9)
A MN ) hih Cih + ~ il ) 9)

The first term is positive because AMIN AMIN. The second term is also positive because
~s,MAX -s ,MAX. sMAX < CIs,MAX
hMAX > MAX. The third term is negative because ",MAX < ,MAX

Now fill in for coal use:

is,MAX Di ^sMAX D, CHh Cs,MAX Di ^sMAX Di Cfh Hh
ih H i H8 i H i -
h h Hih i His

The change in compliance costs resulting from a high sulfur coal contract is the

increase in net allowance purchases minus the cost savings from not switching fuels from

the high sulfur contract coal.

Pj4 V AIN) ) hH- (A-10)
Hih His

Now fill in for the net allowance position:

MIN D I/ I V D ch ih cs c A 'e
IN Srm A Ai D s i + C
His, His, M i

By adding and subtracting mPA ,h combining like terms, and dividing through by

m( H), the expression can be simplified to:
r50 szc ps ps
ih ih ih ih

itHCh [P>( SMh ) + PA CA (A11)
i h i h
iCHh PI ( h ) + PA MCA"] (A- t)
;57h ;571

unit will not buy or sell any allowances either since it has no emissions constraint. The

only possible choice is to install a scrubber at the affected unit (Unit 1).
Case 1: Install No Scrubber

The total costs of not installing a scrubber must be lower than the total costs of

installing a scrubber at the affected unit.

P4AA + PC'h + Pl Cl + PCh + P Cc + PhsC2 + P1 C2 + PJCjh + P C7l

< Pl + P4Ai + PhClh + PlfCPl + PhClh + P/C'l + PhAC + PtlCj2 + PCh + PIF l

From this equation, we can solve for P, for this condition to hold in (3-87).

ps 1 + Ph(C' Ch) + Pi'(C CO-)
(A1 A1)

Case 2: Install One Scrubber

The total costs of installing a scrubber must be lower than the total costs of not

installing a scrubber at the affected unit.

Pz + PAA t + CISlh + Pl + PFClh + P CT + P ChS + PFC + PhFCh + P CIl

From this equation, we can solve for PAs for this condition to hold in (3-88).

ps z + Ph(Ci C +) P(C'1 O-8))
P F > F Cit (3 88)
A (A1- A1)

3.7.12 Summary of Plant Level Results

A plant's decision-making process may not minimize costs for each generating unit

because a plant's concern is based on the combined costs of all generating units under its

operational control. The choice to install a scrubber is based on the characteristics of all

the units at a plant, not just the unit at which the scrubber may be installed. Once the







s,MAX D_ i s,MAX Di CizHH z Cs,MAX Di Cs,MAX D, -C H-
h "i ih ii ilH

By filling these expressions into (2-86), the expression for c can be simplified to parameters

for coal price, sulfur content, heat content, and scrubber capture rate.

C, H(P D(m [(r h h () H %S]H
[[ HH Hl ] HH \(2 87)

A more interpretable expression is derived by multiplying through by ((A" AfMAX))

and dividing through by m(h H ).


((H, WA i) $- (PA MCAn S' (289)
ih TI
From the initial assumption that a unit uses low sulfur coal when it does not install a

scrubber, it is known that (PS > MCA"'). Since (> < h ), then e < 0 and its
il ih
magnitude increases as the size of the low sulfur coal contract (Ci,) increases.

A binding low sulfur coal contract is more likely to impact units under CAIR. An

example using recent data reflective of the coal availability and delivered prices for a

unit in Alabama in 2000 will help to show e < 0.27 Under the assumptions, a unit

will prefer to switch fuels to abate emissions if it does not install a scrubber because

MCAs's = -'11.40, which is much lower than the assumed allowance price PA $700.00.

If a scrubber is installed, a unit prefers to use high sulfur spot market coal because the

marginal cost of abatement is much greater than the allowance price (MCA'"8 = $11, 228).

A unit will not install a scrubber because the allowance price at which a unit is indifferent

to installing a scrubber (PA = $731.46) is higher than PA. A low sulfur coal contract

27 The data can be found in Table 15.A of the 2000 Electric Power Annual Volume II.

Second, consider how fuel contract constraints will impact the total industry

compliance costs. Simulation 5 introduces fuel contract constraints to the the emissions

constrained model assuming the 46 scrubbers installed as in Simulation 3. The simulation

results in a higher allowance price of I_'il. 70, which could be the result of a greater

demand for allowances due to more high sulfur coal use resulting from high sulfur coal

contracts. Total industry costs are $8.76 billion, where 29 scrubbers were installed in

response to Title IV. Assuming these scrubber choices, the minimum compliance costs

relative to the unconstrained model in Simulation 1 are $1.07 billion, or $531 million

('-I'.) higher than if contracts are not taken into consideration.

Simulation 6 introduces fuel contract constraints to the the emissions constrained

model while allowing generating units to make their scrubber choice. The simulation

results in an allowance price of $210.74. Total industry costs are $8.63 billion where 44

scrubbers are installed in response to Title IV. The minimum compliance costs when

compared to Simulation 1 are $939 million, or -.'I,1 million higher than if contracts are not

taken into consideration. The minimum compliance costs are close to the actual cost found

in Sotkiewicz and Holt (2005) at $990 million and Carlson et al. (2000) at $910 million.

Allowing generating units to choose whether to install a scrubber allows the industry

to lower its total costs by $132 million. As would be expected, introducing the contract

constraint results in more scrubber installations due to Title IV from 27 to 44 because

high-sulfur fuel contracts increase the incentive for a constrained generating unit to install

a scrubber. As in Simulation 4, the allowance market does not clear due to the discrete,

endogenous scrubber choice. However, the excess demand of 7,797 allowances account for

less than 0."' of the 5+ million allowance market, and could be assumed to be bought

from an allowance broker.

The actual total industry compliance costs are found in Simulation 7. By using

the actual emissions and electricity production for each unit, it is possible to determine

each unit's actual coal mix. These actual decisions resulted in an allowance price range






















Dependent and Control Variables . ......

Regressions Results . .............

Policy Variables . ... .. .. .. .. ... .

Variable Effects of Significant Variables . ..

Phase I Compliance Cost Estimates . ....

High Sulfur Coal Contract: Assumptions . .

High Sulfur Coal Contract: Results . ....

Low Sulfur Coal Contract Examples: Assumptions .

Low Sulfur Coal Contract Examples: Results .

Example Epsilon Magnitude: Case 1 . ....

Example Epsilon Magnitude: Case 2 . ....

Example Epsilon Magnitude . ........

Example: Contract Coal Distribution . ...

Sulfur Conversion by Fuel Type . ......

Simulation Results . ..............

Impact of Contract Constraint on Scrubber ('C! ..i .

Simulations with Engineering Data . .....

Impacts of a Reduction in the Allowance Allocation o

Math Example: Two Affected Units . ....



.. . 43

.. . 44

.. . 45

.. . 46

. .. . 13

.. . 114

.. . 114

. . 114

... . 15

. .. . 16

.. . 117

. .. . 18

.. . 190

.. . 190

.. . 19 1

... . 92

. .. . 9 2

f 10' . . 192

. .. . 9 2

3 except if the allowance market price falls in the price range (Ps, Ps e) and < .

If the allowance market price falls in this range, there are two countering effects on excess

demand, the increase in excess demand resulting from no scrubber installation and the

decrease in excess demand from the lower sulfur content of low sulfur contract coal relative

to low sulfur spot market coal.

capacity. NET METERING is interacted with GEN to control for the policy's effect based

on market size.

Interconnection standards (INTERCON STANDARDS) are a set of guidelines used

to safely and effectively connect individual renewable generating units to the electric

utility power grid. Some have technical requirements, such as generator type and size

limits, mandatory safety and performance standards, and insurance requirements that

must be met before a net metering customer can connect to the utility's network.

Interconnection standards must be met by any commercial, industrial, residential, or

government customer that decides to connect to the grid. Without these state policies,

the net metering connections could cause 1n i, i problems for the grid, power producers,

and other purchasers. Interconnection standards increase the costs of hooking up to

the grid for net metering and may offset some of the negative effect from net metering.

INTERCON STANDARDS is also interacted with GEN to control for market size.30

State Government Green Power Purchasing policies require that some percentage

of a state government's electricity purchases be from renewable sources. These purchase

agreements range from 5'. to 50' of a state government's electricity purchases. Similar

to Renewables Portfolio Standards with Sales Requirements, a State Government Green

Power Purchasing agreement increases the need for renewable-based electricity generation.

As state government electricity use rises, the renewable generation needed to meet the

requirement increases. If the new generation needs cannot be met by current renewable

capacity, power producers will need to construct new renewable energy capacity. The

size of the State Government Green Power Purchasing requirement, in terms of a

percentage of the state government's electricity purchases, is interacted with GEN to

30 Since only four observations have interconnection standards and no net metering, the
interaction term measures the effect of interconnection standards on states that already
have net metering policies. Only 86 of the 187 observations ( I'.' ) with net metering also
have interconnection standards, which removes concerns of multicollinearity.

and without a coal contract ((A-5)-(A-2)) and split it into two components, the change

in compliance costs and the change in fuel costs. For simplicity, assume the conditions in

Proposition l(i) hold.

4P, + PI(A* A*() C+ P +P ( -C + P(' C}*) (A-16)

Assume that a unit faces a high sulfur coal contract, and prefers to switch to low sulfur

coal use to meet its emissions requirement instead of purchasing allowances or installing a

P/ ,iV AMIN) + p ps(s,MAX I,MAX) (A7)
i js ilc (il

To be able to interpret this expression, it is necessary to add and subtract (P ,hJi ).

C (P PH ) P( V AMIN) + Ps h + P s,MAX s,MAX) (A 18)
-- ich (P ~h i h T + p (A i )+ Ai ihrr 1H isil ~ pi ) "-,;)
ih ih

The first term is the change in high sulfur coal costs from using the contract coal instead

of spot market coal. These are not changes in compliance costs because they will occur

with or without the program. The remaining terms are the change in compliance costs

resulting from the program.

H h ,AIN), Al Pc -9)
ch(Pich Ph ) + P( i" V AMIN )rCH (A 19)
hh ihrs Hsh Hi (iA 1

Change in Fuel Costs Change in Compliance Costs

By filling in for the coal use, the last terms give the same expression for the change in

compliance costs as in the proof of Proposition l(i).

ih ih
Ch Hh [P( ) + PA MCA] (A 20)
ih il
A unit's compliance costs increase if > H-.
Hih ih

The first-order conditions are identical to those of the individual generating unit in

C'! lpter 2 when neither coal contract constraints bind. So a generating unit's cost

minimizing choices of C[*, C,* at PJ for each generating unit also minimize total industry

compliance costs at Pj.

These first-order conditions can be used to solve for Ail, the allowance shadow price,

which is also the same as for the generating unit problem. Since the compliance choices

that minimize the costs for all generating units result in the equilibrium P!, the allowance

choices also minimize the total industry costs.
PS" P5"
Ail MCA' --7Hl h Th (3-29)
(t Z r) (m)( |-h -| )
ih 7T

The same approach can be used under the iiili i-I i compliance cost problem

with coal contract constraints to show that contract constrained minimum total industry

compliance cost will differ from the unconstrained costs.

+ (p p, s, MAX 0 1
z ~ P + max{(P + Phi -h Pfih CM )0
Zi,Ai,Gi,G^ T^
n n

i=i i=1
(C H, + Ch Hh + CizH + CiH,) > Di

Cih, Cl >_ 0

z e {0, 1}

The first-order conditions. For high sulfur spot market coal...


Vi e { ,...,n}

Vi e {1,..., n}

Vi e {1,..., n}

1i + Ail(1 ziri)(m )(S,')) Ai2H,8 > 0,= 0 if C67 > 0


For low sulfur spot market coal...

Pil + Ai1(t zi)(m)(Sf) 2Hf, > 0, 0 if C'i, > 0


First, it is important to generalize the indifference price at which a generating

unit will install a scrubber (Pf), which can be derived by setting a unit's costs when it

does not install a scrubber, which includes net allowance purchases (Ai) and fuel costs

(PijC + PiC,), to a unit's costs when it does install a scrubber, which includes net
allowance purchases (AMAX), fuel costs FCs s,MAX), and costs of a scrubber (P).

Assuming that high sulfur spot market coal is cheaper than low sulfur spot market coal, a

unit uses all high sulfur spot market coal if it installs a scrubber because MCA'"8 > PA. It

is uncertain if a unit will use high or low sulfur coal if it does not install a scrubber, and

will depend on the relationship between PA and MCA'".

Piz + PSASMAX + pih CMAX PAAi + P, i, + PiCi (2-59)

A unit that faces a coal contract will face a different indifference allowance price of

(PfA ) because parameters values on both sides of the equality will change. e could

be positive or negative depending on several conditions, including the type of coal under

contract. The new values that solve this equality are the contract constrained cost

minimizing parameter values. The fuel costs for the contract coal will be the same both

with and without a scrubber and will cancel out.

P. + (Pf AMA + PMh ( A s)AXA, + PsC + PC (2-60)

By solving for the constant P, in (2-59) and (2-60) and setting the two expressions equal

to each other, the sign and value of c can be derived. Impact of a binding high sulfur coal contract

The impacts of a high sulfur coal contract on the two pieces of the excess demand

correspondence will be the same as in Section 6.7 where the scrubber choice is given

except there will be an additional impact on excess demand from the contract on the

allowance price at which a unit is indifferent to installing a scrubber.


3.1 Introduction

The U.S. SO2 Trading Program created by Title IV of the 1990 Clean Air Act

Amendment led to lower compliance costs than what would have occurred under a

Command-and-Control approach. However, all compliance cost savings were not realized

in the early years of the program. There have been several conjectures as to why the

hypothetical outcome was not obtained, including short-run rigidities from fuel contracts.

C'! lpter 2 shows how fuel contracts could alter a generating unit's compliance decision

in the U.S. SO2 Trading Program, but ignores any .I -: regate allowance market and

iiillli-1i v-.1-ide compliance cost impacts. This paper expands on C!i lpter 2 by looking at

the allowance market equilibrium impacts and total industry compliance costs from fuel

contracts through analytics and empirical modeling.

Given the scrubber choice, an allowance market equilibrium will exist. Allowing

for the scrubber choice makes it impossible to guarantee an equilibrium, but one still

may exist. Binding fuel contracts may lead to altered unit-level excess demands and,

in so doing, the allowance market price (PA). Meanwhile, binding contracts can alter

compliance decisions and increase total industry compliance costs.

This paper uses generating unit-level simulations to replicate results from previous

studies and show that short-run fuel contracts appear to explain a large portion of the

previously unexplained excess compliance costs found in previous simulations. Simulating

the least-cost compliance choices without including fuel contract constraints results in

minimum annual industry compliance costs of _'-- 3 million, which varies greatly from

the actual compliance costs of $1.30 billion found in these simulations. Once fuel contract

constraints are introduced into the simulation, the minimum annual industry compliance

costs become $1.01 billion. Based on these results, fuel contract constraints explain -i.'1.1

million, or 6 !'. of the excess compliance costs realized in the program for 1996.


-E C

* Co



n^ n

S*0 e


ee ee

0 e
a t

- 0 -O




v scv


E .2


the contract constraint is large enough it will shift A AX to the left of Ai = 0 and force

a unit to be a net seller. If the contract constraint forces the generating unit to use all

low sulfur coal, then the excess demand is a vertical line where the unit's only choice is to

purchase the minimum amount of allowances. This case may occur during Phase II if some

generating units purchased 101t' low sulfur coal through contracts.

In Figure 2.8, the net compliance costs can be seen in the shaded area (a) that

represents the costs of purchasing allowances to cover the unit's emissions above its

allowance allocation.

Now consider how a low sulfur coal contract will impact the excess demand

correspondence, compliance costs, and total costs. As has already been shown above

and can be see in Figure 2.8, a low sulfur coal contract will decrease the maximum excess

demand from A"AX to AAX and may increase or decrease the minimum excess demand

depending on the relative sulfur to heat content ratio of contract to spot market coal.

These shifts in a unit's excess demand may have two distinct effects on a unit's costs.

The first cost impact is the additional compliance costs to a unit from switching from high

sulfur spot market to low sulfur contract coal instead of purchasing allowances, which is

represented by area (c) in Figure 2.8. A unit must use some low sulfur coal, which results

in a unit abating additional emissions instead of purchasing allowances. Area (c) is the

difference between the costs of switching fuels to decrease emissions, area (b + c), and the

decrease in costs from allowance purchases, area (b). Net compliance costs increase from

(a+b) to (a + b +c).

The second cost impact results from different prices for low sulfur contract and spot

market coal, which is represented by area (d). Higher priced contract coal causes a unit

to have additional fuel costs to meeting electricity demand, which increases a unit's total


The graphical description of coal contract impacts on excess demand and costs

can be seen in the example defined in Table 2-4. Given the coal characteristics, demand,

correspondences with and without a scrubber with a discontinuity representing the

discrete choice separating the two pieces.

A generating unit's excess demand correspondence can be derived from its optimal

compliance choices as the market allowance price changes. For allowance prices MCAj'" <

PA, a generating unit installs a scrubber because PS < PA, and switches fuels from high
sulfur spot market to low sulfur spot market coal because MCA'"8 < PA. So a generating

unit will have minimum excess demand when a scrubber is installed (AMIN), which has

already been derived in (2-36) given that zi 1.

For allowance prices PA < PA < MCA>', a generating unit installs a scrubber

because Pf < PA, and uses high sulfur coal because PA < MCA'A". So a generating

unit will have maximum excess demand when a scrubber is installed (AfMAX), which has

already been derived in (2-39) given zi 1.

For allowance prices MCAj'8 < PA < Pf, a generating unit does not install a scrubber

because PA < P, and switches fuels from high to low sulfur coal because MCA"'8 < PA.

So a generating unit will have minimum excess demand without a scrubber installed

(AMIN), which has already been derived in (2-36) given zi = 0.

For allowance prices 0 < PA < MCA"', a generating unit does not install a scrubber

because PA < Pf, and high sulfur coal because PA < MCA"'j. So a generating unit will

have maximum excess demand without installing a scrubber (AMAX), which has already

been derived in (2-39).

AAX if 0 < PA < MCA"' < P or 0 PA P< PX < MCA'
OAMAX _(1 )AMIN if (MCA = PA < P)V e [0,1]

A AIN if MCA>'" < PA< PA
Ai =S
AMAX if P < PA < MCAj'"
OASMAX (1- O)AMIN if MCA = PA V e [0,t1]

AfIN if MCA'" < PA

compliance choices on which pollution markets rely to improve cost-effectiveness in utility

decision-making. It may well be the case the presence of long-term contracts are driving

part, or most, of the deviations from least-cost that have been simulated or estimated in

the literature for Phase I. If long term coal contracts did in fact lead to inefficiencies under

Phase I, contracts could have similar effects under the newly enacted Clean Air Interstate

Rule (CAIR) of 2005 that further restricts SO2 emissions.

In this paper a model of unit-level S2O compliance is constructed that incorporates

the presence of coal contracts to examine how long-term coal contracts affect utility

compliance choices and a unit's compliance costs. As expected, the presence of coal

contract constraints leads to compliance costs in excess of the hypothetical least-cost

solution. The presence of binding high sulfur contract constraints that were likely in Phase

I of the Title IV SO2 Program may explain the lower than expected allowance prices

in Phase I that accompanied compliance costs that were above the least-cost solution.

It is also found that the presence of binding low sulfur coal constraints that may exist

under CAIR, which may lead to allowance prices that are higher than without the binding

constraint. The effects of the contract constraint seem counter-intuitive: binding high

sulfur coal constraints leading to lower excess demand for allowances, which could reduce

the allowance market price. Binding low sulfur coal constraints leading to higher excess

demand for allowances, which could increase the allowance market price. The interaction

between the contract constraints and the discrete nature of the scrubber choice leads to

these unexpected results.

2.2 Policy Background

2.2.1 Title IV of the Clean Air Act Amendment

Under the Title IV S02 emission trading program, affected units are allocated

allowances, which permit the holder to emit one ton of S02 in the year in which the

allowance is issued or any year thereafter, and that may be traded (bought or sold) in the

market or banked for future use. At the end of each year, generating units are required to

[14] C. C'!, i, R. Wiser, M. Bolinger, 2007. Weighting the costs and benefits of state
renewables portfolio standards: A comparative analysis of state-level policy impact
projections, Lawrence Berkeley National Laboratoy.

[15] J.S. Coc-:iii-, J.R. Swinton. The price of pollution: A dual approach to valuing SO2
allowances, J. Environ. Econ. Manage., 30 (1996) 58-72.

[16] Database of State Incentives for Renewable Energy. State renewable energy policy
summaries, Website, Accessed March 2004-March 2006,

[17] Database of Utility-Scale Renewable Energy Projects. Lawrence Berkeley National
Laboratory and Clean Energy State Alliance (CESA), Website, Accessed December

[18] A.D. Ellerman, R. Schmalensee, P.L. Joskow, J.P. Montero, E.M. Bailey. Emissions
trading under the U.S. acid rain program: Evaluation of compliance costs and
allowance market performance, Center for Energy and Environmental Policy
Research, Massachusetts Institute of Technology.

[19] A.D. Ellerman, et al. Market for clean air: the U.S. acid rain program, Cambridge
University Press, New York, 2000.

[20] A.D. Ellerman, J.P. Montero. The declining trend in sulfur dioxide emissions:
Implications for allowance prices, J. Environ. Econ. Manage., 36 (1998) 26-45.

[21] Energy Information Administration. Electric power annual 1996: Vol.2,
DOE/EIA-0348(96)/2, February 1996.

[22] Energy Information Administration. The effects of Title IV of the Clean Air Act
Amendments of 1990 on electric utilities: An update, DOE/EIA-0582(97), March

[23] Energy Information Administration. Energy Policy Act transportation rate study:
Final report on coal transportation, DOE/EIA-0597(2000), October 2000.

[24] Energy Information Administration. EIA historical state electricity databases
and FERC Form-423 database, Website, Accessed March 2004 March 2007,

[25] Energy Information Administration. Annual Energy Review 2005.
Report #DOE-EIA-0384(2005). Website, Accessed February 2007,

[26] Environmental Protection Agency. Reducing power plant emissions for cleaner
air, healthier people, and a strong America. Office of Air and Rediation,, March 2005.

If renewable capacity is being constructed on economic grounds, a rise in the retail

price of electricity makes renewable energy more profitable and should have a positive

effect on renewable capacity.20 However, retail prices in a state may be simultaneously

determined with renewable capacity because using more renewable capacity increases the

average costs of production, which could lead to higher prices. Using the state's retail

price could also lead to multicollinearity problems with fossil fuel costs because higher

fuel costs will lead to higher electricity prices. To control for this endogeneity and possible

multicollinearity, the model must use a proxy for a state's retail price. A proxy must be

correlated to the endogenous variable and have no impact itself on the dependent variable.

The weighted average real retail price per kilowatt-hour of the bordering states (BORDER

PRICE) is an ideal proxy for retail prices because it meets both of these requirements.21

capacity data for each type of renewable energy are not available from the EIA, making
net summer capacity the closest available alternative measure. Even though there is a cost
of energy estimate for both solar thermal and solar Photovoltaic, the solar capacity data
are not segregated into these two types. A non-weighted average of solar thermal and PV
is taken to get the levelized cost for total solar capacity. Since all solar power accounts
for less than 2.5'. of total non-hydro renewable capacity in the U.S., it is unlikely that
using some weighted average of solar thermal and solar PV would make any significant
difference. Summer capacity refers to the maximum output generating equipment is
expected to supply to a system demonstrated by tests at the time of summer peak
demand. Nameplate and summer capacity have high correlation, but are not identical
due to different operating conditions across utilities. Definition of nameplate is in Footnote
20 Average retail price is based on all sales in the market: residential, commercial,
industrial, and other customers. Data are available from the EIA Historical Databases.
Average retail price data are originally in nominal terms for each month. Two steps have
to be taken to adjust the data into real terms for each year. First, the monthly data are
divided by the CPI for all goods to get the monthly data into real terms. Second, monthly
electricity sales are used to get a weighted average price for each year. The resulting
variable is the real average retail price for each state and year in January 2002 dollars.
21 Bordering states are all states that either share a border, such as Arizona and New
Mexico, or meet at a corner, such as Arizona and Colorado. The prices are weighted by
sales in the bordering states. The correlation of retail price to BORDER PRICE is 0.836.









o AsrE -AS T r -"


Figure 2-6. High Sulfur Contract: Relative Savings from Contract Coal



Figure 2-7. Cost Savings from Using Allowances Over Fuel Switching

MCs, c




Figure 2-8. Low Sulfur Contract

MC4S, -



- J I



C,4 -


' _Jc4SS




_._ -

Table 1-2. Regressions Results
Total Non-Hydro Renewable Capacity








GEN (1 TWh)
















State Fixed-Effects
Robust Standard Errors in Parentheses;* significant at




309.103 413.745
(7.689)*** (55.340)**
400 400
50 50
0.598 0.634
10%;** significant at 5%;'



significant at 1%

< < <









1- 1- Co V Co r Co

~ 0~0c~
0 ~


0 H..
^ ^




n o


. m
0 0

price at which a unit will initially install a scrubber is PA = $1, 592.61, which is much

higher than the assumed allowance price (PA = .-1-.00). As the size of a contract coal

increases from 50'- to 7".' of coal use, c increases from c = $1, 210.61 to c = $1, 349.82,

respectively. With a coal contract of 50' a unit still prefers to use low sulfur coal if it

does not install a scrubber because (PAS e) = $382.00 > MCA'8 = $244.26. However,

as the coal contract increases in size to 7,' a unit now prefers to use high sulfur coal if

it does not install a scrubber because (Ps e) = $242.79 < MCA' = $244.26. Also, a

unit will now install a scrubber because (PAS e < PA), which may have a large impact on

compliance costs.

In the third case, assume that a unit, both with and without a high sulfur coal

contract, prefers to use high sulfur coal both with and without a scrubber because

(PA < MCA"') and (PF < P < MCA'"). Given that a unit that has no high sulfur

contract coal (Ch = 0), a unit is indifferent to installing a scrubber at allowance price PAS:

P + PS MAX sMAX MAX + ps sMAX AX + P AX(2-76)
PAz i + FihAih A' i ihih (2

(2-76) can be rearranged to find an expression for Pi, where the only change in costs is

the change in a unit's net allowance position:

SP, P1A(AMAX ASMAX) (2-77)

A high sulfur coal contract will change the indifference allowance price by some value

c because it alters the costs from net allowance purchases and fuel costs:

P, + (P )nSMAX + P;sMA + P (PS )A" + Ps~M AX + Pit (2 78)

(2-78) can be rearranged to find an expression for Pi,:

SPi = (P e) (A MAX) (2-79)

If a unit that has no high sulfur contract coal (C~ = 0), a unit uses the maximum

amount of low sulfur coal if a scrubber is not installed, but uses the maximum amount of

high sulfur coal if a scrubber is installed. The unit is indifferent to installing a scrubber at

allowance price Pj:

z + PSMAX I ps sMAX p PAMIN psCMAX (2 68)
Piz+ Al i + Pis 7 h ACi + Vi [1-6)

(2-68) shows that the costs with and without a scrubber are equal for a given Pf, and can

be rearranged to find an expression for Pi:

SF, PS(AMIN- ASMAX) ps s, MAX ps sMAX (269)
A- ihih il'il

Since a unit uses low sulfur coal without a scrubber, a high sulfur coal contract will

change the coal use by a unit without a scrubber and change the indifference allowance

price by some value e:

PLi + (PS c)MAX + P ICDS AX + S (P i e i + MPAX Pich (2-70)
A 7,h (PA ihvh +lC1ih

(2-70) can be rearranged to find an expression for Pi,:

SPi = (PS )(Ail A AS MAX) (2-71)

The two expressions for Pi, in (2-69) and (2-71) can be set equal to solve for the

value of e:

Sih i (2 72)

By filling in for allowances and coal use, it is possible to determine the sign of e.

AMIN (D i(f)(m)- A iA ;'" (Di c A
n A/e;l Ah + Cc Sih)Tf -
Hisi Hsh ih

MAXih Di )() )- i MAX) (D ih- c H) ( )(- r_) A'
Hih ih

Table 3-1. Example: Contract Coal Distribution
Plant-Level Constraint Total mmBtu
High Sulfur Contract 300,000
Low Sulfur Contract 300,000
Demand 600,000

Unit Demand
1 200,000
2 200,000
3 200,000

Coal Dist'n by Unit
Unit 1
Unit 2
Unit 3

High Sulfur Coal

Low Sulfur Coal

Table 3-2. Sulfur Conversion by Fuel Type

Fuel Oil #2
Fuel Oil #6
Natural Gas

Emissions Conversion Factor
(38 Sulfur Content)/(mmBtu per ton)
(35 Sulfur Content)/(mmBtu per ton)
(39 Sulfur Content)/(mmBtu per ton)
(30 Sulfur Content)/(mmBtu per ton)
(144 Sulfur Content)/(mmBtu per 1,000 bbl.)
(162 Sulfur Content)/(mmBtu per 1,000 bbl.)
(0.60 lbs. S02/mmCF)/(mmBtu per mmCF)

lbs. S02/mmBtu
bs. S02/mmBtu
bs. S02/mmBtu
bs. S02/mmBtu
l Ibs. SO2/mmBtu
l Ibs. SO2/mmBtu
l Ibs. S02/mmBtu


Complementing federal policies such as the production tax credit, state governments

have taken actions to increase renewable energy capacity and generation, with 41 of

the 50 states enacting policies to encourage the use of renewable energy in their state.

Individual state policies show a great deal of variance. The objective of this paper is to

determine which state policies have led to increased deployment of .-.-regate non-hydro

renewable energy capacity into a state's electric power industry.3 The literature on state

renewable energy policies consists mainly of case studies on policy effectiveness. Only one

previous paper uses econometric methods to estimate the effects of various state policies

on renewable capacity. Menz and Vachon (2006) measure the impacts on wind capacity

in 39 states for 1998-2002. In contrast, my paper uses panel data from all 50 states for

1996-2003 to estimate the effects on total nonhydro renewable capacity deployment, not

just wind power capacity deployment. It estimates the effects of additional policies, and

also controls for differences in the market and political environments.

Three distinctly different types of policies are found to be effective at expanding

non-hydro renewable capacity deployment: a command-and-control policy known as a

Renewables Portfolio Standard (RPS), a tax-and-subsidy scheme facilitated through a

Public Benefits Fund (PBF) or Clean Energy Fund (CEF), and a market-based policy

where consumers can express their preferences to buy power from renewable resources at a

premium price.

The command-and-control policy targets the utility by mandating a specified level

of capacity that must come from renewable energy, and is generally referred to as a

Renewables Portfolio Standard. The tax-and-subsidy scheme collects an additional charge

per unit of electricity consumed from all customers in a state and places the proceeds into

this Public Benefits Fund or Clean Energy Fund. Monies from the PBF/CEF are used to

3 The electric power industry accounted for ,i 1', of renewable energy production in

The two expressions for PiF in (2-91) and (2-93) can be set equal to solve for the

value of e:

A i ~ t +A i ) ih 1 (2-94)

By filling in for allowances and coal use, it is possible to determine the sign of e.

AfAX ( (Hs)(mn) -A AA D sH S+CS" S) n- An

ASMAX ( (Sh)()(1- -A MAX D( c iHhAS+ A A S,rh)(m)(-ri) A7
ih Iihh
sMAX Di CzH, sMAX D, i C
^l -s sMAX
Hiil HI Hih
By filling into (2-94), the expression for e can be simplified to parameters for coal price,

sulfur content, heat content, and scrubber capture rate:

MA4 Aih MAX) H+ ri( -4 _H Sih Hc Jil c )

Di[mPf'( Si Si) (Pi f P ) (2-95)
Hfi Hi8 Hi8 HtJ
An interpretable form is derived by multiplying through by (AjIN AMAX) and dividing

through by mQ(, il ) and combining like terms:
Hih i
i ih i ih
Ais S, S, c (2 96)
e("1 MAX)i i + (DO- Hi C) PA h H (2-96)c
(Hh i H h 7i

( sa sh
( e '- 1AfMA) '- P r i + (D H }ICI)(PA MCA"'8) (2-97)
ih HT
From our initial assumption that a unit uses all high sulfur spot market coal if it does

not install a scrubber if it does not have a high sulfur coal contract, then PAS < MCA"'8.

The total amount of heat content from the contract coal (HicCic) is weakly less than the

total heat content needed to meet demand (Di). So the second term is weakly negative.
Meanwhile, the first term is negative because which means
Meanwhile, the first term is negative because (- < ) and ($ > ), which means
ff"l Tih T' h Tl' 1

of 8.'1.33-4'1.38 during 1996, total industry costs of $8.98 billion, and total industry

compliance costs of $1.30 billion relative to Simulation 1. If the contract constraints

are excluded from the model and the actual costs are compared to Simulation 3 and

Simulation 4, the excess compliance costs are $757 million and $1.01 billion, respectively.

However, including the contract constraint into the model results in excess compliance

costs of $226 million and :.8 million. Contract constraints explain -'. ;1 million and ,

million of excess compliance costs, respectively.

Third, the "true" compliance costs will differ from these estimates because the

appropriate baseline was not used. As shown analytically in Section 4, comparing

Simulation 5 and Simulation 6 to Simulation 1 is not the most appropriate measure

of compliance costs. The contract constraints should be included in both the baseline

simulation and the policy-restricted simulation. Simulation 2 runs the same model as

in Simulation 1 except that it includes contract constraints and results in total industry

costs of $8.27 billion. The difference between total industry costs in Simulation 1 and

Simulation 2 are the additional costs due to contract constraints, which are $582 million.

These additional costs would have resulted with or without the SO2 Trading Program

and should not be considered compliance costs. This is a key result because these costs

were labeled compliance costs by previous studies even though these costs are a result of

generating units locking in prices to protect from the uncertainty of higher coal prices in

the future.

By comparing results in Simulation 5 to Simulation 2, the minimum compliance costs

considering contracts and given the scrubber choice are found to be $490.1 million. "True"

minimum compliance costs are much lower ("1 10 million lower) once this additional

constraint is included in the model. In Simulation 6, the "true" minimum compliance costs

are 7- ;:..8 million, or $581.6 million less than if contracts are excluded from the model.

These are similar to the least-cost results found by Sotkiewicz (2003) at $340- 2-'7 million,

Sotkiewicz and Holt (2005) at $423-$553 million, and Carlson et al. (2000) at $571 million.

spot market coal if it does install a scrubber because (MCA''8 < PA < MCA'8) and

(MCAf'" < PA).

If a unit has no low sulfur contract coal (C^i = 0), a unit uses the maximum amount

of high sulfur coal with a scrubber and the maximum amount of low sulfur coal without a

scrubber, and is indifferent to installing a scrubber at allowance price PS:

Piz + PAsi + 7ihih

Ai + ilil


(2-82) can be rearranged to find an expression for Pi,:

z = p M IN ASMAX) P h sMAX C ssMAX
P?, iz TAs (Ai A 7,h + ~, Ii i C IS,


A low sulfur coal contract will change a unit's coal use and allowance purchases, which

changes the indifference allowance price by some value e:

P + (P ) SMAX pi s ^s,MAX c _c /S i s i s ,MAX + cC (284)
(2z (P ih ih +rearra d to find an exprsion + (2-4)

(2-84) can be rearranged to find an expression for Pi,:

ASMAX) + P^sls,MAX _s is,MAX
Ai ) + 7, ~ ,h C1


The two expressions for Pi, in (2-83) and (2-85) can be set equal to solve for the value of



Filling in for allowances and coal use, it is possible to determine the sign of e.

AIN (D ( A D Cc," H ) (S)m) A
nil= j( l

AiDMAX (Di H i ,)(S)+sC s (m)(1- r)- A

A X ( T ) (Sh) (T,) (t- ri)- A

=> C

= > z = (PA 0) (i" '

The "true" compliance costs to the industry appear to be lower relative to the

previous compliance cost estimates. Actual compliance decisions resulted in total

industry compliance costs that were higher than the least cost choices given the contract

constraints ($226 million compared to Simulation 5 and :.S million compared to

Simulation 6).

3.7 Plant Level Decision-Making Process

3.7.1 Introduction

The generating unit model in C'! plter 2 does not account for the fact that two

or more generating units are often owned and operated by one firm at the same plant.

A plant's decision-making process may not minimize costs for each generating unit

because a plant's concern is based on the combined costs of all generating units under

its operational control. This model derives the plant level problem, which allows us to

analyze differences in high versus low sulfur coal, spot versus contract coal, allowance

excess demand, scrubber installation, and the positioning of contract coal use based on the

different generating unit characteristics at a particular plant.

A plant level decision-making model is more realistic than a generating unit level

model to determine compliance decisions because coal deliveries are made at the plant

level where there are often multiple generating units. All generating units at that plant

facing the same coal use options, including sulfur contents, heat contents, and delivered

prices for both spot market and contract coal. The contract constraints become more

complex in this model where the sum of contract coal use for all generating units at a

plant must cover the contract requirement, 7 1 Cf > Cf Vi E {1, 2, ..., }. A plant

with multiple units has greater degrees of freedom in it's choices as to what fuel types to

purchase, in what quantities, and at which unit to burn the fuel based on their emissions,

demand, and coal contract constraints. The improved freedom in choice variables should

lead to lower compliance costs.

compliance costs versus a unit's total costs, which is something to keep in mind for the

remainder of the paper.

2.6.7 Generating Unit's Net Allowance Position: Excess Demand Correspon-

Assume there are no contract constraints (C, = 0, C, = 0). A generating unit's net

allowance position, or excess demand, is the difference between a unit's initial allowance

allocation and the unit's actual allowance use as governed by (3-2). From (3-2) and (3-3),

the minimum and maximum excess demand for allowances can be formally derived.

If PA < MCA"'8 a unit will use the maximum amount of high sulfur spot market coal

which can be derived from (3-3):
CMAX Di (2-34)
Hh (2 34)

The use of all high sulfur spot market coal leads to the maximum emissions level:

EMAX (1 Zii(m)(S)( ) (2-35)
H ih

Inserting C,,MAX in for h in (3-2) gives an expression for the maximum allowance excess

demand, which is the difference between the maximum emissions level (E MAX) and the

initial allowance allocation (A:):

AMAX MAX Ae (1 z Lr)(m)(St)) A' (2-36)

If a unit's initial allocation cannot cover its maximum possible emissions, then it will have

a positive net allowance position and be a net buyer of allowances.

If PA > MCA"', a unit will use the maximum amount of low sulfur spot market coal,

which can be derived from (3-3):

CSMAX (2-37)
The use of all low sulfur spot market coal leads to the minimum emissions level:

E" = (1 zrj)(T)(S)( ) (2-38)

control for both the state's purchase requirement and the state's market size (PCT STATE


A Required Green Power Option requires utilities to offer customers the option to

purchase renewable power at a premium. There are two versions of how these options

are implemented. The most common type gives consumers the option to make voluntary

contributions, called voluntary renewable energy tariffs in return for the guarantee that

some of the consumer's electricity consumption is produced from renewable sources.

Consumers purchase electricity at the market price and then p i,- a premium for blocks of

green electricity, usually about $2 per 100 kWh. The second type allows the producers to

charge consumers a higher rate per kilowatt-hour, but only to cover the additional costs

for electricity from renewable sources. Both the premium block rate and premium per

kilowatt-hour rate must be approved by the state's Public Utilities Commission (PUC).

Required Green Power Options elicit customer preferences and a crude measure of

willingness to p iv for renewable energy by allowing consumers to voluntarily p i,- higher

prices for the knowledge that they are supporting renewable-based electricity. The creation

of this niche market for renewable energy generation should have a positive impact on

renewable capacity. The variable REQ GREEN POWER OPT is a dummy variable, which

is interacted with GEN in the model to measure the effect of the policy based on the

state's market size (REQ GREEN POWER OPT*GEN).

1.5 Statistical Specifications and Empirical Analysis

Ordinary Least Square regressions with state fixed-effects and robust standard errors

are used in this paper to estimate total non-hydro renewable capacity. Robust standard

errors are used to account for heteroskedasticity, which was found to exist in the model

by using a Breusch-Pagan/Cook-Wesiberg Heteroskedasticity Test.31 Table 3 reports

31 The result was a Chi-Sq 448 and P(*)>Chi-Sq 0.0000, so there is a significant
difference in the variance of the dependent variable, which creates heteroskedasticity.

(ii) When a unit faces P < PA < MCA7'8, a unit prefers to install a scrubber and use
all high sulfur coal. From Proposition 2(ii), given (, < ), a high sulfur coal
contract decreases the maximum emissions level, which will weakly decrease a unit's
allowance excess demand.

(iii) If c > 0, when a unit faces (PAS ) < PA < PA, a high sulfur coal contract decreases
a unit's indifference allowance price of installing a scrubber below the allowance
price, which leads to a unit installing a scrubber where it initially would not and
decreases a unit's emissions and a unit's excess demand. From Proposition 2(ii),
Given the scrubber choice and ( < ), a high sulfur coal contract will weakly
decrease a unit's emissions and excess demand. The combined net effect is weakly
negative and weakly decreases excess demand.

(iv) From Proposition 2(i), when a unit faces MCA"'8 < PA < (PA e), a unit does
not install a scrubber and prefers to use low sulfur coal. A high sulfur coal contract
increases a unit's minimum emissions level and excess demand.

(v) From Proposition 2(ii), when a unit faces 0 < PA < MCA'S a unit does not install
a scrubber and prefers to use high sulfur coal. Given (>h < h ), a high sulfur coal
Hih i- Hih
contract decreases a unit's maximum emissions level and excess demand.E

Proposition 5(b) is shown graphically in Figure 2.8(ii).

There are conditions under which some of these allowance price ranges do not exist.

For example, PA' < MCA'"8 in the third case described above. So there is no price

range (MCA'", PAS) in Figure 2.8. However, Propositions 5(a) and 5(b) still hold for the

price ranges that do exist. In Case 2, c is large enough to shift the allowance price from

PA > MCA"'8 to (PA c) < MCA"'j and causes the visual representation of the excess

demand correspondence to shift from Figure 2.8 to 2.8. Impact of a binding low sulfur coal contract

Proposition 5 can be proven using the same approach that was used to determine the

sign of c with a high sulfur coal contract is used to show that c is alv--, less than or equal

to zero, and may increase the allowance price at which a unit is indifferent to installing a

scrubber. Once again there will be three case under consideration.

In the first case, assume that both with and without a low sulfur coal contract, a unit

prefers to use low sulfur spot market coal if it does not install a scrubber and high sulfur

may lead to only a decrease in the allowance price. However, taking into account the

scrubber choice causes the sign of the impact on the allowance price to become ambiguous.

When you consider the scrubber choice in the decision-making process, it is uncertain

how a coal contract will affect the allowance market because the contract may increase

or decrease excess demand depending on the allowance price ranges derived in ('! Ilpter

2. Although it is certain that if the coal contract binds, then there will be a shift to a

sub-optimal excess demand for three different allowance price ranges. High sulfur coal contract binds

Each of the three price ranges derived in C'!i lpter 2 must be discussed to understand

how a high sulfur coal contract will effect the allowance market, both in terms of the

allowance market supply, allowance market demand, and the allowance price. Assume

ih ih

If a generating unit has a high sulfur coal contract, the unit's excess demand increases

for two allowance price ranges: (PA > MCA7'8) and (P' e > PA > MCA"'").

The increase in excess demand will be from AfMIN to AI when (PA > MCA"'j).

The increase in excess demand will be from AIN to AIN or AIN to AAX when

(PX e > PA > MCA,'"), depending on the relationship between (PS c) and MCA"'.

It will be the latter in the special case where the generating unit ah--,v- prefers to use high

sulfur coal.

When the market allowance price is in these two ranges that result in an increase in

a generating unit's excess demand, there will be either an increase in the market demand

for allowances, or both an increase in market demand and a decrease in market supply for

allowances. In both situations, the allowance price will be driven higher.

The third price range that has a shift in the excess demand is (Pf e, PA). A

generating unit decreases its excess demand from AMIN to ASMAX, which is a result of

a unit installing a scrubber for a price range for which it initially would not install a


4ML ;4-fMAX N A4 s a

Figure 2-17. Impact of a High Sulfur Coal Contract: MCA'" < PS


AMCA7's -




A -

Figure 2-18. Impact of a High Sulfur Coal Contract: MCA > P

A (i

AI '- MCf -

ACA'- ,, MCA4s



Figure 2-19. Impact of a Low Sulfur Coal Contract: MCAi'" > PA

_ _,dA-x SMSX A- 4 uMMy

Petersik (2004) provides a non-econometric analysis of the effectiveness of different

types of Renewables Portfolio Standards as of 2003 for the United States Energy

Information Association (EIA). He finds that only Renewables Portfolio Standards

that mandate a certain level of capacity (number of megawatts) have had any significant

impact on renewable capacity deployment. Policies with renewable generation or sales

requirements as well as voluntary policy programs were found to have no significant effect.

C'!, i, et al. (2007) compares the results from 28 policy impact projections for state

or utilit,---1. ,l Renewables Portfolio Standards and finds that (1) the impact on electricity

prices is minimal, (2) wind power is expected to be the primary renewable used to

meet policy requirements, and (3) the benefit-cost estimates rely heavily on uncertain

assumptions, such as renewable technology costs, natural gas prices, and possible carbon

emissions policy in the future.

Bolinger et al. (2001) describe in detail 14 different state Clean Energy Funds,

enumerating the regulatory background, funding approaches, the current status of the

fund, and the resulting impacts on renewable energy. Programs that fund utility-scale

projects are found to be the most effective at increasing renewable capacity deployment.4

Bolinger et al. (2004, 2006) summarize the same 14 Clean Energy Funds. They find that

due to d-.1 i- and cancelled projects actual capacity often is much lower than initially

obligated capacity.

Wiser and Olson (2004) examine participation in 66 utility green power programs.

They find local green power programs have residential participation rates ranging from

0.02'. to 6.45'. and averaging 1.t:''. However, this study does not look at any state-level

Required Green Power Options that require all utilities in a state to offer consumers

the option to purchase renewable energy. The paper focuses on participation rates of

4 Funding is usually based on actual production, but it is paid in a lump sum once the
capacity has been constructed.

cap-and-trade approach. Sotkiewicz and Holt (2005) find that due to PUC regulation, not

only is there a greater number of scrubbers actually installed at the beginning of Phase I

relative to the least cost solution (18 scrubbers), but only nine of those actually installed

are at units that install scrubbers under the least cost solution. An increase in the number

and inefficient location of scrubber installations increases the total costs of compliance

because installing a scrubber is the most expensive compliance option under Phase I.

In the initial years of Phase I, many firms were not active participants in the

allowance market, choosing to switch fuels and bank allowances or shift allowances

between only their own units (Hart (1998); Ellerman et al. (1998)). The firms that did

participate mainly traded allowances within the same utility company. Bohi and Burtraw

(1997) find that intra-utility trading accounts for two-thirds of the allowance transactions

while the remaining one-third were inter-utility trades. Since most trades were made

between units owned by the same company, trading between two generating units at

the same plant would be a common occurrence. Many studies -ii-.-. -I. 1 state public

utility regulations and other state laws as a reason for the inefficiencies resulting from this

self-sufficient behavior (Bohi (1994); Bohi and Burtraw (1997); Swift (2001)).

Arimura (2002) uses econometric approaches to study the impact of PUC regulation

on compliance choices, and finds that utilities that face PUC regulation are more likely to

switch fuels instead of purchasing allowances for compliance.

Winebrake et al. (1995) estimated the cost inefficiencies from state government

restrictions on a utility's allowance trading, and estimates the total cost estimates for the

first ten years of Title IV (1995-2005). A command-and-control approach was estimated to

result in compliance costs of $4.19 billion greater than in the unrestricted permit trading

system (-. n_- billion, or an average of A 2- million/year) and an estimated allowance price

of $143/ton.

Winebrake et al. (1995) simulates the additional costs from restrictions on between-state

trading that were under consideration by both New York and Wisconsin. Both states were

abatement costs through discrete technology choices and introducing uncertainty into

the Stavins' model. Montero finds that in an allowance trading system with transaction

costs, non-continuous marginal abatement costs resulting from discrete technology choices

can cause the initial allocation of allowances to matter for efficiency, even with constant

marginal costs of abatement and certainty.

The importance of the allowance allocation in these previous studies relies on the

existence of transaction costs. However, this model has assumed no allowance transaction

costs, which allows it to show that even with coal contract constraints the allowance

allocation distribution will not impact a generating unit's compliance choices or the total

industry compliance costs.1

Consider the first-order conditions for both high sulfur and low sulfur coal, which are

independent of a generating unit's allowance allocation. The type of coal a unit will use

will not depend on Ae.

Pih + AiI(l ziri)(m)(Si%) Ai2H > 0 (3-34)

Pi4 + Ai(li zjrj)(m)(Sjf) Ai2H^I > 0 (3-35)

The choice between switching fuels or purchasing allowances is based solely on a

generating unit's relative marginal cost of purchasing the next allowance compared to the

effective marginal cost of abating the next unit of emissions. As can be seen, Ae has no

impact on a generating unit's compliance decisions. A generating unit will use allowances

1 It is assumed that generating units are unable to break their contracts. Generating
units could realistically break their contracts at a very high price and create the additional
freedom in its compliance choices. In such a case, contract constraints would result in very
high transaction costs for trading those additional allowances.

Full Text








Ithankmysupervisorycommitteechair,LawrenceKenny,alongwithmyothercommitteemembers(SanfordBerg,JonathanHamilton,andJaneLuzar)fortheircommentsandadviseregardingmydissertation.SpecialthanksgoestoPaulSotkiewiczforhisinterest,guidance,andsupportinmyresearch.Ialsothankmyparentsfortheirnever-endingsupport. 4


page ACKNOWLEDGMENTS ................................. 4 LISTOFTABLES ..................................... 9 LISTOFFIGURES .................................... 10 ABSTRACT ........................................ 12 CHAPTER 1EFFECTSOFSTATEGOVERNMENTPOLICIESONELECTRICITYCAPACITYFROMNON-HYDROPOWERRENEWABLESOURCES ..... 14 1.1Introduction ................................... 14 1.2LiteratureReview ................................ 16 1.3Model ...................................... 18 1.4VariablesandData ............................... 21 1.4.1EconomicandPoliticalVariables:Wit 21 1.4.2RegulatoryPolicyVariables:Rit 27 1.5StatisticalSpecicationsandEmpiricalAnalysis ............... 33 1.5.1EconomicandPoliticalVariables:Wit 34 1.5.2RegulatoryPolicyVariables:Rit 37 1.5.3StateFixed-EectsVariables:Sit 40 1.5.4YearVariables:Tit 40 1.6Conclusions ................................... 41 2EFFECTSOFCOALCONTRACTCONSTRAINTSONSO2TRADINGPROGRAMCOMPLIANCEDECISIONS ..................... 47 2.1Introduction ................................... 47 2.2PolicyBackground ............................... 48 2.2.1TitleIVoftheCleanAirActAmendment .............. 48 ...................... 49 ...................... 50 2.2.2CleanAirInterstateRule ........................ 51 2.3LiteratureReview ................................ 52 2.3.1TitleIV:PhaseI ............................ 52 2.3.2Utility-LevelModelsofComplianceCosts ............... 55 2.3.3Long-TermCoalContracts ....................... 56 2.4InecienciesResultingfromCoalContractConstraints ........... 57 2.5ModelandParameters ............................. 60 2.6GeneratingUnitLevelDecision-MakingProcess ............... 62 2.6.1GeneratingUnit'sProblem ....................... 63 2.6.2First-OrderConditions ......................... 64 5


................ 65 ........................... 65 ...... 65 ....... 66 2.6.4CoalUseUnderaHighSulfurCoalContractConstraint ....... 66 2.6.5CoalUseunderaLowSulfurCoalContractConstraint ....... 68 2.6.6GeneratingUnit-LevelComplianceCosts ............... 70 2.6.7GeneratingUnit'sNetAllowancePosition:ExcessDemandCorrespondence ....................... 72>MCAs;si 74 .............................. 74

.. 136 3.4.3ComparativeStatics:EectofCoalContractsontheAllowanceMarketwithEndogenousScrubberChoice .............. 136 ............... 137 ................ 138 3.5ComplianceCosts ................................ 139 3.5.1ComplianceCostswithCoalContractsRelativetoComplianceCostsfromPreviousStudies .......................... 139 3.5.2TotalIndustryComplianceCost .................... 141 3.5.3ImpactofAllowanceAllocationonComplianceCostsGivenScrubbertheChoice ................................ 144 3.5.4ImpactofAllowanceAllocationwithEndogenousScrubberChoice 146 3.6SimulationModel ................................ 147 3.6.1Introduction ............................... 147 3.6.2Data ................................... 147 ........................... 147,actualemissions,anddemanddata ....... 150 ........... 151 3.6.3SimulationModelDesign ........................ 152 3.6.4SimulationResults ........................... 152 ..... 153 ............ 157 ........ 158 ..................... 160 ................ 160 3.7PlantLevelDecision-MakingProcess ..................... 161 3.7.1Introduction ............................... 161 3.7.2Plant-LevelProblem .......................... 163 3.7.3First-OrderConditions ......................... 164 3.7.4CharacterizingaUnit'sSpotMarketFuelChoices .......... 165 ............... 166 ............................ 167 ............................ 167 3.7.5ExcessDemandCorrespondence .................... 168 3.7.6CharacterizingaGeneratingUnit'sContractFuelChoices ...... 170\i" .................. 170\i" .................. 171 3.7.7\Non-Aected"GeneratingUnitsatan\Aected"Plant ...... 171 7


........................ 172\Non-aected"generatingunitsandhighsulfurcoalcontracts .................. 173 3.7.8ScrubberInstallationChoice ...................... 174 174 3.7.9APlant'sPreferredOrderofScrubberInstallation .......... 176 177 ................... 178 .... 179 ..... 181 3.7.10ScrubberInstallationExample:PlantwithTwoAectedGeneratingUnits ................................... 182 ................. 183 ................. 184 ................ 185 3.7.11ScrubberInstallationExample:PlantwithOneAectedandOneNon-AectedGeneratingUnits ..................... 186 3.7.12SummaryofPlantLevelResults .................... 187 3.8CONCLUSIONS ................................ 188 ACONTRACTIMPACTSONCOSTSANDSCRUBBERINSTALLATIONINDIFFERENCEPRICE .............................. 199 A.1ImpactsonTotalCostsandComplianceCostsfromaCoalContractConstraint ................................. 199 A.2DerivationofCost-MinimizingInputUsetoFindPSA 205 BMARKETEQUILIBRIUMANDSIMULATIONDESIGN ............ 206 B.1ConditionsforExistenceofanEquilibrium .................. 206 B.2TechnicalDetailsofSimulationModelDesign ................ 208 REFERENCES ....................................... 211 BIOGRAPHICALSKETCH ................................ 216 8


Table page 1-1DependentandControlVariables .......................... 43 1-2RegressionsResults .................................. 44 1-3PolicyVariables .................................... 45 1-4VariableEectsofSignicantVariables ....................... 46 2-1PhaseIComplianceCostEstimates ......................... 113 2-2HighSulfurCoalContract:Assumptions ...................... 114 2-3HighSulfurCoalContract:Results ......................... 114 2-4LowSulfurCoalContractExamples:Assumptions ................ 114 2-5LowSulfurCoalContractExamples:Results ................... 115 2-6ExampleEpsilonMagnitude:Case1 ........................ 116 2-7ExampleEpsilonMagnitude:Case2 ........................ 117 2-8ExampleEpsilonMagnitude ............................. 118 3-1Example:ContractCoalDistribution ........................ 190 3-2SulfurConversionbyFuelType ........................... 190 3-3SimulationResults .................................. 191 3-4ImpactofContractConstraintonScrubberChoice ................ 192 3-5SimulationswithEngineeringData ......................... 192 3-6ImpactsofaReductionintheAllowanceAllocationof10% ............ 192 3-7MathExample:TwoAectedUnits ......................... 192 9


Figure page 2-1TheSO2AllowancePrice .............................. 119 2-2ExcessDemandCorrespondenceandComplianceCostSavingsfromFuelSwitchingOverAllowancePurchasing ............................. 119 2-3HighSulfurContract:ShiftinMinimumExcessDemand ............. 120 2-4NoContract:ComplianceCosts ........................... 120 2-5HighSulfurContract:ComplianceandTotalCosts ................ 120 2-6HighSulfurContract:RelativeSavingsfromContractCoal ............ 121 2-7CostSavingsfromUsingAllowancesOverFuelSwitching ............. 121 2-8LowSulfurContract ................................. 121 2-9NoContract:ComplianceCosts ........................... 122 2-10LowSulfurCoalContract:MCAs;ci 122 2-11LowSulfurCoalContract:MCAs;ci 122 2-12ComplianceCosts:PA=MCAs;si 123 2-13HighSulfurCoalContract .............................. 123 2-14LowSulfurCoalContract .............................. 123 2-15ExcessDemandCorrespondence:MCAs;siPSA 125 2-19ImpactofaLowSulfurCoalContract:MCAs;si>PSA 125 2-20ImpactofalowsulfurCoalContract:MCANSi

........................ 195 3-7GivenScrubberChoice:ShiftfromHighSulfurContract ............. 196 3-8GivenScrubberChoice:ShiftfromLowSulfurContract ............. 196 3-9WithScrubberChoice:ShiftfromHighSulfurContract ............. 197 3-10WithScrubberChoice:ShiftfromLowSulfurContract .............. 198 B-1UpperSemi-ContinuousCorrespondence ...................... 210 B-2CorrespondencewithScrubberChoice ....................... 210 11


Environmentalissueshavebecomeakeypoliticalissueoverthepastfortyyearsandhasresultedintheenactmentofmanydierentenvironmentalpolicies.Thethreeessaysinthisdissertationaddtotheliteratureofrenewableenergypoliciesandsulfurdioxideemissionstrading. Therstessayascertainswhichstatepoliciesareacceleratingdeploymentofnon-hydropowerrenewableelectricitygenerationcapacityintoastateselectricpowerindustry.Aswouldbeexpected,policiesthatleadtosignicantincreasesinactualrenewablecapacityinthatstateeithersetaRenewablesPortfolioStandardwithacertainlevelofrequiredrenewablecapacityoruseCleanEnergyFundstodirectlyfundutility-scalerenewablecapacityconstruction.AsurprisingresultisthatRequiredGreenPowerOptions,apolicythatmerelyrequiresallutilitiesinastatetooertheoptionforconsumerstopurchaserenewableenergyatapremiumrate,hasasizableimpactonnon-hydrorenewablecapacityinthatstate. Thesecondessaystudiesthetheoreticalimpactsfuelcontractconstraintshaveonaelectricitygeneratingunit'scompliancecostsofmeetingtheemissionscompliancerestrictionssetbyPhaseIoftheTitleIVSO2EmissionsTradingProgram.Fuelcontractconstraintsrestrictautility'sdegreesoffreedomincoalpurchasingoptions,whichcanleadtotheuseofamoreexpensivecomplianceoptionandhighercompliancecosts. 12




1 14


Threedistinctlydierenttypesofpoliciesarefoundtobeeectiveatexpandingnon-hydrorenewablecapacitydeployment:acommand-and-controlpolicyknownasaRenewablesPortfolioStandard(RPS),atax-and-subsidyschemefacilitatedthroughaPublicBenetsFund(PBF)orCleanEnergyFund(CEF),andamarket-basedpolicywhereconsumerscanexpresstheirpreferencestobuypowerfromrenewableresourcesatapremiumprice. Thecommand-and-controlpolicytargetstheutilitybymandatingaspeciedlevelofcapacitythatmustcomefromrenewableenergy,andisgenerallyreferredtoasaRenewablesPortfolioStandard.Thetax-and-subsidyschemecollectsanadditionalchargeperunitofelectricityconsumedfromallcustomersinastateandplacestheproceedsintothisPublicBenetsFundorCleanEnergyFund.MoniesfromthePBF/CEFareusedto 15


TheeconometricresultssupportmanyoftheconclusionsfromvariouscasestudieswithrespecttoRenewablesPortfolioStandardandCleanEnergyFundpolicies.Moreover,theresultspresentedherealsoshow,unlikepreviousstudies,thatthepotentialforoeringconsumerstheoptiontopurchaserenewableelectricityatahigherpricethanconventionallyproducedelectricitycanincreaserenewablecapacityinastate. Wiseretal.(2004)consideredallRenewablesPortfolioStandardsandfoundthepitfallsinthecurrentpolicydesigns.Somekeyproblemsinpolicydesignsincludeinsucientdurationandstabilityoftargets,weakenforcement,andnarrowapplicabilityofthepolicy.Otherconditionsthatmayimpactapolicy'seectivenessarethepresenceoflong-termpowerpurchasersandpoliticalandregulatorystability. 16


Chenetal.(2007)comparestheresultsfrom28policyimpactprojectionsforstateorutility-levelRenewablesPortfolioStandardsandndsthat(1)theimpactonelectricitypricesisminimal,(2)windpowerisexpectedtobetheprimaryrenewableusedtomeetpolicyrequirements,and(3)thebenet-costestimatesrelyheavilyonuncertainassumptions,suchasrenewabletechnologycosts,naturalgasprices,andpossiblecarbonemissionspolicyinthefuture. Bolingeretal.(2001)describeindetail14dierentstateCleanEnergyFunds,enumeratingtheregulatorybackground,fundingapproaches,thecurrentstatusofthefund,andtheresultingimpactsonrenewableenergy.Programsthatfundutility-scaleprojectsarefoundtobethemosteectiveatincreasingrenewablecapacitydeployment. WiserandOlson(2004)examineparticipationin66utilitygreenpowerprograms.Theyndlocalgreenpowerprogramshaveresidentialparticipationratesrangingfrom0.02%to6.45%andaveraging1.39%.However,thisstudydoesnotlookatanystate-levelRequiredGreenPowerOptionsthatrequireallutilitiesinastatetooerconsumerstheoptiontopurchaserenewableenergy.Thepaperfocusesonparticipationratesof 17


Birdetal.(2005)summarizefederalrenewableenergypolicies,generalmarketfactors,andstate-specicfactors,suchasstatepolicies,thataredrivingthedeploymentofwindpower.Thekeymarketfactorsarethevolatilityinnaturalgaspricesduringtheearly2000sandtheloweredwindenergygenerationcostsduetolargerwindturbines,whichhavecombinedtomakewindpowermorecompetitivewithnaturalgas-redgeneration. Onlyonepaperhasattemptedtoeconometricallyestimatetheeectsofstaterenewableenergypolicyonrenewablecapacity.MenzandVachon(2006)useordinaryleastsquarestoestimatestatepolicyeectsonwindpowercapacityandgenerationwithapaneldatasetfor39statesfor1998-2002whilecontrollingforwindpoweravailability,retailchoice,andpolicydummyvariablesforPublicBenetsFund,RenewablesPortfolioStandard,RequiredGreenPowerOption,andfuelmixdisclosure. 18


Thedependentvariableisthetotalnon-hydropowerrenewablenameplatecapacityintheelectricpowerindustry(Cit),whichincludesallnameplatecapacityofutilities,independentpowerproducers(IPPs),andindustrialorcommercialcombinedheatandpowerproducersthatusesolar,wind,geothermal,orbiomassasanenergysource. 19


Theeectsofstaterenewableenergypoliciesarebestestimatedusingtotalstatenon-hydrorenewablecapacityasthedependentvariablebecauseseveralpoliciesmandateorfundaspecicamountofrenewablecapacity.Policiesthatdonotsetspecicrenewablecapacityrequirementscanbemeasuredincapacitytermsbycontrollingforeachstate'smarketsize,whichwillbediscussedinmoredetailinSection4. Alargeamountofrenewablecapacitycreatedbefore1996originatedfromthePublicUtilitiesRegulatoryPolicyAct(PURPA),afederalpolicypassedin1978requiringutilitiestopurchaseelectricityfromQualifyingFacilities(QFs),whichareIPPsthatmeetspecicrequirementsandincluderenewable-basedfacilities.Foravarietyofreasons,theeectsofPURPAvariedfromstatetostate.Statedummyvariables(Si)measuretheseeectsandotherunchangingstatefactors,suchasrenewableresourceavailability. 7 20


1.4.1EconomicandPoliticalVariables:Wit 1-1 summarizesthedataforthedependentvariable(RENEWABLECAPACITY)andthecontrolvariables. withoutanyconsiderationofcostsorenvironmentalimpacts.AnycapacityshutdownduetoPURPAcontractexpirationafter1996willdecreasethepositiveeectsofanyenactedpolicy.ThereisalsothepossibilityofastatechangingitsinterpretationandenforcementofPURPAafter1996,whichwouldnotbecapturedinthemodel.9 21


Thefollowingthreevariablesareincludedinthemodeltocontrolformarketstructure.Twoofthesevariablesarehydropowercapacity(PCTHYDROPOWER)andnuclearpowercapacity(PCTNUCLEAR)asapercentageoftotalcapacityexcludingnon-hydrorenewables.Hydropowershouldleadtolessnon-hydrorenewablecapacitybecausehydropowerhaslowmarginalproductioncosts,andthecapacitytypicallywasconstructedmanyyearsago.Withlowermarginalcostsandsunkcapitalcostsassociatedwithhydropower,hydropowerwillbetherstrenewableenergytobeimplementedbecauseitismoreeconomicallycompetitivethanmostnon-hydropowerrenewablesavailabletotheelectricpowerindustry.Consumerand/orpolicydrivendemandforrenewable-basedelectricitymaynotdierentiatebetweenhydropowerandotherrenewablesources,whichallowshydropowertobeasubstituteofnon-hydrorenewables. Similartohydropower,nuclearpowerhaslowmarginalcostsofproducingbaseloadelectricity,hassunkcapitalcosts,andhasnoemissions.Ifnon-hydrorenewablecapacityisdeployedbasedoneconomicfactors,givensimilaremissionsproles,greaternuclearorhydropowercapacityshoulddecreasetheamountofnon-hydrorenewablecapacity. Analternativepossibilityisthatregulatorsinstateswithlargeamountsofnuclearpowerencouragepowerproducerstouseotherresourcetypestomeetnewdemand.Renewableenergymaybeusedbyutilitiestoalleviatepressurefromenvironmentalistsovernuclearpower,thusleadingtogreaterdeploymentofrenewableenergycapacityinstateswithlargeamountsofnuclearcapacity.ThesignofPCTNUCLEARwilldependonwhichofthesetwofactorshasthelargereectonpowerproducers. 22




16 24




Florida,Hawaii,Louisiana,andTexasusethebyproductofsugarproductionfromsugarcaneasabiomassfuel.Forexample,inHawaiisugarcaneisoneoftheprimarysourcesofbiomass.DuetomarketconditionsmostofthesugarcanefarmsinHawaiiwereshutdownoverthe1990s,removingthefuelsourceformuchofthebiomasscapacityinthestate.Changesinsugarcaneproductionarelikelytohaveanimpactontheamountofbiomasscapacityinastate.Thechangeintotaltonsofsugarcaneproductionfrom1996levels(SUGARCANEPRODCHANGE)isincludedinthemodeltocontrolforitsimpactonrenewablecapacity.SUGARCANEPRODCHANGEistheonlycontrolvariablenotinteractedwithgeneration. Apoliticalvariableisincludedtomeasurechangesinrenewableenergypreferencesinastate.TheLeagueofConservationVoters(LCV)ratingisusedtodetermineifpolicypreferencesforenvironmentalprotectionincreaserenewableenergycapacityindependentfromitspolicyeects.TheLeagueofConservationVoters(LCV)annuallypublishestheNationalEnvironmentalScorecard,whichratesallcongressionalvotesonconservationalissuesbyeachrepresentative. Anaverageofallthevotesbyastate'srepresentativesistakentogettheaverageHouseofRepresentativesscore(LCVSCORE).ThescoresfromtheHouseofRepresentativesareusedinsteadoftheSenatebecauserepresentativeshaveashorterterminoce 26


AhighLCVratingforastateindicatesthatthestate'sconstituentsareenvironmentallyfriendlyandaremorelikelytodemandelectricityfromrenewableenergy,allotherthingsbeingequal.ConsumersorenvironmentalgroupsinstateswithhigherLCVratingsmaybemorelikelytopressureutilitiestousegreateramountsofrenewableenergynomatterwhich,ifany,policieshavebeenenactedbythestate. PoliciesmaybeendogenoustohigherLCVratingsbecausestateswithcongresspersonswhovoteforfederalpro-environmentalpoliciesmaybemorelikelytoenactstatepro-environmentalpolicies.ThepolicyendogeneityissueisnotaddressedinthebodyofthispaperbecauseLCVSCOREisnotastrongenoughpredictorofstatepoliciestobeasatisfactoryinstrument.NotealsothatremovingLCVSCOREfromtheregressiondoesnotchangetheotherresults. 27


RegulatorypoliciesdescribedbelowincludeaRenewablesPortfolioStandardwithaCapacityRequirement,RenewablesPortfolioStandardwithaGeneration/SalesRequirement,CleanEnergyFund,NetMetering,InterconnectionStandards,StateGovernmentGreenPowerPurchasing,andRequiredGreenPowerOptions.Table 3-2 summarizesthedataforthepolicyvariables. TherstpolicythatwillbediscussedisaRenewablesPortfolioStandard,whichspeciesanamountofastate'selectricityproduction,sales,orcapacitythatmustberenewable-based.RenewablesPortfolioStandardscanbedierentiatedintothreemainstructuralforms,policiesthatset(1)mandatoryrenewablegenerationorsaleslevels,(2)voluntaryrenewablegenerationorsalesgoals,and(3)mandatoryrenewableenergycapacityrequirements. ThersttypeofRenewablesPortfolioStandardsetsapercentageoftotalgenerationorsalesforeachpowerproducer/retailerthatmustoriginatefromrenewablesources,usuallyincreasingeveryyearoreveryfewyears.Forexample,Arizona'stieredrenewablelevelsthathavetobemetbeganat0.2%in2001andincreasedby0.2%eachyear, 28


ThedierencesbetweenRenewablesPortfolioStandardscanbeaccountedforinthemodelbytwovariables:avariablethatmeasuresthesizeoftherenewablegenerationorsalesrequirement(RPS:SALESREQ)andavariablethatmeasuresthesizeofthecapacityrequirement(RPS:CAPREQ). 29


Althoughthispolicydoesnotdirectlyrequiretheconstructionofrenewablecapacity,anincreaseintherequiredamountofrenewablegenerationmayleadtoaneedformorerenewablecapacity.Ifcurrentlevelsofrenewablecapacitycannotmeetafuturegeneration/salesrequirement,additionalcapacitywillneedtobeconstructed. CleanEnergyFundsarepaidforthroughSystemBenetsCharges(SBCs),whichareadditionalchargespaidbyallconsumersontheirelectricityconsumption.SBCscanbe 30




Interconnectionstandards(INTERCONSTANDARDS)areasetofguidelinesusedtosafelyandeectivelyconnectindividualrenewablegeneratingunitstotheelectricutilitypowergrid.Somehavetechnicalrequirements,suchasgeneratortypeandsizelimits,mandatorysafetyandperformancestandards,andinsurancerequirementsthatmustbemetbeforeanetmeteringcustomercanconnecttotheutility'snetwork.Interconnectionstandardsmustbemetbyanycommercial,industrial,residential,orgovernmentcustomerthatdecidestoconnecttothegrid.Withoutthesestatepolicies,thenetmeteringconnectionscouldcausemajorproblemsforthegrid,powerproducers,andotherpurchasers.Interconnectionstandardsincreasethecostsofhookinguptothegridfornetmeteringandmayosetsomeofthenegativeeectfromnetmetering.INTERCONSTANDARDSisalsointeractedwithGENtocontrolformarketsize. 32


ARequiredGreenPowerOptionrequiresutilitiestooercustomerstheoptiontopurchaserenewablepoweratapremium.Therearetwoversionsofhowtheseoptionsareimplemented.Themostcommontypegivesconsumerstheoptiontomakevoluntarycontributions,calledvoluntaryrenewableenergytarisinreturnfortheguaranteethatsomeoftheconsumer'selectricityconsumptionisproducedfromrenewablesources.Consumerspurchaseelectricityatthemarketpriceandthenpayapremiumforblocksofgreenelectricity,usuallyabout$2per100kWh.Thesecondtypeallowstheproducerstochargeconsumersahigherrateperkilowatt-hour,butonlytocovertheadditionalcostsforelectricityfromrenewablesources.Boththepremiumblockrateandpremiumperkilowatt-hourratemustbeapprovedbythestate'sPublicUtilitiesCommission(PUC). RequiredGreenPowerOptionselicitcustomerpreferencesandacrudemeasureofwillingnesstopayforrenewableenergybyallowingconsumerstovoluntarilypayhigherpricesfortheknowledgethattheyaresupportingrenewable-basedelectricity.Thecreationofthisnichemarketforrenewableenergygenerationshouldhaveapositiveimpactonrenewablecapacity.ThevariableREQGREENPOWEROPTisadummyvariable,whichisinteractedwithGENinthemodeltomeasuretheeectofthepolicybasedonthestate'smarketsize(REQGREENPOWEROPT*GEN). 33


Thepercentageofothercapacitycomprisedofhydropowerinteractedwithgeneration(PCTHYDRO*GEN)isnotstatisticallysignicant.However,thecoecientforthepercentageofothercapacitycomprisedofnuclearpowerinteractedwithgeneration(PCTNUCLEAR*GEN)ispositiveandstatisticallysignicant.Aonestandarddeviation(12.46%)increaseinthepercentageofnon-renewablecapacitycomprisedofnuclearpowerleadstoanincreaseof2.09MWperterawatt-hourofgenerationinastate.Sothisonestandarddeviationchangeinastatewithamediangenerationlevel(51.15TWh)leadstoanincreaseof107MW.Itispossiblethatutilitieswithmorenuclearpoweraredeployingmorerenewablecapacitybecausetheutilitiesarefocusedondiversifyingitsgenerationmix,eithertodecreasetheutilities'useoffossilfuelsandloweremissionsortoalleviatepressurefromenvironmentalistswhoareupsetabouttheuseofnuclearpower. ThecoecientsontheaverageLCVscorefortheHouseofRepresentativesinteractedwithgeneration(LCVSCORE*GEN)arepositiveandsignicant.Aonestandarddeviationincrease(26.51points)inastate'sLCVscoreleadstoanincreaseof0.663MWperterawatt-hourofgeneration.Aonestandarddeviationincreaseinastatewithmedian 34


Asexpected,renewableenergycostinteractedwithgeneration(RENEWCOST*GEN)hasanegativeandstatisticallysignicantcoecient.Aone-centperkWhdecreaseinrenewableenergycostleadstoanincreaseof0.712MWperterawatthourofgeneration.Inastatewithmediangeneration,aonecentdecreaseinRENEWCOSTleadstoanincreaseof36MW.RENEWCOSTdecreasedby1.79centsfrom1996to2003,whichimpliesanincreaseof65MWforastatewithmediangeneration.Thiseectdoesnotjustincludethetechnologicalchangesinrenewableenergy.AsmentionedinSection4,allfederalproductionincentivesareincludedinthecostsofproductionforeachrenewablesource,capturingthefederalpolicychangesaswellasthetechnologicaladvances.Theyearvariablecoecients,whichexplainthesameimpactsasRENEWCOST,areexplainedindetailinSection5.4. Thecoecientonaverageborderstateretailpriceinteractedwithgeneration(BORDERPRICE*GEN)isnegativeandismarginallystatisticallysignicantonlyinSpecication3ofTable3.Higherelectricitypricesdonotappeartoresultinmorerenewableenergycapacityconstruction.Infact,thenegativecoecientimpliesthatanincreaseofonecentinthepriceofelectricityleadstoasmalldecreaseinrenewablecapacityby13MWinastatewithmediangeneration.Aonestandarddeviation(2.07cents)increaseinpriceleadstoadecreaseofonly27MW.Itispossiblethatconsumersinastatewithhighelectricitypriceshavelessofanappetiteforfurtherincreasesinpricesthroughmoreexpensiverenewablegeneration. 33 35


ToalleviateanyconcernsaboutFUELCOST,anadditionalspecicationisestimatedreplacingFUELCOSTwithvevariables:averagecostofcoal,averagecostofoil,averagecostofnaturalgas,percentofnon-renewablecapacitycomprisedofcoal,andpercentofnon-renewablecapacitycomprisedofnaturalgas.Naturalgasandcoalcapacityaretreatedinthesamemannerashydropowerandnuclearpowercapacityinthemodel.EachvariableisinteractedwithGEN,justliketheothercontrolvariables. Ahigherpercentageoftotalnon-renewablecapacitycomprisedofcoaliscorrelatedwithmorerenewablecapacity,whichgivessomesupportthatstateswithdirtierconventionalcapacityusemorerenewablecapacity.Asthepriceofcoalincreases,lessrenewablecapacityisconstructed.Thesetworesultssupporttheideathattheemissionsrequirementsutilitiesmustmeetareadrivingforcetorenewabledeployment,whileeconomiccompetitivenessinthemarketdoesnothavemuchofanimpact.Cautionisnecessaryininterpretingtheresultswiththeadditionalsetoffuelvariablesbecausetherearemanymissingobservationsthatmustbeextrapolated.Thedummyvariablesthatcontrolformissingobservationsforcoalandnaturalgasarebothstatistically 36


ThecoecientonSUGARCANEPRODCHANGEisinsignicant.Themissingfuelcostdummyvariableinteractedwithgeneration(MISSINGFUELCOST*GEN)controlsforanymeasurementerrorcausedbytheextrapolationofthe38missingdatapointsandisinsignicantaswell. CEF:CAPFUNDED,whichmeasurestheamountofcapacitythatthefundhasagreedtohelpnance,hasamarginallystatisticallysignicantcoecient.Thisincludescapacitythathasbeenagreedupon,buthasnotyetbeenbuilt,eitherbecausetheprojecthasnotbeennishedortheprojectislatercanceled.ForeachmegawattofcapacitythattheCleanEnergyFundhasfundedoragreedtofundinthenearfuture,approximately0.206MWhasbeenconstructed.Thisisnotsignicantlydierentthanthefractionofcapacitythathasactuallybeenconstructedasof2003,whichwas0.33MWper1MW.Eventhoughactualrenewablecapacityisprobablynotconstructedlinearlyoverthelifetimeofthepolicy,theestimatesfromthelinearinterpolationseemtoberepresentativeofactualcapacityconstructionduetotheCleanEnergyFunds. 37


ThemostinterestingresultfromthemodelistheeectthatRequiredGreenPowerOptionshaveonrenewablecapacity.ThecoecientforREQGREENPOWEROPT*GENhasapositiveandstatisticallysignicantcoecientandhassomeofthelargesteectsonrenewablecapacityofanyvariable,whereenactmentleadstorenewablecapacityincreasingby3.46MWperterawatt-hourofgeneration.Astatewithamediangenerationlevel(51.15TWh)thatenactsaRequiredGreenPowerOptionwouldhaveanincreaseof177MW.WashingtonhasthelargestelectricitymarketofstatesthathaveenactedaRequiredGreenPowerOption(100.1TWh),whichwouldleadtoanincreaseinrenewablecapacityof346MW.Togivesomeperspectiveontheseresults,theestimatedimpactsintotalrenewablecapacitycanbeexpressedintermsofapercentageoftotalcapacityinastate.Dependingonthestate,aRequiredGreenPowerOptionleadstoanincreaseofabout1.2-1.6%. 34 38


TherearemultiplereasonsfortheinsignicanceofRPS:SALESREQ*GEN.First,mostRenewablesPortfolioStandardswithsalesrequirementshavebeenenactedfairlyrecentlyandcurrentlyhavelowrequirements.Powersuppliersmaybeabletomeettheirlowinitialrequirementsbyusingcurrentrenewablecapacity.Second,someofthestatesimplementingthesepolicieshaveavailablehydropowercapacityalreadyinplacethatisconsideredeligibletomeetaportionofthesecurrentlylowmandates,whichdecreasesthepolicies'eectivenessinencouragingnewnon-hydrorenewablecapacitydeploymentinastate.Third,somestatesallowthepurchaseofRenewableEnergyCredits(RECs),whicharecerticatesthatrepresenttheenvironmentalrightsofrenewableelectricity,insteadofactualgeneration.Manyofthesesamestatesallowpowerproducers/retailerstopurchaseRECsfromout-of-statepowerproducerstomeettheirin-staterequirements.Allthreeofthesefactorswilldecreasethepolicy'simpactsonin-staterenewablecapacitydeployment.AfewmoreyearsofdatashouldresultinRPS:SALESREQ*GENtohavestatisticallysignicantimpactsonrenewablecapacityinitsstate. Neithernetmeteringnorinterconnectionstandardsappeartohaveanimpactonrenewablecapacity.ThecoecientsonNETMETERINGandINTERCONSTANDARDSarestatisticallyinsignicant. 39




Threeregulatorypoliciesappeartobeeectiveatincreasingrenewablecapacitydeploymentinastate.Thesignicantresultsfromtheseregulatorypoliciesconrmmanyofthendingsfrompriorcasestudies,whichndRenewablesPortfolioStandardswithCapacityRequirementsandCleanEnergyFundshaveincreasedrenewablecapacity.Anadditionalpolicy,MandatoryGreenPowerOptions,isalsofoundtoincreasecapacitydeploymentinastateaswell. ThepreviousempiricalstudyfoundPublicBenetsFunds,whichincludeanyCleanEnergyFundinastate,tobeinsignicantintheirmodel.MypaperndsthatCleanEnergyFundswithutility-scaleprojectsincreasethedeploymentofrenewablecapacityinastate.ByusingSystemBenetsCharges(SBCs)astatecaneectivelymakeconsumerspayforcleanerenergywithoutcreatingadierentmarketforrenewableenergydemand.SimilartothecasestudyndingsbyBolingeretal.(2001,2004,2005),largerutility-scaleprojectsmakeCleanEnergyFundsmoreeectiveatincreasingrenewablecapacitydeploymentinastate. ThispaperndsthatdierenttypesofRenewablesPortfolioStandardshavedierenteectsonrenewablecapacity.EachmegawattofcapacitymandatedbyRenewablesPortfolioStandardswithCapacityRequirementsresultsinthedeploymentofonemegawattofadditionalrenewablecapacityinastate.ButrecentRenewablesPortfolioStandardsthatmandategenerationorsaleslevelsappearnottohavestatisticallysignicanteects.TheseresultsmirrorPetersik'scasestudyinthatonlyRenewables 41


StatewideRequiredGreenPowerOptionsappeartohavebeenaseectiveasanyotherpolicy.Forcingutilitiestooercustomerstheoptiontopurchaserenewable-basedelectricityatareasonablepremiumratedrasticallyincreasesrenewablecapacityinastate.Thepolicyhasagreaterimpactinlargerelectricitymarketsandappearstobeeectiveregardlessofastate'spoliticalenvironment. Therearemajorrenewablepolicyimplicationsiftheseresultsholdwhenadditionalyearsareeventuallyincludedinthemodel.OnlyvestateshavecurrentlyimplementedRequiredGreenPowerOptionseventhoughcreatingastatewidegreenpowermarketappearstobeaseectiveatincreasingrenewableenergycapacityinastateasacommand-and-controlschemeofaRenewablesPortfolioStandardsortax-and-subsidyschemeofaCleanEnergyFund.Stategovernmentpurchasingagreementsofrenewableenergyappeartobenomorethanwindowdressingforpoliticianstoshowtheirsupporttotheenvironmentalcommunity,andadditionalfundingtorenewablepowerproducers. Theremainingpoliciesinthemodeldonotappeartoimpactrenewableenergycapacityconstructioninthetheelectricpowerindustry.Stategovernmentgreenpowerpurchasingdoesnotincreaserenewableelectricitydemandenoughtodrivecapacityconstruction.Netmeteringandinterconnectionstandardstargetresidentialandcommercialcapacityanddonotimpactelectricpowerindustrydecisions. Theimportantpolicyimplicationsthatarisefromtheresultsindicatepolicymakershaveawidearrayoftoolsattheirdisposaltopromoterenewableenergydeploymentinastatetomeetenvironmentalandenergysecuritypolicygoals.ThearrayofpolicymechanismswillbecomeevenmoreusefultostategovernmentsiftheprospectofU.S.climatechange/carbonemissionspolicybecomesareality. 42


DependentandControlVariables RENEWABLECAPACITY(MW)348.7827.60.006177.4178.5GEN(TWh)73.8264.984.95385.6351.15PCTHYDROPOWER(Percentage)14.1420.680.0091.596.26PCTNUCLEAR(Percentage)11.1312.460.0056.207.54BORDERPRICE(2002cents/kWh)7.582.074.8214.496.68RENEWCOST(2002cents/kWh)6.930.5856.007.796.94FUELCOST(2002dollars/MMBtu)2.0860.9870.6017.4311.861LCVSCORE(0to100)43.0626.510.0100.038.0SUGARCANEPRODCHANGE88.95599.33-1707.04882.00.0


RegressionsResults CEF:CAPFUNDED(MW)0.1980.1440.206(0.107)*(0.145)(0.124)*RPS:CAPREQ(MW)1.2451.1441.142(0.175)***(0.168)***(0.168)***RPS:EFFECTIVEGENREQ0.1530.1080.100(0.056)***(0.077)(0.071)PCTSTATEGREENPOWERPURCHASING*GEN0.0260.013-0.002(0.014)*(0.017)(0.016)REQUIREDGREENPOWEROPT*GEN3.5393.3183.457(0.667)***(1.207)***(1.166)***NETMETERING*GEN0.093-0.237-0.033(0.205)(0.214)(0.227)INTERCONSTANDARD*GEN-0.246-0.241-0.240(0.211)(0.195)(0.210)SUGARCANEPRODUCTIONCHANGE(Tons)-0.011-0.007(0.023)(0.023)GEN(1TWh)-0.184(1.150)FUELCOST($/mmBtu)*GEN-0.154-0.124(0.107)(0.109)FUELCOSTMISSING*GEN-1.261-1.167(0.843)(0.802)BORDERPRICE(Cents/kWh)*GEN0.199-0.259(0.207)(0.117)**RENEWCOST(Cents/kWh)*GEN-0.712(0.211)***LCVSCORE*GEN0.0240.025(0.011)**(0.011)**PCTHYDRO*GEN0.018-0.027(0.028)(0.024)PCTNUCLEAR*GEN0.1430.168(0.043)***(0.052)***YR1997*GEN-0.048(0.151)YR1998*GEN-0.142(0.186)YR1999*GEN-0.187(0.182)YR2000*GEN-0.269(0.199)YR2001*GEN0.322(0.247)YR2002*GEN0.250(0.199)YR2003*GEN0.530(0.226)**CONSTANT309.103413.745267.888(7.689)***(55.340)***(50.741)***Observations400400400StateFixed-Eects505050R-squared0.5980.6340.646 RobustStandardErrorsinParentheses;*signicantat10%;**signicantat5%;***signicantat1%




VariableEectsofSignicantVariables VariableMeanStd.Dev.IncreasePerUnitperTWhStd.Dev.inMedianGen.State RENEWCOST(2002cents/kWh)6.930.59-0.725MW-36MWBORDERPRICE(2002cents/kWh)7.582.07-0.235MW-27MWPCTNUCLEAR11.1312.460.170MW107MWLCVSCORE(0-100)43.0626.510.025MW34MW EectofEnactingaPolicyMeanStd.Dev.PerUnitEectImpactinMedianGen.State PBF:CAPFUNDED8.2061.210.206MWper1.0MW20.6%ofFundedCapacityRPS:CAPREQ14.0578.301.142MWper1.0MW114%ofReq.Cap.REQGREENPOWEROPT0.020.153.457MWperTWh177MW Note1:ResultsareforSpec.3excludingRENEWCOST,whichisfromSpec.2.Note2:Allothervariablesarestatisticallyinsignicant.




Inthispaperamodelofunit-levelSO2complianceisconstructedthatincorporatesthepresenceofcoalcontractstoexaminehowlong-termcoalcontractsaectutilitycompliancechoicesandaunit'scompliancecosts.Asexpected,thepresenceofcoalcontractconstraintsleadstocompliancecostsinexcessofthehypotheticalleast-costsolution.ThepresenceofbindinghighsulfurcontractconstraintsthatwerelikelyinPhaseIoftheTitleIVSO2ProgrammayexplainthelowerthanexpectedallowancepricesinPhaseIthataccompaniedcompliancecoststhatwereabovetheleast-costsolution.ItisalsofoundthatthepresenceofbindinglowsulfurcoalconstraintsthatmayexistunderCAIR,whichmayleadtoallowancepricesthatarehigherthanwithoutthebindingconstraint.Theeectsofthecontractconstraintseemcounter-intuitive:bindinghighsulfurcoalconstraintsleadingtolowerexcessdemandforallowances,whichcouldreducetheallowancemarketprice.Bindinglowsulfurcoalconstraintsleadingtohigherexcessdemandforallowances,whichcouldincreasetheallowancemarketprice.Theinteractionbetweenthecontractconstraintsandthediscretenatureofthescrubberchoiceleadstotheseunexpectedresults. 2.2.1TitleIVoftheCleanAirActAmendment 48


Anadditional168unitsparticipatedinPhaseIin1996basedontherulesestablishedbyEPAallowingaplantto\opt-in"units(7units),designatesubstitutionunits(160units),ordesignatecompensatingunits(1unit)aspartoftheirPhaseIcomplianceplans.Thevoluntaryparticipationoftheseadditionalunitsresultedinatotalof431aectedgeneratingunitsunderPhaseI.A\substitutionunit"isaunitthatwouldeventuallybeaectedinPhaseIIthatvoluntarilyenrolledintoPhaseItomeetsomeoralloftherequiredemissionsreductionsforaPhaseIunit(SotkiewiczandHolt,2005).Substitutionunitsreceiveanallowanceallocationbasedonitshistoricalheatinput.AutilitymaydecidetoreduceitselectricityproductionataPhaseIaectedunit.Todoso,theutilitymusthavea\compensationunit"fromthePhaseIIunitstheutilityoperatestocoverthenecessaryadditionalelectricity.ThiscompensationunitisthenbroughtintoPhaseIandgivenanallowanceallocationbasedonitshistoricalheatinput.IndustrialsourcesofSO2emissionscouldusetheopt-inprovisionandvoluntarilyenrollintoPhaseIandreceive 49


ThebankingprovisionhasallowedutilitiestotradeintertemporallywithutilitiesusingthesubstantialallowancebankaccumulatedthroughPhaseIforcomplianceinPhaseIIleadingtoannualemissionlevelsinexcessof10milliontonsineachyearfrom2000to2005. 50


GeneratingunitsstillreceivetheirallowanceallocationasdenedunderPhaseII.Beginningin2010,theemissionsvalueoftheallowancesforunitsintheCAIRregioniscutinhalffrom1.2to0.6lbs.SO2/MMBtuofheatinputwhichimpliesaunitmustholdtwoTitleIVallowancesforeachtonofactualemissions.Startingin2015unitsmusthold2.86TitleIVallowancesforeverytonofemissions,whichtranslatestoanallocationofapproximately0.4lbs.SO2/MMBtu.MeanwhilegeneratingunitsoutsidetheCAIRregionwillcontinueoperateundertheTitleIV,PhaseIIProgramwithtradesallowedtotakeplacebetweenCAIRandPhaseIIunits.UnitsunderPhaseIIandCAIRparticipateinthesameallowancemarketandfacethesamemarketallowanceprice. Thespikeinallowancepricesuptoover$1,600in2006seeninFigure 2.8 mayhavebeenaresultoftheproposalandenactmentofCAIR.UtilitiescouldhavechosentobankallowancesinsteadofsellingallowancesinthemarkettoensuretheirabilitytocoverrequirementsatthebeginningofCAIR.Itisstilluncertainwhatledtothetemporaryspikeinallowanceprices,butitappearslikelythatitwasatemporaryreactiontothepolicyenvironmentorothermarketconditionsastheallowancepricehasquicklydecreasedtoitscurrentpriceoflessthan$600/ton. 51


2.3.1TitleIV:PhaseI 2-1 ,itcanbeseenthat,ingeneral,thecompliancecostestimatesbeforePhaseItookeectwerehigherthantheestimatesmadeafterPhaseIbecameeectiveandactualdatacouldbeusedintheestimates.Thepre-policyestimatesrangeashighas$1.34billion/yearwithmostestimatesatleast$860million/year.Theactualcostestimatesaretowardsthelowerendofthisrangebetween$730-$990million/year. Thereareseveralreasonsforthedierencesbetweeninitialestimatesandactualaggregateindustrycompliancecosts.Themostimportantfactorwasthedecreaseindeliveredlowsulfurcoalprices.Attheunitlevel,lowerlowsulfurcoalpricesdecreasedthemarginalcostofreducingemissionsthroughfuelswitching,whichwasthecomplianceoptionchosenby52%ofallaectedunits,while32%ofaectedunitschosetopurchaseallowances,10%installedascrubber,3%shutdown,and3%choseothermethods. Keohane(2002)simulateswhichgeneratingunitswouldhaveinstalledscrubbersunderauniformemissions-ratestandardandndsthatthetotalnumberofscrubberswouldhavebeenone-thirdhigherthantheactualnumberofinstalledscrubbersunderthe 52


IntheinitialyearsofPhaseI,manyrmswerenotactiveparticipantsintheallowancemarket,choosingtoswitchfuelsandbankallowancesorshiftallowancesbetweenonlytheirownunits(Hart(1998);Ellermanetal.(1998)).Thermsthatdidparticipatemainlytradedallowanceswithinthesameutilitycompany.BohiandBurtraw(1997)ndthatintra-utilitytradingaccountsfortwo-thirdsoftheallowancetransactionswhiletheremainingone-thirdwereinter-utilitytrades.Sincemosttradesweremadebetweenunitsownedbythesamecompany,tradingbetweentwogeneratingunitsatthesameplantwouldbeacommonoccurrence.Manystudiessuggestedstatepublicutilityregulationsandotherstatelawsasareasonfortheinecienciesresultingfromthisself-sucientbehavior(Bohi(1994);BohiandBurtraw(1997);Swift(2001)). Arimura(2002)useseconometricapproachestostudytheimpactofPUCregulationoncompliancechoices,andndsthatutilitiesthatfacePUCregulationaremorelikelytoswitchfuelsinsteadofpurchasingallowancesforcompliance. Winebrakeetal.(1995)estimatedthecostinecienciesfromstategovernmentrestrictionsonautility'sallowancetrading,andestimatesthetotalcostestimatesforthersttenyearsofTitleIV(1995-2005).Acommand-and-controlapproachwasestimatedtoresultincompliancecostsof$4.19billiongreaterthanintheunrestrictedpermittradingsystem($5.02billion,oranaverageof$502million/year)andanestimatedallowancepriceof$143/ton. Winebrakeetal.(1995)simulatestheadditionalcostsfromrestrictionsonbetween-statetradingthatwereunderconsiderationbybothNewYorkandWisconsin.Bothstateswere 53


SomestudiesfurtherexplaintheinecienciesbyexaminingtheactuallostcostsavingsthatarespecicallyaresultofstatePUCregulationunderPhaseI.Carlsonetal.(2000)ndthattheactualcompliancecostswere$339million(59%)greaterin1996thantheleast-costsolution.Thestudyconcludesthedierencebetweenactualcompliancecostsandtheleast-costcompliancemaybeattributableto\adjustmentcostsassociatedwithchangingfuelcontractsandcapitalexpendituresaswellasregulatorypolicies."SotkiewiczandHolt(2005)ndthatPUCregulationsresultedin$131millionoftheadditionalcompliancecostsrelativetotheleastcostsolution.However,thereisasignicantamountofcompliancecoststhatremainedunexplained. 54


Swinton(2004)followsthesameapproachexceptitexpandsthestudyto40plantswithdatafrom1994to1998,andintroducesthepossibilitythatlong-termcoalcontractsmayprohibitutilitiesfromswitchingcoaltypes,althoughitisnotmodeled.Bothstudiesndtheactualutility-levelcompliancecoststobemuchhigherthantheestimatedleast-costsolution. CogginsandSwinton(1996)useanoutputdistancefunctiontoestimatetheshadowpriceofSO2emissionsabatementforelectricpowerplantsinWisconsin.Thestudyestimatedtheallowanceshadowpricetobegreaterthantheobservedallowancepricesatthetime,whichtheyassertmaybepartiallyexplainedbyWisconsin'sstrictstateregulationsonSO2emissions. SeveralstudieshaveshownanalyticallyorthroughsimulationmodelsthatstatePUCregulationscanleadtoinecienciesattheutility-level.BohiandBurtraw(1992)developamodelofutilitydecision-makinggiventwocomplianceoptions,purchasingallowancesorinstallingemissionscontroltechnology.BohiandBurtrawderivetworecommendationssothatstateregulationdoesnotresultininecientcompliancechoicesbyautility.First,ifautility'sallowedreturnislessthanitscostofcapitalwithrespecttobothcomplianceoptions,symmetricalcostrecoveryrulesarerecommendedasuneventreatmentofcostrecoverymaycreateincentivesforautilitytomakesuboptimalcompliancechoices.Second,ifautilityisallowedtoearnmorethanitscostofcapitalwithregardtoboth 55


Joskow(1988,1990)ndsthatduringperiodsinwhichthespotmarketcoalpriceswerelowerthanthecontractedprices,thecontractpricesfailedtoadjustdownward.Thisdownwardrigidityofcoalpricescanleadtoutilitycoalcostsbeinghigherthanisoptimalintheshortrun.Somerenegotiation,breachofcontract,andlitigationhasoccurred,butnearlyallcontractsappeartohavecontinuedunchanged.Themainreasonfortheconstraintsinalteringthesecoalcontractsisthatlessthan15%ofcoalconsumedbyutilitiesissuppliedbyacoalcompanyownedbythesameutility(Joskow,1987).Firmshavehighlegalornegotiationcostsofbreakingacoalcontractwhentheagreementismadewitharmthathasnonancialtiestotheutility.Coalcontractsmayalsobearesultofregulationsprotectingthelocalcoalindustry(Arimura2002).Duetotheinabilityofcontractedcoalpricestodecreasewithspotmarketprices,largecoalprice 56




Third,underCAIRsomeutilitiesmaybelockedintolowsulfurcoalcontractsenteredintoforTitleIVcomplianceandmaybeunableswitchbacktohighsulfurcoalandscrubifitiscost-eectivetodoso.Consequently,thecontractconstraintpushesautilityintosub-optimalcompliancechoicessuchasallowancepurchasesorscrubberinstallationwhileusinglowsulfurcoal. AnexaminationofthedataforPhaseIaectedunitsindicatesthatofthe26scrubbersinstalledbytheendof1996inresponsetothepassageofthe1990CleanAirActAmendments,23ofthosescrubberswereinstalledatfacilitieswith40percentormoreofitscoaldeliveriesbycontractandwith20ofthemhavingaweightedaverageSO2emissionratesgreaterthanthePhaseIallowedlevelof2.5lbs.SO2/mmBtu(poundspermillionBtusofheat). 58


12 59








subjectto...Aei+Ai(1ziri)(m)(CsihSsih+CcihScih+CsilSsil+CcilScil)i1 (CsihHsih+CcihHcih+CsilHsil+CcilHcil)Dii2 (2{6) Equation( 3{1 )representstheunit'scostfunction.Thesecostsincludethecostofscrubberinstallation(ziPiz),netcostsofallowancepurchases(PAAi),andcostsofcoalpurchases(PsihCsih+PsilCsil+Pcih 3{2 ),wherethenumberofallowancesheld(Aei+Ai)mustbeaslargeastheamountoftotalemissionsbythegeneratingunit[(1ziri)(m)(CsihSsih+ 3{3 ).TheLagrangemultiplieronthedemandconstraintisrepresentedbyi2.Coalcontractconstraintsrequiretheunittouseaspecicamountofeachcontractcoaltype, 3{4 )and 3{5 ).Aunitwilluseexactlythecontractedamountbecause(1)ifthecontractcoalismoreexpensivethanspotmarketcoal,thena 63


SinceAicanbeeitherpositiveornegativebasedonthenetallowanceposition,( 2{8 )willholdwithequality.Theadditionalcosttothermofemittingonemoretonofemissionsisequivalenttotheallowanceprice,i1=PA. Letf2fh;lgrepresentthetypeofcoalandg2fs;cgrepresentthetypeofpurchase.Eachcoaltypehasitsownsulfurcontent(Sgif),heatcontent(Hgif),anddeliveredprice(Pgif).ThepartialderivativeswithrespecttoCgifrepresenttheimpactaoneunitchangein\f"typesulfurcoal(highorlow)froma\g"typepurchasingagreement(spotmarketorcontract)hasontheunit'stotalcosts. ThecostofusingonemoreunitofCgifcanbedisaggregatedintofourdierentcostchanges:Pgifistheadditionalcostsofpurchasingonemoreunitofcoal,i1(1ziri)(m)(Sgif)istheadditionalcostsoftheextraemissionsfromonemoreunitofcoal, 64


3{37 )or( 2{10 )holdswithequality.Sincethecontractconstraintholdswithequality,( 2{11 )and( 2{12 )alwaysholdwithequality. 3{37 )and( 2{10 )holdwithequality,or ( 2{13 )canberearrangedtoisolatetheshadowpriceofallowancesoremissions,i1,toderivetheMarginalCostofAbatementfromSwitchingFuelsfromhighsulfurspottolowsulfurspotmarketcoal(MCAs;si)in( 2{14 ).Exploiting( 2{8 ),theallowancepriceequalsMCAs;si: (2{14) Theshadowpriceisequaltothedierenceinpriceperunitofheatdividedbythedierenceinemissionsperunitofheat. 65


3{37 )holdswithequalitywhile( 2{10 )holdswithweakinequality,or ( 2{15 )canberearrangedtoshowtheallowancepriceisweaklylessthanMCAs;si: 2{10 )holdswithequalitywhile( 3{37 )holdswithweakinequality,or ( 2{17 )canberearrangedtoshowthattheallowancepriceisweaklygreaterthanMCAs;si: Assumeaunithasahighsulfurcoalcontractandusesonlylowsulfurspotmarketcoal(nohighsulfurspotmarketcoal).So( 2{10 )and( 2{11 )holdwithequality.Ithasbeenshownin( 2{17 )thatifaunitusesonlylowsulfurspotmarketcoal,PAMCAs;si: 66


2{19 )andexploiting( 2{8 ),itcanbeshownthattheallowancepriceisequaltoMarginalCostofAbatementofswitchingfromhighsulfurcontracttolowsulfurspotmarketcoal(MCAc;si)plusanadditionaltermrepresentingthebenetsofmeetingthecontractconstraint,whichisweaklygreaterthanMCAs;si: From( 2{20 ),ifMCAs;siMCAc;si>0,thenih>0andhighsulfurcontractcoalismoreexpensivethanspotmarkethighsulfurcoal.IfMCAs;siMCAc;si<0,thenihcanbepositiveornegative.Ifih<0,thencontractcoalischeaperthanspotmarketcoalandtheadditionalcompliancecostsduetoemissionsdominatethesavingsfromthelowerfuelcosts.Ifih>0,thencontractcoalischeaperthanspotmarketcoalandthesavingsfromthelowerfuelcostsdominatetheincreasedcompliancecostsduetoemissions. Nowassumeaunithasahighsulfurcoalcontractandusesonlyhighsulfurspotmarketcoal(nolowsulfurspotmarketcoal).( 2{11 )and( 2{12 )holdwithequalitywhile( 2{10 )holdswithweakinequality.Also,ithasbeenshownthatwhenaunitusesonlyhighsulfurspotmarketcoal,PAMCAs;si.Usingthesameapproachasinthepreviouscase,itcanbefoundthattheallowancepriceisweaklylessthanMCAc;siplusanadditionaltermrepresentingthebenetsofmeetingthecontractconstraint: ThesignofihcannotbedeterminedbycomparingMCAs;siandMCAc;si,buttherst-orderconditionsforhighsulfurcontractandspotmarketcoalcanbeusedtodetermineitssign.Psih+i1(1ziri)(m)(Ssih)


Iftheadditionalcosts,bothfromfuelcostsandemissions,ofusinghighsulfurcontractcoalismoreexpensivethanusinghighsulfurspotmarketcoal,thenih>0.Iftheadditionalcosts,bothfromfuelcostsandemissions,ofusinghighsulfurcontractcoalislessthanusinghighsulfurspotmarketcoal,thenih<0.Iftheadditionalcosts,bothfromfuelcostsandemissions,ofusinghighsulfurcontractcoalisisthesameasusinghighsulfurspotmarketcoal,thenih=0. Nowassumeaunithasahighsulfurcoalcontractandusesbothhighsulfurandlowsulfurspotmarketcoal.So( 2{10 )and( 2{11 )holdwithequality.Also,ithasbeenshownin( 2{15 )thatwhenaunitusesbothhighandlowsulfurspotmarketcoal,PA=MCAs;si: Bysolvingforihin( 2{24 ),thesignofihcanbedetermined.IfMCAs;siMCAc;si=0,thenih=0andhighsulfurcontractcoalhasthesamecostashighsulfurspotmarketcoal.IfMCAs;siMCAc;si>0,thenih>0andcontractcoalismoreexpensivetousethanspotmarketcoal.IfMCAs;siMCAc;si<0,thenih<0andcontractcoalischeapertousethanspotmarketcoal. Assumeaunithasalowsulfurcoalcontractandusesonlyhighsulfurspotmarketcoal.Ithasalreadybeenshownin( 2{16 )thatwhenaunitusesonlyhighsulfurspot 68


3{37 )and( 2{12 )holdwithequalitywhile( 2{10 )holdswithaweakinequality.Usingthesameapproachasintheprevioussectionyields: Using( 2{26 ),thesignofilcanbedetermined.IfMCAs;siMCAc;si>0,thenil>0andlowsulfurcontractcoalismoreexpensivetousethanlowsulfurspotmarketcoal.IfMCAs;siMCAc;si<0,thenilcaneitherbenegativeorpositive.IfMCAs;siMCAc;si<0andil>0,thenthehigherfuelcostsdominatethelowercostsfromemissions.IfMCAs;siMCAc;si<0andil<0,thenthethelowercostsfromemissionsdominatehigherfuelcosts. Nowassumeaunithasalowsulfurcoalcontractandusesonlylowsulfurspotmarketcoal.So( 2{10 )and( 2{12 )holdwithequalitywhile( 3{37 )holdswithaweakinequality.Ithasalreadybeenshownin( 2{17 )thatwhenaunitusesonlylowsulfurspotmarketcoal,PAMCAs;si. AlthoughthesignofilcannotbedeterminedbycomparingMCAs;siandMCAs;ci.Therstorderconditionsforlowsulfurcontractandspotmarketcoalcanbeusedtodetermineitssign.Psil+i1(1ziri)(m)(Ssil) Iftheadditionalcosts,bothfromfuelcostsandemissions,ofusinglowsulfurcontractcoalismoreexpensivethanusinglowsulfurspotmarketcoal,thenil>0.Ifthe 69


Nowassumeageneratingunithasalowsulfurcoalcontractandusesbothhighsulfurandlowsulfurspotmarketcoal.So( 3{37 ),( 2{10 ),and( 2{12 )holdwithequalityandPA=MCAs;si: Bysolvingforil,thesignofilcanbedetermined.IfMCAs;ciMCAs;si=0,thenil=0andcontractcoalisascostlytouseasspotmarketcoal.IfMCAs;ciMCAs;si>0,thenih>0andcontractcoalismoreexpensivetousethanspotmarketcoal.IfMCAs;ciMCAs;si<0,thenih<0andcontractcoalischeapertousethanspotmarketcoal. Thecostsofaunit'snetallowancepositionisthedierencebetweenageneratingunit'sinitialallowanceallocationanditsactualemissionsmultipliedbytheallowanceprice(PAAi). Thecostsofswitchingfuelsisthelargeroftwovalues:(1)totalcostsofactualcoalpurchases(PsihCsih+PsilCsil+Pcih 70


Thelatterwillonlyoccurifitisweaklycheaperforthegeneratingunittouselowsulfurcoalwithouttheemissionsrestrictions(Psih Combiningeachofthethreecostcomponentsresultsinaunit'stotalnetcompliancecosts.Eventhoughthecontractcoalhasnodirectaect,thecontractswillindirectlyaectcompliancecoststhroughaunit'sallowancepositionandscrubberchoice.Proposition1showsthesucientconditionsunderwhichacoalcontractwilleitherincreaseordecreasecompliancecosts. (i) IfPA>MCAs;siandthesulfurtoheatcontentratioofhighsulfurcontractcoal(Scih (ii) IfPAMCAs;siandthesulfurtoheatcontentratioofhighsulfurcontractcoal(Scih (iii) IfPAMCAs;siandthesulfurtoheatcontentratiooflowsulfurcontractcoal(Scil (iv) IfPA

3{2 ).From( 3{2 )and( 3{3 ),theminimumandmaximumexcessdemandforallowancescanbeformallyderived. IfPAMCAs;si,aunitwillusethemaximumamountoflowsulfurspotmarketcoal,whichcanbederivedfrom( 3{3 ): Theuseofalllowsulfurspotmarketcoalleadstotheminimumemissionslevel: 72


2{37 )forCsilin( 3{3 )givesanexpressionfortheminimumallowanceexcessdemand,whichisthedierencebetweentheminimumemissionslevel(EMINi)andtheinitialallowanceallocation(Aei): Ifaunit'sinitialallocationcancoveritsminimumpossibleemissions,thenitwillhaveanegativenetallowancepositionandbeanetsellerofallowances. IfPA=MCAs;si,aunitmayuseanycombinationofhighsulfurspotmarketcoalandlowsulfurspotmarketcoal,whichleadstoanylevelofexcessdemandintherange(EMINiAei;EMAXiAei).TheallowanceexcessdemandcanberepresentedbyAi=EMAXi(1)EMINiAeiwheretheconstant2[0;1].Aunitthatisindierentbetweenfuelswitchingandallowancespurchasescouldbeeitheranetbuyeroranetseller. CombiningtheexcessdemandsforeachofthethreecasescreatestheExcessDemandCorrespondence:Ai=8>>>><>>>>:AMAXiifPA>MCAs;siAMAXi(1)AMINiifPA=MCAs;si82[0;1]AMINiifPAMCAs;si.ThecasewhereageneratingunitusessomecombinationoflowsulfurspotmarketcoalandhighsulfurspotmarketcoalisrepresentedbythehorizontallineatwhichPA=MCAs;si. 73


2{39 ),wheretheexcessdemandwillactuallybenegative. Theunit'scostsavingsfromswitchingfuelsoverpurchasingallowancesis(PAMCAs;si)(AMAXiAMINi),whichisthearea\a+b"seeninFigure 2.8 (ii).Thedark-shadedarea(a)isthecostsavingsforthegeneratingunitfromabatingemissionsthroughfuelswitchinginsteadofpurchasingadditionalallowances.Thelight-shadedarea(b)isthecostsavingsfromabatingmorethanitsinitialallocationandsellingtheextraallowances. 2.8 ,whichissummarizedinProposition2. (i) FortherangeofallowancepricesPAMCAs;si,ahighsulfurcoalcontractwillweaklyincreaseaunit'sallowanceexcessdemand. (ii) FortherangeofallowancepricesPAMCAs;si,ahighsulfurcoalcontractwillweaklydecreaseexcessdemandifScih 74


FortherangeofallowancepricesPAMCAs;si,ahighsulfurcoalcontractwillweaklyincreaseexcessdemandifScih ThedecreaseinlowsulfurspotmarketcoaluseincreasesemissionsfromEMINitobEMINi,whichistheminimumemissionsgiventhehighsulfurcoalcontractconstraint: Higheremissionsmustbecoveredbyadditionalallowances,whichresultsintheminimumexcessdemandwithahighsulfurcoalcontracttobegreaterthantheminimumexcessdemandwithnohighsulfurcoalcontract: Therefore,ahighsulfurcoalcontractwillincreaseexcessdemandforallowancepricesPA>MCAs;si. 75


Loweremissionsresultinadecreaseinaunit'smaximumexcessdemandfromAMAXitobAMAXiinFigure 2.8 (i): Therefore,ahighsulfurcoalcontractforcoalwithalowersulfurtoheatcontentratiowilldecreaseexcessdemandforallowancepricesPA

2.8 ,aunitpreferstouselowsulfurcoalbecausePA>MCAs;si,andthecompliancecostsfromswitchingfuelstoabateemissionscanbeseeninarea(a+b).Area(b+c)istherevenuegained(negativecost)fromsellingtheremainingallowancesthatareavailableduetoabatingemissionsbelowaunit'sinitialallowanceallocation.Thenetcompliancecostsforaunitwillbethecostsofswitchingfuelsminustherevenuesfromallowancesales,or(a+b)(b+c)=(ac).If(ac)<0,thenaunitwillhavenegativecompliancecosts. Nowconsiderhowahighsulfurcoalcontractwillimpacttheexcessdemandcorrespondence,compliancecosts,andtotalcosts.AshasalreadybeenshownaboveandcanbeseeninFigure 2.8 ,ahighsulfurcoalcontractwillincreasetheminimumexcessdemandfromAMINitobAMINiandmayincreaseordecreasethemaximumexcessdemanddependingontherelativesulfurtoheatcontentratioofcontracttospotmarketcoal. Theseshiftsinaunit'sexcessdemandmayhavethreedistincteectsonaunit'scosts.Therstcostimpactistheadditionalcompliancecoststoaunitfromlostallowancesales,whichisrepresentedbyarea(d)inFigure 2.8 .Aunitmustusesomehighsulfurcoal,whichresultsinaunitcoveringadditionalemissionsthroughmoreexpensiveallowancesinsteadofswitchingtolowsulfurcoal. Thesecondimpactresultsfromthedierenceinthesulfurtoheatcontentratiobetweenhighsulfurcontractandspotmarketcoal.IfScih 2.8 (i)forreducingemissionstotheallowanceallocation 77


2.8 (ii)forreducingemissionstotheallowanceallocationemissionslevel. Considerhowtheseshiftsinexcessdemandwillimpactaunit'snetcompliancecosts.Theincreaseintheminimumexcessdemandwillincreasenetcompliancecostsbydecreasingarea(b)andarea(c).Ifmaximumexcessdemanddecreases,netcompliancecostsdecreasebyarea(f)inFigure 2.8 (i).Combiningthetwoimpactsresultsinnetcompliancecostsof(ac)inFigure 2.8 (i).Ifmaximumexcessdemandincreases,netcompliancecostsincreasebyarea(f)inFigure 2.8 (ii).Combiningthetwoimpactsresultsinnetcompliancecostsof(a+fc)ininFigure 2.8 (ii). Thethirdcostimpactresultsfromdierentpricesforhighsulfurspotmarketandcontractcoal,whichisrepresentedbyarea(e).Higherpricedcontractcoalcausesaunittohaveadditionalfuelcoststomeetingelectricitydemand,whichisanincreaseaunit'stotalcosts. Thegraphicaldescriptionofahighsulfurcoalcontract'simpactsonexcessdemandandcostscanbeseenintheexampledenedinTable 3-1 .Giventhecoalcharacteristics,demand,allowanceallocation,andallowanceprice,andusingahighsulfurcontractcoalpriceof$1.50/mmBtu,theminimumandmaximumallowanceuseandMCAs;sicanbecomputedifaunitfacesnohighsulfurcoalcontract( Theminimumallowanceusebasedonthelowsulfurcoalcharacteristicsis11,400tonswhilethemaximumallowanceusebasedonthehighsulfurspotmarketcoalcharacteristicsis38,000tons.TheMCAs;siis$270.68,whichislowerthantheallowancepriceof$300.00.Soaunitwillswitchtoalllowsulfurcoaltominimizeitstotalcosts.Indoingso,itwilluseuse11,400ofitsallowanceallocationtocoveritsminimumemissions 78


Nowconsiderthesameunitwithahighsulfurcoalcontractfor500,000tonsofcoal,whichaccountsforhalfoftherequiredheatinput.Themaximumallowanceuseremainsat38,000tons,buttheminimumallowanceusedecreasesto24,700tonsbecauseofthecoalcontract.TheMCAc;siis$90.23,whichismuchlowerthantheallowancepriceof$300.00andMCAs;siof$270.68.Thecoalcontractresultsincompliancecosts(costsofswitchingfuelsminusallowancesales)of$5.01millionandtotalcoststomeetingelectricitydemand(lowsulfurspotmarketcoalpurchasesplushighsulfurcontractcoalpurchasesplusallowancepurchases)of$38.61million.Compliancecostsincreasedby$390,000becausetheunitcouldnotswitchallitscoalusetolowsulfurcoal.Totalcoststotheunitincreasedby$2.79millionbecauseofthehigherpriceforthecontractcoal. Anotherpossibilityisthatthehighsulfurcontractcoalcouldactuallybecheaper,whichisreasonablebecausethemainreasonformakingacoalcontractagreementisforprotectionagainstfuturecoalpriceuctuations.ThiscasecanbeseeninFigure 2.8 ,whereMCAc;si>MCAs;siandthereisactuallyacostsavingstotheunitfromusinghighsulfurcontractcoaloverhighsulfurspotmarketcoal.Thenetcompliancecostsremainthesameasinthepreviousexample,area(ac). 79


2{36 ),wheretheexcessdemandwillbepositive. GiventhataunitpreferredtousehighsulfurcoalbeforeSO2constraints,theunit'scostsavingsfromusingallowancesoverswitchingfuelsis(MCAs;siPA)(AMAXiAMINi),whichisthearea\a+b"seeninFigure 2.8 .Thedark-shadedarea(a)isthecostsavingsforthegeneratingunitfrompurchasingadditionalallowancesinsteadofswitchingfuels.Thelight-shadedarea(b)isthecostsavingsfromusingtheallocatedallowancesinsteadofabatingemissionsandsellingtheextraallowances. 2.8 ,whichissummarizedinProposition3. 80


FortherangeofallowancepricesPAMCAs;si,alowsulfurcoalcontractwillweaklydecreaseaunit'sallowanceexcessdemand. (ii) FortherangeofallowancepricesPAMCAs;si,alowsulfurcoalcontractwilldecreaseexcessdemandifScil (iii) FortherangeofallowancepricesPAMCAs;si,alowsulfurcoalcontractwillincreaseexcessdemandifScil Thedecreaseinmaximumhighsulfurcoalusedecreasesaunit'smaximumemissionsfromEMAXitobEMAXi: Loweremissionsdecreasetheallowancesused,whichresultsinthemaximumexcessdemandwithalowsulfurcoalcontracttobelowerthanthemaximumexcessdemandwithnolowsulfurcoalcontract: Therefore,ahighsulfurcoalcontractwillincreaseexcessdemandforallowancepricesPA

IfScil Loweremissionsresultinadecreaseinaunit'sminimumexcessdemandfromAMAXitobAMAXiinFigure 2.8 (i): Therefore,alowsulfurcoalcontractforcoalwithalowersulfurtoheatcontentratiowilldecreaseexcessdemandforallowancepricesPA>MCAs;si. Greateremissionsresultinanincreaseinaunit'sminimumexcessdemandfromAMINitobAMINiinFigure 2.8 (ii): Therefore,ahighsulfurcoalcontractforcoalwithahighersulfurtoheatcontentratiowillincreaseexcessdemandforallowancepricesPA>MCAs;si. 2.8 .If 82


InFigure 2.8 ,thenetcompliancecostscanbeseenintheshadedarea(a)thatrepresentsthecostsofpurchasingallowancestocovertheunit'semissionsaboveitsallowanceallocation. Nowconsiderhowalowsulfurcoalcontractwillimpacttheexcessdemandcorrespondence,compliancecosts,andtotalcosts.AshasalreadybeenshownaboveandcanbeseeinFigure 2.8 ,alowsulfurcoalcontractwilldecreasethemaximumexcessdemandfromAMAXitobAMAXiandmayincreaseordecreasetheminimumexcessdemanddependingontherelativesulfurtoheatcontentratioofcontracttospotmarketcoal. Theseshiftsinaunit'sexcessdemandmayhavetwodistincteectsonaunit'scosts.Therstcostimpactistheadditionalcompliancecoststoaunitfromswitchingfromhighsulfurspotmarkettolowsulfurcontractcoalinsteadofpurchasingallowances,whichisrepresentedbyarea(c)inFigure 2.8 .Aunitmustusesomelowsulfurcoal,whichresultsinaunitabatingadditionalemissionsinsteadofpurchasingallowances.Area(c)isthedierencebetweenthecostsofswitchingfuelstodecreaseemissions,area(b+c),andthedecreaseincostsfromallowancepurchases,area(b).Netcompliancecostsincreasefrom(a+b)to(a+b+c). Thesecondcostimpactresultsfromdierentpricesforlowsulfurcontractandspotmarketcoal,whichisrepresentedbyarea(d).Higherpricedcontractcoalcausesaunittohaveadditionalfuelcoststomeetingelectricitydemand,whichincreasesaunit'stotalcosts. ThegraphicaldescriptionofcoalcontractimpactsonexcessdemandandcostscanbeseenintheexampledenedinTable 2-4 .Giventhecoalcharacteristics,demand, 83


Theminimumallowanceusebasedonthelowsulfurcoalcharacteristicsis11,400tonswhilethemaximumallowanceusebasedonthehighsulfurspotmarketcoalcharacteristicsis38,000tons.TheMCAs;siis$270.68,whichishigherthantheallowancepriceof$200.00.Soaunitwillpurchaseallowancestominimizeitstotalcosts.Indoingso,itwilluseusetheentire20,000tonallowanceallocationandpurchaseanadditional18,000allowancestocoveritsmaximumemissionslevel.Netcompliancecostsaretheadditionalcostsfrompurchasingallowances,whichis$3.6million.Totalcostsarethetotalcostoffuelpurchasesplusallowancepurchases,whichis$34.8million. Nowconsiderthesameunitwithalowsulfurcoalcontractfor500,000tonsofcoal,whichaccountsforhalftherequiredheatinputtomeetdemand.Theminimumallowanceuseremainsat11,400tons,butthemaximumallowanceusedecreasesto24,700tonsbecauseofthecoalcontract.TheMCAs;ciis$451.13,whichismuchhigherthantheallowancepriceof$200.00andMCAs;siof$270.68.Thecoalcontractresultsinnetcompliancecosts(additionalcostsofallowancepurchasesandfuelswitching)of$4.54millionandtotalcoststomeetingelectricitydemand(highsulfurspotmarketcoalpurchasespluslowsulfurcontractcoalpurchasesplusallowancepurchases)of$38.14million.Compliancecostsincreasedby$940,000becausetheunitcouldnotuseallhighsulfurcoal.Totalcoststotheunitincreasedby$3.34millionbecauseofthehigherpriceforthecontractcoal. Anotherpossibilityisthatthelowsulfurcontractcoalcouldactuallybecheaperthanlowsulfurspotmarketcoal,whichisreasonablebecausethemainreasonformakingacoalcontractagreementisforprotectionagainstfuturecoalpriceuctuations.Thiscasecan 84


2.8 ,whereMCAc;siMCAs;si.However,theimpactsoncompliancecostsandtotalcostswillbedierent.Thereasoningforthisisthatthecontractdoesnotforceaunittouseamoreexpensivecomplianceoption.BycomparingFigure 2.8 toFigure 2.8 ,thesedierencescanbederived.Theshiftinaunit's 85



2{56 )canbeusedtosolveforPSA,theminimumallowancepriceatwhichageneratingunitwillinstallascrubber: (bAieAi)(2{57) AunitwillprefertoinstallascrubberatPSAiftheaveragecostsofabatementfromusingascrubberisweaklylessthanthecostsofpurchasinganallowance. ScrubberinstallationdecreasesthesizeofthedenominatorofMeCAs;siby(ri)(m)(Ssih 87



TheexcessdemandcorrespondenceseeninFigure 2.8 (i)includesageneratingunit'snetallowancepositionunderthefourdierentallowancepricerangesif(Psih 2.8 (ii). 2.8 89


Aunitthatfacesacoalcontractwillfaceadierentindierenceallowancepriceof(PSA)becauseparametersvaluesonbothsidesoftheequalitywillchange.couldbepositiveornegativedependingonseveralconditions,includingthetypeofcoalundercontract.Thenewvaluesthatsolvethisequalityarethecontractconstrainedcostminimizingparametervalues.Thefuelcostsforthecontractcoalwillbethesamebothwithandwithoutascrubberandwillcancelout. BysolvingfortheconstantPizin( 2{59 )and( 2{60 )andsettingthetwoexpressionsequaltoeachother,thesignandvalueofcanbederived. 90


Intherstcase,withoutahighsulfurcoalcontract,aunitpreferstouselowsulfurspotmarketcoalwithoutascrubberandhighsulfurcoalwithascrubber.Thetwosubcaseswillbedeterminedbythegeneratingunit'scharacteristicsandthesizeofthecontractconstraint. Intherstsubcase,bothwithorwithoutahighsulfurcoalcontract,aunitpreferstouselowsulfurspotmarketcoalifitdoesnotinstallascrubberandhighsulfurspotmarketcoalitdoesinstallascrubber.Forthistohold,(MCAs;si

2{62 )and( 2{63 )canbesetequaltosolveforthevalueof: (bAMINibASMAXi)(2{65) Fillinginforallowancesandcoaluseparameters,itispossibletodeterminethesignof.AMINi=Di 2{65 ),theexpressionforcanbesimpliedtoparametersforcoalprice,sulfurcontent,heatcontent,andscrubbercapturerate. Aninterpretableformisderivedbyseparatingterms,multiplyingthoughby(bAMINibASMAXi),anddividingthroughbym(Ssih Fromtheinitialassumptionthataunituseslowsulfurcoalwhenitdoesnotinstallascrubber,itisknownthatPSA>MCAs;si.Soif(Scih 92


2-6 isbasedondatafromtheaunitthatinstalledascrubberunderPhaseI. 2-6 ,aunitwillprefertouselowsulfurcoalifitdoesnotinstallascrubberbecauseMCAs;si=$244:26andPA=$250.Ifascrubberisinstalled,aunitpreferstousehighsulfurspotmarketcoalbecausethemarginalcostofabatementincreasesabovetheallowancepricetoMeCAs;si=$4;885:20.TheannualizedscrubbercostsarePiz=$15:886million,whichresultsinPSA=$1;572:56.Ahighsulfurcoalcontractfor50%ofcoaluseresultsinadecreaseintheindierenceallowancepriceof=$1;210:61to(PA)=$382:00(a76%decreaseinPSA).Althoughaunit'scompliancechoicesarenotalteredatPA=$250,thelargedecreaseinencouragesscrubbingatmuchlowerallowancepricesthanwithoutthecoalcontract.Thissubcasewillholdforacontractconstraintoflessthan73.5%ofcoaluse.Thisexampleshowsthatformanyunits,highsulfurcoalcontractswouldnothaveenoughofanimpacttoresultinscrubberinstallationinPhaseI.Onlyabout10%ofunitsactuallyinstalledscrubberstocomplywithPhaseI,buttheseunitsaccountforoverhalfoftotalabatementbyaectedunits. Inthesecondsubcase,assumethatwithoutahighsulfurcoalcontract,aunitpreferstouselowsulfurspotmarketcoalifitdoesnotinstallascrubberandhighsulfurspotmarketcoalitdoesinstallascrubberbecause(MCAs;si

( 2{68 )showsthatthecostswithandwithoutascrubberareequalforagivenPSA,andcanberearrangedtondanexpressionforPiz: Sinceaunituseslowsulfurcoalwithoutascrubber,ahighsulfurcoalcontractwillchangethecoalusebyaunitwithoutascrubberandchangetheindierenceallowancepricebysomevalue: ( 2{70 )canberearrangedtondanexpressionforPiz: ThetwoexpressionsforPizin( 2{69 )and( 2{71 )canbesetequaltosolveforthevalueof: Byllinginforallowancesandcoaluse,itispossibletodeterminethesignof.AMINi=Di


2{72 ),theexpressionforcanbesimpliedtoparametersforcoalprice,sulfurcontent,heatcontent,andscrubbercapturerate: Thenewformin( 2{73 )iseasiertodeterminethesignofbecausethesignoftheleft-handsideisunchangedwhilesimplifyingtheright-handside.Nowgettheright-handsideintoaformthatisinterpretablebyseparatingtermsanddividingthroughbym(Ssih (Ssih Thelasttwotermsin( 2{74 )canbecombinedtogetDi(PSAMCAs;si): Fromourinitialassumptionthataunitusesalllowsulfurspotmarketcoalifitdoesnotinstallascrubber,ifitdoesnothaveahighsulfurcoalcontract,thenPSA>MCAs;si.If(Scih AssumingthedatainTable 2-6 ,thiscasewilloccurifthecontractisgreaterthan73.5%ofcoaluse.Forthiscasetooccur,ahighsulfurcoalcontractmustresultinaunitpreferringtousehighsulfurcoalwithoutascrubberinstalledwhen,withoutacontract,itinitiallypreferstouselowsulfurcoalwithoutascrubber(PSA>MCAs;siandPSA


2{77 )and( 2{79 )canbesetequaltosolveforthevalueof: (bAMAXibASMAXi)(2{80) Fillinginforallowancesandcoaluse,itispossibletodeterminethesignof.AMAXi=Di 2{80 ),theexpressionforcanbesimpliedtoparametersforcoalprice,sulfurcontent,heatcontent,andscrubbercapturerate: Ifhighsulfurcontractcoalhasaweaklyhighersulfurtoheatcontentratiothanhighsulfurspotmarketcoal(Scih Undertheinitialassumptions,PSA

(i) (ii) Thesignofisunknownif(Scih GiventheimpactahighsulfurcoalcontracthasonexcessdemanddenedinProposition2andthesignofdenedinProposition4,theimpactofahighsulfurcoalcontractisderivedinProposition5(a)and5(b). (i) FortherangeofallowancepricesPAMeCAs;si,ahighsulfurcoalcontractwillweaklyincreaseaunit'sexcessdemand. (ii) FortherangeofallowancepricesPSAPAMeCAs;si,ahighsulfurcoalcontractwillweaklyincreaseexcessdemand. (iii) Fortherangeofallowanceprices(PSA)PAPSA,ahighsulfurcoalcontractwillweaklydecreaseexcessdemandifAMINibASMAXiandweaklyincreaseexcessdemandifAMINibASMAXi. (iv) FortherangeofallowancepricesMeCAs;siPA(PSA),ahighsulfurcoalcontractwillweaklyincreaseaunit'sexcessdemand. (v) Fortherangeofallowanceprices0PAMCAs;si,ahighsulfurcoalcontractwillweaklyincreaseaunit'sexcessdemand. WhenaunitfacesPAMeCAs;si,aunitpreferstoinstallascrubberandusealllowsulfurcoal.FromProposition2(i),ahighsulfurcoalcontractincreasestheminimumemissionslevel,whichwillweaklyincreaseaunit'sallowanceexcessdemand. (ii) WhenaunitfacesPSAPAMeCAs;si,aunitpreferstoinstallascrubberanduseallhighsulfurcoal.FromProposition2(iii),given(Scih (iii) FromProposition4(i),whenaunitfaces(PSA)PAPSA,ahighsulfurcoalcontractdecreasesaunit'sindierenceallowancepriceofinstallingascrubber 98


(iv) Whenaunitfaces(Scih (v) Whenaunitfaces0PAMCAs;si,aunitdoesnotinstallascrubberandpreferstousehighsulfurcoal.FromProposition2(iii),given(Scih 2.8 (i).Proposition5(b)expressestheimpactahighsulfurcoalcontractwillhaveonaunit'sexcessdemandcorrespondencebyassumingthat>0. (i) FortherangeofallowancepricesPAMeCAs;si,ahighsulfurcoalcontractwillweaklyincreaseaunit'sexcessdemand. (ii) FortherangeofallowancepricesPSAPAMeCAs;si,ahighsulfurcoalcontractwillweaklydecreaseexcessdemand. (iii) Fortherangeofallowanceprices(PSA)PAPSA,ahighsulfurcoalcontractwillweaklydecreaseexcessdemand. (iv) FortherangeofallowancepricesMeCAs;siPA(PSA),ahighsulfurcoalcontractwillweaklyincreaseaunit'sexcessdemand. (v) Fortherangeofallowanceprices0PAMCAs;si,ahighsulfurcoalcontractwillweaklydecreaseexcessdemand. WhenaunitfacesPAMeCAs;si,aunitpreferstoinstallascrubberandusealllowsulfurcoal.FromProposition2(i),ahighsulfurcoalcontractincreasestheminimumemissionslevel,whichwillweaklyincreaseaunit'sallowanceexcessdemand. 99

PAGE 100

WhenaunitfacesPSAPAMeCAs;si,aunitpreferstoinstallascrubberanduseallhighsulfurcoal.FromProposition2(ii),given(Scih (iii) If>0,whenaunitfaces(PSA)PAPSA,ahighsulfurcoalcontractdecreasesaunit'sindierenceallowancepriceofinstallingascrubberbelowtheallowanceprice,whichleadstoaunitinstallingascrubberwhereitinitiallywouldnotanddecreasesaunit'semissionsandaunit'sexcessdemand.FromProposition2(ii),Giventhescrubberchoiceand(Scih (iv) FromProposition2(i),whenaunitfacesMeCAs;siPA(PSA),aunitdoesnotinstallascrubberandpreferstouselowsulfurcoal.Ahighsulfurcoalcontractincreasesaunit'sminimumemissionslevelandexcessdemand. (v) FromProposition2(ii),whenaunitfaces0PAMCAs;si,aunitdoesnotinstallascrubberandpreferstousehighsulfurcoal.Given(Scih 2.8 (ii). Thereareconditionsunderwhichsomeoftheseallowancepricerangesdonotexist.Forexample,PSAMCAs;sito(PA)
PAGE 101

Ifaunithasnolowsulfurcontractcoal( ( 2{82 )canberearrangedtondanexpressionforPiz: Alowsulfurcoalcontractwillchangeaunit'scoaluseandallowancepurchases,whichchangestheindierenceallowancepricebysomevalue: ( 2{84 )canberearrangedtondanexpressionforPiz: ThetwoexpressionsforPizin( 2{83 )and( 2{85 )canbesetequaltosolveforthevalueof: (bAMINibASMAXi)(2{86) Fillinginforallowancesandcoaluse,itispossibletodeterminethesignof.AMINi=Di

PAGE 102

2{86 ),theexpressionforcanbesimpliedtoparametersforcoalprice,sulfurcontent,heatcontent,andscrubbercapturerate. Amoreinterpretableexpressionisderivedbymultiplyingthroughby((bAMINibASMAXi))anddividingthroughbym(Ssih (Ssih (m)(Ssih (Ssih Fromtheinitialassumptionthataunituseslowsulfurcoalwhenitdoesnotinstallascrubber,itisknownthat(PSA>MCAs;si).Since(Scil AbindinglowsulfurcoalcontractismorelikelytoimpactunitsunderCAIR.AnexampleusingrecentdatareectiveofthecoalavailabilityanddeliveredpricesforaunitinAlabamain2000willhelptoshow<0. 102

PAGE 103

PAGE 104

2{91 )and( 2{93 )canbesetequaltosolveforthevalueof: Byllinginforallowancesandcoaluse,itispossibletodeterminethesignof.AMAXi=Di 2{94 ),theexpressionforcanbesimpliedtoparametersforcoalprice,sulfurcontent,heatcontent,andscrubbercapturerate:)(bAMINibASMAXi)=mPSAHcilCcilh(Ssih Aninterpretableformisderivedbymultiplyingthroughby(bAMINibASMAXi)anddividingthroughbym(Ssih (Ssih (Ssih Fromourinitialassumptionthataunitusesallhighsulfurspotmarketcoalifitdoesnotinstallascrubberifitdoesnothaveahighsulfurcoalcontract,thenPSA
PAGE 105

ThesecondexampleinTable 2-8 usesdatathatreectsthiscase.Withoutacoalcontract,PSA=$520:35,whichislessthanMCAs;si=$657:90andPA=$700:00,andaunitwillusehighsulfurcoalbothwithandwithoutascrubberinstalled.Inthiscase,aunitwillinstallascrubberbecausePSA
PAGE 106

2{99 )and( 2{101 )canbesetequaltosolveforthevalueof: (bAMAXibASMAXi)(2{102) Fillinginforallowancesandcoaluse,itispossibletodeterminethesignof.AMAXi=Di 2{102 ),theexpressionforcanbesimpliedtoparametersforcoalprice,sulfurcontent,heatcontent,andscrubbercapturerate. Since(Scil ThethirdexampleinTable 2-8 usesdatareectiveofdeliveredcostsandcoalcharacteristicsforaunitinFlorida.Aunitinitiallypreferstousehighsulfurcoalbothwithandwithoutascrubberbecause(PA=$600:00)<(MCAs;si=$3;887:56).AunitwillprefertoinstallascrubberinthisexamplebecausetheindierentpriceisPSA=$565:60.Alowsulfurcoalcontractfor50%ofcoalusewillincreasetheindierencepriceto(PSA)=$642:42,whichwillresultinaunitnotinstallingascrubber. Basedontheabovethreecases,thesignofepsiloncanbesummarizedinProposition6. GivenProposition3andProposition6,theimpactofalowsulfurcoalcontractisderivedinProposition7(a)and7(b). 106

PAGE 107

FortherangeofallowancepricesPAMeCAs;si,alowsulfurcoalcontractwillweaklyincreaseexcessdemand. (ii) Fortherangeofallowanceprices(PSA)PAMeCAs;si,alowsulfurcoalcontractwillweaklydecreaseexcessdemand. (iii) FortherangeofallowancepricesPSAPA(PSA),alowsulfurcoalcontractwillweaklyincreaseexcessdemand. (iv) FortherangeofallowancepricesMeCAs;siPAPSA,alowsulfurcoalcontractwillweaklyincreaseexcessdemand. (v) Fortherangeofallowanceprices0PAMCAs;si,alowsulfurcoalcontractwillweaklydecreaseexcessdemand. WhenaunitfacesPAMeCAs;si,aunitpreferstoinstallascrubberandusealllowsulfurcoal.FromProposition3(iii),givenScil (ii) Whenaunitfaces(PSA)PAMeCAs;si,aunitpreferstoinstallascrubberanduseallhighsulfurcoal.FromProposition3(i),alowsulfurcoalcontractdecreasesthemaximumemissionslevel,whichwillweaklydecreaseaunit'sallowanceexcessdemand. (iii) FromProposition6,whenaunitfacesPSAPA(PSA),alowsulfurcoalcontractincreasesaunit'sindierenceallowancepriceofinstallingascrubberabovetheallowanceprice,whichleadsaunittonotinstallascrubberwhereitinitiallywouldhavedonesoandincreasesaunit'semissionsandaunit'sexcessdemand.FromProposition3(iii),giventhescrubberchoiceandScil (iv) WhenaunitfacesMeCAs;siPAPSA,aunitdoesnotinstallascrubberandpreferstouselowsulfurcoal.FromProposition3(iii),givenScil (v) Whenaunitfaces0PAMCAs;si,aunitdoesnotinstallascrubberandpreferstousehighsulfurcoal.FromProposition3(i),alowsulfurcoalcontractdecreasesaunit'smaximumemissionslevelandexcessdemand. 2.8 (i). 107

PAGE 108

FortherangeofallowancepricesPAMeCAs;si,alowsulfurcoalcontractwillweaklydecreaseexcessdemand. (ii) Fortherangeofallowanceprices(PSA)PAMeCAs;si,alowsulfurcoalcontractwillweaklydecreaseexcessdemand. (iii) FortherangeofallowancepricesPSAPA(PSA),alowsulfurcoalcontractwillincreaseexcessdemandASMAXibAMINiandweaklydecreaseexcessdemandifASMAXibAMINi. (iv) FortherangeofallowancepricesMeCAs;siPAPSA,alowsulfurcoalcontractwillweaklydecreaseexcessdemand. (v) Fortherangeofallowanceprices0PAMCAs;si,alowsulfurcoalcontractwillweaklydecreaseexcessdemand. WhenaunitfacesPAMeCAs;si,aunitpreferstoinstallascrubberandusealllowsulfurcoal.FromProposition3(ii),givenScil (ii) Whenaunitfaces(PSA)PAMeCAs;si,aunitpreferstoinstallascrubberanduseallhighsulfurcoal.FromProposition3(i),alowsulfurcoalcontractdecreasesthemaximumemissionslevel,whichwillweaklydecreaseaunit'sallowanceexcessdemand. (iii) FromProposition6,whenaunitfacesPSAPA(PSA),alowsulfurcoalcontractincreasesaunit'sindierenceallowancepriceofinstallingascrubberabovetheallowanceprice,whichleadsaunittonotinstallascrubberwhereitinitiallywouldhavedonesoandincreasesaunit'semissionsandaunit'sexcessdemand.FromProposition3(ii),giventhescrubberchoiceandScil (iv) WhenaunitfacesMeCAs;siPAPSA,aunitdoesnotinstallascrubberandpreferstouselowsulfurcoal.FromProposition3(ii),givenScil (v) Whenaunitfaces0PAMCAs;si,aunitdoesnotinstallascrubberandpreferstousehighsulfurcoal.FromProposition3(i),alowsulfurcoalcontractdecreasesaunit'smaximumemissionslevelandexcessdemand.

PAGE 109

2.8 (ii).Aswithahighsulfurcontract,someofthepricerangesmaynotexistforaparticularcase.However,theremainingpartsofthepropositionshold. Ashasbeenshowinseveralexamples,highsulfurcoalcontractsforalargefraction(50-100%)ofcoalusecangreatlyreduceaunit's\indierenceprice"toinstallingascrubber.SomeunitsunderPhaseIinitiallyappeartohaveinstalledascrubberwhenitwasnotaunit'soptimalcomplianceoption,increasingaunit'scompliancecosts.Additionalscrubberinstallationsshouldhaveresultedingreateremissionsreduction,whichshouldsimultaneouslylowerdemandandincreasesupplyofallowancesasaunitswitchesfromanetdemandertoanetseller.Indoingso,theequilibriumallowancemarketpriceshouldbedrivenlower,whichmayexplainthelowerthanexpectedallowancepricesrealizedduringPhaseI.Eventhoughtheallowancemarketpricewaslowerthanexpected,theinecientunitcompliancechoicesresultedinhigherthanexpectedtotalindustrycompliancecosts. UnderfutureCAIRregulation,aunit'scomplianceoptionsmayberestrictedbylowsulfurcoalcontractsagreeduponduringthe1990stomeetTitleIVemissionsrequirements.Aunitmayndinstallingascrubberandusinghighsulfurcoaltobeitbestcomplianceoption.However,lowsulfurcoalcontractsmayleadaunittochooseasuboptimalcompliancechoice,suchasswitchingfuelsorinstallingascrubberwhileusing 109

PAGE 110

Ashasbeenshowinseveralexamples,lowsulfurcoalcontractsforalargefraction(50-100%)ofcoalusecangreatlyincreaseaunit's\indierenceprice"toinstallingascrubber.SomeunitsunderCAIRmaynotinstallascrubberwhenitisoptimalforthemtodoso,increasingaunit'scompliancecosts.Fewerscrubberinstallationswouldresultingreateremissions,whichshouldsimultaneouslyincreasedemandanddecreasesupplyofallowancesasaunitwouldbeanetdemanderinsteadofanetseller,andtheallowancemarketpriceshouldbedrivenhigherthanwouldbeexpected.Inthiscase,ahigherthanexpectedallowancemarketpricewouldoccurwithhigherthanexpectedtotalindustrycompliancecosts. First,giventhescrubberchoice,itiseasytodeterminehowacoalcontractwillimpactaunit'sexcessdemandforallowances.Ahighsulfurcoalcontractwillweaklyincreaseaunit'sexcessdemandwhilealowsulfurcoalcontractwillweaklydecreaseaunit'sexcessdemand. Second,somecoalcontractsmayactuallydecreaseaunit'stotalcostsrelativetousingonlyspotmarketcoalwhileincreasingaunit'scompliancecostsifacoalcontractallowsaunittolockinalowerpriceforagiventypeofcoal.Thecoalcontractwillrestrictcompliancechoices,whichmayresultinhighercompliancecosts.Ifthefuelcostsavingsisgreaterthantheincreaseinaunit'scompliancecosts,thenthecoalcontractlowersaunit'stotalcosts.Sinceaunitonlycaresaboutitstotalcosts,aunit'scompliancedecisionsdonotnecessarilyminimizeaunit'scompliancecosts. 110

PAGE 111

Fourth,coalcontractconstraintschangetheallowancepriceatwhichaunitwillprefertoinstallascrubber.InthecaseofPhaseI,ahighsulfurcoalcontractmayincreaseaunit's\indierenceprice",whichcreatesagreaterincentiveforaunittoscrubandsellitsextraallowances.Thisresultmayexplainwhysomescrubberswereinstalledatsub-optimalunitswhiletheallowancemarketpricewasmuchlowerthanexpectedduringPhaseI.TheoppositemayoccurunderCAIR,wherealowsulfurcoalcontractmayincreaseaunit's\indierenceprice"andlowertheincentiveforaunittoinstallascrubberevenifitwouldbetheoptimalcompliancechoice.Ineithercase,asuboptimalcompliancechoicewillbemadeandaunit'scompliancecostswillweaklyincrease. Fifth,thereiscertaintyhowacoalcontractwillimpactaunit'sexcessdemandforallowancesevenwhenthescrubberchoiceisconsideredformostallowancepriceranges.However,duetothediscretescrubberchoiceandthechangeinthe\indierenceprice"duetoacoalcontract,itisuncertainhowacoalcontractwillalteraunit'sexcessdemandiftheallowancemarketpricefallsinoneparticularpricerange.AhighsulfurcoalcontractwillshiftexcessdemandasderivedinProposition2exceptiftheallowancemarketpricefallsinthepricerange(PSA;PSA)andScih 111

PAGE 112


PAGE 113

PhaseIComplianceCostEstimates SourceMCAi/tonPred.CostsLeast-CostActualCost ICF(1989,1990)$199-$226$450-860millionN/AN/AEPRI(1993,1995)$879-$1238$900-1,340millionN/AN/AGAO(1994)$299$1,170millionN/AN/AWinebrake,etal.(1995)$143$502millionN/AN/APre-PolicyEstimates$143-$1238$450-$1340millionN/AN/AEllermanetal.(1997)$200-300N/AN/A$730millionSotkiewiczandHolt(2005)$150-$180N/A$423-$553million$990millionCarlsonetal.(2000)$71N/A$571million$910millionPost-PolicyEstimates$71-$800N/A$423-$571million$730-$990million Sources:BohiandBurtraw(1997);Carlsonetal.(2000);SotkiewiczandHolt(2005);SmithandEllerman(1998)

PAGE 114

HighSulfurCoalContract:Assumptions CoalTypePrice/mmBtuHgifPrice/TonSgif Table2-3. HighSulfurCoalContract:Results CostsComplianceTotalCostsComplianceTotalCostsCosts(Ex.1)(Ex.1)Costs(Ex.2)(Ex.2) Unconstrained$4.62million$35.82million$4.62million$35.82millionConstrained$5.01million$38.61million$5.01million$35.01millionChange$390,000$2.79million$390,000-$810,000 AllowanceUse(Tons)MCAi Table2-4. LowSulfurCoalContractExamples:Assumptions CoalTypePrice/mmBtuHgifPrice/TonSgif 114

PAGE 115

LowSulfurCoalContractExamples:Results CostsComplianceCosts(Ex.1)TotalCosts(Ex.1)ComplianceCosts(Ex.2)TotalCosts(Ex.2) Unconstrained$3.6million$34.8million$3.6million$34.8millionConstrained$4.54million$38.14million$4.54million$34.54millionChange$940,000$3.34million$940,000-$260,000 AllowanceUse(Tons)MCAi

PAGE 116

ExampleEpsilonMagnitude:Case1 ScrubberCharacteristicsGeneratingUnitCharacteristics CapitalCost$260/kWCapacity365.3MWOandMCost2mills/kWhHeatRate10,000Btu/kWhPiz$15,886,295Cap.Factor75%ri95% ParametersCharacteristicsValueResults LowSulfurCoalPrice$1.50/mmBtu50%Contract(Spot)HeatContent24mmBtu/tonPSA=$1,592.61=$1,210.61SulfurContent0.65%PSA=$382.00Pct.#PSA=76%HighSulfurCoalPrice$1.30/mmBtu75%Contract(Spot)HeatContent24mmBtu/tonPSA=$1,592.61=$1,349.82SulfurContent2.5%PSA=$242.79Pct.#PSA=85%HighSulfurCoalPrice$1.50/mmBtu(Contract)HeatContent24mmBtu/tonSulfurContent4.0%

PAGE 117

ExampleEpsilonMagnitude:Case2 ScrubberCharacteristicsGeneratingUnitCharacteristics CapitalCost$260/kWCapacity365.3MWOandMCost2mills/kWhHeatRate10,000Btu/kWhPiz$15,886,295Cap.Factor75%ri95% ParametersCharacteristicsValuesResults LowSulfurCoalPrice$2.79/mmBtuBLEngland(Spot)HeatContent24mmBtu/ton50%ContractSulfurContent0.5%PSA=$460.80=$44.66HighSulfurSpotCoalPrice$1.52/mmBtuPA=$416.14Pct.#PSA=10%(Spot)HeatContent24mmBtu/tonSulfurContent1.91%HighSulfurCoalPrice$1.72/mmBtu(Contract)HeatContent24mmBtu/tonSulfurContent2.32%

PAGE 118

ExampleEpsilonMagnitude CapitalCost$141.34/kWCapacity365.3MWOandMCost1.23mills/kWhHeatRate10,000Btu/kWhPiz$9,016,616.85Cap.Factor75%ri95% ParametersCharacteristicsValuesResults LowSulfurCoalPrice$1.50/mmBtuCase1(Alabama)(Alabama)HeatContent24mmBtu/tonPSA=$731.46=-$42.00SulfurContent0.7%PA=$773.46Pct:"PSA=6%HighSulfurSpotCoalPrice$1.10/mmBtu(Alabama)HeatContent24mmBtu/tonSulfurContent1.6%HighSulfurSpotCoalPrice$2.00/mmBtuCase2(CAIR)(CAIRExample)HeatContent25mmBtu/tonPSA=$520.35=-$352.99SulfurContent1.0%PA=$873.34Pct."PSA=68%LowSulfurCoalPrice$2.20/mmBtu(CAIRExample)HeatContent25mmBtu/tonSulfurContent0.6%HighSulfurSpotCoalPrice$2.30/mmBtuCase3(Florida)(Florida)HeatContent24mmBtu/tonPSA=$565.60=-$76.82SulfurContent1.4%PA=$642.42Pct:"PSA=14%LowSulfurCoalPrice$3.50/mmBtu(Florida)HeatContent25mmBtu/tonSulfurContent0.7%

PAGE 119

TheSO2AllowancePrice.DatawasgiventomebyDallasBurtrawofResourcesfortheFuture Figure2-2. ExcessDemandCorrespondenceandComplianceCostSavingsfromFuelSwitchingOverAllowancePurchasing 119

PAGE 120

HighSulfurContract:ShiftinMinimumExcessDemand Figure2-4. NoContract:ComplianceCosts Figure2-5. HighSulfurContract:ComplianceandTotalCosts 120

PAGE 121

HighSulfurContract:RelativeSavingsfromContractCoal Figure2-7. CostSavingsfromUsingAllowancesOverFuelSwitching Figure2-8. LowSulfurContract 121

PAGE 122

NoContract:ComplianceCosts Figure2-10. LowSulfurCoalContract:MCAs;ci LowSulfurCoalContract:MCAs;ci

PAGE 123

ComplianceCosts:PA=MCAs;si HighSulfurCoalContract Figure2-14. LowSulfurCoalContract 123

PAGE 124

PAGE 125

ImpactofaHighSulfurCoalContract:MCAs;siPSA ImpactofaLowSulfurCoalContract:MCAs;si>PSA

PAGE 126

PAGE 127

Giventhescrubberchoice,anallowancemarketequilibriumwillexist.Allowingforthescrubberchoicemakesitimpossibletoguaranteeanequilibrium,butonestillmayexist.Bindingfuelcontractsmayleadtoalteredunit-levelexcessdemandsand,insodoing,theallowancemarketprice(PA).Meanwhile,bindingcontractscanaltercompliancedecisionsandincreasetotalindustrycompliancecosts. Thispaperusesgeneratingunit-levelsimulationstoreplicateresultsfrompreviousstudiesandshowthatshort-runfuelcontractsappeartoexplainalargeportionofthepreviouslyunexplainedexcesscompliancecostsfoundinprevioussimulations.Simulatingtheleast-costcompliancechoiceswithoutincludingfuelcontractconstraintsresultsinminimumannualindustrycompliancecostsof$288.3million,whichvariesgreatlyfromtheactualcompliancecostsof$1.30billionfoundinthesesimulations.Oncefuelcontractconstraintsareintroducedintothesimulation,theminimumannualindustrycompliancecostsbecome$1.01billion.Basedontheseresults,fuelcontractconstraintsexplain$651.1million,or64%oftheexcesscompliancecostsrealizedintheprogramfor1996. 127

PAGE 128

Thepaperwillbestructuredinthefollowingmanner:Section2willlookattheconditionsforanallowancemarketequilibriumwhileSection3willlookatthecomparativestaticsoftheallowancemarket.Section4willlookatcompliancecosts,bothforanindividualgeneratingunitandtheentireindustry.Section5andSection6willexplainthedatausedinthegeneratingunitlevelsimulationsandtheresultsfromtheactualsimulations,respectively.Section7willanalyzetheplantleveldecisionmakingprocess,howitdiersfromthegeneratingunitlevelmodel,andwhatadditionalinformationthemodelwilladdtotheliterature. 3.2.1GeneratingUnitProblem subjectto...Aei+Ai(1ziri)(m)(CsihSsih+CcihScih+CsilSsil+CcilScil)i1 (CsihHsih+CcihHcih+CsilHsil+CcilHcil)Dii2 128

PAGE 129

(3{6) Equation( 3{1 )representstheunit'scostfunction.Thesecostsincludethecostofscrubberinstallation(ziPiz),netcostsofallowancepurchases(PAAi),andcostsofcoalpurchases(PsihCsih+PsilCsil+PcihCcih+PcilCcil).Theemissionsconstraintisshownin( 3{2 ),wherethenumberofallowancesheld(Aei+Ai)mustbeaslargeastheamountoftotalemissionsbythegeneratingunit[(1ziri)(m)(CsihSsih+CcihScih+CsilSsil+CcilScil)].Totalemissionsisafunctionoftheamountofeachcoaltypeusedaswellastheemissionsreductionduetoascrubber,ifoneisinstalled.TheLagrangemultiplierontheemissionsconstraintisrepresentedbyi1.Thedemandconstraintrequiresthattheamountofheatinputtogenerateelectricity(CsihHsih+CcihHcih+CsilHsil+CcilHcil)mustcovertheconsumerdemand(Di)forelectricityexpressedasheatinput,whichisseenin( 3{3 ).TheLagrangemultiplieronthedemandconstraintisrepresentedbyi2.Coalcontractconstraintsrequiretheunittouseaspecicamountofeachcontractcoaltype, 3{4 )and 3{5 ).Aunitwilluseexactlythecontractedamountbecause(1)ifthecontractcoalismoreexpensivethanspotmarketcoal,thenaunitwillnotwanttouseanymorecontractcoalthanisnecessaryand(2)ifcontractcoalischeaperthanspotmarketcoal,thecoalproducerwouldprefertosellanyadditionalnon-contractedcoalthroughthespotmarket.TheLagrangemultiplierforeachcontractconstraintoneachcoaltypeisrepresentedbyihforhighsulfurcontractcoalandilforlowsulfurcontractcoal. 129

PAGE 130

Introducingahighsulfurcoalcontractconstraintwillrestrictspotmarketcoaluseby Introducingalowsulfurcoalcontractconstraintwillrestrictspotmarketcoaluseby 130

PAGE 131

Allowingthescrubberchoicetobeendogenousresultsinmorecompleximpactsofcoalcontractsoncompliancedecisions.Ageneratingunitthatdoesnotfaceanycoalcontractconstraintswillmakeitsoptimalchoicesbasedontwoconditions:(1)therelationshipofPArelativetotheallowancepriceatwhichaunitisindierenttoinstallingascrubber(PSA)and(2)therelationshipofPArelativetoMCAs;sigiventhescrubberchoice. IfPSA>PA,aunitwillnotinstallascrubberbecausetotalcoststothegeneratingunitwillbelowerwithoutinstallingascrubber.Theresultingcoaluseandexcessdemandwillbethesameasdescribedabove:Csih=Di,Csil=0,andAi=AMAXiifPAMCAs;siwhereMCAs;siisthemarginalcostofabatementfromswitchingfuelswithoutascrubber. IfPSAMeCAs;si. Theresultisacombinationofthetwoexcessdemandcorrespondenceswhereatsomeallowanceprice(PSA)whereascrubberwillbeinstalledthereisalargenon-continuousdecreaseinaunit'sexcessdemand. Introducingahighsulfurcoalcontractconstraintwillhavethesamedirectimpactsaswhenscrubbersaregiven.Coalusewillberestrictedby 131

PAGE 132

Introducingalowsulfurcoalcontractconstraintwillhavethesamedirectimpactsaswhenscrubbersaregiven.Coalusewillberestrictedby Insummary,coalcontractswillrestrictaunit'scoaluseoptions,whichmayalteraunit'scompliancechoicesandexcessdemandforallowances.Lessfreedomincompliancedecisionsmayresultinanincreaseinaunit'scompliancecosts. 132

PAGE 133

minzi;Ai;Csih;CsilziPiz+PAAi+PsihCsih+PsilCsil+PcihCcih+PcilCcil subjectto...Aei+Ai(1ziri)(m)(CsihSsih+CcihScih+CsilSsil+CcilScil) (3{9) (CsihHsih+CcihHcih+CsilHsil+CcilHcil)Di (3{13) (3{15) (2)Theallowancemarketclears.AtPA, FromKakutani'sFixedPointTheorem,axedpointexistsifthemarketexcessdemandcorrespondenceuppersemi-continuous,compact-valued,andconvex-valued.Eachgeneratingunit'sexcessdemandcorrespondencesareuppersemi-continuous,compact-valued,andconvex-valued,whichimpliesthatthemarketexcessdemandisalsouppersemi-continuous,compact-valued,andconvex-valued.Giventheseconditionsthereisaxedpoint,andtherefore,anequilibriumexists.Notethattheremaybemultipleequilibria,butthereisatleastoneequilibrium.TheProofofTheorem1isinAppendixB. Theexistenceofanallowancemarketequilibriumisnolongerguaranteedoncethediscretescrubberchoiceisintroducedintothemodel.Thenon-convexnatureofthescrubberdecisionvariablemakesitimpossibletoguaranteethatanequilibriumexistsbecausetheexcessdemandisnolongerconvexfortherange(ASMAXi;AMINi).Insuch 133

PAGE 134

3.4.1ComparativeStatics:EectofRelativeFuelCostontheAllowanceMarket 3{17 ). ConsidertheimpactofanincreaseinthepriceoflowsulfurspotmarketcoalontheMCAs;si.Anincreaseinthepricedierencebetweenfuelsperunitofheatfrom(Psil 3{17 )andresultsinanincreaseinthemarginalcostofabatementfromMCAs;sitoMCAs;si+where= Hsil HsihSsil Hsil).AhigherMCAs;simakestherelativecosttothegeneratingunitpurchasingallowanceslower,andwillleadtoaweakincreaseinaunit'sexcessdemand.ExcessdemandwillincreaseifMCAs;siPAandnotchangeifMCAs;si+PA.Greaterexcessdemandintheallowancemarketresultsinanincreaseinthetotalallowancemarketdemandandpossiblyadecreaseinallowancemarketsupply,bothofwhichwillweaklyincreasePA. 134

PAGE 135

IfPAMCAs;si,ageneratingunitpreferstoswitchfuelsfromhightolowsulfurcoal,createitsminimumpossibleemissions,andhaveanexcessdemandofAMINi.However,thehighsulfurcoalcontractwouldforcegreaterthantheminimumamountofemissionsandweaklyincreaseexcessdemandtobAMINi.Ahigherexcessdemandwillleadtoadecreaseintheallowancemarketsupplyorbothadecreaseinallowancemarketsupplyandanincreaseintheallowancemarketdemand.InbothcasesPAweaklyincreases. 135

PAGE 136

IfPA>MCAs;si,ageneratingunitpreferstoswitchfuelsfromhightolowsulfurcoal,createitsminimumpossibleemissions,andhaveanexcessdemandofAMINi.Assumingthesamerelativesulfurcontentforspotandcontractcoal,alowsulfurcoalcontractwillnotchangetotalemissionsorexcessdemand(bAMINi=AMINi)andPAremainsunchanged. IfPA
PAGE 137

Whenyouconsiderthescrubberchoiceinthedecision-makingprocess,itisuncertainhowacoalcontractwillaecttheallowancemarketbecausethecontractmayincreaseordecreaseexcessdemanddependingontheallowancepricerangesderivedinChapter2.Althoughitiscertainthatifthecoalcontractbinds,thentherewillbeashifttoasub-optimalexcessdemandforthreedierentallowancepriceranges. Ifageneratingunithasahighsulfurcoalcontract,theunit'sexcessdemandincreasesfortwoallowancepriceranges:(PA>MeCAs;si)and(PSA>PA>MCAs;si).TheincreaseinexcessdemandwillbefromASMINitobAMINiwhen(PA>MeCAs;si).TheincreaseinexcessdemandwillbefromAMINitobAMINiorAMINitoAMAXiwhen(PSA>PA>MCAs;si),dependingontherelationshipbetween(PSA)andMCAs;si.Itwillbethelatterinthespecialcasewherethegeneratingunitalwayspreferstousehighsulfurcoal. Whenthemarketallowancepriceisinthesetworangesthatresultinanincreaseinageneratingunit'sexcessdemand,therewillbeeitheranincreaseinthemarketdemandforallowances,orbothanincreaseinmarketdemandandadecreaseinmarketsupplyforallowances.Inbothsituations,theallowancepricewillbedrivenhigher. Thethirdpricerangethathasashiftintheexcessdemandis(PSA;PSA).AgeneratingunitdecreasesitsexcessdemandfromAMINitoASMAXi,whichisaresultofaunitinstallingascrubberforapricerangeforwhichitinitiallywouldnotinstallascrubber. 137

PAGE 138

Ifageneratingunithasalowsulfurcoalcontract,theunit'sexcessdemanddecreasesfortwoallowancepriceranges:(MeCAs;si>PA>PSA+)and(minfPSA;MCAs;sig>PA).ThedecreaseinexcessdemandwillbefromASMAXitobAMAXiwhen(MeCAs;si>PA>PSA+).ThedecreaseinexcessdemandwillbefromAMAXitobAMAXiwhen(minfPSA;MCAs;sig>PA). Whenthemarketallowancepriceisintheserangesthatresultinadecreaseinageneratingunit'sexcessdemand,therewillbeeitherandecreaseinthemarketdemandforallowances,orbothandecreaseinmarketdemandandaincreaseinmarketsupplyforallowances.Inbothsituations,PAwillbedrivenlower. Thethirdpricerangethathasashiftintheexcessdemandis(PSA;PSA+).AgeneratingunitincreasesitsexcessdemandfromeitherASMAXitoAMINiorASMAXitobAMAXidependingontherelationshipbetween(PSA+)andMCAs;si.Thisincreaseisaresultofaunitnotinstallingascrubberforapricerangeforitinitiallywouldhaveinstalledascrubber. Whenthemarketallowancepriceisinthisrangethatresultsinanincreaseinageneratingunit'sexcessdemand,therewillbeeitheranincreaseinthemarketdemandforallowances,orbothanincreaseinmarketdemandandadecreaseinmarketsupplyforallowances.Inbothsituations,PAwillbedrivenhigher. 138

PAGE 139

3{18 )). Compliancecostsmaybepositiveornegativedependingonitscompliancedecisionsanditsinitialallowanceallocation.ThescrubberinstallationcostsarerepresentedbyPiz,andwillonlyattributetoaunit'scompliancecostsifascrubberisinstalledinresponsetotheprogram(zi=1).Thecostsofaunit'snetallowancepositionisthedierencebetweenageneratingunit'sinitialallowanceallocationanditsactualemissionsmultipliedbytheallowanceprice(PAAi).Thecostsofswitchingfuelsisthelargeroftwovalues:(1)totalcostsofactualcoalpurchases(PsihCsih+PsilCsil+Pcih 139

PAGE 140

(bziPiz+PAbAi+PsilbCsil+PsihbCsih+Pcih Theminimumcompliancecostchoiceswithnocontractconstraintforpreviousstudiesaredenotedby\"whilethecontractconstrainedminimumcompliancecostchoicesaredenotedby\^"in( 3{20 ).Theassumed\c"isthetotalcostsofmeetingdemandassumingnocontractconstraintoremissionsconstraint.Thecostdierencebetweenthesetwosetsofchoiceswillbethecombinationofchangesincompliancedecisionsandfuelcosts.Someofthese\excesscompliancecosts"maynotbecompliancecosts,whichmakesthederivedcompliancecoststoohigh.Soitisimportanttoincludethecontractconstrainedcoaluseinthebaseline\c"togettheappropriatecompliancecosts.Assumingthatcontractedcoalisatleastasexpensiveasspotmarketcoal,thecompliancecostswillbegreaterforthecontractconstrainedcaseasshownin( 3{20 )andasexpressedin( 3{21 ). Considerasimpleexamplewherehighsulfurcontractcoalismoreexpensivethanhighsulfurspotmarketcoal,lowsulfurcoalispreferredoverhighsulfurcoal,andnoscrubberwouldbeinstalledifageneratingunithadtotalfreedominitscoalusechoices.Nowcompareageneratingunit'scompliancecoststothecompliancecostsaunitwould 140

PAGE 141

3{22 )includetheadditionalcostsfromanincreaseinallowancepurchasesandadecreaseincoalcostsfromusingcheaper,butdirtierfuel. Sincethecontractcoalwouldhavetobeusedregardlessofenactmentoftheprogram,theadditionalcostsforhighsulfurcontractcoaluseinsteadofhighsulfurspotmarketcoal(PcihCc;MAXihPsihCs;MAXih)shouldnotbeincludedincompliancecosts.However,previousstudieswouldrecognizethesefuelcostsascompliancecostsbasedontheassumedcostminimizationwithoutcontractconstraints. Theappropriateunit-levelcompliancecostswilltaketheformshownin( 3{23 )wherethebaselinetotalcostswilltakethevalue\bc",whichrepresentsthecostsofmeetingelectricitydemandgiventhecoalcontractconstraint.Thesecosts\bc"willbedierentthan\c"becausethespotmarketcoalusewillbealteredifthecontractedcoaldiersrelativethespotmarketcoalitreplacesinproduction. 3{25 ). 141

PAGE 142

3{26 )arethesumofthecompliancecostsforeachgeneratingunit.Similartothegeneratingunit'sprobleminChapter2,therearetwoconstraintsthatmustbemet.First,thesumofallowanceallocationsmustcoverthesumofemissionsthatareemittedbyallaectedgeneratingunits.Second,eachgeneratingunitmustproduceenoughelectricitytocoveritselectricitydemandrequirement. minzi;Ai;Csih;CsilnXi=1ziPiz+maxf(PsilCsil+PsihCsihPsihCs;MAXih);0g Solvefortherst-orderconditionsgivenzi.Forhighsulfurcoal: Forlowsulfurcoal: 142

PAGE 143

Theserst-orderconditionscanbeusedtosolvefori1,theallowanceshadowprice,whichisalsothesameasforthegeneratingunitproblem.SincethecompliancechoicesthatminimizethecostsforallgeneratingunitsresultintheequilibriumPA,theallowancechoicesalsominimizethetotalindustrycosts. (3{29) Thesameapproachcanbeusedundertheindustry-widecompliancecostproblemwithcoalcontractconstraintstoshowthatcontractconstrainedminimumtotalindustrycompliancecostwilldierfromtheunconstrainedcosts. minzi;Ai;Csih;CsilnXi=1[ziPiz+maxf(PsilbCsil+PsihbCsihPsihbCs;MAXih);0g] (3{30) Forlowsulfurspotmarketcoal... 143

PAGE 144

(3{33) Theserst-orderconditionscanbeusedtosolvefori1,theallowanceshadowprice,whichisalsothesameasfortheindividualgeneratingunitproblem.Ifageneratingunituseshighsulfurcoal,PAMCAs;si.Ifageneratingunituseslowsulfurcoal,PAMCAs;si.Ifageneratingunitusesbothhighandlowsulfurcoal,PA=MCAs;si.Soageneratingunit'scostminimizingchoicesofzi;Csih;Csil;Asiforeachgeneratingunitalsominimizetotalindustrycompliancecosts.IfthecontractconstrainedcompliancechoicesthatminimizethecostsforageneratingunitresultsinanequilibriumPA,thechoicesalsominimizethecontractconstrainedtotalindustrycosts.Notethattheoptimalparametercombinationforthecontractconstrainedcasewillnotbethesameasundertheunconstrainedcase. Ithasalreadybeenshownthatanybindingcontractconstraintforanindividualgeneratingunitwillresultinasub-optimalcombinationofcoaluse,whichwillweaklyincreasecompliancecostsforthecontractconstrainedgeneratingunit.Anysub-optimalchoicesmadebyonegeneratingunitwillweaklyincreasethecostsfortheentireindustry. 144

PAGE 145

Theimportanceoftheallowanceallocationinthesepreviousstudiesreliesontheexistenceoftransactioncosts.However,thismodelhasassumednoallowancetransactioncosts,whichallowsittoshowthatevenwithcoalcontractconstraintstheallowanceallocationdistributionwillnotimpactageneratingunit'scompliancechoicesorthetotalindustrycompliancecosts. Thechoicebetweenswitchingfuelsorpurchasingallowancesisbasedsolelyonageneratingunit'srelativemarginalcostofpurchasingthenextallowancecomparedtotheeectivemarginalcostofabatingthenextunitofemissions.Ascanbeseen,Aeihasnoimpactonageneratingunit'scompliancedecisions.Ageneratingunitwilluseallowances 145

PAGE 146

(3{36) (bAieAi)PSA (cEi+AeifEiAei)PSA(3{37) Nowconsiderthecompliancechoicesmadeafterthescrubberinstallationdecision.ItwasshowninChapter2thatageneratingunit'scompliancechoicesarebasedontherelationshipbetweenthePAandMCAs;siorMeCAs;si,dependingontheunit'sscrubberchoice,foraparticularunitofcoal.Similartowhatwasshownabove,ageneratingunit'sMeCAs;siandMCAs;siforagivenunitofcoalarenotaectedbyAeibecauseallowancesarenotaparameterinthemarginalcostofabatement.MeCAs;si=Pil 146

PAGE 147

3.6.1Introduction Therstportionofthissectionwilllookatthedatausedtoparameterizethesimulationmodel,whichincludesadescriptionofwherethedatawasobtained,thetechniquesusedtocreatetheparameters,andsomeissuesregardingthedata.Thesecondportionwilllookatthemodeldesignandapproach.Thelastportionwillsummarizethesimulationresultsintermsofthetotalindustry,individualstates,andindividualgeneratingunits. Uptothispointallgeneratingunitsareassumedtobecoal-redunitsbecauseitistheprimaryfueloptionsforelectricitygenerationintheU.S.However,thereare24unitsofthe431unitsthatusedfueloilornaturalgasforelectricitygeneration.ThetreatmentoftheseunitsisdescribedinSection6.2.1. 147

PAGE 148

TheheatcontentofdeliveredcoalisinBtusofheatpertonofcoal,barrelofoil,or1,000cubicfeetofnaturalgas.Allfuelparametersareconvertedintermsofaheatcontentbaselinetoallowforcomparisonsacrossunits. Thepurchasingagreementislabeledastwopossibleparameters:(1)spotmarketand(2)contractpurchases.Anagreementisconsideredaspotmarketpurchaseiftheagreementisforlessthan2years.Purchasingagreementsoftwoyearsorgreaterareconsideredtobeundercontract.Theamountoffuelundercontractisintons,barrels,or1,000cubicfeetdependingonthefuel.Thedataismanipulatedintototalheatcontentbymultiplyingtheheatcontentperunitoffuelbytheamountoffueldelivered. Sincecoaldeliveriesaremadeattheplantlevel,thesizeofthecontractconstraintisnoteasilyderivedforeachgeneratingunit.Theparameterfortheamountofcontractcoalforeachgeneratingunitwasderivedinthefollowingmanner.First,anyunitatagivenplantthatactuallyinstalledascrubberisallocatedasmuchhighsulfurcontractcoalaspossiblewhilereceivingaslittlelowsulfurcontractcoalaspossible.Thereasoningforthisisthataunitwithascrubberwillhavealargermarginalcostofabatementfromswitchingfuels(MCAs;si
PAGE 149

3-1 andwouldbeconsideredcoalstoredattheplantforusethefollowingyear. Sulfurcontentisthepercentageofeachtonofcoal,barrelofoil,or1,000cubicfeetofnaturalgasthatissulfur.Thedatamustbemanipulatedtocreatethedesiredvariable,whichispoundsofsulfurdioxidepermillionBtusofheat.PhaseIofTitleIVdistributesallowanceallocationsbasedon2.5poundsofSO2permillionBtusofheat.Sofuelisconsidered\highsulfur"ifcontainsgreaterthen2.5lbs.ofSO2permmBtuand\lowsulfur"ifitcontainslessthan2.5lbs.ofSO2permmBtu.Anycoalusethathasahigher(lower)sulfurcontentwillincrease(decrease)emissionsabove(below)itsallocationallows. 149

PAGE 150

AlthoughnearlyallgeneratingunitsaectedbyPhaseIwerecoal-redunits,24unitsusefueloilornaturalgasforelectricitygeneration.Fortheseunits,fueloilisdesignatedashighsulfurfueloptionwhilenaturalgasisdesignatedasthelowersulfurfueloptionfornon-coalredunits. Thedeliveredpriceisthecentsperton,barrel,or1,000cubicfeetpaidatthetimeofdelivery.Thepriceincludesthepurchasingofthefuelandthetransportationcostsofshipment.ThepriceisconvertedtothepriceoffuelincentspermillionBtus.Aseparatedeliveredpriceisrequiredforbothhighsulfurandlowsulfurfuel,whichisaweightedaverageofalldeliveriesofeachcategoryofcoal.Sothehigh(low)sulfurpriceusedinthesimulationsistheweightedaveragedeliveredpriceforallhigh(low)sulfurcoaldeliveries.TheparameterusedinthesimulationsisactuallydollarspermmBtu,orcents/mmBtudividedby100. Anotherissueisthatnotallgeneratingunitspurchasedbothhighsulfurandlowsulfurfuel.Proxieswererequiredtodeterminethefuelcharacteristicsfacingeachunit,andwerersttakenfromothergeneratingunitsthatwereownedbythesamecompanywhereavailable.Ifaunitdidnothaveanotherunitunderthesamecompany,aproxywastakenfromthegeographicallyclosestplantbecauseshippingcostsareanimportantfactorindeterminingthedeliveredprice.RegionalvariationinpriceswillbemuchsmallervariationinpricesacrosstheU.S. 150

PAGE 151

ScrubbercostdatawasfoundintheEIA's\ElectricPowerAnnual1997".Thedatawasthehistoricalscrubbercostdata,includingstate-by-stateaverageinstalledcostsperkilowattofcapacity,averageoperationandmaintenancecostsintermsofmillsperkilowatt-hour,megawattsofcapacityforunitswithscrubbersinstalled,andsulfurremovaleciency. 151

PAGE 152

Anendogenousscrubberchoicecreatesanon-convexitythatmaynotresultinanallowancemarketequilibrium.Insuchacase,therewillbeeitherapositiveoranegativeexcessdemandforallowancesandanallowancepriceatwhichageneratingunitisindierenttoinstallingascrubber.Thiscouldbeconsidereda\quasi-equilibrium"wheretheexcessdemandiseitherboughtfromorsoldtoanallowancebrokeratthe\quasi-equilibrium"allowanceprice.ThetechnicalexplanationofthebisectioniterativeprocessusedtoconvergetoanequilibriumallowancepriceisdescribedindetailinAppendixB. 152

PAGE 153

First,considertheabilityofthesimulationmodelstorecreatetheresultsfrompreviousstudies.Simulation1recreatestheunconstrainedcost-minimizationresultsfromSotkiewicz(2003)andSotkiewiczandHolt(2005).Therewere17scrubbersthatwereinstalledasaresultoftheNewSourcePerformanceStandards(NSPS)andwereunrelatedtoTitleIV.Assumingthesescrubbersasgiven,thetotalcoststotheindustryofmeetingelectricitydemandwere$7.69billion.Thesecostswillbeusedasthebaselinetotalindustrycoststodeterminetotalindustrycompliancecostscomparabletopreviousstudies. Simulation3recreatestheemissionsconstrainedresultsfromSotkiewiczandHolt(2005)assumingthe46scrubbersthatwereinstalledin1996asgiven.Thesimulationresultsinanallowancepriceof$149.64,whichexactlyreplicatestheallowancepricefoundbySotkiewicz(2003)andSotkiewiczandHolt(2005).Totalindustrycostsare$8.23billionwhere29scrubberswereactuallyinstalledinresponsetoTitleIV.ThedierencebetweenSimulation3andSimulation1isthetotalindustrycompliancecostsofmeetingtheemissionsconstraint,whichis$541millionandsimilartotheresultsfoundinprevious 153

PAGE 154

TheallowancemarketdoesnotclearinSimulation4becausegeneratingunitsareallowedtomaketheirscrubberchoice.Thediscretenatureofscrubberinstallationleadstoanon-negativeexcessdemand.However,theexcesssupplyof53,576allowancesinSimulation4accountforlessthan1%ofthe5+millionallowancemarketandcouldbeassumedtobebankedforfutureuseorsoldtoanallowancebroker,whichcouldbeconsideredaquasi-equilibrium. 154

PAGE 155

Simulation6introducesfuelcontractconstraintstothetheemissionsconstrainedmodelwhileallowinggeneratingunitstomaketheirscrubberchoice.Thesimulationresultsinanallowancepriceof$210.74.Totalindustrycostsare$8.63billionwhere44scrubbersareinstalledinresponsetoTitleIV.TheminimumcompliancecostswhencomparedtoSimulation1are$939million,or$651millionhigherthanifcontractsarenottakenintoconsideration.TheminimumcompliancecostsareclosetotheactualcostfoundinSotkiewiczandHolt(2005)at$990millionandCarlsonetal.(2000)at$910million. Allowinggeneratingunitstochoosewhethertoinstallascrubberallowstheindustrytoloweritstotalcostsby$132million.Aswouldbeexpected,introducingthecontractconstraintresultsinmorescrubberinstallationsduetoTitleIVfrom27to44becausehigh-sulfurfuelcontractsincreasetheincentiveforaconstrainedgeneratingunittoinstallascrubber.AsinSimulation4,theallowancemarketdoesnotclearduetothediscrete,endogenousscrubberchoice.However,theexcessdemandof7,797allowancesaccountforlessthan0.2%ofthe5+millionallowancemarket,andcouldbeassumedtobeboughtfromanallowancebroker. TheactualtotalindustrycompliancecostsarefoundinSimulation7.Byusingtheactualemissionsandelectricityproductionforeachunit,itispossibletodetermineeachunit'sactualcoalmix.Theseactualdecisionsresultedinanallowancepricerange 155

PAGE 156

Third,the\true"compliancecostswilldierfromtheseestimatesbecausetheappropriatebaselinewasnotused.AsshownanalyticallyinSection4,comparingSimulation5andSimulation6toSimulation1isnotthemostappropriatemeasureofcompliancecosts.Thecontractconstraintsshouldbeincludedinboththebaselinesimulationandthepolicy-restrictedsimulation.Simulation2runsthesamemodelasinSimulation1exceptthatitincludescontractconstraintsandresultsintotalindustrycostsof$8.27billion.ThedierencebetweentotalindustrycostsinSimulation1andSimulation2aretheadditionalcostsduetocontractconstraints,whichare$582million.TheseadditionalcostswouldhaveresultedwithorwithouttheSO2TradingProgramandshouldnotbeconsideredcompliancecosts.Thisisakeyresultbecausethesecostswerelabeledcompliancecostsbypreviousstudieseventhoughthesecostsarearesultofgeneratingunitslockinginpricestoprotectfromtheuncertaintyofhighercoalpricesinthefuture. BycomparingresultsinSimulation5toSimulation2,theminimumcompliancecostsconsideringcontractsandgiventhescrubberchoicearefoundtobe$490.1million.\True"minimumcompliancecostsaremuchlower($610millionlower)oncethisadditionalconstraintisincludedinthemodel.InSimulation6,the\true"minimumcompliancecostsare$357.8million,or$581.6millionlessthanifcontractsareexcludedfromthemodel.Thesearesimilartotheleast-costresultsfoundbySotkiewicz(2003)at$340-$527million,SotkiewiczandHolt(2005)at$423-$553million,andCarlsonetal.(2000)at$571million. 156

PAGE 157

Simulation4usesatotalof2,651,410,732mmBtuofhighsulfurcoal(41.4%)and3,750,168,009mmBtuoflowsulfurcoal(58.6%).Simulation6usesatotalof2,614,773,732mmBtuofhighsulfurcoal(40.8%)and3,786,804,109mmBtuoflowsulfurcoal(59.2%).Includingcontractconstraintsintothemodelresultsina0.6%decreaseinhighsulfurcoalrelativetoSimulation4.Overallcoalusedoesnotappeartohavebeensignicantlyaltered.However,thisdoesnottellthewholestory.Contractconstraintsledto27units(6.3%ofaectedunits)usingasuboptimalcoalcombination.5ofthe27unitshadachangeofatleast98%ofcoaluseand15ofthe27hadatleasta50%changeincoaluse.14ofthe27unitshadanincreaseinhighsulfurcoalusewhile13unitshadandecreaseofhighsulfurcoaluse.Theconcernwiththecontractsisnotnecessarilythattheentire 157

PAGE 158

Second,comparingscrubberinstallationsinSimulation6toSimulation5willshowhowscrubberdecisionswouldbealteredifthescrubberdecisionismadeendogenousandcontractconstraintsareincludedinthemodel.Allowingthescrubberchoicetobeendogenousresultsin53generatingunitsalteringtheirscrubberchoice,including19scrubberstoberemovedand34tobeaddedforatotalof15moreinstalledscrubbers.27generatingunitsmaintainthesamescrubberchoices,but17ofthosewereinstalledforNSPS.So10scrubberinstallationsremainedthesame. Finally,itisimportanttoconsidertheimpactscontractconstraintshaveonaunit'sendogenousscrubberinstallationdecision,whichcanbedeterminedbycomparingSimulation6toSimulation4.33scrubberchoicesarealteredasaresultofcontractconstraints,including25unitsthatwillnowinstallascrubberand8unitsthatnolongerinstallascrubber.Ofthe25unitsthatchosetoinstallascrubber,20ofthemhadahighsulfurcoalcontract.Ofthe8unitsthatchosetonotinstallascrubber,all8ofthemhadalowsulfurcoalcontract.So28of33scrubberchoicesappeartohavebeendirectly 158

PAGE 159

3-4 ,contractconstraintsincreasescrubberinstallationsinOhio,Alabama,Florida,Indiana,Mississippi,andMissourianddecreasescrubberinstallationsinWestVirginia,NewYork,Pennsylvania,Wisconsin,Georgia,Kentucky,andNewJersey. AnunexpectedresultinSimulation6isthatscrubberinstallationsactuallyincreaserelativetoSimulation5.Thesearelikelyaresultoftheassumedscrubbercostestimatesbecausethehistoricalscrubbercostdatausedinthesimulationsislowerthantheengineeringcostestimates.Scrubbersappearcheapertoinstallthantheengineeringestimatesstate.Ofthe404generatingunitswithatleasta90%removalrate,31historicalcapitalcostestimatesarehigher,84arethesame,and289arelowerthantheestimatedengineeringcapitalcosts. AscanbeseeninTable 3-5 ,thereisasignicantdierenceinscrubberinstallations.Theuseofengineeringcostestimatesforscrubberinstallationresultsinadecreaseinscrubberinstallationsfrom44to25inSimulation4andfrom61to38inSimulation6becauseofthehighercostsinvolved.Noticethatintroducingcontractconstraintsresultsinadditionalscrubberinstallationsfrom25to38,whichisclosetotheactualinstallationsof46scrubbers.Compliancecostinterpretationsremainsimilartosimulationsusing 159

PAGE 160


PAGE 161

3.7.1Introduction Aplantleveldecision-makingmodelismorerealisticthanageneratingunitlevelmodeltodeterminecompliancedecisionsbecausecoaldeliveriesaremadeattheplantlevelwherethereareoftenmultiplegeneratingunits.Allgeneratingunitsatthatplantfacingthesamecoaluseoptions,includingsulfurcontents,heatcontents,anddeliveredpricesforbothspotmarketandcontractcoal.Thecontractconstraintsbecomemorecomplexinthismodelwherethesumofcontractcoaluseforallgeneratingunitsataplantmustcoverthecontractrequirement,Pni=1Ccif 161

PAGE 162

Threeadditionalfactorswillbeofgreatimportanceattheplantlevel.First,plantsareabletotradeallowancesbetweengeneratingunitsasneededtocoveremissionsatnoadditionalcosts.Theliteraturehasstatedthattradingallowancesbetweenitsunitsatthesameplanthasbeenacommonoccurrence,whichsupportstheuseofamodelthatconsidersdecisionsattheplantlevelinsteadofthegeneratingunitlevel. AplantmayhaveoneormoregeneratingunitsatitslocationthatarenotaectedbyPhaseIanddoesnothavetomeetanyemissionsrequirements.Owninganon-aected 162

PAGE 163

minzi;Ai;Csih;Csil;Ccih;CcilnXi=1ziPiz+PAAi+PshCsih+PslCsil+PchCcih+PclCcil subjectto...Aei+Ai(1ziri)(m)[CsihSsh+CcihSch+CsilSsh+CcilScl]i1 (CsihHsh+CcihHch+CsilHsl+CcilHcl)Dii2 (3{44) TheLagrangemultiplieronageneratingunit'semissionsconstraintisrepresentedbyi1.TheLagrangemultiplieronageneratingunit'sdemandconstraintisrepresentedbyi2.Coalcontractconstraintsrequireaplanttouseaminimumamountofeachcontractcoal 163

PAGE 164

SinceAicanbeeitherpositiveornegativebasedonthenetallowanceposition,( 3{46 )willholdwithequality.Theadditionalcosttothermofemittingonemoretonofemissionsisequivalenttotheallowanceprice,i1=PA,whichisthesameresultasfromthepreviousmodels. ThepartialderivativewithrespecttoCsihrepresentstheimpactaoneunitchangeinhighsulfurspotmarketcoalhasontheunit'stotalcosts. Ifthegeneratingunitusessomeamountofhighsulfurspotmarketcoal(Csih>0),then( 3{47 )holdswithequality. ThepartialderivativewithrespecttoCsilrepresentstheimpactaoneunitchangeinlowsulfurspotmarketcoalhasontheunit'stotalcosts. Similartohighsulfurspotmarketcoal,ifthegeneratingunitusessomelowsulfurspotmarketcoal(Cil>0),then( 3{48 )holdswithequality. 164

PAGE 165

Unlikewithhighsulfurspotmarketcoal,highsulfurcontractcoalwillbeimpactedbythehighsulfurcoalcontractconstraint(h).Ifthegeneratingunitusessomeamountofhighsulfurcontractcoal(Ccih>0),then( 3{49 )holdswithequality. ThepartialderivativewithrespecttoCcilrepresentstheimpactaoneunitchangeinlowsulfurcontractcoalhasontheunit'stotalcosts. Unlikewithlowsulfurspotmarketcoal,lowsulfurcontractcoalwillbeimpactedbythelowsulfurcoalcontractconstraint(l).Similartohighsulfurcontractcoal,ifthegeneratingunitusessomelowsulfurcontractcoal(Ccil>0),then( 3{50 )holdswithequality. 165

PAGE 166

3{47 )and( 3{48 )holdwithequality.Theadditionalcoststothegeneratingunitofusingonemoreunitoflowsulfurspotmarketcoalisequaltotheadditionalcostsofusingonemoreunitofhighsulfurspotmarketcoal. Sinceboth( 3{47 )and( 3{48 )holdwithequality,itispossibletondanexpressionfor(i2)ineachequationandsetthetwoexpressionsequaltoeachotherandderivetheallowanceshadowprice(i1)in( 3{52 ). Asdiscussedearlier,theallowanceshadowprice(i1)isanexpressionforageneratingunit'sMarginalCostofAbatementfromSwitchingfromhighsulfurspotmarkettolowsulfurspotmarketcoal(MCAs;si).From( 3{46 ),wealsoknowthattheactualallowancepriceequalstheallowanceshadowprice(PA=i1).Ascanbeseenin( 3{52 ),theallowanceprice(PA)equalstothemarginalcostofabatement(MCAs;si),whichistheincreaseincoalcostsfromswitchingfromhighsulfurspottolowsulfurspotmarketcoalperunitofreducedemissions. Themarginalcostsofabatementmayormaynotdieracrossgeneratingunits.Ifaplantdoesnotinstallanyscrubbers,thenMCAs;si=MCAs;sj8fi;jgforallgeneratingunitsandtheplant'schoiceswillbeidenticaltothegeneratingunits'decisionsinSection5.Thiscanbeseenin( 3{52 )wheretheonlywaythatthemarginalcostsofabatementcandieracrossgeneratingunitsisthroughthescrubberchoiceandascrubber'semissionscapturerate.Ifallgeneratingunitsatagivenplantbothinstallascrubberwhereri=rj, 166

PAGE 167

Marginalcostsofabatementwillonlydieracrossgeneratingunitsataplantif(1)aplantinstallsascrubberatsomebutnotallofitsgeneratingunits,or(2)aplantinstallsascrubberatallitsgeneratingunitsbutthescrubbershavedierentcapturerates(ri6=rj9fi;jg).Wewillintroducethescrubberchoice,andderiveitsimpactonaplant'schoiceslater. 3{47 )holdswithequalityand( 3{48 )holdswithweakinequalityresultingin( 3{53 ). Equation( 3{53 )statesthatadditionalcoststothegeneratingunitofusingonemoreunitofhighsulfurspotmarketcoalisweaklylessthanorequaltotheadditionalcostsofusingonemoreunitoflowsulfurspotmarketcoalinclusiveofemissionsanddemandrequirements. FollowingSection6.4.1wecanderivetherelationofPAtoMCAs;si.Bycomparingthesetwoexpressionsandsolvingfori2,yougetaninequalitycomparingi1=PAandMCAs;si.Sincethegeneratingunituseshighsulfurspotmarketcoal,theallowancepriceisweaklylessthanthemarginalcostofabatement(PAMCAs;si)asshownin( 3{54 ). 3{47 )holdswithequalityand( 3{48 )holdwithweakinequalityresultingin( 3{55 ).Theadditionalcoststothegeneratingunitofusingone 167

PAGE 168

OnceagainwecanderivetherelationofPAtoMCAs;si.Bycomparingthesetwoexpressionsandsolvingfori2,yougetaninequalitycomparingi1=PAandMCAs;si.Sincethegeneratingunituseslowsulfurspotmarketcoal,theallowancepriceisweaklygreaterthanthemarginalcostofabatement(PAMCAs;si)asshownin( 3{56 ). First,ifageneratingunitfacesPA
PAGE 169

3{41 ).Assumingonlylowsulfurspotmarketcoalusetomeetdemand,themaximumamountoflowsulfurcoalisexpressedin( 3{59 ). ReplacingCsilin( 3{41 )withtheexpressionin( 3{59 )forCs;MAXilgivesanexpressionfortheminimumallowanceexcessdemandin( 3{60 ).Ageneratingunit'sminimumexcessdemandmustcoverthedierencebetweenitsinitialallowanceallocation(Aei)andtheamountofallowancesneededtocovertheunit'sminimumactualemissions[EMINi=(1ziri)(m)(Ssil)(Di IfPA=MCAs;si,ageneratingunitmayuseanycombinationofhighsulfurspotmarketcoalandlowsulfurspotmarketcoalandleadtoanylevelofexcessdemandintherange(EMINiAei;EMAXiAei).TheallowanceexcessdemandcanberepresentedbyAi=(EMAXi(1)EMINiAei)wheretheconstant2[0;1].Aunitthatisindierentbetweenfuelswitchingandallowancespurchasingcouldbeeitheranetbuyeroranetseller. Combiningtheexcessdemandsforeachofthethreecasescreatestheexcessdemandcorrespondenceshownbelow.Ai=8>>>><>>>>:AMAXiifPA>MCAs;siAMAXi(1)AMINiifPA=MCAs;si82[0;1]AMINiifPA
PAGE 170

Theuseofcontractcoalwillbebasedontherelativemarginalcostsofusingcontractcoalateachgeneratingunitoperatedbyaplant.Aplantwillchoosetousecontractcoalatthegeneratingunitthatwillresultinthelowestincreaseintheplant'stotalcosts.Tomakethesecostcomparisons,itisnecessarytoderivethemarginalcostsofabatementforthegivencombinationofspotandcontractcoalandcomparethemacrossgeneratingunits.Initiallywewillignoreaplant'sscrubberchoiceforeachgeneratingunitandaplantmakesthesamechoiceforallitsgeneratingunits(zi=zj=0,orzi=zj=1andri=rj).Undertheseconditions,themarginalcostsofabatementareidenticalacrossallgeneratingunits,anditdoesnotmatteratwhichofthesegeneratingunitsthecontractcoalisused.Thiswillbeshownseparatelyforbothahighsulfurandlowsulfurcoalcontractbecausetheuseofhighsulfurcontractcoalisindependentoflowsulfurcontractcoalandvisaversa. 3{61 ).Themarginalcostofabatementfromswitchingfromhighsulfurcontractcoaltolowsulfurspotmarketcoal(MCAc;si)willbethesameforallgeneratingunitsthatmakethesamescrubberinstallationchoice(zi),includingthesamescrubber 170

PAGE 171

Allunitsatwhichaplantdoesnotinstallascrubberhavethesamemarginalcostofabatementofswitchingfromhighsulfurcontractcoaltolowsulfurspotmarketcoal(MCAc;si=MCAc;sj).AlsoallunitsataplantthatinstallascrubberwiththesamecaptureratewillhaveidenticalMCAc;si. 3{62 ).TheMCAs;ciwillbethesameforallgeneratingunitsthatmakethesamescrubberinstallationchoice,includingthesamescrubbertechnologywiththesamecapturerate.Psh+i1(1ziri)(m)(Ssh)+i2Hsh=Pcl+i1(1ziri)(m)(Scl)+i2Hcll Allunitsatwhichaplantdoesnotinstallascrubberhavethesamemarginalcostofabatementofswitchingfromhighsulfurspotmarketcoaltolowsulfurcontractcoal(MCAs;ci=MCAs;cj).AlsoallunitsataplantthatinstallascrubberwiththesamecaptureratewillhaveidenticalMCAs;ci. 171

PAGE 172

Theemissionsconstraintdoesnotbind(i1=0)fortheseunits,whichmeansthat\non-aected"generatingunitswillprefertousethecoaltypewiththelowestdeliveredprice.Formostplantsthiswillbehighsulfurcoal,particularlythoselocatedintheEasternU.S. 3{63 )and( 3{64 ). @Csih=Pshi2Hsh0;=0ifCsih>0(3{63) and @Csil=Psli2Hsl0;=0ifCsil>0(3{64) Assumingthatageneratingunitusesbothhighandlowsulfurspotmarketcoal,weuse( 3{63 )and( 3{64 )toderive( 3{65 ).Ifageneratingunitusesbothhighandlowsulfurspotmarketcoal,thenthepriceperunitofheatinputisequalforbothcoaltypes.Aplantwillnolongermakeit'sgeneratingunitlevelfuelchoicesbasedonitsmarginalcostofabatementbecauseitnolongerisconcernedaboutabatingemissions. Byusingtherst-orderconditions,wecandeterminewhenageneratingunitwillonlyuseonetypeofspotmarketcoal.Ifageneratingunitusesonlyhighsulfurspotmarket 172

PAGE 173

3{63 )holdswithequalityand( 3{64 )remainsweaklygreaterthanzeroandlowsulfurcoalisweaklymoreexpensive(Psh 3{64 )holdswithequalityand( 3{63 )remainsweaklygreaterthanzeroandhighsulfurspotmarketcoalismoreexpensive(Psh First,assumethattwogeneratingunits(Unit\i"andUnit\j")arebothaectedunits.Aplantwillbeindierenttousinghighsulfurcontractcoalateitherunitwhentherst-orderconditionsforusinghighsulfurcontractcoalareequalacrossgeneratingunits.Pch+i1(1ziri)(m)(Sch)i2Hchh=Pch+j1(1zjrj)(m)(Sch)j2Hchh NowassumethatUnit\i"isanaectedunitanditsemissionsconstraintbinds(i1>0)whileUnit\j"isnotaectedandtheemissionsconstraintwillnotbind(j1=0).Underthiscondition,theadditionalcostsofusinghighsulfurcontractcoalatanaectedunitisgreaterthanatanon-aectedunit.Pch+i1(1ziri)(m)(Sch)i2Hchh>Pchj2Hchh)i1(1ziri)(m)(Sch)i2Hch>j2Hch 173

PAGE 174


PAGE 175

First,considerthemarginalcostofabatementofswitchingfromhighsulfurspotmarketcoaltolowsulfurspotmarketcoal(MCAs;si)withascrubberin( 3{68 ),andcompareittothemarginalcostofabatementwithoutascrubberin( 3{69 ).AsinChapter2,ascrubberdecreasesthesavingsfromswitchingspotmarketfuelsbecauseascrubbercausestheemissionsreductiontobesmallerthanwithoutascrubber. Second,considerthemarginalcostofabatementofswitchingfromhighsulfurcontractcoaltolowsulfurspotmarketcoal(MCAc;si)withascrubberin( 3{70 ),andcompareittothemarginalcostofabatementwithoutascrubberin( 3{71 ).Asinthecaseofonlyspotmarketcoal,installingascrubberdecreasesthesavingsfromswitchingfuels.Inthecaseofahighsulfurcontract,itislesscostlytoaplanttousehighsulfurcoalataparticulargeneratingunitifithasascrubber. Third,considerthemarginalcostofabatementofswitchingfromhighsulfurspotmarketcoaltolowsulfurcontractcoal(MCAs;ci)withascrubberin( 3{72 ),andcompareittothemarginalcostofabatementwithoutascrubberin( 3{73 ).Asinthetwocasesabove,installingascrubberdecreasesthesavingsfromswitchingfuels.Inthecaseofalow 175

PAGE 176

Foreachcombinationofcoaltypes,installingascrubberdecreasesthesizeofthedenominator,whichincreasesMCAi.AsinChapter2,ascrubbergreatlydecreasesthesavingsfromswitchingfuels.AhigherMCAiincreasestherangeofallowancepricesatwhichaplantwillprefertopurchaseallowancesinsteadofswitchingfuelsbyloweringthepriceatwhichageneratingunitwillbeindierentbetweenpurchasingallowancesandswitchingfromhighsulfurtolowsulfurcoal.Anyunitoperatedbyaplantthathasascrubberwillhavealargermarginalcostofabatementthananyunitwithoutascrubber(MeCAi>MCAj). Thedecisionisbasedsolelyonaunit'sACAibecausethemarginalcostsofabatementareequalacrossgeneratingunitswithnoscrubbersinstalledattheplant. 176

PAGE 177

PAGE 178

PAGE 179

Assumingnocontractcoal,wecanderivetheallowancepriceatwhichaplantisindierenttoinstallingascrubberforagivengeneratingunitin( 3{75 ). (bAieAi)(3{75) Aspecialcaseexistsfor( 3{75 )inwhichthegeneratingunitisanon-aectedunit,whichresultsinPA!1becausetheunitdoesnothavetocoveritsemissionswithallowancesandgainsnothingfrominstallingascrubber.AnindierencepriceofinnityimpliesthatascrubberwillneverbeinstalledatUnit\i"ifitanunaectedunit. (bAieAi)=Piz+Psh(eCsihbCsih)+Psl(eCsilbCsil) 0=1(3{76) 179

PAGE 180

Aplantwithabindinghighsulfurcoalcontracthasagreaterincentivetoinstallascrubberateachofitsgeneratingunitsbecauseitmustusesomehighsulfurcoal,evenifitwouldprefertouselowsulfurcoalatallgeneratingunits.AhighsulfurcoalcontractdecreasestheindierencepriceatwhichaunitwillinstallascrubberfromPAto(PA)foreachgeneratingunit.TheincentivemaynotbelargeenoughtoresultinascrubberinstallationatitsgeneratingunitwiththelowestACAi.Insuchascase,aplantisindierenttousinghighsulfurcontractcoalatanyofitsgeneratingunitsbecauseallunitshavethesameMCAc;si. AbindinghighsulfurcoalcontractresultsinadecreaseinPSA. AtPSAwhere... 180

PAGE 181

ThisisthesameresultasinChapter2whereahighsulfurcoalcontractresultsinaninecientcoalusecombinationofhighandlowsulfurcoal.Theindierencepriceatwhichaplantwillinstallascrubberatagivengeneratingunitwillweaklydecreasewhenaplantchoosestousehighsulfurcontractcoalatthatunit. 181

PAGE 182

AbindinglowsulfurcoalcontractresultsinanincreaseinPSA. AtPSAwhere... Iflowsulfurcontractcoalisatleastasexpensiveaslowsulfurspotmarketcoal,weknowthat... ThisisthesameresultasinChapter2.Alowsulfurcoalcontractresultsinaninecientcoalusecombinationofhighandlowsulfurcoal.Theindierencepriceatwhichaplantwillinstallascrubberatagivengeneratingunitwillweaklyincreasewhenaplantchoosestouselowsulfurcontractcoalatthatunit. 182

PAGE 183

3{81 ).Theamountofcontractcoalisthesamenomatterthescrubberchoiceateachgeneratingunitmadebyaplant. (bA1eA1) (3{81) Second,thetotalcoststotheplantofnotinstallinganyscrubbersmustbelessthanthetotalcostsofinstallingscrubbersatbothgeneratingunits. 3{82 ). (bA1eA1)+(bA2eA2) (3{82) TheminimumofthetwoPSAwillbetheallowancepriceatwhichaplantisindierenttoinstallingascrubberandnotinstallinganyscrubbers.Noticethattheallowanceprice 183

PAGE 184

3{83 ). Second,thetotalcostsofinstallingonegeneratingunitmustbelowerthanthetotalcostsofinstallingtwoscrubbers. 3{84 ). (bA2eA2)>PSA Theallowancepricemustbebetweenthesetwoindierenceprices,whichresultsinPSAPAforUnit2. 184

PAGE 185

3{85 ). (bA2eA2)
PAGE 186

3-7 Aplantwithnocontractcoalwillprefertoinstallnoscrubbersiftheallowancepriceisbelow$194.48.Anallowancepricebetween$194.48and$230.89willleadaplanttoinstallascrubberatUnit1,butnotatUnit2.Aplantwillinstallscrubbersatbothunitsiftheallowancepriceisgreaterthan$230.89. Nowconsidertheplanthasahighsulfurcoalcontractfor28.92%ofcoaluse(13,855,269mmBtu),whichwillaltertheindierenceatwhichaplantwillinstallascrubberatUnit1.WehavealreadyshownthathighsulfurcontractcoalwillbeusedatthegeneratingunitwiththelowestACAi,whichisUnit1.TherequireduseofhighsulfurcoalwilllowertheallowanceindierentpricetoinstallingascrubberatUnit1to$140.69.TherewillbenochangeintheindierencepricetoinstallingascrubberatUnit2becausethehighsulfurcoalcontractdoesnotbindforthatgeneratingunit. Nowconsidertheplanthasahighsulfurcoalcontractforallofitscoaluseinbothgeneratingunits(47,913,973mmBtu).Theindierencepricesforbothgeneratingunitswillbothdecreaseasaresult.TheallowancepriceatwhichtheplantwillinstallascrubberatUnit1willagainbe$140.49.TheindierencepriceforUnit2willdecreaseto$153.61.Ahighsulfurcoalcontractforallofageneratingunit'scoalrequirementsresultsinlargereductionsoftheallowancepriceatwhichaplantwillbeindierenttoinstallingascrubberforbothUnit1($54or28%)andUnit2($77or33%). 186

PAGE 187

3{87 ). (bA1eA1) (3{87) 3{88 ). (bA1eA1) (3{88) 187

PAGE 188

Aplantlevelmodelismorerealisticthanageneratingunitlevelmodelbecausecoaldeliveriesaremadeattheplantlevelwherethereareoftenmultiplegeneratingunits.Operatingmultipleunitsallowsaplanttorelaxitscontractconstraintbecausethereareadditionaldegreesoffreedominfuelusedependingoneachunit'semissions,demand,andcoalcontractconstraints.Also,plantsareabletotradeallowancesbetweengeneratingunitsasneededtocoveremissionsatnoadditionalcost.Aplantmayalsohaveoneormore\non-aected"generatingunits,whichfacenoemissionsconstraint.Anyhighsulfurcontractcoalcanbeusedatthesenon-aectunitswithoutanynegativenancialrepercussionsduetothepolicy.Thesefactorscanleadtoplant-levelchoicesthatdonotminimizeeachgeneratingunit'stotalcosts. Anallowancemarketequilibriumwillexistonlyifthediscretescrubberchoiceisgiven.Allowingforanendogenousscrubberchoicemakesitimpossibletoguaranteeanequilibrium,althoughonemaystillmayexist.Bindingfuelcontractsmayalteraunit'scompliancedecisionsandexcessdemand.Alteringcompliancedecisionscouldleadtobothanalteredallowancemarketpriceandanincreasetotalindustrycompliancecosts. Generatingunit-levelsimulationswereabletoeectivelyreplicatetheresultsfrompreviousstudiesandshowthatfuelcontractscanexplainaportionofthepreviouslyunexplainedexcesscompliancecosts.Simulatingtheleast-costcompliancechoiceswithout 188

PAGE 189

Aplant'sdecision-makingprocessmaynotminimizecostsforeachgeneratingunitbecauseaplant'sconcernisbasedonthecombinedcostsofallgeneratingunitsunderitsoperationalcontrol.Thechoicetoinstallascrubberisbasedonthecharacteristicsofalltheunitsataplant,notjusttheunitatwhichthescrubbermaybeinstalled.OncetheorderofpreferredscrubberinstallationsisdeterminedbasedontheACAi,aplantisabletomakeitscostminimizingfuelchoices. Aplantlevelmodelismorerealisticthanageneratingunitlevelmodelbecausecoaldeliveriesaremadeattheplantlevelwherethereareoftenmultiplegeneratingunits.Operatingmultipleunitsallowsaplanttorelaxitscontractconstraintbecausethereareadditionaldegreesoffreedominfuelusedependingoneachunit'semissions,demand,andcoalcontractconstraints.Also,plantsareabletotradeallowancesbetweengeneratingunitsasneededtocoveremissionsatnoadditionalcost.Aplantmayalsohaveoneormore\non-aected"generatingunits,whichfacenoemissionsconstraint.Anyhighsulfurcontractcoalcanbeusedatthesenon-aectunitswithoutanynegativenancialrepercussionsduetothepolicy.Thesefactorscanleadtoplant-levelchoicesthatdonotminimizeeachgeneratingunit'stotalcosts. 189

PAGE 190

Example:ContractCoalDistribution Plant-LevelConstraintTotalmmBtuUnitDemandScrubber HighSulfurContract300,0001200,000YesLowSulfurContract300,0002200,000NoDemand600,0003200,000No CoalDist'nbyUnitHighSulfurCoalLowSulfurCoal Unit1200,0000Unit250,000150,000Unit350,000150,000 Table3-2. SulfurConversionbyFuelType FuelEmissionsConversionFactorBaseline Bituminous(38*SulfurContent)/(mmBtuperton)=lbs.SO2/mmBtuSub-bituminous(35*SulfurContent)/(mmBtuperton)=lbs.SO2/mmBtuAnthracite(39*SulfurContent)/(mmBtuperton)=lbs.SO2/mmBtuLignite(30*SulfurContent)/(mmBtuperton)=lbs.SO2/mmBtuFuelOil#2(144*SulfurContent)/(mmBtuper1,000bbl.)=lbs.SO2/mmBtuFuelOil#6(162*SulfurContent)/(mmBtuper1,000bbl.)=lbs.SO2/mmBtuNaturalGas(0.60lbs.SO2/mmCF)/(mmBtupermmCF)=lbs.SO2/mmBtu 190

PAGE 191

SimulationResults Sim.EmissionsContractScrubberScrubbersPAAiTotalCostsCompl.CostsComp.CostsConstraintConstraintChoiceInstalledvs.(Sim.1)vs.(Sim.2) (1)NONOGiven17NANA$7,685,800,000NANA(2)NOYESGiven17NANA$8,267,400,000NANA(3)YESNOGiven46$149.640$8,226,500,000$540,700,000NA(4)YESNOChosen44$155.58-53,576$7,974,100,000$288,300,000NA(5)YESYESGiven46$206.700$8,757,500,000$1,071,700,000$490,100,000(6)YESYESChosen61$210.747,797$8,625,200,000$939,400,000$357,800,000(7)--Actual46NANA$8,983,500,000$1,297,700,000$716,100,000

PAGE 192

ImpactofContractConstraintonScrubberChoice StateDecreaseStateIncrease WV-4MO2NY-2MS2PA-2IN3WI-2FL4GA-1AL7KY-1OH10NJ-1 Table3-5. SimulationswithEngineeringData Sim.AiPAScrubbersIndustryCostsComp.CostsComp.CostsInstalled(vs.1)(vs.2) 1NANA17$7,685,800,000NANA2NANA17$8,267,400,000NANA30$149.6446$8,239,000,000$553,200,000NA41,972$214.8325$8,010,200,000$324,400,000NA50$207.6046$8,770,900,000$1,085,100,000$503,500,00062,611$238.5538$8,686,300,000$1,000,500,000$418,900,0007NANA46$8,995,600,000$1,309,800,000$728,200,000 Table3-6. ImpactsofaReductionintheAllowanceAllocationof10% Simulation4InitialAllocationAllocationMinus10%Increase IndustryComplianceCosts$288,300,000$381,900,000$93,600,000ScrubbersInstalled446117AllowancePrice$155.58$189.43$33.85 Simulation6InitialAllocationAllocationMinus10%Increase IndustryComplianceCosts$357,800,000$476,600,000$118,800,000ScrubbersInstalled617312AllowancePrice$210.74$222.22$11.48 Table3-7. MathExample:TwoAectedUnits LowSulfurSpot$1.200.625%24mmBtuHighSulfurSpot$1.102.580%24mmBtuHighSulfurContract$1.152.580%24mmBtu GeneratingUnitPizriDemand(mmBtu) Unit1$2,719,32090%13,855,269Unit2$7,298,41290%34,058,704 ScrubberInstallationPSA:NoContractPSA:Unit1DemandPSA:Unit1andUnit2Demand None<$194.48<$140.69<$140.69Unit1($194.48,$230.89)($140.69,$230.89)($140.69,$153.61)Unit2$230.89$230.89$153.61

PAGE 193

ExcessDemandCorrespondence Figure3-2. ImpactofHighSulfurCoalContract 193

PAGE 194

ImpactofLowSulfurCoalContract Figure3-4. ExcessDemandCorrespondencewithScrubberChoice 194

PAGE 195

ImpactsofHighSulfurCoalContract Figure3-6. ImpactsofLowSulfurCoalContract 195

PAGE 196

GivenScrubberChoice:ShiftfromHighSulfurContract Figure3-8. GivenScrubberChoice:ShiftfromLowSulfurContract 196

PAGE 197

WithScrubberChoice:ShiftfromHighSulfurContract 197

PAGE 198

WithScrubberChoice:ShiftfromLowSulfurContract 198

PAGE 199

Withtheemissionsconstraintfromtheprogram,aunitwillminimizeitscostsofcoaluse,netallowancepurchases,andscrubberinstallation. Thedierencebetween( A{1 )and( A{2 )isthetotalcompliancecostsresultingfromtheprogram,whichincludesthechangeincoalcosts,changeinthenetallowancepurchases,andscrubberinstallationcosts. Second,consideraunit'stotalcostswithacoalcontractconstraint.Withouttheprogramrestrictionsonemissions,aunitwillsimplyminimizeitscostsofmeetingelectricitydemandbyusingallthethecoalundercontract,andcovertheremainderofitscoaldemandwiththecoalwiththelowestpriceperunitofheatcontent. Withtheemissionsconstraint,aunitwillminimizeitscostsforcoaluse,netallowancepurchases,andscrubberinstallationgivenitsemissionsandcoalcontractconstraint. 199

PAGE 200

A{4 )and( A{5 )isthetotalcompliancecostsresultingfromtheprogram,whichincludesthechangeinspotmarketcoalcosts,netallowancepurchases,andscrubberinstallation.Thecostsfromcontractcoalcanceloutbecausecontractcoalusewillbethesamebothwithandwithouttheprogram. Thesucientconditionsunderwhichacoalcontractconstraintwillincreaseordecreaseaunit'scompliancecostscanbederivedfromthedierenceincompliancecostswithandwithoutacoalcontract( A{6 )minus( A{3 ).hbziPiz+PAbAi+Psih(bCsihbCs0ih)+Psil(bCsilbCs0il)i Forsimplicity,assumethatthescrubberchoiceasagivenandhighsulfurspotmarketcoalisrelativelycheaperthanlowsulfurspotmarketcoal.Sowithoutanemissionsconstraintaunitwillprefertousethecheaperhighsulfurcoal.Proposition1canbeprovenbyconsideringthechangeincompliancecostsin( A{8 )resultingfromacoalcontract.FirstconsiderahighsulfurcoalcontracttoshowProposition1(i)andProposition1(ii)hold. 200

PAGE 201

Thechangeincompliancecostswillbe: ThersttermispositivebecausebAMINiAMINi.ThesecondtermisalsopositivebecauseCs;MAXihbCs;MAXih.ThethirdtermisnegativebecausebCs;MAXilCs;MAXil. Nowllinforcoaluse:Cs;MAXih=Di Nowllinforthenetallowanceposition:AMINi=Di 201

PAGE 202

Allcoaluseremainsthesame,whichmeansthechangeincompliancecostswillbetheincreaseincostsfromadditionalallowances.IfScih 202

PAGE 203

Thechangeincompliancecostswillbetheincreaseincostsfromtheincreaseinaunit'snetallowanceposition.IfScil Thechangeincompliancecostsisthechangeincostsfromthechangeinnetallowanceposition.Nowllinforthenetallowanceposition:AMAXi=Di SinceScil 203

PAGE 204

A{5 )-( A{2 ))andsplititintotwocomponents,thechangeincompliancecostsandthechangeinfuelcosts.Forsimplicity,assumetheconditionsinProposition1(i)hold. Assumethataunitfacesahighsulfurcoalcontract,andpreferstoswitchtolowsulfurcoalusetomeetitsemissionsrequirementinsteadofpurchasingallowancesorinstallingascrubber. Tobeabletointerpretthisexpression,itisnecessarytoaddandsubtract(Psih Thersttermisthechangeinhighsulfurcoalcostsfromusingthecontractcoalinsteadofspotmarketcoal.Thesearenotchangesincompliancecostsbecausetheywilloccurwithorwithouttheprogram.Theremainingtermsarethechangeincompliancecostsresultingfromtheprogram. {z }ChangeinFuelCosts+PA(bAMINiAMINi)+(Psih {z }ChangeinComplianceCosts(A{19) Byllinginforthecoaluse,thelasttermsgivethesameexpressionforthechangeincompliancecostsasintheproofofProposition1(i). Aunit'scompliancecostsincreaseifScih 204

PAGE 205

Assumingascrubberisinstalled,thecost-minimizingcombinationofinputsisexpressedbelow. Noticethatcontractcoaluseandthecostofinstallingascrubbercanbeignoredbecauseallareconstants. 205

PAGE 206

Theorem5.1:Assumingthescrubberchoice(zi)asgiven,amarketequilibriumexists. Kakutani'sFixedPointTheoremthatstates:IfXisanon-empty,compact,convexsubsetofRmandiffisanuppersemi-continuouscorrespondencefromXintoitselfsuchthat(8x2X)thesetf(x)isnon-emptyandconvex,thenfhasaxedpoint(thereisanx2f(x)). Itisnecessarytoshowthatthesetofpricesiscompact(closedandbounded).Denethelowerboundonpricetobezerobecauseitisnecessarytohaveapositiveprice.Denetheupperboundonpricetobe 206

PAGE 207

PA],whichisnon-emptybydesign.Sincethereisanupperbound,alowerbound,andbothboundsareincludedintheset(closed),thesetiscompact. Sincethesetofpricesandmarketexcessdemandsarecompactandthemarketexcessdemanddependsontheallowanceprice,themarketexcessdemand(Am(PA))isamappingfrom[0; PA]intoX.Itisnecessarytoshowthatexcessdemandisnon-empty,convexvaluedateachPA2[0; PA],anduppersemi-continuous.Itissucienttoshowthateachgeneratingunit'sexcessdemandcorrespondenceisnon-empty,convex,anduppersemi-continuousbecausethesumofanitenumberofnon-empty,convex,uppersemi-continuouscorrespondencesisalsonon-empty,convex,anduppersemi-continuous. Ageneratingunits'excessdemand(Ai(PA))aredenedtobemappingsfrom[0; PA]intoX,whichmakesthemnon-emptybyconstruction.Ageneratingunit'sexcessdemandcorrespondenceisclosedandboundedintheinterval[AMINi;AMAXi],whichmakesitacompactset.Thesetofexcessdemandsisconvexbecausetheaverageofanytwoexcessdemandvaluesisalsointheset.Foracorrespondencetobeuppersemi-continuous,theconvergenceexcessdemandvalueofanypricesequencemustalsobeinthecorrespondence.Sinceeverypossiblepricesequenceconvergestoavaluethatisintheexcessdemandcorrespondence(seeFigure B.2 ),eachunit'sexcessdemandcorrespondenceisuppersemi-continuous. NowwemustdenethemappingfromXintothesetofprices[0; PA]as(x)where... PA]:PAx=maxQ2[ PA]:PAx=0ifotherwise Since[0; PA]isnon-empty,(x)mustbenon-emptyaswell.Themappingisconvexvaluedsinceformarketexcessdemandequaltozero(x=0),(x)=[0; PA].Ifthereisa 207

PAGE 208

PA]iscompact,thegraphof(x)isclosed,whichimplies(x)isuppersemi-continuous. NowdeneF(x;PA)=(x)Am(PA).Since(x)andAm(PA)satisfyallpropertiesneededtoapplyKakutani'sFixedPointTheorem,thereexistsaxedpoint(x;PA)suchthatPA2[0; PA]andx2XsuchthatPA2(x)andx2Am(PA). Oncethescrubberchoiceisintroducedintothedecision-makingprocess,thecorrespondencebecomesmorecomplex,asseeninFigure B.2 .Anequilibriummayinfactexist,butthereisnowaytoguaranteeanequilibriumbecausetheexcessdemandcorrespondenceisnolongeraconvexset.Theaverageofthetwoexcessdemandvalues(ASMAXiandAMINi)isnotintheexcessdemandcorrespondence. Ifmarketexcessdemandispositive,theallowancepriceistoolowandtheallowancepriceisincreasedbyone-halfthedierencebetweentheupperandlowerlimits.Theoldpricenowbecomesthenewlowerlimitwhiletheupperlimitremainsthesame.Ifthemarketexcessdemandisnegative,theallowancepriceistoohighandtheallowancepriceisdecreasedbyone-halfthedierencebetweentheupperandlowerlimits.Theoldpricebecomesthenewupperlimitandthelowerlimitremainsthesame.Inthiscase,theupper 208

PAGE 209

Theprogramisthenrunagainwiththenewallowanceprice,eachiterationdecreasingthedierencebetweentheupperandlowerlimitsbyhalfuntiltheprogramconvergestoanallowanceprice.Oncetheprogramconverges,itmustensureamarketexcessdemandofzero.Aconcernisthattheprogramtendstopushaunit'schoicestowardsacornersolutionwherethemarketmaynotclear. Giventhescrubberchoice,theallowancepricewillconvergetoavalueequaltoatleastonegeneratingunit'sMCAs;si,whichallowsthosermschoicestobeshiftedtoaninteriorsolutiontoclearthemarketwithoutalteringtheunit'stotalcostsortotalindustrycostsbecausetheunitisindierenttopurchasingallowancesorswitchingfuelsfromhightolowsulfurcoalattheequilibriumallowanceprice(PA). AllowingforthescrubberchoicewillresultintheconvergenceofPAatanallowancepricewhereageneratingunitisindierenttoinstallingascrubber.Inthiscase,thereisnotatrueequilibriumandmustconsideritaquasi-equilibriumasdescribedabove. 209

PAGE 210

UpperSemi-ContinuousCorrespondence FigureB-2. CorrespondencewithScrubberChoice 210

PAGE 211

[1] T.Arimura.AnempiricalstudyoftheSO2allowancemarket:eectsofPUCregulations,J.Environ.Econ.Manage.,44(2)(2002)271{289. [2] R.E.Baldwin,C.S.Magee,Istradepolicyforsale?Congressionalvotingonrecenttradebills,PublicChoice105(2000)79{101. [3] L.Bird,M.Bolinger,T.Gagliano,R.Wiser,M.Brown,B.Parsons.PoliciesandmarketfactorsdrivingwindpowerdevelopmentintheUnitedStates,EnergyPol.33(11)(2005),1397{1407. [4] D.R.Bohi.Utilitiesandstateregulatorsarefailingtotakeadvantageofemissionallowancetrading,ElectricityJ.,7(2)(1994)20{27. [5] D.R.Bohi,D.Burtraw.Utilityinvestmentbehaviorandtheemissiontradingmarket,ResourcesandEnergy,14(1992)129{153. [6] D.R.Bohi,D.Burtraw.SO2allowancetrading:Howexperienceandexpectationsmeasureup,DiscussionPaper97{24,ResourcesfortheFuture,Washington,D.C.,1997. [7] M.Bolinger,R.Wiser.2004.Theimpactofstatecleanenergyfundsupportforutility-scalerenewableenergyprojects,LawrenceBerkeleyNationalLaboratoryandCleanEnergyStatesAlliance. [8] M.Bolinger,R.Wiser,G.Fitzgerald.2006.Theimpactofstatecleanenergyfundsupportforutility-scalerenewableenergyprojects,LawrenceBerkeleyNationalLaboratoryandCleanEnergyStatesAlliance. [9] M.Bolinger,R.Wiser,G.Fitzgerald.Anoverviewofinvestmentsbystaterenewableenergyfundsinlarge-scalerenewablegenerationprojects,ElectricityJ.18(1)(2005)78{84. [10] M.Bolinger,R.Wiser,L.Milford,M.Stoddard,K.Porter.Cleanenergyfunds:Anoverviewofstatesupportforrenewableenergy,EnergyEciencyandRenewableEnergy,OceofPowerTechnologies,DepartmentofEnergy,2001. [11] D.Burtraw.Costsavingssansallowancetrades?EvaluatingtheSO2emissiontradingprogramtodate,DiscussionPaper95{30,ResourcesfortheFuture,Washington,D.C.,1995. [12] D.Burtraw,Evans,A.Krupnick,K.Palmer,R.Toth.EconomicsofpollutiontradingforSO2andNOX,DiscussionPaper05{05,ResourcesfortheFuture,Washington,D.C.,2005. [13] C.Carlson,D.Burtraw,M.Cropper,K.Palmer.Sulfurdioxidecontrolbyelectricutilities:Whatarethegainsfromtrade?,J.ofPoliticalEconomy,108(6)(2000)1292{1326. 211

PAGE 212

C.Chen,R.Wiser,M.Bolinger,2007.Weightingthecostsandbenetsofstaterenewablesportfoliostandards:Acomparativeanalysisofstate-levelpolicyimpactprojections,LawrenceBerkeleyNationalLaboratoy. [15] J.S.Coggins,J.R.Swinton.Thepriceofpollution:AdualapproachtovaluingSO2allowances,J.Environ.Econ.Manage.,30(1996)58{72. [16] DatabaseofStateIncentivesforRenewableEnergy.Staterenewableenergypolicysummaries,Website,AccessedMarch2004{March2006, [17] DatabaseofUtility-ScaleRenewableEnergyProjects.LawrenceBerkeleyNationalLaboratoryandCleanEnergyStateAlliance(CESA),Website,AccessedDecember2006, Renewables Database.xls. [18] A.D.Ellerman,R.Schmalensee,P.L.Joskow,J.P.Montero,E.M.Bailey.EmissionstradingundertheU.S.acidrainprogram:Evaluationofcompliancecostsandallowancemarketperformance,CenterforEnergyandEnvironmentalPolicyResearch,MassachusettsInstituteofTechnology. [19] A.D.Ellerman,etal.Marketforcleanair:theU.S.acidrainprogram,CambridgeUniversityPress,NewYork,2000. [20] A.D.Ellerman,J.P.Montero.Thedecliningtrendinsulfurdioxideemissions:Implicationsforallowanceprices,J.Environ.Econ.Manage.,36(1998)26{45. [21] EnergyInformationAdministration.Electricpowerannual1996:Vol.2,DOE/EIA{0348(96)/2,February1996. [22] EnergyInformationAdministration.TheeectsofTitleIVoftheCleanAirActAmendmentsof1990onelectricutilities:Anupdate,DOE/EIA{0582(97),March1997. [23] EnergyInformationAdministration.EnergyPolicyActtransportationratestudy:Finalreportoncoaltransportation,DOE/EIA{0597(2000),October2000. [24] EnergyInformationAdministration.EIAhistoricalstateelectricitydatabasesandFERCForm{423database,Website,AccessedMarch2004{March2007, [25] EnergyInformationAdministration.AnnualEnergyReview2005.Report#DOE{EIA{0384(2005).Website,AccessedFebruary2007, [26] EnvironmentalProtectionAgency.Reducingpowerplantemissionsforcleanerair,healthierpeople,andastrongAmerica.OceofAirandRediation, les/cair nal presentation.pdf,March2005. 212

PAGE 213

EnvironmentalProtectionAgency.Acidrainprogramprogressreport,EPA{430{R{06{015,CleanAirMarketsDivisionOceofAirandRadiation,October2006. [28] EnvironmentalProtectionAgency.InformationontheCleanAirInterstateRule.Website,AccessedSeptember2007, [29] FederalReserveBranchofSt.Louis.Consumerpriceindexdata.Website,AccessedDecember2004, [30] D.Fullerton,S.McDermott,J.P.Caulkins.Sulfurdioxidecomplianceofaregulatedindustry,J.Environ.Econ.Manage.,34(1)(1997)32{53. [31] GDSAssociates,Inc.AnalysisofrenewablesportfoliostandardoptionsforHawaii,TheStateofHawaiiDepartmentofBusiness,EconomicDevelopment,andTourism,2001 [32] P.L.Joskow.Verticalintegrationandlong-termcontracts:Thecaseofcoal-burningelectricgeneratingplants,J.ofLaw,Econ.,andOrg.,1(1)(1985)33{80. [33] P.L.Joskow.Contractdurationandrelationship-specicinvestments:Empiricalevidencefromcoalmarkets,AmericanEconomicReview,77(1)(1987)168{185. [34] P.L.Joskow.Priceadjustmentinlong-termcontracts:Thecaseofcoal,J.ofLawandEconomics,31(1)(1988)47{83. [35] P.L.Joskow.Assetspecicityandthestructureofverticalrelationships:Empiricalevidence,J.ofLaw,Econ.,andOrg.,4(1)(1988)95{117. [36] P.L.Joskow.Theperformanceoflong-termcontracts:Furtherevidencefromcoalmarkets,RANDJ.ofEcon.,21(2)(1990)251{274. [37] J.P.Kalt,M.A.Zupan.Captureandideologyintheeconomictheoryofpolitics,AmericanEconomicReview74(3)(1984),279{300. [38] N.O.Keohane.Environmentalpolicyandthechoiceofabatementtechnique:Evidencefromcoal-redpowerplants.YaleSchoolofManagement,2002. [39] O.Langniss,R.Wiser.TherenewablesportfoliostandardinTexas:Anearlyassessment,EnergyPol.31(6)(2003),527{535. [40] LeagueofConservationVoters.DescriptionanddatafromtheLCV\NationalScorecard".Website,AccessedMarch2004, [41] J.McVeigh,D.Burtraw,J.Darmstadter,K.Palmer.Winner,loser,orinnocentcictim?Hasrenewableenergyperformedasexpected?,DiscussionPaper99{28,ResourcesfortheFuture,Washington,D.C.,1999. [42] F.C.Menz,S.Vachon.TheEectivenessofdierentpolicyregimesforpromotingwindpower:Experiencesfromthestates,EnergyPol.34(2006)1786{1796. 213

PAGE 214

J.P.Montero.Marketablepollutionpermitswithuncertaintyandtransactioncosts,ResourceEnergyEcon.,20(1997)27{50. [44] D.W.Montgomery.Marketsinlicensesandecientpollutioncontrolprograms,J.Econ.Theory,5(1972)395{418. [45] G.Morris.ThestatusofbiomasspowergenerationinCalifornia.SubcontractorReport.NationalRenewableEnergyLaboratory.GreenPowerInstitute,PacicInstituteforStudiesinDevelopment,Environment,andSecurity,2003. [46] J.P.Nelson.Greenvotingandideology:LCVscoresandroll-callvotingintheU.S.Senate,1988{1998,ReviewofEcon.andStat.84(3)(2003),518{529. [47] T.Petersik.Staterenewableenergyrequirementsandgoals:statusthrough2003.EnergyInformationAssociation,2004. [48] B.G.Rabe.Racetothetop:TheexpandingroleofU.S.staterenewableportfoliostandards,PewCenteronGlobalClimateChange,2006. [49] A.E.Smith,J.Platt,A.D.Ellerman.ThecostsofreducingutilitySO2emissions-Notaslowasyouthink,CharlesRiverAssociates,ElectricPowerResearchInstitute,andMassachusettsInstituteofTechnology. [50] D.C.Smith,R.C.Grace,R.Wiser.Massachusettsrenewableportfoliostandard:Costanalysisreport,LaCapraAssociatesandSustainableEnergyAdvantage,LLCundercontracttoMassachusettsDivisionofEnergyResources,2000. [51] P.M.Sotkiewicz.Theimpactofstate-levelpublicutilitycommissionregulationonthemarketforsulfurdioxideallowances,compliancecosts,andthedistributionofemissions.DoctoralDissertation,Dept.ofEconomics,UniversityofMinnesota,January2003postedat [52] P.M.Sotkiewicz,L.Holt.Publicutilitycommissionregulationandcost-eectivenessofTitleIV:LessonsforCAIR,ElectricityJ.,October2005,68{80. [53] R.Stavins.Transactioncostsandtradeablepermits,J.Environ.Econ.Manage.,29(1995),133{148. [54] B.Swift.Howenvironmentallawwork:Ananalysisoftheutilitysector'sresponsetoregulationofnitrogenoxidesandsulfurdioxideundertheCleanAirAct,TulaneEnvironmentalLawJ.,14(1)(2001)309{426. [55] J.R.Swinton.PhaseIcompleted:Anempiricalassessmentofthe1990CAAA,EnvironmentalandResourceEcon.,27(2004)227{246 [56] J.R.Swinton.Thepotentialforcostsavingsfromthesulfurdioxideallowancemarket:EmpiricalevidencefromFlorida,LandEcon.,78:3(2002)390{404. 214

PAGE 215

J.J.Winebrake,A.E.Farrell,M.A.Bernstein.TheCleanAirAct'ssulfurdioxideemissionsmarket:Estimatingthecostsofregulatoryandlegislativeintervention,ResourceEnergyEcon.,17(1995)239{260. [58] R.Wiser,S.Olson.Utilitygreenpricingprograms:Astatisticalanalysisofprogrameectiveness,EnvironmentalEnergyTechnologiesDivision,LawrenceBerkeleyNationalLaboratory,2004. [59] R.Wiser,K.Porter,R.Grace.EvaluatingexperiencewithrenewablesportfoliostandardsintheUnitedStates,EnvironmentalEnergyTechnologiesDivision,LawrenceBerkeleyNationalLaboratory,2004. 215

PAGE 216

JoshuaDavidKneifelwasbornin1981inNorthPlatte,Nebraska.HegrewupinNorthPlatte,graduatingsalutatorianfromHersheyHighSchoolin1999.Joshuareceivedhisbachelor'sdegreesineconomicsandmathematicsin2003fromDoaneCollegeinCrete,Nebraska.HereceivedhisMasteroftheArtsineconomicsin2005fromtheUniversityofFlorida,wherehespecializedinindustrialorganization,publiceconomics,andeconometrics.FromFall2006throughSpring2008,Joshuainstructedfoursemestersofacourseinenvironmentaleconomics.HisclassworkandresearchallowedhimtoobtainhisPhDineconomicsfromtheUniversityofFlorida.UponcompletionofhisPhDprogram,hewilltakeaneconomistpositionattheNationalInstituteofStandardsandTechnology(NIST)inGaithersburg,Maryland.HisresponsibilitiesatNISTwillincluderesearchonthelife-cyclecostandenvironmentalimpactsofindividualproductsusedintheconstructionindustry. 216