EPA-600/5-73-014
DECEMBER 1973
Socioeconomic Environmental Studies Series
Enforcement Economics In
ir Pollution Control
I
55
\
UJ
Washington Environmental Research Center
Office of Research and Development
U.S. Environmental Protection Agency
Washington, D.C. 20460
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and appli-
cation of environmental technology. Elimination of traditional grouping
was consciously planned to foster technology transfer and a maximum inter-
face in related fields. The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the SOCIOECONOMIC ENVIRONMENTAL STUDIES
series. This series includes research that will assist EPA in implement-
ing its environmental protection responsibilities. This includes examining
alternative approaches to environmental protection; supporting social and
economic research; identifying new pollution control needs and alternate
control strategies; and estimating direct social, physical, and economic
cost impacts of environmental pollution.
EPA REVIEW NOTICE
This report has been reviewed by the Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
-------
EPA-600/5-73-014
December 1973
ENFORCEMENT ECONOMICS IN
AIR POLLUTION CONTROL
by
Paul B. Downing
Visiting Associate Professor of Economics
Virginia Polytechnic Institute and State University
Blacksburg, Virginia 24061
and
William D. Watson, Jr.
Washington Environmental Research Center
Implementation Research Division
Environmental Protection Agency
Washington, D.C. 20460
Program Element 1HA094
WASHINGTON ENVIRONMENTAL RESEARCH CENTER
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON D.C., 20460
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.O. 20102 - Price $1.60
-------
ABSTRACT
This report investigates the effects of alternative enforcement
strategies on the pollution control activities of the firm. There are
a number of tradeoffs available to a firm including delay and non-
compliance which allow it to minimize expected pollution control
costs. These are identified within the context of a generalized
behavioral model for the firm and an empirical study is undertaken to
determine their importance.
In a simulation of current enforcement of the federal new source
particulate matter discharge standard for coal-fired power plants
(start-up compliance or certification tests for pollution control
devices plus fines for violating in-operation emission standards) it
is found that cost-minimizing power plants will install relatively
costly pollution control technology and frequently violate federal fly
ash standards. Two alternative enforcement strategies for overcoming
these shortcomings, namely compliance tests in combination with
emission taxes and emission taxes alone, are analyzed.
It is recommended that enforcement agencies give careful consid-
eration to management costs imposed upon the firm and the control
agency by an implementation and enforcement scheme. In the case of
the federal fly ash discharge standard for coal-fired power plants it
is tentatively concluded that emission tax enforcement would probably
result in an approximate minimization of the sum of firm and
enforcement agency resource costs. The general applicability of this
result to other enforcement problems is discussed.
ii
-------
CONTENTS
Page
Abstract ii
List of Figures iv
List of Tables v
Preface vi
Sections
I Summary, Recommendations, and Conclusions 1
II Optimal Control 10
III An Enforcement Model of the Firm's Control Behavior 16
IV A Simulation of Enforcement Alternatives 29
V Policy Analysis 57
VI Policy Recommendations 70
VII Interpreting the Results 78
Footnotes 82
References 85
Mathematical Appendix 87
iii
-------
FIGURES
No. Page
1 Standards Enforcement 12
2 Distribution of Measurement Errors for Two 24
Alternative Emission Monitors
3 Alternative Forms of Penalties for Air Pollution 27
Violations
4 Precipitator Operating Curves 33
5 Simulation Model, Scenarios S1-S4 35
6 Fines and Emission Taxes 38
7 Compliance Test Tradeoffs 40
8 Simulation Model, Scenarios S5 and S6 43
9 Fines and Control Effort 54
10 Cost Comparisons 58
11 Enforcement by Compliance Test and Opacity 61
Standard, 1300 Megawatt Plant
12 Enforcement by Compliance Test and Emission Tax, 64
1300 Megawatt Plant
13 Technology-Cost Tradeoffs (Current Enforcement 67
Practice)
14 Control Effort (Current Enforcement Practice) 71
15 Effective Legal Enforcement 74
16 Effective Emission Tax Enforcement 76
A.I Beta Distribution Shapes Programmed in Monte Carlo 91
Model
iv
-------
TABLES
No. Page
1 Regression Coefficients from Simulation Analyses, 47
SI: Compliance Test with Fine for Violating an
Opacity Standard (Inflexible Technology)
2 Regression Coefficients from Simulation Analyses, 48
S2: Compliance Test with Fine for Violating an
Opacity Standard (Flexible Technology)
3 Regression Coefficients from Simulation Analyses, 49
S3: Compliance Test with Tax on Emitted Fly Ash
(Inflexible Technology)
4 Regression Coefficients from Simulation Analyses, 50
S4: Compliance Test with Tax on Emitted Fly Ash
(Flexible Technology)
5 Regression Coefficients from Simulation Analyses, 51
S5 and S6: Emission Tax Only
6 Current Enforcement Practice 56
A.I Nonmenclature 92-94
A.2 Electrostatic Precipitation Parameters 95
A.3 Compliance Test Failure Probabilities (R - 1.15V, 96
Compliance Test Standard » .04)
A.4 Compliance Test Failure Probabilities (R - 1.1V, 97
Compliance Test Standard = .04)
A.5 Compliance Test Failure Probabilities (R = 1.15V, 98
Compliance Test Standard = .1)
A.6 Compliance Test Failure Probabilities (R = 1.1V, 99
Compliance Test Standard = .1) _L
A.7 Precipitator Characteristics (Inflexible Technology) 100
A.8 Precipitator Characteristics (Flexible Technology) 101
A.9 Characteristic of Cost Parameter Distributions 102
A.10 Fixed "Costing" Parameters 103
A.11 Representative Flue Gas Volumes and Precipitator 104
Sectionalization for Different Sized Power Plants
-------
PREFACE
This report is part of the Implementation Research Division's
(IRD) research program on the implementation aspects of environmental
pollution control. As a part of this program, the Division is engaged
in comprehensive studies of the economics of pollution control
regulatory activities. A major goal of these studies is to provide
information on and insights into the economic aspects of determining
environmental standards and pollution control strategies.
The body of information presented in this report is directed to
those individuals concerned with efficient and cost-effective
enforcement of environmental standards. Its aim is twofold. One is
to determine some of the differing impacts upon firm behavior of legal
enforcement, of economic incentive enforcement, and of mixed
legal-economic enforcement. A second objective is to initiate
identification of enforcement systems which are most likely to
minimize resource costs to firms and enforcement agencies of meeting
environmental standards.
Several other IRD studies deal with related subjects. They
include "An Economic Analysis of Periodic Vehicle Inspection Programs"
by Paul B. Downing (Atmospheric Environment, Dec. 1973), "A Cost
Evaluation of Alternative Air Quality Control Strategies" by Donald H.
Lewis and Scott E. Atkinson (forthcoming) and "Costs and Benefits of
Fly Ash Control" by William D. Watson, Jr. (Journal of Economics and
Business, May 1974).
The authors are grateful to their colleagues in IRD for many
helpful comments at various stages of this research. The study has
benefited substantially from their suggestions.
•
Alan Carlin
Director, IRD
vi
-------
SECTION I
SUMMARY, RECOMMENDATIONS, AND CONCLUSIONS
Current pollution control efforts are directed mainly at
establishing environmental standards which protect human health and
welfare. This is a fundamental step toward improving environmental
quality and it is not surprising that a great deal of attention is
presently focused on this activity. One must be careful, however, not
to overlook important interfaces or couplings in this endeavor. For
example, it is very important to anticipate reactions both to
environmental standards and the methods used to enforce them. Indeed,
in this report we show that standards setting cannot be divorced from
enforcement. Several examples suffice to illustrate the linkages.
For instance, a standard—whether it be tight or lax—which is not
enforced may lead to excessive pollution damages to human health and.
welfare. On the other hand, heavy handed enforcement may lead to very
high enforcement costs which may also reduce human welfare because
large amounts of resources would be devoted to enforcement at the
expense of attractive alternative employments. In actual fact, the
situation may be somewhat more complicated than this since complex
feedbacks may induce counterproductive behavior. A very strict
standard may lead a control agency to engage in vigorous enforcement.
But the firm being regulated may resist via legal maneuverings simply
because this is less costly than controlling pollution. This in turn
could lead the control agency and firm to engage in further legal
battles, all of which results in spiralling enforcement costs but
little pollution control.
Our objective in this report has been to theoretically and
empirically model firm and pollution control agency pollution control
behavior in sufficient detail so that we can determine enforcement
policies which minimize the total resource costs (insofar as they can
be measured) of meeting environmental standards. Resource costs are
defined to include out-of-pocket pollution control costs to the firm
(such as the capital and operating costs of pollution control
-------
equipment), firm management costs (such as the costs of monitoring
discharges and conducting start-up compliance tests for pollution
control devices), pollution control agency enforcement costs (such as
the costs of inspection and preparing legal suits against firms
accused of violating standards), and the damage costs of residual or
after-control pollution. We do consider other firm costs such as
fines and emission charges but since these are transfer costs and not
resource costs they are excluded in identifying enforcement policies
which minimize resource costs. The analysis has both a static and
dynamic dimension. The static analysis investigates enforcement
responses given current technology; the dynamic analysis attempts to
determine the type of enforcement policies which provide incentives to
firms to develop and adopt resource efficient pollution control
technology over time.
It is assumed throughout that firms are primarily motivated by
the desire to minimize their expected costs, or obversely for fixed
prices and outputs, to maximize their expected profits. In the
theoretical sections of this report we comprehensively cover possible
reactions to pollution control enforcement including such alternatives
as "public relations" on the part of both firms and enforcement
agencies. In response to enforcement policies, firms actually have a
wide variety of alternative reactions. These can range all the way
from complete compliance to delay and non-compliance wherein firms
legally challenge enforcement and use public relations to advertise
"their side of the story". Our analysis assumes firms will weigh the
costs of each of these actions and choose the least costly
alternatives. For their part, enforcement agencies also have similar
choices. For instance they may choose to publicly disclose
uncooperative and recalcitrant behavior on the part of non-complying
firms. It is, of course, in the interest of firms to anticipate this
and to act accordingly. Our theoretical analysis allows tradeoffs
along these lines. The empirical part of this report, however, is
constrained by data availabilities. Here we undertake a much less
-------
ambituous analysis of enforcement behavior. We investigate responses
of regulated firms only; enforcement agency behavior is not modeled
due to lack of data. Furthermore, this empirical analysis does not
allow for subtle variations such as those produced by public relations
efforts. Nonetheless our empirical analysis does produce a variety of
interesting and useful policy implications and we do point out the
sensitivity of our conclusions to missing links.
The empirical section considers three alternative enforcement
policies. One, termed current enforcement practice because it is
modeled after enforcement guidelines promulgated by EPA, uses start-up
compliance or certification tests for pollution control devices and
fines for violating in-operation emission standards. A second
enforcement policy which we consider uses compliance tests in
combination with per unit taxes (emission or effluent taxes or fees)
on discharged pollutants. A third policy is emission taxes alone.
Allowance is made in the analysis for firm influence on compliance
test conditions and on fines and probabilities of conviction when
emission standards are violated.
It is useful to differentiate the two different but related
optimizing orientations covered by the analysis in this report. One
is that of economic efficiency. The other is cost-effectiveness. An
economically efficient pollution control level and enforcement policy
is identified as a level of pollution control and an enforcement
policy for which marginal resource costs of control equal marginal
benefits and for which total net benefits are maximized. This is also
equivalent to a policy which minimizes total resource costs when
resource costs are defined comprehensively, that is, when they include
internal resource costs to firms and enforcement agencies and total
external damage costs of discharged pollutants. In a norrower
framework where marginal benefits or pollution damage costs are not
known it is not possible to determine efficient policies;
consequently, in such cases the analysis focuses on cost-effective
policies which are defined as enforcement policies which minimize the
-------
internal resource costs (i.e., excluding external pollution damage costs)
of meeting any given environmental standard. Our theoretical analysis
focuses on both efficient and cost-effective enforcement policies while
our empirical analysis focuses mainly on cost-effective policies.
There are two basic messages of our analysis. One is that there are
many alternatives for setting and enforcing a pollution control standard.
These can range from systems which rely mainly upon legal sanctions to
systems which rely totally upon economic incentives. In any event it is
probably desirable to identify and implement efficient policies if that
is possible, and if not (say due to lack of data on pollution control
benefits) then an attempt should be made to identify and implement
cost-effective policies, that is, enforcement policies which minimize the
sum of resource costs to firms and enforcement agencies of meeting any
given environmental standard. Our analysis provides guidelines for
implementing efficient and cost-effective policies. A second message,
especially relevant for control of stationary source pollution, is that
current enforcement practice is probably not cost-effective. In this
report we identify several alternative methods of enforcement which would
probably substantially reduce internal pollution control resource costs
below the levels achievable under current enforcement practice.
The following specific conclusions and recommendations have emerged
from our work:
— The optimal level of control of emissions depends upon the
•?r
cost of the control devices or process changes, the management costs
imposed on the firm by the control agency, and the management cost of
the control agency itself (and, of course, the benefits of control).
These costs are likely to differ among alternative implementation and
enforcement schemes. In order to determine the optimal implementation
and enforcement scheme it is necessary to determine the optimal
control level, and hence values of policy parameters, for each
alternative. The net benefit of control for each alternative could
then be compared and the scheme with the largest net benefit chosen.
While we cannot prove it without further research, the evidence we
-------
present indicates that an effluent fee enforcement scheme would be
optimal in controlling fly ash emissions from coal-fired power plants.
However, we do not expect this result to apply universally to other
situations. Some form of legal enforcement may be preferred in many
cases. This is especially true in cases where continuous monitoring
of emissions is technically difficult and expensive.
— Our analysis indicates that when information and management
costs are included the optimal effluent fee system consists of a
marginal charge and a lump sum charge. The marginal charge would be
set equal to the firm's marginal control cost, including its internal
management costs, at the point where the optimal control would be
obtained. The fee is less than marginal benefits at the optimal
control level. The lump sum charge would be based upon the control
agency's management costs. Not including the lump sum allows firms to
bear less than the full social cost of control thus leading to
inefficiently large output of final goods and pollution.
— Assuming that firms are expected cost minimizers we find that
different implementation and enforcement techniques imply different
reactions to control agency policy. Under a legal enforcement system
the relevant policy parameters are inspection and monitoring
techniques, emission standards, device certification procedures,
probability of conviction if accused of a violation, fines and
shutdown penalties, and damage to the corporate image. As one would
expect there are tradeoffs among these policy parameters. For
example, in our simulation of fly ash control we find that stricter
compliance tests (a certification of a control device) and less
stringent opacity (emission) standards can yield the same level of
control. The model indicates that a higher marginal fine or penalty
would yield greater control. In our emprirical case, however, we find
that any positive effective fine will have the same effect on the
firm's control decision. This probably is not a general result.
In effluent fee enforcement the relevant policy parameters are
the marginal fee, the device certification process if any, and the
-------
inspection and monitoring system employed. As expected, higher
effluent fees yield greater control. When a certification procedure
is added to the effluent fee we find that a tradeoff between the
certification standard and the effluent fee exists. This is born out
in our empirical test. However, there is a range of effluent fees for
which any feasible compliance test will have no effect on the firm's
control efforts.
— We find that very high accuracies in monitoring devices are
not needed to determine compliance with some desired pollution control
standard. All that is necessary is a monitor which has a known
relationship between what it measures and the pollutant to be
controlled and a known measurement error. Thus, efforts should be
directed toward developing monitoring systems for difficult to measure
emissions rather than improving the accuracy of already adequate
monitors.
With an adequate monitoring device, the control agency can adjust
the emission standard or fee to fully account for measurement errors.
The confidence level at which they decide that a violation has
occurred is a function of the costs of making Type I and Type II
errors. The higher the cost of not stopping violators in terms of
damages from pollution, the lower the confidence level (or higher the
probability of incorrect accusations) the control agency should pick.
— Our analysis shows that when a plant fails a compliance test,
an enforcement agency must be willing at all times to say to the
operator of that plant: "You cannot open your plant." Without this
threat firms will install and operate grossly inadequate pollution
control devices especially under enforcement via start-up compliance
tests of control devices and fines for violating in-operation emission
standards (current enforcement practice), less so under enforcement
via compliance tests combined with, per unit emission taxes.
— Under current enforcement practice, the threat of almost any
positive effective fine when the emission standard is violated is a
necessary condition for enforcement success. Positive effective fines
-------
encourage firms to maintain their pollution control equipment.
— Under enforcement schemes using compliance tests, our
analysis indicates that plants, especially large ones, will vigorously
seek relaxations in the conditions under which compliance tests are
conducted. The reason is that low compliance test flue gas flow
rates, and small numbers of averaged compliance tests, and large
numbers of successive re-runs of the compliance test reduce fail
probabilities, making "shoddy" devices with their smaller costs, least
costly. Obviously, an enforcement agency in seeking effective
compliance should attempt to prevent such relaxations. Unfortunately,
federally promulgated guidelines already permit as few as three
averaged stack samples during compliance tests for fly ash control and
an unlimited number of successive compliance tests.
— Under current enforcement practice, most coal-fired power
plants will not meet federal new source fly ash standards. Since our
analysis is not unusual in any way, we suspect that this
non-compliance result also applies to some degree to similar
enforcement practices for other pollution standards.
— Under current enforcement practice, small power plants in
comparison with large plants will control at higher levels which is
inefficient. This is likely to hold generally for any enforcement
systems which use compliance tests and fines for violating
in-operation emission standards.
— Current enforcement practice for pollution control is likely
to lead to some reductions in pollution, but it does this with a
rather severe dynamic penalty. Our analysis indicates that such
enforcement will probably lead to the selection of inflexible
technology and to negative economic incentives towards the development
and adoption of more flexible and consequently less polluting
abatement technologies including process modification. By 1980, extra
stationary source pollution control resource costs for the U.S.
stemming froci mis-directed technology selection under current
enforcement practice are likely, at the very least, to be running at
-------
the rate of $75 to $150 million per year. There is the further danger
under current enforcement practice (no emission taxes) that damages
are going to be suboptimally high because firms will not be paying the
full social costs of their emissions. Without emission taxes, firms
may produce more that the optimal level of output and emissions.
— Compliance tests with emission taxes or emission taxes only
are two alternative enforcement policies which would overcome most of
the deficiencies of current enforcement practice. While it will not
be universally the case that legal or current enforcement practice is
less preferable, we feel that in a high percentage of the cases it
will be inferior to effluent fee enforcement.
— Effluent fee enforcement provides incentives toward the
adoption of efficient technology. Furthermore, since effluent charges
are immediate there is little the firm can do to avoid compliance.
There is, however, one sense in which firms could avoid or delay
compliance under effluent tax enforcement. This is by initial
challenges to effluent tax legislation. Our simulation results
indicate that effective emission tax enforcement of new source federal
fly ash standards for coal-fired power plants can raise costs by as
much as 25% above costs incurred under current enforcement practice.
This means, of course, that there are substantial cost saving payoffs
to firms from preventing effluent tax enforcement of pollution
standards. The message for pollution control agencies is that
substantial legal resources may have to be devoted to an initial legal
defense of effluent tax legislation.
— Once the initial challenge to effluent tax enforcement has
been met, there is likely to be substantial enforcement cost savings
to those pollution control agencies using effluent tax enforcement of
pollution standards. Right away, compliance tests and the costs of
policing them can be eliminated. There is also almost no need to
retain a staff of enforcement agency lawyers who periodically threaten
to prosecute violating firms under the civil suit provisions of the
Clean Air Act, the effluent charge itself now more effectively plays
8
-------
this role. Firms also have less incentive to retain lawyers for
purposes of delaying enforcement; effluent charges provide immediate
incentives towards control and consequently firms would tend to shift
resources away from delaying and non-compliance tactics towards
pollution control activities. Policing of stack monitoring is the one
activity to which a pollution control agency must devote substantial
resources under effluent tax enforcement. Cost minimizing firms will
achieve high collection levels only if full and proper effluent
charges are levied. Honest and hence carefully policed stack
monitoring is a necessary condition for effluent fee enforcement.
-------
SECTION II
OPTIMAL CONTROL
Overview. The use of the environment by a firm can impose
uncompensated costs on other firms or on individuals. There are two
general methods which may be employed to internalize these costs to
the polluting firm; namely, emission standards and emission charges.—
In assessing the cost of pollution control typical studies look only
at the cost of the control device or process change without concern
for the institutional constraints placed on the firm by the control
agency and the legislature. Yet it is clear that the firm incurs
differential expenses in addition to (or instead of) the actual
installation and operation costs of the control device or process
change itself. These additional expenses can include compliance
testing or other certification expenses, legal expenses, fines, and
other enforcement costs. These expenses are a function of the
implementation and enforcement rules employed by the control agency.
Hence they are likely to vary with the method of internalization
(policy instrument) chosen.
The goal of this paper is to determine the likely effect on a
firm's control actions of alternative implementation and enforcement
policies available to the control agency. Three alternatives are
studied, legal enforcement through the new source performance
standards set forth by EPA and two effluent fee enforcement
alternatives. First, a generalized model of the effects of
implementation and enforcement policies on the firm's control actions
is developed. This model assumes that the firm is an expected cost
minimizer. The model is then applied to the case of particulate
matter discharges from coal-fired power plants in order to estimate
empirically the effect of policy alternatives on the firm's control
efforts. Finally, the results of the model and its empirical
application are used to develop policy functions which relate control
to the values of various policy parameters. These results lead us to
several policy recommendations.
10
-------
Optimal Emission Standards and Taxes. Before we proceed with the
development of our model, a general framework is provided by
investigating how the cost to the firm of complying with control
requirements and the cost to society of insuring that the firm
2/
complies affect the optimal level of pollution control.— It is likely
that both these costs will differ between the two implementation and
enforcement alternatives. Let us first investigate legal enforcement
and then turn our attention to effluent fee enforcement. In Figure la
we plot increasing percent removal of a pollutant (R) on the
horizontal axis and dollar costs on the vertical axis. The marginal
cost of a control device (MCDT.,) increases as removal increases. This
LK
is the cost function measured in the usual control cost study.
However, the cost of the device is not the full cost born by the firm.
Depending upon the form of legal enforcement the firm may have to
conduct compliance tests, incur monitoring costs, keep records and
meet other requirements imposed by the control agency. Interpreting
these curves as planning horizon cost curves it is clear that at least
some of these compliance costs vary with R. Thus the marginal cost of
control for legal enforcement (MCCT1,) which the firm actually faces
Lri
includes both MCDT_ and these other costs and lies above MCDT1,.
LE LE
The marginal social cost of control using legal enforcement
(MSC__) includes the costs to the firm (MCC__) and the cost to the
LiCi li£i
control agency of carrying out enforcement activities in an attempt to
insure that its rules and regulations are carried out (MMCL _). The
control agency must inspect the site to determine that the firm has
the required controls installed and operating and that it does not
cheat by turning the devices off when the control agency personnel are
not around. It is reasonable to assume that at least some of these
costs vary with the level of removal. This is because it is likely
that the payoff to cheating will increase as the required level of
control increases. Control agency enforcement efforts should increase
in an attempt to counteract this incentive.
Assuming the usual declining marginal benefit function (MB), the
11
-------
Figure 1
Standards Enforcement
(a)
Legal Enforcement
LE
R,
MSCLE=MCCLE
MCC
LE
MCD,
LE
%Removal
(b)
Effluent Fee Enforcement
MSCEF=MCCEF
EF,
EF
EF*
2
X
TMC
0
SEF R3
MCC
'EF
MCD
EF
emoval
12
-------
optimal level of control would be where MSC__ = MB or S__ in Figure
LE LE
la. Note that when it is recognized that social control costs are
greater than the cost of the device itself, the optimal level of
control of pollution is less than that usually determined in empirical
studies (R-). The neglect of these costs would lead to the setting of
a standard which is inefficiently stringent.
In Figure Ib the same conceptual set of functions is presented
for the effluent fee enforcement case. However, each of these
functions may differ from their legal enforcement equivalents in their
actual location on the graph. There are compliance costs for the
effluent fee enforcement system as well. The firm must record
emissions, pay the fee, deal with periodic checks by control agency
personnel, etc. It is reasonable to assume that these compliance
costs would increase with the level of removal. Likewise, the
marginal management costs to the control agency are likely to increase
with the level of removal. This is because higher removal and
consequently greater effluent fees makes cheating more profitable to
the firm. This in turn necessitates greater checking by the control
agency. The optimal control level is at $„_ where MSC = MB (which
£j£ CiS
does not change with the enforcement technique employed).
If society's goal is to control pollution at least cost (and if
it wished to neglect distributional issues), it should pick that
institutional form which is least costly. Economists have often
argued that the best institutional form for pollution control is the
effluent fee. For this to be true it is necessary that the net social
benefit of control for the effluent fee enforcement system is greater
than the net social benefit of control for the legal enforcement
system where each is at its optimal level (i.e., MSC,- = MB and MSC F
= MB). In order to determine if the economists argument is correct it
is necessary to know both MCC and MMC under legal enforcement and
effluent fee enforcement. While logical arguments can be made to
support the economist's argument, the other side also has merit. The
determination will probably rest on empirical evidence yet to our
13
-------
knowledge no such estimates exist. This paper attempts to fill part
of this gap by determining the firm's cost ftractions tinder alternative
enforcement policies. The determination of the control agency's cost
functions are left for further determination.
If it is truly the case that effluent fee enforcement systems are
preferable, one might ask why government continues to employ legal
enforcement methods. There are several possible explanations. In
many cases measurement of effluent quantity is technologically
difficult and expensive. This certainly can explain why some
emissions are not controlled through effluent fee enforcement. But
there are other cases where the measurement problem is not severe.
There is another possible explanation for avoiding effluent fees. The
effluent fee system implies that property rights to the air are vested
with the general public. Thus, the firm must pay not only cleanup
costs but also a sum for residual damages (the effluent fee) . This
makes the firm's out-of-pocket costs higher for effluent fee
enforcement than for legal enforcement thus reducing the profitability
of the firm and the wealth of its owners. Furthermore, the effluent
fee system places the burden of control technology research and
development on the firm while legal enforcement places the burden of
proof and hence research and development responsibility on the control
agency. Shifting to an effluent fee system would increase the firm's
research and development outlays and further reduce the owner's wealth
(in some transition period). There are obvious incentives for firm
owners to opt in favor of legal enforcement if they are forced to
control.
The government bureaucracy also prefers the legal enforcement
•
system. This is because it gives them substantial power in the
control decisions of the firm. Also, because they must demonstrate
the availability of technology, their budgets are large compared to
budgets under an effluent fee system. Legal enforcement may also
require a larger enforcement staff. Bureaucrats prefer systems which
increase their power, staff, and money because these lead directly to
14
-------
greater prestige and remuneration.
Before we turn to the model and the empirical analysis one
additional point needs to be made. The economic literature agures
that an effluent fee be set at the point where MSC = MB (EF- in Figure
Ib). This is incorrect since the firm will equate the fee to MCC__
Er
and overcontrol at R . Empirical studies of cost functions imply that
the effluent fee should be set at EF where MCD__ = MB and predict a
Z EF
suboptimally high control level at R~. This also is incorrect since
the firm will actually control at R» which may be greater or less than
S depending upon the costs of implementation and enforcement to the
£ir
firm and to the control agency. The optimal effluent fee is at EF*
since at this point the firm will control at S^- where EF* = MCC-^,.
.Eif Jtr
Note that at this point the effluent fee is less than MB by the cost
of implementation and enforcement to the control agency (MMC.,^) . This
£jr
level of fee provides the correct marginal signals to the firm but it
does not provide the correct total conditions. In addition to the
effluent fee it is necessary to collect an amount equal to the total
management costs imposed on the control agency by the firm (TMC,,,,) in
Jir
a lump sum. This will cause the firm to include the total costs it
imposes on society from using the environment. In a perfectly
competitive world, not paying this full cost would cause an
inoptimally large output for the industry and allow submarginal firms
to continue operation. In the non-perfectly competitive world the
3/
lump sum payment may cause a suboptimally small output.—
To summarize, we find that the firm reacts to a marginal cost
function which includes both the costs of control and implementation
and enforcement imposed upon it. These costs as well as the costs of
implementation and enforcement to the control agency differ for
various institutional forms of control. The preferred institutional
form is the one which maximizes the net social benefit of control.
The optimal effluent fee system is a combination of a marginal
effluent fee equal to MCC__ at the removal rate where MSC__ = MB and a
fir • EF
fixed fee equal to the total annual management costs (TMC,,-) imposed
4/
on the control agency by the firm.—
15
-------
SECTION III
AN ENFORCEMENT MODEL OF THE FIRM'S CONTROL BEHAVIOR
This section derives a model of the firm's reactions to
enforcement strategies. It then explores various cases to determine
the likely reaction of the firm to alternative values of the policy
variables under differing technological and time frame assumptions.
In the following section this model is applied to the case of new
source performance standards for fly ash discharge from coal-fired
power plants.
Becker (1968) developed a model of the economics of crime and
punishment which consists of damages function, an enforcement cost
function, a supply of offenses function, and a punishment function.
Interpreting his results in terms of the air pollution control problem
we address in this paper, his damages function is the dollar value of
the damages dones to society when emissions from a source exceed a
given standard (thus constituting an offense). The corresponding
punishment function reflects this dollar damage and the cost of
enforcement function.— This procedure is conceptually similar to
pollution control through the institution of an effluent charge system
but it differs in two significant ways. First the offense system is a
threshold system which presumes that in the absense of an offense
there are zero dollar damages from pollution. It postulates a
threshold to pollution damages while the effluent charge system
generally assumes the more likely case of a continuous damage
function. Since there are probably residual damages incurred at an
optimally set standard, the threshold concept can lead to an
inefficient solution. Second, the fine Becker suggests would include
both the damages and the cost of enforcement born by society. The
second part of this fine generally is not considered in setting
effluent fees, hence making such fees inefficient.
Becker's supply of offenses function can also be interpreted in
terms of air pollution control. The polluter's supply of offenses
(the number of times he exceeds the standard) are assumed by Becker to
16
-------
be a function of the probability of his being convicted, the fine he
pays per conviction, and what Becker calls "a portmanteau variable"
representing the sum of all other influences. It is this supply of
offenses (emissions) function that we explore for the air pollution
case in this paper. Specifically stated, our goal is to investigate
the reactions of an individual firm to alternative standards,
conviction probabilities, and fines (the policy variables) under
different implementation schemes.
**
Costs of Pollution Control to the Firm. It is assumed for the
purposes of this paper that the firm seeks to minimize the expected
cost of control of pollutants [E(CC)].— These expected costs are the
sum of the expected cost of control devices [E(CD)] and the expected
cost of compliance and enforcement actions imposed on the firm for
compliance or non-compliance with required controls or standards
[E(EC)]. The firm's objective function— is then
(1) minE(CC) =E(CD) + E(EC)
given a fixed set of control regulations (the policy variables). Both
CD and EC are stochastic in this formulation. Device costs include
both capital and installation costs (KG) and operation and maintenance
costs (OM) . For many devices OM will have some distribution about an
expected value because the-device might partially or fully fail during
the period (as when a catalytic reactor gets poisoned). Enforcement
costs are stochastic because the control efficiency of the device is
stochastic causing the incidence of violation to be uncertain. A
complete analysis of E(CD) is not necessary for our purposes. It is
assumed here (and later shown for the electrostatic precipitator case
we explore empirically) that:
3E(CD)/3R> 0
and
32E(CD)/3R > 0
The arguments in the E(EC) function are somewhat different
depending upon the implementation and enforcement method used. For
the legal enforcement method now employed for new sources by EPA the
17
-------
expected enforcement and compliance costs are a function of the
expected number of days the firm is detected to be in non-compliance
during the year [E(N)] times the expected penalty imposed on the firm
for each violation [E(P)].
(2) E(EC) = f[E(N)-E(P)]
E(N) is a function of the expected control efficiency of the device
installed by the firm [E(R)] given the various rules and regulations
imposed upon the firm by the control agency and/or the legislature.
(3) E(N) = g[E(R)|l,S,C]
where
I = the frequency, accuracy, and form of the
inspection and monitoring actions of the
control agency
S = the emission standard set by the control
agency
C = the requirements set by the control agency
for certification of the effectiveness of
the firm's control device (usually through
some sort of compliance testing procedure).
That is, for any given set of control agency policies, the higher E(R)
the lower E(N). If the control agency were to increase its
enforcement efforts by increasing the frequency of inspections,
improving the accuracy of monitoring, or making compliance tests more
strict, any given E(R) would imply a larger E(N). Likewise, a more
stringent emission standard would increase E(N).
The expected penalty is a function of the probability of being
convicted of being in violation (PC), the money fine imposed on the
firm by the courts if convicted of being a polluter (F), the damages
to the firm's image if convicted (DI) and the possible shutdown time
(ST) for required repairs or construction if found in violation by
either the control agency or the courts.
(4) E(P) = h(PC, F, DI, ST)
PC is a function of the legal costs incurred by the firm to defend
18
-------
itself against the control agency (LC). The effectiveness of a dollar
spent on defense depends upon the control agency's prosecution efforts
(CAP).
(5) PC = k(LC|CAP)
The firm will minimize its cost where:
(6) 3E(CC)/3R = 3E(CD)/3R + 3E(EC)/3R = 0
Since enforcement costs decline as removal increases, this condition
can be satisfied. For a set of policy parameter values equation 6
defines the values of MCC__ and MCD__ as equal to the values of
LI& L£I
3E(CC)/3R and 3E(CD)/3R respectively.
In the case of pure effluent fee enforcement the E(EC) function
is less complex. Expected enforcement costs are simply a function of
R and the level of the effluent fee (EF) per unit of emissions given
some monitoring and inspection system and possibly some certification
of the control device as well.
(7) E(EC) = m(R,EF I,C)
where
3E(EC)/3R< 0
and
3E(EC)/3EF> 0
Alternative Enforcement Strategies. Having discussed the
factors which affect the firm's expected cost of environmental
control, we turn our attention to the effects of alternative
enforcement strategies on this expected cost and the firm's reaction
in terms of pollution control.
Let us assume that the control agency has an air quality goal
which it is attempting to reach using the legal enforcement method.
It has several policy tools available by which it can effect the
control efforts of the firm. It can set higher or lower emission
standards, change penalties for non-compliance, make court actions
more prompt, and impose external pressures on the firm through public
statements.
Standard. Local air pollution control agencies are faced with
19
-------
the problem of obtaining control efforts by firms and individuals
sufficient to reach specified air quality goals. They may recognize
that control devices do not work perfectly all the time. Thus, in
order to insure the desired level of total emission control the agency
could set individual standards at a higher level than would be
required if all devices worked perfectly.
The firm will react to the higher standards by installing more
effective devices but only under specified conditions, the firm will
control to the desired level only if the expected penalties and court
costs are higher than the cost of control. It will delay as long as
the court cost of delaying actions Is less than the interest on the
cost of control devices and savings in operation and maintenance
expenses. As we will see below, the savings in O&M expenses may drop
out if enforcement after installation of the device is lax. This
implies that lax enforcement can increase initial compliance by the
firm but this may not yield a net improvement in emissions.
The above argument implies that enforcement is the key to
compliance with an emission standard. However, for a given
enforcement cost to the firm, a higher standard will cause the firm to
attempt more delaying actions. This is because a higher standard
implies higher control costs to the firm thus making court actions
more cost-saving.
Monitoring. The lack of any monitoring of the control actions
of the firm will make any standard set by the control agency
ineffective. It is obvious that if E(EC) is zero the firm will
minimize costs by not controlling. And E(EC) will be zero in the
i •, . ) •
absence of a monitoring effort. The frequency and type of monitoring
will also affect the firm's compliance.
There are two stages of our legal enforcement model. One for the
situation before the firm takes any control action and another for the
situation after the installation of control equipment. This is
because control and enforcement costs differ in the two cases. To
, ' .* , ' " • • - i •
make this distinction clear, equation (1) is rewritten as follows.
20
-------
(8) min E(CC) = KG + E(ECB) + E(OM) + E(EC )
where:
E(ECO = expected enforcement cost before
D
installation of a control device.
E(EC.) = expected enforcement cost after
installation of a control device
(i.e. during operation).
In the before installation case all of equation (8) holds although it
is possible that E(EC.) may be zero in which case the last two terms
A
will drop out. If the control agency were to step up its before
installation monitoring (increase I_), the firm would find it more
D
expensive to delay compliance. However, this result holds only if the
penalty increases with the number of times the firm is found not to
have installed the required devices. Since this is not usually the
case, one inspection to determine non-compliance is sufficient until
the firm claims compliance.
After installation of the required devices the first two terms on
the right hand side of equation (8) drop out. The firm is faced with
the choice of operating the device or not and its decision clearly
depends upon E(EC.). This in turn depends upon I . Assuming that
A A
each violation detected by an inspection is a separate offense (the
usual case in control legislation), an increase in I. will cet. par.
yield more control. The device will be operated more effectively and
more often. But the form as well as the frequency of inspection will
affect this result.
Inspections might be announced ahead of time (either formally or
through indirect means) or they could be unannounced. The
cost-minimizing firm facing announced inspections will operate the
device during the inspection only if the penalty for non-compliance is
greater than the O&M costs. After the inspection I. = 0 and thus
E(EC) = 0. This being the case the firm will not operate the device
until the next announced inspection. Indeed it has been observed that
when control authority personnel go home at night firms take the
21
-------
opportunity to blow the accumulated fly ash out of the stack. This
can be safely done because, in effect, the control agency has
announced non-inspection.
If inspection is unannounced, the firm will operate and maintain
the device as long as E(OM)< E(ECA>. Thus, increased frequency of
inspection cet. par, will cause more effective operation of devices
and more emission control.
Another policy choice available to the control agency is
inspection to determine the actual emissions of the firm rather than
inspection to determine if devices installed are in good operating
condition (no obvious malfunctions). The control efficiency of any
given device depends upon certain design parameters, some random
performance, and the chance that the device will partially or fully
fail to function. If inspection measures actual emissions, the full
model applies. The firm will operate and maintain the device being
conscious of the actual effectiveness of the device as long as the
savings in enforcement costs justify operation. Also, when faced with
this sort of inspection the firm may find it advantageous to install a
device with a larger E(R) than otherwise required. The larger E(R)
will reduce E(N) thus making violations less frequent. The firm will
incur additional installation and associated O&M costs to the point
where the cost of increasing E(R) (marginal cost) equals the savings
in enforcement costs.
A variant of this case is currently being used by EPA in its new
source performance standards. In this case a compliance test is
required which samples the actual emissions during the test period to
determine if the device will control emissions to the required level.
After the device passes the compliance test and the plant is opened, a
continuous monitor is employed to insure that the device is in
operation and that it is not suffering from a serious malfunction.
It is obvious that the frequency of monitoring will affect
control efforts of the firm. The more frequent I the greater E(ECA) .
This is because each violation constitutes a separate offense. For
22
-------
example, if the probability that the observed removal rate is less
than the standard equals 10 percent and this probability is constant
over time, then the firm will expect to be found in violation once if
inspected ten times and ten times if inspected one hundred times or 10
percent of the time if continuously monitored. Thus, the number of
accusations, given some set of design parameters and O&M efforts, is
solely dependent on the frequency of inspection. An increase in this
frequency will lead directly to an increase in E(EC ) which implies
£\
that the firm will control more (either by improving maintenance or
increasing E(R)) in order to avoid these enforcement costs.
The accuracy of a monitoring device has no effect on the firm's
control effort. Any monitoring device has a distribution of
measurement errors about the true emission value. A more accurate
device would be one for which the standard error is smaller than the
alternative. In Figure 2 emissions as measured by a monitoring device
are on the horizontal axis and frequency of a given measurement is on
the vertical axis. Suppose that the true emission at some point in
time were Y. A monitoring device is subject to measurement errors
which are distributed about the true value such as curve 1. If the
standard were at S , then when the monitoring device measured a value
LE
of Y the control agency could assume that the firm is in violation of
the standard. But they will not be 100% certain because the true
emission could have been at or below ST„. The probability that the
L£I
firm really is not in violation is equal to the area under curve 1 to
the left of S . If the standard error of measurement and the shape
LE
of the distribution are known this probability can be calculated.
Alternatively stated, if the control agency observes a reading of Y,
it can be X% confident that the firm is in violation (X given by the
area under curve 1 to the right of S_„) . Curve 2 in Figure 2
L£J
represents a more accurate monitoring device (one with a lower
standard error of measurement). At some reading from device 2 closer
to ST7, (Z; for example) the control agency can also be X% confident
LE
that the firm is in violation. Thus the control agency can choose a
23
-------
Figure 2
Distribution of Measurement Errors
For 'Two Alternative Emission Monitors
Frequency
LE
Emissions
24
-------
confidence level it wishes and determine the monitor reading which
corresponds to this level of confidence given the measurement error of
the monitor. Any reading equal to or greater than this point, which
we will call S g., will be presumed by the control agency to show that
the firm is in violation. A more accurate device would result in a
S that is closer to S__ but if the confidence level remains
LjiuA ijh,
constant, then it has no effect on E(N). The effective policy
parameter is the confidence level chosen by the control agency. The
higher the confidence the control agency chooses, the fewer will be
the citations for a violation given a level of E(R). However, as the
confidence level increases the control agency is more likely to win in
court if the firm contests a citation. Thus, a higher confidence
level will cause PC to increase and E(N) to decline. The net result
on the firm's control actions depends upon its relative cost of
control on the one hand and court action and fines on the other.
The problem for the control agency of correctly setting S
Li£iA
merits more discussion. Increasing the confidence level required
before a citation is issued means that firms which are in violation
will be cited less frequently. This error is costly because true
violations supposedly cause damages. At the same time, higher
confidence levels imply reduced probabilities that a firm which is not
in violation will be incorrectly cited. This type of error is also
costly since the control agency must use its scarce resources to
prepare and prosecute the case. If it looses the case, it will have
wasted its resources (and those of the firm and the court). If it
wins, the firm may be forced to control to an inefficiently high
level. Thus, there is a tradeoff available between Type I and Type II
errors. The control agency will maximize at that confidence level
where the marginal costs of making Type I and Type II errors are
equal. If pollution damages are rapidly increasing with emissions
above the standard, the optimal confidence level will result in a
relatively large number of incorrect citations. On the other hand, if
the economic (and political) costs of issuing many incorrect citations
25
-------
is high, high confidence levels will be chosen. As we will see below,
current control agency policy tends toward this case.
It could often happen that the control agency is faced with a
SLEA wllicl1 is set by statute and would require long delays to change.
If the control agency wishes to increase its confidence level in these
circumstances, it could require a more accurate device be used in
monitoring.
Penalty. It is perhaps obvious that increasing the level of
penalty imposed will increase compliance by the firm. There are,
however, some circumstances under which this will not occur. If there
were no inspection then any level of penalty will have no effect. It
could also be the case that it is less expensive to incur court costs
to fight the penalty than to increase control. This might be the case
either with high or low penalties. In the very high penalty
situation, the firm may feel that it has a good chance of getting the
court to rule that the penalty is excessive.
The form of the penalty can also have an effect on the firm's
8/
control efforts.— The form of penalty imposed can be classified into
the following groups: (1) cease and desist orders, (2) constant
penalty per violation, (3) constant penalty per pound of pollutant
released above the standard, (4) increasing penalty with the number of
pounds of pollutant released above the standard, (5) a shutdown order.
These alternatives are depicted in Figure 3. S _ is the firm's
standard and emissions increase as you move to the right.
Cease and desist orders have no immediate effect because they
impose no penalty at that instant. However, if they imply a larger
penalty for future violations they may induce additional control in
the future. The constant penalty per violation is an all or nothing
case. Since the penalty is independent of the size of the violation,
the firm will spend less to reduce major breakdowns than might be
optimal relative to minor violations, where the device is removing
almost all the pollutants required, considering that breakdowns are
9/
probably more damaging to society.— The use of the constant or
26
-------
Figure 3
Alternative Forms of Penalties
for Air Pollution Violations
(5)
Penalty
($)
(2)
(1)
LE
Emissions (1 - R)
27
-------
increasing charge per pound reduces this effect. Now, since the
penalty for a complete breakdown will be substantially higher than for
a minor violation the firm will spend more effort on reducing
breakdowns. More reliable devices will be sought. However, since
minor violations are relatively less expensive, this reduces the
incentive to install devices with E(R)>ST7,. The non-linear form makes
LiEj
breakdown prevention relatively more important to the firm. The
penalty in the shutdown order case is the loss in profit to the firm
due to not operating. Whether the firm closes down or controls
clearly depends upon which is the least costly of the two options.
The possibility also exists that the firm will operate without .
controls, and take long term legal (or political) action to prevent
the implementation of controls.
The timing of the imposition of a penalty can also have a
substantial effect on the firm's control effort. If the expected
value of the penalty is constant, it will induce firms to employ legal
delaying actions if the legal costs are less than the interest on the
expected value of the penalty. If the penalty were made a fee and
hence payable upon release of the pollution, its present value would
be increased. Thus, an effluent tax is more effective than an
equivalent penalty per pound because it is payable on release rather
than after court action. As a corrollary to this result, the control
agency can make the effective penalty larger by increasing the speed
of bringing accused violators into court.
In addition to the above policy alternatives, the control agency
has two more options. First, it can try to obtain more tightly
written laws which would increase the probability of obtaining a
conviction (make the penalty more certain) and/or improve their
preparation to the same end.— Second, the control agency can increase
the damage to the firm's image by publicly announcing violations.—
In this part of the paper we have identified conceptually the
tradeoffs available to the firm under alternative policies of the
control agency. In the following sections we attempt to assess
empirically the magnitude of these tradeoffs.
28
-------
SECTION IV
A SIMULATION OF ENFORCEMENT ALTERNATIVES
In the preceding section of this paper we have presented a
general theory of a firm's reactions to environmental control
implementation and enforcement alternatives. In order to demonstrate
some of these propositions and determine their empirical significance
a simulation study was conducted for enforcing the federal new source
performance standards for particulate matter discharges from
coal-fired power plants. The simulation model employed allows us to
determine the likely control actions of the firm (and related costs)
resulting from alternative levels of enforcement policy parameters and
implementation schemes. In effect, via this analysis we will be
examining a variety of enforcement "experiments".
Ideally it is desirable to find the set of enforcement policy
parameters which minimize the sum of resource costs for both firms and
enforcement agencies. This analysis, however, covers only costs to
firms since data and information on enforcement agency costs are
almost nonexistant. Nonetheless it will be seen that the partial
results reported here are rich in policy implications.
It is assumed throughout that managers of coal-fired power plants
attempt to minimize expected costs over their planning horizons and
that available cost effective fly ash control technology is
electrostatic precipitation (see Watson (1974)). We deliberately
focus upon interpretations of the model, its results and related
policy issues. Mathematical details of the model can be found in an
attached appendix.
In this section we begin with a discussion of the new source
performance standards for fly ash control. Next we present a
diagrammatic exposition of the simulation model. The results of the
simulation analysis are then compiled. Policy analysis and
recommendations based on these results are presented in a following
section. A final section of this report discusses the impact on the
analysis of some of the key assumptions which underlie our simulation
29
-------
model; this final section also discusses application of the results of
this analysis to other pollution control problems.
New Source Performance Standards. Final rules and regulations
for particulate matter discharges from fossil-fueled steam generators
were issued by the U.S. Environmental Protection Agency on December
23, 1971 (Federal Register December 23, 1971 pp. 24876-24895).
Particulate matter discharges (which are mainly fly ash and unburned
carbon particles) are not to exceed 0.1 Ib. per million B.t.u. heat
input maximum 2-hour average. This standard is applicable to any
power plant unit of more than 250 million B.t.u. per hour heat input
or approximately 25 megawatts in capacity whose construction is
commenced after August 17, 1971. Eventually, with the retirement of
pre-standard plants, every plant will be subject to the standard.
Under these regulations, firms are required to pass compliance
tests on fly ash control devices before new plants go into operation.
A plant is certified for operation when, on the basis of prescribed
stack testing procedures, discharges during the test period are no
greater than the standard. During operation, opacity of stack
discharges is to be continuously monitored by the firm at its expense
and reported to EPA. If the firm violates the opacity standard (20
percent opacity) it can be charged in a civil action under the
provisions of the Clean Air Act and if convicted, fined as much as
$50,000 per day of violation.
These regulations have several peculiar features. For one thing,
the start-up compliance test can be run an unlimited number of times.
Secondly, the conditions under which compliance tests are to be
conducted are not clearly defined:
All performance tests shall be conducted while the
affected facility is operating at or above the maximum
steam production rate at which such facility will be
operated and while fuels or combinations of fuels
representative of normal operation are being burned
and under such other relevant conditions as the
30
-------
Administrator (of EPA) shall specify based on repre-
sentative performance of the affected facility.
(Ibid. p. 24879.)
Beyond these general stipulations, the rules and regulations do not
specify test conditions. Presumably EPA technical personnel will be
on hand to check test conditions. The tests, themselves, will be
conducted by utility company personnel. A strong fraternity of
engineering interests is likely to pervade compliance testing
activities with liberal interpretations of test conditions "being
understood" by the participants. A third feature is that the average
of as few as three compliance test stack samples is the measurement
for comparison with the promulgated standard:
Each performance test shall consist of (at least)
three repetitions of the applicable test method.
For the purpose of determining compliance with
an applicable standard of performance, the average
of results of all repetitions shall apply. (Ibid.
p. 24878.)
As will be seen, the number of successive compliance tests, the
stringency of test conditions, and the number of averaged compliance
test stack samples markedly.influence firm behavior.
A peculiar feature of the federally promulgated opacity standard,
the basis for detecting a violation during operation, is that it
allows roughly twice the quantity of discharges as are allowed by the
particulate matter discharge standard. This too influences firm
pollution control effort.
The Simulation Model. We have simulated six policy scenarios:
Inflexible Flexible
Technology Technology
Compliance Test with Fine
for Violating an Opacity
Standard SI S2
Compliance Test with Tax on
Emitted Fly Ash S3 S4
31
-------
Emission Tax Only S5 S6
Our model describes the firm's least-cost effort to control fly ash
discharges given each of the three enforcement policy sets listed
above and two variants of electrostatic precipitator technology:
inflexible and flexible.
Figure 4 demonstrates the difference between flexible and
inflexible precipitator technology. Expected collection efficiency is
measured on the vertical axis; operating hours are measured on the
horizontal axis. A typical base loaded power plant will operate about
7440 hours per year, the remaining hours in that year will be outage
hours when normal maintenance is performed on generating equipment and
pollution control devices. The two curves labelled "inflexible" and
"flexible" show that precipitator efficiency declines over operating
hours. This occurs because precipitator discharge electrodes fail,
lowering the filtering capacity of the precipitator (Greco and Wynot
(1971)). It is plausibly assumed that the failure rate is negative
exponential which produces an approximately linear (in efficiency)
operating curve for precipitator performance. The dashed-line
sections of the operating curves represent precipitator maintenance
time during scheduled outages of the power plant. On restart,
precipitators again perform at top efficiency. This cycle of
deterioration, maintenance, and re-start at top efficiency produces
the ratcheted performance curves of Figure 4. By comparing the two
performance curves it is seen that a flexible precipitator's
efficiency declines less rapidly during an operating cycle. This
results from having power shunting electronic instrumentation which
optimizes precipitator filtering capacity as discharge electrodes
fail.
As drawn in Figure 4, the flexilbe and inflexible operating
curves produce the same average efficiency (AE) over the operating
cycle. The curves have purposely been drawn in this way to illustrate
the fact that a flexilbe precipitator (for the same
over-the-operating-cycle average efficiency relative to an inflexible
32
-------
Expected
Efficiency
II
Fl
AE
u>
to
Figure 4
Precipitator Operating Curves
1 year=8,760 hours
7,440 «,760
Time (hours)
-------
precipitator) has a lower "first day" collection efficiency (Fl
versus II) and consequently smaller dimensions and smaller
installation cost since "first day" efficiency is a proxy for
precipitator size. Hence, in comparison with a larger inflexilbe
precipitator, a smaller sized flexible precipitator can produce the
same average collection efficiency over an operating cycle. As we
show later, at high collection efficiencies the cost savings from
these smaller dimensions outweigh the extra flexible instrumentation
costs making flexible precipitator technology the less costly of the
two alternatives.
The Legal Enforcement Model. Figure 5 is a diagram of the model
used to analyze implementation and enforcement scenarios SI, 52, S3,
and S4 (the legal enforcement options). The model is basically a cost
minimizing model. It considers a number of precipitators of different
sizes and consequently different expected collection efficiencies.
For each precipitator the model computes the probability of passing a
start-up compliance test at some specified compliance test standard.
(This is described in further detail at a later point.) It also
computes the expected number of days per year when each precipitator
would violate a specified opacity standard. Using these two pieces of
information it then computes and sums costs in order to determine
total expected costs.
The model begins by computing and summing precipitator
installation costs and compliance test costs. Using the probability
of passing the compliance test as a weighting factor it then adds in
operating, maintenance and stack monitoring costs plus fines for
violating the opacity standard, all of these costs, of course, having
been computed for a precipitator of the originally specified size. A
given precipitator, however, may fail the compliance test. If it
fails the model assumes that the precipitator is enlarged to a size
which has virtually no probability of failing a subsequent compliance
test. In such cases, a power plant would then incur the installation
and penalty costs^for an enlarged precipitator and its operating,
34
-------
Figure 5
Simulation Model
Scenarios S1-S4*
of
Pass
Operating
Cost
Days/Year
> Opacity Standard ---
Violated
EffectJve
Fine
"~ or Size
i
Installed Cost
!<• " Incr e
1 Compliance Test Costl
Tons of
Fly Ash 'Discharged
per Year
Tons of
Fly Ash Discha
per Year
'
Minimum Total
Discounted Cost
—
sent # of Compliance Tests*
of
Fail
Enlarged
Installation
Costs
Enlarged
Operating
Costs
Days/Year
rged- Opacity Standard
Violated
Effective
Fine
_
Total Discounted Cost
Effective Fine
•t
Installation Cost + Compliance Test Cost +
(Probability of Pass) x (Operating Cost + Effective Fine) +
(Probability of Fail) x (Enlarged Installation Cost +
Enlarged Operating Cost + Enlarged Effective Fine)
'(Day/Year Opacity Std. Vlolated)x(Fine/Day)x(Prob. of Conv.)»
Scenarios SI and S2
.(Tons of Fly Ash Discharged)x(Tax/Ton), Scenarios S3 and S4
Explicit equations and parameter values for this model are presented in an
attached appendix.
35
-------
maintenance and stack monitoring costs plus fines for violating a
specified opacity standard during operation of the enlarged
precipitator. The model sums these costs and uses the probability of
failing the compliance test as a weighting factor. The sum of the
expected cost for the original precipitator times the probability of
passing the compliance test and the expected cost of the enlarged
precipitator times the probability of failing the compliance test
yield total expected out-of-pocket costs for a precipitator of some
specified size, for a specified compliance test standard and opacity
13/
standard, and for a single compliance test.— The model then allows
successive runs of the compliance test. This changes the probability
of passing and failing the compliance test and changes the weighting
factors in computing total expected costs. For example, if the
probability of fail in one compliance test is .7, the probability of
2
fail in two tests would be (.7) and so on. This lowers expected
enlargement costs which are relatively large but raises expected
original size costs and compliance test costs. The model allows as
many as 15 successive compliance tests based upon the estimate that
each compliance test takes approximately one week and that even if 15
tests were run, this would keep precipitator testing time well within
the normal 6 month shakedown period for a new power plant. At this
stage, the model finds the number of compliance tests at which total
expected out-of-pocket costs are a minimum. It then goes on to
successively larger sized precipitators, computing costs in exactly
the same fashion for the given set of compliance test and opacity
standards. It also holds constant throughout, the flue gas flow rate,
the number of averaged stack samples taken during a compliance test
and the expected fine for violating the opacity standard (see below).
As a final step it finds the precipitator size or efficiency which
minimizes total expected out-of-pocket costs to the firm for the given
set of enforcement policy parameters. The set of exogenous
enforcement policy parameters is then changed and the model rerun.
Each case simulated by the model actually entails 100 iterations.
36
-------
Each cost computation in a particular run is based upon a Monte Carlo
selection of cost factors from Beta distributions which are keyed to
econometric and engineering estimates of the relevant cost factors.
(See Dienemann (1966) and Watson (1973 and 1974).) The minimum total
expected cost is the average of the minimums from the 100
iterations.—' Each iteration, using a different randomly generated
set of costing factors, selects the precipitator design which
minimizes costs. This random selection process is reinitialized at
the same starting value for each new set of enforcement policy
parameters. This prevents firm behavior from being confounded in the
simulation by differential stochastic variation of the costing
parameters. The estimated minimum costs are also total discounted
costs where the discounting reflects usage and electric utility
costing conventions over a 30 year period, this being the normal
lifetime of a new power plant and its precipitator.
The only difference between policy scenarios SI and S2 (similarly
S3 and S4) is the selection of precipitator technology. In going from
SI to S2 (and S3 to S4) everything else is held constant in running
the model including the exogenous enforcement policy paramters.
The difference between scenarios SI and S2 (the fine scenarios)
and scenarios S3 and S4 (the tax scenarios) is illustrated by Figure
6. Under scenarios SI and S2 a given precipitator will have an annual
operating curve such as EE in Figure 6. For a given opacity standard,
OS, (converted to efficiency) there will be some hours (perhaps 0)
when precipitator efficiency violates the opacity standard. This is
shown by FH in Figure 6 accounting for a 40 day lag which represents a
detection-of-violation lag of 10 days plus a lag of 30 days for time
between detection and filing of a civil suit. This 30 day delay is
used to insure that the violation is not just a chance occurrence and
to provide the time necessary to prepare the control agency's case.
Effective fine is computed as FH/24 times fine per day times
probability of conviction.
Under the tax scenarios, S3 and S4, a tax is paid on every ton of
37
-------
Expected
Efficiency
100
Average
Efficiency
(AE)
OS
Figure 6
Fines and Emission Taxes
H
0
E
' I
1.
1
I 1
FH ,
1; 4
3720 7440
Hours
38
-------
emitted fly ash. In terms of Figure 6, annual total emission tax
would be 1-AE times engineering factors which convert efficiency to
tons of fly ash per hour, times hours per year (H), times tax per ton.
Otherwise in going from the fine scenarios to the tax scenarios, all
parameter specifications remain the same including the exogenous
enforcement policy parameters. There is, of course, no opacity
standard in the tax scenarios.
The Compliance Test. The remaining unexplained link in scenarios
SI, S2, S3 and S4 is the compliance test. Earlier discussion
indicated that the impact of compliance testing on firm behavior
depends upon four factors, namely the efficiency standard which must
be met during the compliance test, the number of averaged stack
samples taken during a compliance test, the flue gas flow rate or
boiler load conditions when the test is taken and the number of
successive reruns of the compliance test. So far we have explained
only the impact of the last factor on firm behavior.
The impact of the other factors can be explained with the help of
Figure 7. For a given precipitator and given compliance test standard
(converted to efficiency) the model computes the area under the
probability density function of efficiencies which is below the given
compliance test standard (see Watson (1973)). This area is the
probability of fail, and one minus this area is the probability of
pass. These factors then become the weights which pre-multiply
original-size costs and enlarged costs in the simulation model. An
example is shown in Figure 7a. PI is the probability density function
for precipitator one; P2 is the probability density function for
precipitator two. El and E2 are their respective expected first day
efficiencies. Precipitator two which is larger than precipitator one
(and which consequently has a higher expected first day efficiency)
has a smaller probability (shaded area) of failing the given
compliance test than does precipitator one (the cross hatched area).
Tightening the compliance test standard would increase the probability
of fail for both precipitators one and two. This would provide larger
39
-------
Figure 7
Compliance Test Tradeoffs
El
Compliance
Test Std.
E2
Efficiency
Probability
of
Fail
(b)
~T
I
N is the number of averaged stack
samples taken during a compliance
test.
Compliance
Test Std. •
Expected Efficiency
As N increases, probability of fail decreases for precipitators
with expected efficiencies above the compliance test standard
and increases for precipitators with expected efficiencies below
•the standard.
.(c)
Probability
V is the "normal
load" flue gas
flow rate.
Compliance
Test Std.
Efficiency
As the flue gas flow rate increases, probability of fail in-
creases for a precipitator of fixed size.
40
-------
weighting factors for the relatively large enlargement costs and
would tend, consequently, to induce firms to select more efficient and
more costly precipitators. Thus, other things constant,
cost-minimizing firms would favor lax compliance test standards.
These probabilities and their associated density functions are
actually computed for a varying number of averaged compliance test
stack samples. Recall that federally promulgated regulations require
that the average of at least three separate stack samples must provide
a reading which satisfies the compliance test standard before a power
plant is allowed to begin full time operation. The model simulates
this by repeated sampling from the appropriate density functions,
averaging of the sample efficiencies, and computation of pass and fail
probabilities. In effect, it generates a series of power functions
like those shown in Figure 7b.— As the number of averaged stack
samples is increased, cost minimizing power plants will tend to pick
more efficient precipitators. This occurs because higher numbers of
stack samples provide probabilities of pass and fail (cost weighting
factors in the model) which favor more efficient precipitators. For
example, as the number of stack samples taken during a compliance test
increases, probability of fail decreases for precipitators with
expected efficiencies above the compliance test standard and increases
for precipitators with expected efficiencies below the standard.
Precipitators with expected efficiencies below the standard rather
than above would have relatively higher weighting factors for the
relatively large enlargement costs. This would tend to induce firms
to pick larger and hence more efficient precipitators.
The probability density functions associated with the compliance
tests are also affected by boiler load conditions during compliance
tests. When boilers are loaded at peaking levels, the flue gas flow
rate through a precipitator can be about 15% above the normal level.
Figure 7c shows a representative probability of fail (cross hatched
area) for peak load conditions and probability of fail (shaded area)
for normal load conditions. Clearly probability of fail is less under
41
-------
normal load conditions. A cost minimizing firm would favor low load
conditions during the compliance test since this provides smaller
weighting factors for the relatively large enlargement costs and
consequently lower total expected costs. On the other hand, compliance
tests under high load conditions make the compliance test more
effective in enforcing a given fly ash emission standard. The model
allows for flue gas flow rate variations in simulating compliance tests
and hence in computing probabilities of pass and fail.
The Emission Tax Model. Figure 8 shows the cost model used to
simulate the emission-tax-only scenarios, S5 and S6. For a
precipitator of given size and for a given emission tax per ton of fly
ash discharged, the model computes total emission taxes. To these it
adds installation costs, operating, maintenance and stack monitoring
costs to obtain total expected out-of-pocket costs. Precipitator size
is then incremented and total costs recomputed. Computation is
truncated when the model finds the precipitator size or efficiency
which minimizes the sum of precipitator costs and total taxes for the
given emission tax. The emission tax, which is a constant value per
ton, is then incremented and the model rerun. Unit emission taxes
which vary over time with meteorological conditions for example, and
unit taxes which increase as total emissions increase, are not
Specifically considered. However, such emission taxes would not
change our basic results.
'As before selection of cost minimizing expected precipitator
efficiency (in this case for a given emission tax) is based upon the
average of 100 iterations of the model. Each iteration, using a
randomly generated set of costing factors selects the precipitator
design which minimizes costs.. The random selection process is
reinitialized at the same starting value for each new emission tax so
that firm behavior as a function of emission tax is not muddied in the
simulation by differential stochastic variation of the costing
parameters. Policy scenario 5 assumes inflexible technology, scenario
6 assumes flexible technology; everything else is kept constant
42
-------
Figure 8
Simulation Model
Scenarios S5 and S6*
Preclpitator
Efficiency
or Size
Increment Efficiency
Installed
Cost
Operating
Cost
Increment
Tax
Tax
per
Ton
Total
Emission
Tax
Tons of
Discharged
Fly Ash
Minimum
Total
Discounted
Cost
Total Discounted Cost = Installation Cost + Operating Cost + Total Emission Tax
^Explicit equations and parameter values for this model are presented in an
attached appendix.
43
-------
between the two scenarios.
Simulation Results. The objective of the simulation model is to
provide cost and performance functions for each of the policy
scenarios. The following functions are of interest: expected
out-of-pocket costs to the firm as a function of enforcement policy
parameters, expected precipitator efficiency as a function of
enforcement policy parameters, expected out-of-pocket costs to the
firm as a function of removal efficiency, and expected resource costs
to the firm as a function of removal efficiency.
The following ranges of enforcement policy parameters are covered
in the simulated scenarios:
Compliance Test Parameters Range
Compliance Test Standard (GTS) .04-.14 Ib./million
B.t.u. discharge rate
No. of Averaged Stack Samples (N) 3-50 stack samples
No. of Successive Compliance Tests (M) 3-15 tests
Flue Gas Flow Rate (R) 1V-1.15V (V is the
normal load flue
gas flow rate)
Opacity Standard Parameters
Opacity Standard (OS) 5%-40%
Fine/Day of Violation (F) $500-$50,000/day
Probability of Conviction 0-1
Emission Tax Parameter
Tax/Ton of Fly Ash (T) $5-$180/ton
Scenarios SI through S4 use a combination of structured and
randomly chosen enforcement policy parameters. Our objective was to
uniformly cover a relatively wide range of enforcement policy
combinations. In all, 50 different policy combinations were selected
for these simulations. In the case of the emission-tax-only
scenarios, the model was run for only a maximum of 10 different tax
rates since each emission tax produces a unique least-cost response.
For each set of enforcement policy parameters the model computes
44
-------
expected precipitator efficiency, expected least-costs of fly ash con-
trol and expected fines or emission taxes paid. Furthermore, in order
to provide for differential response due mainly to economies of scale,
the model considers four different plant sizes, 1300 megawatts, 800
megawatts, 200 megawatts, and 25 megawatts. That is, for each power
plant size per scenario, the simulation experiments provide 50
observations (scenarios SI through S4) or a maximum of 10 observations
(scenarios S5 and S6) on firm least-cost behavior as a function of
enforcement policy parameters .
Since the model is too complex to solve analytically regression
analysis has been used to summarize these "experimental" data. In
effect, this "solves" the model. The following functions have been
fitted:
Scenarios SI through S4
a a. MD-aq a, a. a,
(9) C=A(CTS) 1(N) Z e J(R) ^(OS) D(F or T) b
b b, MD-b. b, b b
(10) E=100-100-EXP[-B(CTS) (N) e J(R) * (OS) 3(F or T)
Scenarios S5 and S6
(11) C=A(T) 6 b
(12) £=100-100- EXP[-B(T) 6]
All Scenarios^
d d
(13) C=D(ln(100/(100-E)))
(14) C-FT=G(ln(100/(100-E)))
C is total expected discounted cost. It includes out-of-pocket
firm pollution control costs, associated firm management costs, and
total fines or emission taxes. E is average expected precipitator
collection efficiency (%) during base-load years. FT is total
expected discounted fine or emission tax. MD is a dummy variable
which is one when the maximum number of allowable successive
compliance tests is 3 and zero when greater than 3.
Assuming appropriate signs for coefficients, equations (10) and
(12) indicate that collection efficiency approaches 100% as
enforcement is made extremely stringent. This is consistent with the
45
-------
known operating characteristics of electrostatic precipitators (White
(1963)). Cost equations (13) and (14) are consistent with optimizing
behavior whereby accounting costs for pollution control are minimized
subject to an efficiency function (Watson (1970)). Equations (9) and
(11) are reduced form cost minimizing equations derived by
substituting equations (10) and (12) into a form-(13) cost equation.
These six equations provide a consistent set of equations for
determining the functions which are of interest. Equations (9) and
(11) are used to derive out-of-pocket pollution control and management
costs for the firm as a function of enforcement policy. Equations
(10) and (12) are used to derive precipitator efficiency as a function
of enforcement policy. Equation (13) is used to derive the firm's
out-of-pocket control and management cost including fines or emission
taxes as a function of precipitator efficiency and equation (14) is
used to derive resource costs—pollution control and mangement cost
excluding fines or emission taxes—as a function of precipitator
efficiency.
Since the basic simulation model contains functional forms which
are similar to equations (9) through (14) (see appendix), high
multiple correlations from the regression analyses should not be
surprising. Thus the regressions cannot be regarded as a test of the
goodness of fit of equations (9) through (14) . On the other hand,
individual enforcement policy coefficients within the indicated
functional forms are not constrained in the simulation model. They
may or may not be significant depending upon least cost tradeoffs.
Therefore in "solving" the model the regressions can help to determine
which enforcement policy coefficients are significant and therefore
which exert an influence on the firm's control efforts.
Estimates of the regression coefficients are listed in Tables 1
through 5. Prior expectation is that the regression coefficients for
the compliance test standard and the opacity standard will be negative
and that all the other regression coefficients will be positive.
Lower numbers for the compliance test and opacity standards represent
46
-------
SI:
Table 1
Regression Coefficients from Simulation Analyses*
Compliance Test with Fine for Violating an Opacity Standard
(Inflexible Technology)
Dependent
Variable**
Cost (9)
Efficiency (10)
Cost (13)
Cost-Fine (14)
Cost(9)
Efficiency(lO)
Cost (13)
Cost-Fine (14)
Cost (9)
Efficiency (10)
Cost (13)
Cost-Fine(14)
Cost?(9)
Efficiency (10)
Cost (13)
Cost-Fine (14)
Constant In .....
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
(8
(1
(8
(8
(8
(1
(8
(8
(7
(1
(7
(7
(6
(1
(6
(6
.86)
.24)
.46) .368
.46) .368
.45)
.22)
.06) .371
.06) .371
.42)
.12)
.07) .349
.07) .349
.46)
.19)
.25) .198
.25) .198
Compliance
Test
Standard
1300MW
-.073
-.168
800MW
-.075
-.17
200MW
-.072
' -.185
25MW
-.043
-.194
No. of
Test
Samples
.012
.032
.012
.032
.011
.034
.013
NS
.015
.015
No. of
Compliance
Tests
.01***
.05
.01****
.05
NS
NS
NS
NS
Flue
Gas
Rate
.495
1.26
.505
1.26
.459
1.21
.275
1.18
Opacity Finfi R2
Standard
-.04 NS .91
-.134 NS. .88
.97
.97
-.037 NS .92
-.128 NS .89
.97
.97
-.024 NS .94
-.097 NS .92
.96
.96
-.008***** NS .88
-.06 NS .87
.84
.84
*Signifleant at the .01 level or smaller unless otherwise indicated
NS Non-significant
**Dollar amounts are in thousands of 1967 dollars. The numbers in parentheses indicate
the functional forms fitted.
***Significant at the .05 level
****Significant at the .10 level
*****Signifleant at the .02 level
-------
S2:
Table 2
Regression Coefficients from Simulation Analyses*
Compliance Test with Fine for Violating an Opacity Standard
(Flexible Technology)
Dependent
Variable**
Cost (9)
Efficiency (10)
Cost (13)
Cost-Fine (14)
Cost (9)
Efficiency (10)
Cost (13)
Cost-Fine (14)
Cost (9)
Efficlency(lO)
Cost (13)
Cost-Fine (14)
Cost (3)
Efficiency (10)
Cost (13)
Cost-Fine (14)
Constant In VQQ_E
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
(8.79)
(.917)
(8.49) .346
(8.49) .346
(8.38)
(.97)
(8.07) .357
(8.07) .357
(7.39)
(.917)
(7.1) .317
(7.1) .317
(6.46)
(1.24)
(6.23) .202
(6.23) .202
Compliance
Test
Standard
1300MH
-.082
-.223
800MW
-.083
-.21
200MW
-.074
-.226
25MW
-.043
-.187
No. of
Test
Samples
.011
.042
.011
.041
.01
.042
.011
NS
.012
.012
No. of
Compliance
Tests
.0?
.07
.01
.06
NS
.03****
NS
NS
Flue
Gas
Rate
.53
1.34
.549
1.31
.515
1.43
.322
1.47
Opacity 2
Standard tlne K
-.013 NS .93
-.049 NS .88
.97
.97
-.012 NS .94
-.049 NS .86
.95
.95
-.008*** NS .94
-.033 NS .92
.95
.95
NS NS .81
-.029 NS .86
.76
.76
*Significant at the .01 level or smaller unless otherwise indicated
NS Non-significant
**Dollar amounts are in thousands of 1967 dollars. The numbers in parentheses indicate.
the functional forms fitted.
***Significant at the .02 level
****Significant at the .03 level
48
-------
Table 3
Regression Coefficients from Simulation Analyses*
S3: Cor,)liance i'est with Tax on Kmitted Fly Ash
(Inflexible Technology)
Dependent
Variable**
Cost (9)
Efficiency (10)
Cost (13)
Cost-Tax(14)
Cost (9)
Efficiency (10)
Cost (13)
Cost-Tax(lA)
Cost (9)
Efficiency (10)
Cost (13)
Cost-Tax(14)
Cost (9)
Efficiency
CostU3)
Cost-Tax(14)
Constant In Tggi-g
Exp (8.82)
Exp (.88)
Exp (8.43) .427
Exp (8.46) .367
Exp (8.41)
Exp (.89)
Exp (8.02) .431
Exp (8.05) .370
Exp (7.39)
Exp (.89)
Exp (7.05) .376
Exp (7.08) .329
Exp (6.43)
Exp (1.07)
Exp (6.23) .212
Exp (6.20) .224
Compliance
Test
Standard
1300MW
-.017
-.077
800MW
-.02
-.081
200MW
-.027
-.108
25MW
-.031
-.14
No. of
Test
Samples
NS
NS
NS
NS
.006
NS
.005***
.005
.014
NS
.017
.017
No. of Flue
Compliance Gas Tax
Tests Rate
NS .058*** .064
NS .378 .132
NS .074**** .062
NS .425 .124
NS .118 .047
NS .573 .1
NS .131 .019
NS .79 .045
R2
.97
.89
.93
.99
.96
.87
.92
.98
.92
.83
.89
.93
.77
.71
.71
.79
*Significant at the .01 level or smaller unless otherwise indicated
NS Non-sipnificant
**Dollar air.oun'cs are in thousands of 1967 dollars. The numbers in parentheses indicate
the functional forns fitted.
***Significant at the .08 level
****Significant at the ,04 level
49
-------
Table A
Regression Coefficients from Simulation Analyses*
SA: Compliance Test with Tax on Emitted Fly Ash
(Flexible Technology)
Dependent
Variable**
Cost (9)
Efficiency (10)
Cost (13)
Cost-Tax(lA)
Cost (9)
Efficiency (10)
Cost (13)
Cost-Tax(lA)
Cost (9)
Efficiency (10)
Cost (13)
Cost-Tax(14)
Cosc(9)
Efficiency (10)
Cost (13)
Cost-Tax(lA)
Constant
Exp (8.85)
Exp (.96)
Exp (8.51)
Exp (8.A7)
Exp (8.8A)
Exp (.95)
Exp (8.11)
Exp (8.08)
Exp (7.A2)
Exp (.96)
Exp (7.13)
Exp (7.08)
Exp (6. AS)
Exp (1.2 A)
Exp (6.2A)
Exp (6.16)
Tn IPO
ln 100-E
.353
.349
.35
.34
.315
.323
.189
.23
Compliance
Test
Standard
I300MW
-.028
-.12
800MW
-.032
-.123
200MW
-.OA
-.15
25MW
-.03A
-.16
No. of
Test
Samples
US
MS
MS
MS
.006
No. of
Compliance
Tests
NS
.OA***
MS
.04***
MS
.01A**** MS
.013
MS
.017
.016
MS
MS
Flue
Gas Tax
Rate
.132 .045
.617 .096
.155 .043
.66 .093
.211 .031
.85 .069
.208 .01
1.12 ,02*****
R2
.92
.83
.88
.97
.91
.82
.90
.95
.87
.84
.88
.90
.78
.77
.75
.79
*Significant at the .01 level or smaller unless otherwise indicated
MS Mon-significant
**Dollar amounts are in thousands of 1967 dollars. The numbers in parentheses indicate
the functional forms fitted.
***Significant at the .02 level
****Signlfleant at the .05 level
*****Significant at the .03 level
-------
Table S
Regression Coefficients from Simulation Analyses*
S5 and Sd: Eaission Tax Only
S5i
Dependent
Variable**
Cost(ll)
E££iclency(12)
Cost (13)
Cost-Tax(lA)
Cost (11)
Efficiency (12)
Coat (13)
Cost-Tax(U)
Cost(ll)
Efficiency (12)
Cost (13)
Cost-Tax(14)
•
Cost (11)
Efficiency(12)
Cost (13)
Cost-Tax(14)
Inflexible
Constant
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
(8.8)
(.466)
(8.7)
(8.5)
(8.4)
(.45)
(8.3)
(8.1)
(7.4)
(.40)
(7.3)
(7.1)
(6.4)
(.28)
(6.4)
(6.3)
Technology
*!§& "« R*
1300HW
.27
.355
800MW
.271
.354
200MW
.254
.325
25MW
.135
.164
.081 .99
.298 .98
.99
.99
.081 .99
.296 .98
.99
.99
.075 .99
" .291 .98
.99
.99
.04 .99
.296 .98
.99
.99
S6; Flexible Technology
Constant In ^ Tax R*
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
Exp
(8
(.
(8
(8
(8
(.
(8
(8
(7
(.
(7
(7
(6
(.
(6
(6
.8)
52)
.7)
.5)
.4)
51)
.3)
.1)
.4)
45)
.3)
.1)
.4)
31)
.4)
.3)
1300MW
.244
.33
800MW
.246
.331
200MW
.232
.305
25MW
.12
.151
.072 .99
.289 .95
.99
.99
.072 .99
.287 .95
.99
.99
.067 .99
.282 .96
.99
.99
.0365 .99
.299 .96
.99
.99
*Significant at the .001 level or smaller
**Dollar amounts are in thousands of 1967 dollars. The numbers in parentheses indicate
the functional forms fitted.
51
-------
tighter standards. With tight standards, firms will tend to pick
relatively large precipitators in order to avoid enlargement costs and
fines. Hence the compliance test and opacity standard will be
negatively related to precipitator efficiency and cost. Previously it
was shown that the probability of failing the compliance test
increases (1) as the number of averaged compliance test samples
increases (given a "shoddy" device), (2) as the allowed number of
successive compliance tests declines, and (3) as the flue gas flow
rate during the compliance test increases. Accordingly, increases in
these enforcement parameters will provide relatively large weighting
factors for large enlargement costs. To avoid this, firms will tend
to pick more efficient and higher cost precipitators, producing a
positive relationship between these enforcement parameters and
precipitator efficiency and cost. A positive relationship between
emission tax and precipitator cost and efficiency, and between cost
and logarithmic transformation of efficiency is also expected. The
signs of the estimated regression coefficients (see Tables 1 through
5) are consistent with these prior expectations and the fits are very
good.
The role of effective fine (days x fine/day x probability of
conviction) in scenarios SI and S2 needs further elaboration. Note
that fine appears to be an insignificant determinant of behavior in
scenarios Si and S2. This is misleading. The probability of
conviction has been subsumed into the opacity standard. Whenever
probability of conviction is zero, opacity standard was set equal to
40% in the regression analyses, in effect making the opacity standard
non-operative since 40% is a large value or a relatively lax opacity
standard. Obversely the role of. a positive effective fine is to help
make the opacity standard operative. In the model itself, costs
(excluding effective fines) are nearly constant over a wide range of
precipitator sizes. Consequently, the impact of any positive
effective fine is to usually induce a cost minimizing firm to pick a
fine-avoiding precipitaltor. Furthermore, increasing the dollar fine
52
-------
per conviction usually makes the cost curve more steep around the
least cost precipitator size, but does not shift the least cost point
(see Figure 9). Hence, the impact of effective fine on firm behavior
is a "zero-one" effect. If the effective fine is any positive value
(fine positive, probability of conviction positive) then the
promulgated opacity standard is operative (i.e., the opacity standard
impacts firm behavior in relationship to its specified value). A
positive effective fine, of course, also promotes maintenance of
pollution control devices since even very lax opacity standards would
be violated if firms did not maintain their control devices. Annual
maintenance cost for a precipitator ranges from about $10,000 per year
(small plant) to $40,000 per year (large plant). When opacity
standard violations occur a firm might have to hire legal resources to
defend .itself and it might also have to pay a fine. These expenses
would probably far exceed maintenance costs. On the other hand, if
the effective fine is zero (a fine of zero or probability of
conviction zero), opacity standard violations will produce no cost
penalties for the firm and hence will have no impact on firm behavior.
This is the rationale for setting opacity standard equal to 40% for
those policy simulations in which probability of conviction is zero.—
The nonsignificance of fine in the regression analyses of scenarios
SI and S2 merely reflects the fact that, increasing the value of a
positive dollar fine has no impact on firm behavior; positiveness of
the fine itself and not the degree of positiveness influences firm
behavior.
One final result of our simulation analysis is of interest. Our
best assessment of EPA's current choice of policy parameters for the
enforcement of the new source .performance standards for coal-fired
power plants is:
Compliance Test Standard = 0.1 Ib./million B.t.u.
No. of Successive Compliance Tests = 15 or less
No. of Averaged Stack Samples = 3
Flue Gas Flow Rate = 1.1V
53
-------
Figure 9
Fines and Control Effort
Total
Expected
Cost
Fine KFine 2
-------
Opacity Standard = 30%
Fine/Day of Violation = $500-$50,000/day
Using these values in the model, we find that most plants will control
to less than the standard and almost never be cited for a violation.
In fact, plants larger than 100 megawatts will be in violation from 50
to 70 percent of the time depending upon plant size (see Table 6). The
reason why plants are not cited for a violation is that the enforced
opacity standard allows about three times the emissions of the
compliance test standard. We also find that small plants control to a
higher level than large plants even though it is relatively more
expensive for them to do so. This is because large plants enjoy
economies of scale which allow them (relative to small plants) to make
more favorable cost-reducing tradeoffs against enforcement policy
parameters. Furthermore, all firms choose inflexible technology since
its out-of-pocket cost to the firm is less than flexible technology.
This is an inefficient choice for society, however, since the real
resource cost of the same level of average control using flexible
technology is less. In fact, savings in the resource costs of control
are probably an underestimate of the societal savings since flexible
technology has a higher last day efficiency than inflexible
technology. Thus, if marginal damages decline with control as is
usually the case, then the increased damages due to a lower first day
efficiency for the flexible device are more than offset by the higher
damage savings due to its greater last day efficiency. Furthermore,
damages are likely to be suboptimally high because the current
combination of policy parameters yields a level of control below the
fly ash standard and there are some indications that the standard is
approximately correct in terms of benefits and costs (Watson (1974)).
55
-------
Table 6
Current Enforcement Practice
Plant
Size
(Megawatts)
Expected
Average Efficiency*
(%, Inflexible
Technology)
Expected
Cost
(1000 fs of 1967
Dollars, Discounted)
Expected
Time in
Violation*
(%)
25
200
800
1300
99.1%
98.0
97.7
97.7
$ 720
1,900
5,200
7,800
0%
61
70
70
*During base load year at normal flue gas flow rates. Time in violation
would be higher and average efficiency lower to the extent that plants
are operated above normal loads, for example, under peak load demand con-
ditions .
56
-------
SECTION V
POLICY ANALYSIS
Cost Comparisons. We can now use our simulation results,
summarized by our regression equations, to investigate tradeoffs among
the alternative enforcement schemes.
Four straightforward results evolve from a comparison of
out-of-pocket costs to the firm over the different enforcement schemes
and from a comparison of resource costs (cost minus total fine or
total tax) over the different enforcement schemes.
First, at high collection efficiencies the expected resource costs
of flexible technology are generally less than those of inflexible
technology at all plant sizes and for each of the three enforcement
schemes. Figures lOa, lOb, and lOc show some representative curves for
a 1300 megawatt plant. Under enforcement schemes using compliance
tests, firms will incur enlargement costs weighted by the probability
of failing the compliance test. These enlargement costs tend to be
quite large while their weighting factors—the probabilities of
compliance test failure—tend to decline at high efficiencies. This
produces relatively small expected enlargement costs at high collection
efficiencies. Hence at high efficiencies flexible devices have
smaller expected resource costs than inflexible devices (for the same
average efficiency) because the savings from their smaller original-
size costs exceed the sum of their extra instrumentation costs, their
higher power input costs, and their larger (but relatively small)
enlargement costs. This is demonstrated by Figures lOa and lOb: flex-
ible costs are less than inflexible when collection efficiency is
approximately 97% or greater. Under emission-tax-only enforcement
and at high collection efficiencies a flexible precipitator also has
smaller expected resource costs than an inflexible precipitator (see
Figure lOc). The reason is that the smaller original-size costs for
flexible precipitators provide savings which exceed their extra
instrumentation and power costs. In this case there is no question of
a plant failing a compliance test and incurring enlargement costs.
57
-------
Figure 10
Cost Comparisons
_ _ .Total Res_ource Cost£ Q-SOO^Megawatt^lajQt)
10
Millions
of
1967
Dollars'
0
0
(a)
Compliance Test
and Opacity Standard
Inflexible
Flexible
97 96
Efficiency—
99 100
<>
o>>
Compliance Test
and Emission Tax
•Flexible
97
98
99
.Cc)
Emission Tax Only
flexible -^/y
\
100
97
98 99
100
Millions
of
1967
Collars
9-5
8.5
Millions
of
1967
Dollars
,„
10
(d)
Total Firm Costs
(1300 Megawatt Plant)
Emission Tax Only Jl
(Flexible) .^
—-5^5^
Compliance Test
and Emission Tax
(Flexible)
Compliance Test and
Opacity Standard
(Inflexible)
(e)
Total Resource Costs
1300 Megfcwatt Plant
(Flexible Technology)
97 98 99 100
Efficiency
(f)
Marginal Resource Costs
2.51- 1300 Megawatt Plant
(Flexible Technology)
MCC (Compliance Test
'and Opacity Std.)
Millions
of
1967
Dollars/Eff.
1.5
MCD
(Compliance Test
and Emission Tax)
v
98.5
99 99.5 100
Efficiency
99
99.5
100
Efficiency
58
-------
Therefore since enlargement costs need not be overcome by flexible
cost savings, flexible precipitators enjoy an even greater cost
advantage over inflexible under emission-tax-only enforcement than they
do under compliance test (with an opacity standard or emission tax)
enforcement. This is demonstrated by the relatively larger cost ad-
vantage for flexible technology in Figure lOc; in Figures lOa and lOb
flexible technology enjoys a relatively smaller cost advantage.
Second, the lowest out-of-pocket cost to the firm occurs with
enforcement via a compliance test and opacity standard (with
inflexible technology), the next from the lowest is a compliance test
with emission tax (with flexible technology), and the third from the
lowest is the emission tax only (with flexible technology). Out-of-
pocket costs, of course, include fines and emission taxes paid under
each of the enforcement schemes. Figure lOd presents each of these
costs for a 1300 megawatt plant. On the other hand, a comparison of
resource costs (all for flexible technology since we have just seen
above that flexible is cheaper) gives the exact opposite ordering (see
Figure lOe). Hence the enforcement schemes which use emission taxes
and resource-saving flexible technology and which consequently are
attractive to a cost-minimizing resource manager are unattractive to
18/
the firms being regulated and vice versa.— An implication is that
there will be some resistance by firms to a shift toward enforcement
schemes which use emission taxes even though this is desirable from
the viewpoint of resource cost minimization. We will have more to
say about these matters at a later point.
In our earlier discussion of efficient enforcement (p. 11) a
distinction was made between resource costs of control only (MCD) and
marginal resource costs of control including marginal firm management
costs (MCC). We now have quantitative measures of these costs. Using
our simulation regressions we have plotted marginal resource costs for
a 1300 megawatt plant for each of our three alternative enforcement
schemes (see Figure lOf). The emission-tax-only curve is the marginal
resource cost of control only curve since under this enforcement
59
-------
scheme there are no differential management costs such as those
associated with compliance testing. The other two curves are the
marginal resource costs of control including firm management costs for
the indicated enforcement schemes. On the average (at high efficiency
levels) there is about a 6% difference between MCC and MCD under
compliance-test-with-emission-tax enforcement and about a 6.6%
difference under compliance-test-with-opacity-standard enforcement.—'
It would appear that if a marginal benefit curve crosses these cost
curves at high efficiency levels, using one or the other to determine
"efficient" control levels results in approximately the same control
level. It is well to recall however, (see p. 13) that the proper
inclusion of marginal enforcement agency costs could significantly
impact determination of efficient control levels.
Policy Frontiers. Particular technologies were deliberately
specified in the above ordering of preferred costs by the firm. This
is necessary because the firm in reacting to enforcement policy
parameters chooses the precipitator size and technology which
minimizes its costs. Indeed, different mixes of enforcement policy
parameters will induce it to pick flexible technology in some cases
and inflexible technology in others. We proceed now to investigate
the conditions governing technology selection.
The curve labeled AA in Figure 11 is the locus of compliance test
standards and opacity standards for which flexible technology control
out-of-pocket costs (for a 1300 megawatt plant) equal inflexible
technology control out-of-pocket costs. This locus is determined by
setting costs as a function of enforcement policy parameters from
scenarios SI and S2, equal to each other. The dashed perpendiculars
and the area to the northeast of these perpendiculars indicate
approximate feasible choices for the compliance test and opacity
standards. A compliance test standard of .2 and an opacity standard
of 5 are factor increases of 5 and 4 respectively in current
standards. It is doubtful that stricter nominal standards could be
promulgated without serious legal challenges by affected industries.
60
-------
Compliance
Test
Standard
(Ibs. of fly
discharged -15
per million B.t.u. .14
heat input) ,o
.12
.11
.10
.09
.08
.07
.06
.05
.04
.03
.02
.01
Flexible
Figure 11*
Enforcement by Compliance Test and Opacity Standard
1300 Megawatt Plant
(N=3,MD=0,R=1.1V)**
Inflexible
Technology
97.7
I •>
• — 99.67
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Opacity Standard
(% Opacity of Stack Discharge)
^Similar tradeoffs occur at ot.her plant sizes.
**An average of three stack samples (N=3), no limits on the number of
successive compliance tests (MD=0), and an intermediate level for the
flue gas flow rate (1.1V) arc. representative of current enforcement
practice.
-------
The shaded area to the left of AA is the policy area within which
flexible technology is cheaper. To the right, inflexible technology
is cheaper. The curve labeled 99.54 is the locus of compliance test
and opacity standards (given flexible technology) which would induce a
cost minimizing firm to select a 99.54% efficient precipitator. The
efficiency 99.54% is the average expected efficiency during base-load
years at normal flue gas flow rates. The curvfe labeled 99.67 is a
similar locus given inflexible technology. Note that the
iso-efficiency curves are only relevant for the policy areas where
their technologies are less costly. A 99.54% efficient flexible
precipitator and a 99.67% efficient inflexible precipitator are
devices which would meet the new source fly ash discharge standard of
.1 Ib./million B.t.u. (two-hour average). That is, these devices have
sufficient capacity to meet the new source standard even on the last
day of their operating cycles at peak load flue gas flow rates
(1.15V). Current legal enforcement practice is somewhere in the
vicinity of the point labeled Q (compliance test standard of .1,
20/
opacity standard 6f 30%) .— As indicated by the iso-efficiency curve
passing through Q, a cost minimizing 1300 megawatt plant would install
a precipitator having a base-load efficiency of about 97.7%. This is
substantially below 99.67%, the base-load efficiency needed to meet
the federally promulgated new source fly ash standard.
Furthermore, as is clearly indicated, current legal enforcement
practice induces the firm to pick inflexible technology even though
its resource costs are greater than flexible technology. This can be
explained as follows. For a relatively tight compliance test standard
a cost minimizing firm will pick roughly the same sizes of flexible
and inflexible precipitators to avoid high enlargement costs.
Therefore the "first day" efficiencies of the two devices will be
approximately the same while the installation costs of the equivalent
size flexible precipitator will be higher because of extra flexible
instrumentation costs. Moreover, the flexible precipitator will have
a higher average operating efficiency and consequently higher
62
-------
operating costs. Thus, for a given set of SI and S2 enforcement
parameters (and specifically a relatively tight compliance test
standard) a cost minimizing firm would pick an inflexible precipitator
of lower average operating efficiency but the same first day
efficiency.
Similar analysis has been carried out for scenarios S3, S4, S5
and S6. The results are summarized in Figure 12. Feasible
enforcement policies lie in the inclusive area between the dashed
perpendiculars and the line CB extended beyond B. Emission taxes
below $5 per ton are likely to be less than the minimum average cost
of pollution control and therefore they probably would not impact
pollution control effort. A compliance test of less than .02 Ibs. per
million B.t.u. is a very strict standard which would probably
encourage legal challenges by affected industries. FF is the locus of
compliance standards and emission taxes for which total flexible and
inflexible precipitator out-of-pocket costs for a 1300 megawatt plant
are equal. This locus or policy frontier is determined by setting
1300 megawatt costs as a function of enforcement policy parameters
from scenarios S3 and S4, equal to each other. In the shaded area to
the left of FF, inflexible technology is cheaper. To the right,
flexible is less costly. BC is the locus of compliance test standards
and emission taxes using compliance-test-with-emission-tax enforcement
and emission-tax-only enforcement for which precipitator efficiency is
equal in a comparison of these two alternative enforcement schemes. It
is determined by setting 1300 megawatt efficiencies as a function of
enforcement policy parameters from scenarios S4 and S6 equal to each
other. The curve labeled 99.54 is the 99.54% or "law abiding"
iso-efficiency curve for a flexible precipitator under emission tax
enforcement. The curve labeled 99.67 is a similar curve under
compliance-test-tax enforcement where inflexible technology is
cheaper. The point labeled G is the compliance test standard and
emission tax combination where total expected costs to the firm for a
99.54% efficient precipitator are equal to tax-only enforcement at H.
63
-------
Compliance . 16
Test
Standard
(Ibs. of .14
fly ash dis-
charged per
million .12
B.t.u. heat
input)
.10
.08
.06
.04
.02
Figure 12*
Enforcement by Compliance Test and
Emission Tax
1300 Megawatt Plant
(N-3,MD-0,R"1.1V)**
Flexible
Technology
99.67- —
20
40
, 60
80
100 120 140
Emissioif Tax'
($/Ton of Discharged Fly Ash)
160
^Similar tradeoffs occur at other plant sizes.
**An average of three stack samples (N=3), .no limits on the number of successive
compliance tests (MDB0), and an intermediate level for the flue gas flow rate
(1.1V) are representative of current enforcement practice.
-------
To the left of G on the 99.54% iso-efficiency curve, the test-tax
policy combinations result in smaller costs to the firm while to the
right of G they are more expensive than tax-only enforcement
(indicated by point H). At point I, compliance standard of .1 (the
current EPA standard) combined with an emission tax of $56/ton would
induce a cost minimizing firm to pick a "law-abiding" 99.54% efficient
precipitator. However, note that an emission tax alone of the same
amount would produce the same level of control at less expected cost
to the firm. Point K is the least cost point for the firm under
compliance-test-tax enforcement.
Figure 12 also indicates that flexible technology enjoys a
relative policy advantage under emission tax enforcement. This occurs
because increased flexibility allows the firm, for a given
precipitator size, to reduce total emission taxes. Loosely speaking,
flexible technology will cost less than inflexible as long as this
emission tax savings (offset by some additional fly ash disposal
costs) exceeds the additional flexible instrumentation costs. This
may, of course, not occur if the emission tax rate is relatively
small or if the compliance test standard is relatively tight. In
these cases enlargement costs dominate technology selection and inflex-
ible technology clearly has a cost advantage over flexible technology.
Enforcement Policy and Technology Development. The model
contains two "types" (really degrees) of precipitator technology,
labeled, for convenience flexible and inflexible. These particular
variants were modeled because they are feasible choices in today's
technology choice set. Over time though, one would expect that
precipitators even more efficiency-flexible than these could be
developed. This raises an important issue, namely, do different
enforcement schemes either encourage or discourage the development and
adoption of efficiency-enhancing technology?
The answer is that emission tax enforcement schemes encourage
such developments while enforcement by compliance test and opacity
standard discourages them. We proceed now to investigate the reasons
65
-------
for this. It is assumed throughout this discussion that everything
except operating flexibility remains constant including, most
particularly, the installed instrumentation costs associated with
flexibility.
Figure 13a shows that increasing flexibility reduces emission
taxes from an amount proportional to area ABCD to an amount
proportional to area AFCD for a given precipitator size. Collected
fly ash disposal costs are, in turn, increased in proportion to area
ABF. However, the net outcome is usually a reduction in out-of-pocket
costs to the firm. Under compliance-test-emission-tax enforcement of
environmental standards, increased flexibility may also allow the firm
to install a smaller precipitator since the net savings in emission
taxes and collected fly ash disposal costs can be traded off against
decreased original size costs and increased enlargement costs. Thus
under emission tax enforcement of environmental standards, the firm
can reduce its costs by adopting technology of greater and greater
flexibility. Thus, it will pay them (or their suppliers) to expend
resources to develop more flexible technologies.
Assume now that Figure 13a is for enforcement by compliance test
and opacity standard. Assume also that the compliance test is
relatively tight and that the firm may choose operating curve AB or
AF. AF, or a flexible operating curve, would (in addition to
instrumentation costs) increase collected fly ash disposal costs in
proportion to area ABF. Thus out-of-pocket costs and average
collection efficiency for the firm would rise and consequently
cost-minimizing firms would not opt for flexible technology.
On the other hand, if the compliance test is lax relative to the
opacity standard the firm may be able to choose a flexible operating
curve like GB. This reduces its fly ash disposal costs costs and its
installation costs by trading off smaller original-size costs against
larger enlargement costs. Hence when the compliance test standard is
relatively lax, cost minimizing firms may opt for flexible technology.
As shown by Figure 13b development of new more flexible technologies
66
-------
Expected
Efficiency
100
Figure 13
Technology-Cost Tradeoffs
(Current Enforcement Practice)
(a)
_ __ .__ i C
B
7440
Annual Operating Hours
Compliance
Test
Standard
0
(b)
Inflexible
Opacity Standard
67
-------
would shift the equal cost technology locus from AB to AC. But since
collection efficiency increases in the southwest direction such a
shift is unlikely to produce very many instances at high efficiency
standards when compliance-test-opacity-standard enforcement will lead
to situations where it pays the firm to develop and adopt more
flexible technology.
The crux of the matter is that enforcement by compliance test and
opacity standards tends, for the most part, to encourage good "first
day" performance by firms. Hence, flexible technology development
which improves over-the-operating-cycle efficiency is not cost
effective for the firm under these enforcement circumstances.
Moreover, improving flexibility generally shrinks the relevant policy
area within which flexible technology would be adopted under such an
enforcement scheme. In comparison, emission tax enforcement rewards
over-the-operating-cycle performance. Hence, costs to the firm tend
to fall as flexibility increases, given emission tax enforcement of
environmental standards. This is true over a wide policy range even
when emission taxes are combined with compliance tests. Or in terms
of Figure 12, gains in precipitator flexibility would cause the
technology policy frontier, FF, to shift toward the origin.
The important conclusion of the discussion is that the resource
costs of pollution control fall as technology is made more flexible
and so it is important to devise enforcement schemes which encourage
firms in this direction. We have seen that compliance-test-opacity-
standard enforcement will usually fail in this regard while emission
tax enforcement schemes will generally succeed. Later we provide
estimates of the extra resource costs which would occur as a result of
cost minimizing firms choosing inflexible technology under compliance-
test-opacity-standard enforcement.
One might ask why EPA could not develop more flexible
technologies to counteract this bias in legal enforcement. They could
perform the necessary research and development but firms would have no
incentive to adopt this new more expensive (in out-of-pocket costs)
68
-------
technology unless policy parameters were adjusted to fall to the left
of the new AC curve which would result from the developed technology.
69
-------
SECTION VI
POLICY RECOMMENDATIONS
Current EPA Policy. At this point we can analyze current EPA
policy in greater detail. We have already seen that under current
enforcement practice a 1300 megawatt plant would control fly ash at
approximately the 97.7% efficiency level and choose inflexible
precipitator technology. Using our regression results we have also
computed collection efficiencies for other plant sizes. These
results, shown in Figure 14a, indicate that smaller plants, for the
same set of current practice enforcement policies would control at
progressively higher efficiencies. The different performance at
different plant sizes is due mainly to the existence of economies of
.scale. Figure 14b converts these efficiencies to time-in-violation of
21/
the federal fly ash standard.— In general, under current EPA policy
all plants choose inflexible precipitator technology, small plants
control at higher efficiencies than do large plants even though large
plants have smaller marginal costs of control and significant
economies of scale, and all but small plants will be violating the
federal fly ash standard for considerable amounts of time.
There are several implications of the differential impact by
plant size. One is that it is inefficient and hence undesirable to
have small plants with their higher marginal costs controlling fly ash
at relatively high levels (assuming equal marginal benefits of control
22/
at all plant sizes) .— A second implication arises with respect to
industry structure and industry competitiveness. In the electric
power industry, installation of larger power plants would be
encouraged by legal enforcement (compliance tests and opacity
standards). Competitiveness, on.the other hand, would probably not be
affected; this industry is a "natural" monopoly with regulated prices;
increased costs would merely be passed on to consumers. Since our
analysis has general applicability, it is likely that such
differential impacts will also arise in other industries as legal
enforcement of pollution control is undertaken. Here smaller plants
70
-------
Expected 10°
Efficiency*
99
98
97
96
Time in
Violation*
100
80
60
40
20
Figure 14
Control Effort
(Current Enforcement Practice)
(a)
0 200 400 600 800 1000 1300
Plant Size (Megawatts)
(b)
0 200 400 600 800 1000 1300
Plant Size (Megawatts)
*PQrformance during base load years, normal load conditions.
71
-------
will be at a definite disadvantage. When pollution control costs are
substantial this could result in the more frequent closing down of
smaller plants and subsequent trends towards industry dominance by
larger plants and firms. A third implication is that large power
plants in comparison with small plants will put greater pressure on
EPA for lax enforcement of compliance test conditions and lax stack
monitoring. It is likely that there will be substantial effort by
large power plants to cultivate "good working relationships" with EPA.
23/
The financial rewards for doing this can be substantial.— A fourth
implication is that the bias toward larger plants could increase
damages. The standard is stated in terms of a quantity of emissions
per unit of output. Higher output implies a larger number of allow-
able tons of emissions, hence a greater concentration of particulates
near the larger plant. Since total and marginal damages are probably
a function of concentrations, a larger plant would cause greater total
and marginal damage while meeting the standard than would a small
plant, cet. par.. Thus, the bias in favor of large plants is doubly
damaging. Because of the reduction in marginal cost of control for
large plants and the greater marginal damages implied by control to a
given percentage removal, efficient control dictates that large plants
control at a higher percentage removal than small plants.
The implications for technology selection are disturbing. At
control levels which would prevail under current enforcement practice,
our data indicate that inflexible precipitation costs about 1.5% more
than flexible precipitation. By 1980, stationary source, air and
water pollution control costs will be running at the rate of $5 to $10
billion per year. If standards are enforced legally and if similar
technology selection incentives operate in other cases (and this is
certainly possible), then control costs by 1980 would be higher by an
amount between $75 and $150 million per year. Obversely, this is the
approximate annual control expense which could be saved if instead,
emission taxes were used to enforce environmental standards at
promulgated levels. Moreover, this extra control cost is probably a
72
-------
lower bound estimate. Our analysis of precipitation technology is
based upon a marginal change in technology. If in fact, firms had
incentive to control pollution at very high levels as they would under
emission tax enforcement, they then would probably be much more
innovative. This would imply that technology selection losses are
larger than $75 to $150 million per year.
Correcting Policy Deficiencies. There are several alternatives
for potentially correcting current policy deficiencies. One is to
tighten the compliance test and opacity standard under current
practice enforcement, another is to switch to enforcement by
compliance test and emission tax, a third is to switch to
emission-tax-only enforcement. We now explore the implications of
each of these alternatives.
At the compliance test and opacity standards to the left of point
B on the 99.54% curve in Figure 15a (a tightening of current
enforcement at point Q), power plants would be meeting the federal fly
ash standard and installing flexible technology of the type simulated
24/
in our model.— This policy however, has two major drawbacks. Firms
would probably not be given the incentive to install devices of
greater flexibility (i.e., new devices even more flexible than those
now modeled) since under this enforcement scheme, such devices would
tend to increase firm costs. An improvement in flexibility, for
example, would shift the equal cost frontier AA toward the northwest
(see Figure 13b). Hence the technology now identified in the model as
flexible technology would continue to be less costly since point B
would probably lie to the right of a new equal cost frontier.
Furthermore, both the compliance test standard and most particularly
the opacity standard would have to be substantially tightened. This
could lead to much higher enforcement costs for the enforcement
agency. Picking point H instead (see Figure 15a) would probably
reduce enforcement costs but has the very undesirable feature of
inducing the firm to pick inflexible technology. A desirable feature
is that either one of these policies would result in more equal
73
-------
Compliance
Test
Standard
.1
.048
.02
Figure 15
Effective Legal Enforcement
(a)
99.54|
5.6 9.6
30
Opacity Standard
100
Expected gg 4
Efficiency
99
98
97
(b)
Expected 100
Efficiency 99.67
99
98
97
Point B
Enforcement
(Flexible)
Current
Enforcement
Practice
200 400 600 800 1000 1300
Plane Size (Megawatts)
(c)
Point H
Enforcement
(Inflexible)
Current
Enforcement
Practice
200 400 600 800 1000 1300
Plant Size (Megawatts)
74
-------
pollution control effort across plant size (see Figure 15b and 15c).
Of course, a complete analysis of firm and control agency costs would
be necessary to determine the optimal point in the policy space.
We now consider enforcement by compliance-test-emission-tax and
emission-tax-only strategies. Figure 16a shows the relevant policy
space and iso-efficiency curves. In this policy space, point K and
any point on the 99.54% efficiency curve to the right of point K would
meet the federal fly ash standard and all plants would be using
flexible technology. Furthermore, the incentive to develop even more
flexible technology would be operative since firm costs (given this
enforcement scheme) decrease as technology becomes more
efficiency-flexible. Figure 16b shows that pollution control effort
remains about the same across plant size. A similar pattern evolves
from an investigation of emission-tax-only enforcement (point H).
Between the two emission tax enforcement schemes there are,
however, some differences. One possible difference is in enforcement
costs to the enforcement agency. But since our model does not include
such costs we are unable to say what this cost difference might be.
Another difference is that compliance-test-emission-tax enforcement
can have lower total out-of-pocket costs to the firm so that industry
reductions in growth of output, another way to reduce pollution, will
be somewhat less with this enforcement scheme. The major difference,
however, is probably in acceptability of the two enforcement schemes.
Enforcement agency personnel are usually lawyers and engineers.
Understanding of the role of emission taxes in enforcement by such
individuals has improved but there is still the tendency to cling to
legal and technical approaches. Compliance-test-emission-tax
enforcement has the obvious "acceptability" advantage that it keeps
part of the technical enforcement approach intact and consequently may
overcome some of the resistance to use of emission taxes in
25/
enforcement.—'
In comparison with stricter current enforcement practice,
emission tax enforcement alone or in combination with compliance tests
75
-------
Compliance
Test
Standard
.066
Figure 16
Effective Emission Tax Enforcement
(a)
0
Inflexible
1 I
99.67
34
56
Emission Tax
Expected 100
Efficiency 99.54
99
98
(b)
Point K
"Enforcement
(Flexible)
Current
•Enforcement
Practiqe
97
<
0
>•
200 400 600 800 1000 1300
Plant Size (Megawatts)
76
-------
seems to be a better method of effective enforcement. Most
importantly, emission tax enforcement provides incentives toward the
adoption of resource saving flexible technology. Indeed savings in
control cost due to emission-tax-induced accelerated technology
development are likely to be very large. Furthermore, since emission
charges are immediate there is little the firm can do to avoid
compliance. There is, however, one sense in which firms could avoid
or delay compliance under emission tax enforcement. This is by
initial challenges to emission tax legislation. Our results indicate
that effective emission tax enforcement can raise out-of-pocket costs
to the firm by as much as 25% above costs incurred under
current-practice enforcement. Thus there are substantial cost savings
to firms from preventing emission tax enforcement of pollution
standards. An implication is that substantial resources may have to
be devoted to passage and legal defense of emission tax legislation.
Policing of stack monitoring is the one activity to which a
pollution control agency must devote substantial resources under
emission tax enforcement. Cost minimizing firms will achieve high
collection levels only if full and proper emissions charges are
levied. Honest and hence carefully policed, stack monitoring is a
necessary condition for this.
77
-------
SECTION VII
INTERPRETING THE RESULTS
This study has attempted to produce a generalized theory of
enforcement and a specific application. As with any such endeavor,
simplifying assumptions have had to be made to make the problem
tractable. This limits our ability to interpret the simulation
results and generalize them into recommended policies for EPA. In
this section we explore the limits to our study, discuss the
generality of our results, and suggest further research needed to
solve unanswered questions.
Qualifications. A number of simplifications have been made in
our model which are worth mentioning at this point. The important
aspect is the extent and direction of their impact on enforcement
policy characteristics and comparisons.
Recall that the model assumes firms attempt to minimize expected
costs. But if in actual fact firms are risk averse, they may tend to
control at somewhat higher levels than the model predicts. For
example, a firm facing a very large fine for violating the opacity
standard may "play it safe" and pick a rather large precipitator to
avoid the remote possibility of such a heavy penalty even though the
expected fine is relatively small. The same thing can be said about
firm reaction to failing the compliance test if enlargement costs are
relatively large. The exact opposite behavior occurs of course, if
firms are risk seeking. However, it should be pointed out that in the
actual runs of the simulation model (expected cost minimizing
behavior) firms almost never selected precipitators which violated the
opacity standard. Hence in our specific application it is mainly
reaction to the compliance test which could generate firm behavior
different from what we have predicted.
Another important qualification to the model is the absence of
firm challenges to enforcement. For example, some runs of the
simulation model indicated that for given policy parameters, firms
would control at relatively high efficiency levels. In actual fact
78
-------
when firms are faced with such situations they may use legal and
political resources to "purchase" lax enforcement. The reason is that
control costs increase at an increasing rate and consequently such
countervailing expenditures look increasingly attractive to the firm.
Behavior of this sort which is not now included in the model would
generate lower control levels than predicted.
On balance, we think that the simulation model probably
overstates firm pollution control effort especially for large plants.
However, this qualification does not alter relative comparisons
between the alternative enforcement schemes. These important and
useful results remain intact.
Recall also that the model uses Monte Carlo selection of costing
and engineering parameters from specified distributions to compute
costs. This procedure, designed to capture circumstantial and
geographic variation in costing conditions, produces a distribution of
results for each set of simulated enforcement policy parameters. By
comparing distributions it is possible to determine the frequency of
deviation from the model's expected results. Important considerations
here are technology selection and relative rankings of firm and
resource costs under the different sets of enforcement policies. In
general, the model is very robust. There are few cases—5% at the
most—when specific parameter selections produce results
qualitatively different than the reported expected results.
In addition the Monte Carlo distributions show that variation in
control effort is relatively small. Roughly speaking, about 97.5% of
the down-side variation in collection efficiency for a given set of
enforcement policy parameters is less than one-half of a percentage
point. Up-side variation at the 97.5% confidence level is covered by
an even narrower range especially at high efficiency levels where
distributions are pushed up against 100% efficiency.
Generalizing the Results. Many of the results of our
theoretical and empirical analyses are general in character. For
example, we would expect that differing combinations of compliance
79
-------
test standards and operating emission standards could yield identical
control levels whether we are analyzing fly ash control or automotive
emission controls. Likewise, we would expect the technology selection
reversal phenomenon to occur in other control situations as well.
However, our model and its application does not necessarily imply that
these results will occur in the same relevant policy space. Nor can
we imply that one form, of implementation and enforcement is
universally better than another. There most certainly are cases where
the technological problems of more-or-less continuous monitoring
generally associated with emission tax enforcement would be
prohibitively costly and a legal enforcement scheme would be optimal.
In the case we analyze we strongly suspect (but cannot prove without
additional research) than an effluent fee system is preferable.
However, this is almost the ideal case. Monitoring is easy and
available. Control technology is developed and well understood. In
many other cases this will not be so. Each case will have to be
analyzed individually to determine the correct implementation and
enforcement technique and the correct levels for the relevant policy
parameters.
Additional Research. Our model, as indicated earlier, is
incomplete. We have not attempted to simulate enforcement agency
costs and behavior. This is the major missing cost link in our
analysis. Consequently, we cannot yet determine total marginal costs,
that is the least-cost sum of firm marginal control costs, firm
marginal management costs, and enforcement agency marginal management
costs.
Completing the cost analysis, however, is a straightforward
conceptual problem although one which will probably require extensive
empirical analysis. Conceptually one would proceed by integrating
enforcement agency cost and behavioral equations (all as functions of
enforcement policy parameters) into the existing simulation model.
Least cost functions could then be determined, as before, by running
the model for a variety of enforcement policy sets. Summarization
80
-------
would be accomplished through regression analysis on only the optimum
envelope points. In effect, this would estimate reduced form
behavioral and cost equations (as a function of enforcement policy
parameters) which would reflect the simultaneous minimization of firm
control and management costs and enforcement agency management costs.
In turn, these equations would allow comparison of marginal and total
conditions for alternative enforcement schemes. The final result
would be specification of a complete cost minimizing enforcement policy.
This, after all, is the core problem of enforcement economics.
81
-------
FOOTNOTES
1. Other possible control instruments such as subsidies and marketable
permits have been neglected in this study.
2. Anderson and Crocker (1971) suggest that these issues are of vital impor-
tance in control instrument decisions but do not cite any literature
which explores their effects on control.
3. See Watson (1972).
4. See Becker (1968) who suggests a similar but not quite correct point.
5. Becker (1968, p. 199) has noted that: "fines should exceed the harm
done if the probability of conviction were less than unity. The
possibility of avoiding conviction is the intellectual justification
of punative, such as triple, damages against those convicted." Since
it is obvious from past experience that polluters are not always
convicted, penalties greater than damages and enforcement costs
may be justified.
6. While our model does not specifically consider the tradeoffs involved
in the interrelationships between control costs and total product
output of the firm, the conclusions reached here do hold in the
general case. For a model which relates pollution control costs to
the optimal output of the firm see Fan and Froehlich (1972).
7. This objective function can easily be translated into Becker's supply
of offenses function. However, it is stated in stochastic form rather
than deterministic form since many of the terms are stochastic in
nature. The first derivative of this function represents the value to
the firm of a violation and hence under perfectly competitive conditions
the opportunity cost to society of pollution control.
8. Stigler (1970, p. 528) argues strongly for a variable penalty. He con-
cludes that "marginal costs are necessary to marginal deterrence,"
Thus, penalties such as cases 3 and 4 are to be preferred.
9. For a discussion of the use of reliability in standards see Blumstein,
et al, (1972).
10. Tittle (1969) has shown that greater certainty of punishment for a crime
is associated statistically with lower offense rates.
82
-------
11. It has another option, to shift to an alternative enforcement scheme.
This may be preferable since in the current legal enforcement scheme
non-compliance is "... enforced by criminal process, probably the most
cumbersome coercive tool we have. The violator is protected by all
the constitutional protections which apply to any criminal trial. He
can demand a trial by jury and unanimous verdict (and this against
the heavy burden of proof faced by the prosecution)." (Krier, 1970,
p. 5-29).
12. Penalty costs in this case are the increased costs of producing the
power from alternative sources and the interest on investment in the
plant during the six months that would be required to complete the
expansion.
13. Two very computationally complicated variants of this model were investi-
gated. One was least cost selection of load shedding or fines when
the opacity standard was violated. Another was least cost selection
of serial enlargement or a single stage enlargement. In a sensitivity
analysis, both variants in combination produced results approximately
equal to those of the simpler basic model.
14. In a few cases 1000 iteration, Monte Carlo simulations were completed
and compared to the results of the 100 iteration runs. In each case
the results were identical to six or more decimal places.
15. The shifts in the power, function are due to the central limit theorem.
16. N was included in equations (13) and (14) only when it was a significant
determinant of cost and when N was not intercorrelated with E. This
occurred mostly for the 25 megawatt plants.
17. One might logically ask why a firm would bother to operate the device
if PC actually were zero. We assume it operates so as to avoid a fla-
grant violation of the law and the political and social costs of doing
so. However, it could be the case that the firm would stop operating
the device when PC=0.
18. The relevant economic costs for resource management are resource costs;
fines and taxes paid by firms are transfer payments which should not
influence resource allocation decisions.
19. Cost differences of about the same relative magnitude occur at other
plant sizes. Note that we have assumed that record keeping and fee
paying costs do not vary with the removal rqte.
83
-------
20. The enforced opacity standard is likely to be 30% or higher, rather
than the promulgated 20%. In the past courts have levied fines only
when violations were considerably greater than the relevant standards
and when firms were uncooperative and incalcitrant.
21. Time in violation (base load years, normal load conditions) is compu-
ted using the following equations:
IT = SQRT[(99 - AE)/.0026], 96.4=AE=99
100, AE<96.4
SQRT[(99
0, AE>99
100, AE<97.7
FT = (98.8 - AE)/.011 97.7=AE=98.8
0, AE>98.8
where IT=inflexible precipitator time-in-violation (base load
years, normal load conditions)
FT=flexible precipitator time-in-violation (base load years,
normal load conditions)
AE=average expected efficiency (base load years, normal load
conditions)
22. Becker (1968, p. 189 and 196) derived a similar result on theoretical
grounds. He argues that penalties (or standards) should be less for
smaller violators (plants) and that high income firms should be prosecuted
more thoroughly rather than less thoroughly as we have found in our
analysis.
23. This implies that the curve in Figure 14b represents a lower bound
estimate to the actual time in violation.
24. At point B, the firm would be indifferent between technologies.
25. It*also reduces the power lost by the control agency bureaucrats thus
further increasing its acceptability to them (see Section II).
84
-------
REFERENCES
1. Anderson, Robert J. Jr. and Thomas D. Crocker, "The Economics of
Air Pollution: A Literature Assessment" in Paul B. Downing (ed),
Air Pollution and The Social Sciences (New York: Praeger Publishers,
1971).
2. Becker, Gary A., "Crime and Punishment: An Economic Approach,"
Journal of Political Economy. 76: 169-217 (March/April, 1968).
3. Blumstein, Alfred, et al, "Optional Specifications of Air-Pollution-
Emission Regulations Including Reliability Requirements," Operations
Research, 20: 752-763 (July/August, 1972).
4. Dienemann, Paul F., Estimating Cost Uncertainty Using Monte Carlo
Techniques (Santa Monica, Calif.: Rand Corp., 1966).
5. Fan, Liang-Shing and B. R. Froehlich, "Pollution Control and the
Behavior of the Firm," The Engineering Economist, 17: 261-267 (Summer
1972).
6. Federal Register, Thurs. Dec. 23, 1971, Vol. 36, No. 247, Part II.
7. Greco, Joseph and Wynot, William A., "Operating and Maintenance
Problems Encountered with Electrostatic Precipitators," Proceedings
of the American Power Conference, 33: 345-353 (1971).
8. Krier, James, "Air Pollution and Legation Institutions," in P. B.
Downing (ed), The Contribution of the Social Sciences to the Solution
of the Air Pollution Problem, Taskforce Assessments, Project Clean
Air (Riverside: Statewide Air Pollution Research Center, University
of California, 1970).
9. Stigler, George J., "The Optimum Enforcement of Laws," Journal of Poli-
tical Economy, 78: 526-536 (May/June, 1970).
10. Tittle, Charles R., "Crime Rates and Legal Sanctions," Social Problems,
16: 409-422 (Spring, 1969).
11. Watson, William D. Jr., Costs of Air Pollution Control in the Coal-Fired
Electric Power Industry, Ph. D. Dissertation, University of Minnesota
(Ann Arbor: University Microfilms, 1970).
12. Watson, William D. Jr., "External Diseconomies, Corrective Taxes and
Market Structure: Comment," unpublished paper, Environmental Protection
Agency, Sept. 1972.
85
-------
13. Watson, William D. Jr., "Stochastic Operating Characteristics and
Cost Functions of Electrostatic Precipitators," The Engineering
Economist, 18: 79-98, (Winter 1973).
14. Watson, William D. Jr., "Costs and Benefits of Fly Ash Control,"
Journal of Economics and Business, May 1974.
15. White, Harry J., Industrial Electrostatic Precipitation, (Reading,
Mass.: Addison-Wesley, 1968).
86
-------
MATHEMATICAL APPENDIX
Figures 2 and 5 show the efficiencies, costs and probabilities
which enter the simulation model. Equations and parameter values
for these are presented below. Symbols are defined in Table A.I.
Expected Precipitator Efficiency. The following equation (Watson
(1970 and (1973b))) is used to compute expected precipitator efficiency
(EPE):
(A.I) EPE = Expectation {100-[l-exp(-z-exp u)]}
where z = a0(A/V)ai(KW/V)a2(S/Ah)a3(a4)MF
and u is N(0, a£)
Values for parameters aQ through a^ and the variance of u are listed
in Table 2.A. These have been estimated via regression analysis
using cross section data on 37 precipitator systems (Watson (1970)).
Equation A.I has two key functions in the simulation models.
One is to determine expected precipitator efficiency during a compliance
test. This establishes the probabilities of pass and fail for
each of the precipitator sizes considered in simulating the compliance
test. The following Monte Carlo procedure is used (Dienemann (1966)):
1. Designate a value for z.
2i Randomly select a value for u from its distribution.
3. Compute efficiency using equation A.I (without taking
the expectation).
4. Repeat step 2 through 3, j times where j^3.
5. Average the computed efficiencies, j in number.
6. Repeat steps 2 through 5, 200 times.
In simulating the compliance test the model considers fourteen
different precipitators of increasing size, four different compliance
test standards, and four different flue gas flow rates. Some representa-
tive probabilities of fail are listed in Tables A.3 through A.6.
These values reflect the relationships demonstrated in Figures 7a, 7b
and 7c.
The second function of equation A.I is to determine the days
per year when a designated opacity standard is violated (scenarios
SI and S2) and the tons of fly ash discharged per year (scenarios
87
-------
S3 and S4) . Linear operating curves for both flexible and inflexible
technology are computed using the following variant of equation A.l:
(A. 2) EPE = Expectation {100[l-exp(-z.exp u)]}
where "first day" z = z and
r\
"last day" z - (.808) 2 (base load years inflexible technology)
or "last day" z = (,808)'6z (base load years flexible technology)
The deterioration factor, .808, is from Greco and Wynot (1971).
The exponent 2 for "last day'1 inflexible efficiency indicates that
both available collecting plate and power input deteriorate under
inflexible technology. The exponent .6 for "last day" flexible
efficiency indicates that power input only deteriorates under flexible
technology. Numerical integration is used in the actual computation
of expected efficiency. Operating efficiencies for inflexible
and flexible precipitators are shown in Tables A. 7 and A. 8 respectively.
Expected Precipitator Costs . Installed precipitator costs (IPC)
discounted over n years at r% are computed using the following
equation (Watson (1973b)):
1/2 1/2
(A.3) IPO 2V(lnz) • (1/203) •
.4)'
n
/SIXMO + r)*
t=l * r
dac
'T-R
.6
.3
+200 MW
The term 200MW is the discounted instrumentation cost for flexible
technology (scenarios S2, S4, and S6 only).
Compliance tests costs (CTC) are computed as:
(A. 4) CTC = (XI + S1-X2)M
Operating costs are the sum of discounted labor and maintenance
costs (DLMC), discounted fan power costs (DFPC), discounted fly ash
disposal costs (DFADC), and discounted opacity monitoring costs (DMC)
88
-------
These are computed using the following equations (Watson (1973))
(A. 5) DLMC = dmQ + iyV-
+r)
t =
•60
(A.6) DFPC =V 1 /Pressure drop in inches of water • 5.202 -1000-V-h.-flX
t^O+r)H 44,250 Fan Efficiency I
(A.7) DFADC =
=£_L/Mw-CF-h.-1000-HR\/5:FCfl-Ah.AE
ffiS-M 5 )( FA
\ HC . 2000 / \
(A. 8) DMC = dMC
The same equations, with some exceptions to be noted, are
used to compute enlarged installation and operating costs. One
exception is that enlarged costs are keyed to precipitator sizes
(certainty size) which have almost no probability of failing the
compliance test. For example, Table A.5 shows under test conditions
of 1.15V and a compliance test standard of .1 that precipitator
13 (10 stack samples) is of certainty size. When computing enlarged
installation and operating costs for smaller precipitators, given
a compliance test standard of .1 and flue gas flow rate of 1.15V,
the simulation model uses this precipitator size. Similar procedures
are followed under other compliance test conditions. A second
exception is that enlarged installation costs are premultiplied
by a factor, Y3, which adjusts for the extra structural costs required
for enlargement. A third exception is that enlarged operating
costs include an appropriate compliance test cost for the enlarged
89
-------
precipitator and certain penalty costs. These penalty costs (PC)
are estimated as follows:
(A.9) PC = (MW-lOOO-Yl-r/2) + (MW-1000-CF-H1-Y2)
The first bracketed term is additional interest on plant investment
due to a six month delay for precipitator enlargement. This can
be thought of as an opportunity cost to the firm. The second bracketed
term is the higher costs of power from an alternative generator
during the shutdown-enlargement period. The fourth exception is that
the "normal" enlarged operating costs are discounted with a half-year
delay to account for operating delays.
The remaining costs are total discounted fines (F) and emissiojn
taxes. Fines are computed using the following equation:
(A. 10)
F =t|1Dayt-Fine/Day-Probability of Conviction/(l+r)t
Emission taxes are computed using equation A.7 except that average
\
expected efficiency is replaced by one minus average expected effi-
ciency (see Figure 3) and average disposal cost per ton (AC) is re-
placed by emission tax per ton.
Parameter Values. In computing costs the simulation model
randomly selects values from Beta distributions of key costing para-
meters (Dienemann (1966)). Table A.9 lists their low modal and high
values and distribution types. Representative probability density
curves for these distributions are shown in Figure A.I.
A number of parameters retain fixed values throughout the simula-
tions. These are listed in Table A.10.
Other values change as the simulation model considers different
power plant sizes. These are shown in Table A.11.
90
-------
Figure A.I
Beta Distribution Shapes Programmed in Monte
Carlo Model*
High
variance
Skewed left
Type 1
Type 2
Skewed right
Type 3
Medium
variance
Type 4
a=/3=2.75
Type 5
Type 6
Low
variance
Type 7
a=/3=4.0
Type 8
Type?
*Taken from Dienemann (1966).
91
-------
TABLE A.I
NOMENCLATURE
Symbol
Definition
Unit of Measurement
A
aO»al»a2»a3»a4
a
AC
AE
AH
CF
C0
cl
Collecting plate area of an electro-
static precipitator
Empirically determined parameters for
explaining precipitator performance
Annual capital charge rate
Average cost of disposing of collected
fly ash
Average expected precipitator effi-
ciency
Percent ash by weight of combusted
coal
Costs of a kilowatt-hour of electric
power
Fixed cost for installing collecting
plates
Installed cost per square foot of
collecting plate area
Capacity factor of associated
generator during operating hours
Installed cost per BBS
Installed cost per each KW of power
input capacity to the discharge
electrodes
Empirically determined parameter for
determining precipitator performance
1000's of sq. ft.
Undimensioned
$/$1000
$/ton in 1967 $
$/KW-Hr in 1967 $
$1000 in 1967 $
$/ft.2'in 1967 $
Average load in mega-
watts divided by capa-
city load in megawatts
$1000/EBS in 1967 $
$1000/KW in 1967 $
Undimensioned
92
-------
TABLE A.I (Continued)
NOMENCLATURE
Symbol
Definition
Unit of Measurement
Dayt
d
E
EBS
HI
HC,
HR
KW
Number of days opacity standard is
violated in year t
Weight by fly ash collected divided
by weight of fly ash entering an
electrostatic precipitator, multi-
plied by 100
Average efficiency of the trans-
former-rectifier sets in a precipi-
tator
Electric bus sections in an electro-
static precipitator
Number of coal burning hours of
associated boiler in year t
Shutdown hours for enlargement
Average heat content of combusted
coal
Average heat rate of associated
generator
Average rate of power capacity
utilization in year t
Power input to the discharge elec-
trodes of an electrostatic preci-
pitator
Fixed operating and maintenance cost
Operating and maintenance cost per
each cubic foot of flue gas treated
Days/year t
Und imens ioned
Undimensioned
The number of EBS's
Hours/year t
Number of hours
BTU/lb. of coal
BTU/KW-hr.
Und imens ioned
Kilowatts
1967 $
$/ft.3 in 1967 $
93
-------
TABLE A.I (Continued)
NOMENCLATURE
Symbol
Definition
Unit of Measutement
M
MW
MC
n
r
S
SI
u
XI
X2
Yl
Y2
Y3
Number of Compliance Tests
Output rating at full capacity of
associated generator
Opacity monitoring dosts
Number of years of operation
Discount rate
Percent sulfur by weight of com-
busted coal
Number of Compliance Test Stack
Samples
Fly ash emission factor for com-
busted coal
Random error term in the regres-
sion equation for efficiency
Normal load volumetric flue gas
flow rate through a precipitator
Setup cost for compliance test
Cost per stack sample (compliance
test)
Installed capital cost of an
electric generating unit
Penalty cost for alternative power
Cost penalty multiplier for pre-
cipitator enlargement
Number of tests
Megawatts
1967 $/yr.
Number of years
Undimensloned
Z
Number of samples
Tons of fly ash generated
from 1% ash coal per ton
of combusted coal
Undimensioned
1000's of cubic feet/
minute
1967 $/test
1967 $/sample
1967 $/KW
1967 $/Kwh
Undimensioned
94
-------
TABLE A.2
ELECTROSTATIC PRECIPITATION PARAMETERS
Parameter
Estimated Value
Standard Error
Exp(5.06) = 157.6
1.4
.6
.22
Exp(.252) = 1.29
.12
.43
.165
.1
.0975
.1477
95
-------
TABLE A.3
COMPLIANCE TEST FAILURE PROBABILITIES
(R • 1.157, Compliance Test Standard - .04)
Sample
Size.
3
5
10
15
vo
ON 20
25
30
35
40
45
50
1
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
2
1.000
1.000
i.qoo
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
3
1.000
1.000
1.000
1.000
ilooo
1.000
1.000
1.000
1.000
1.000
1.000
4
1.000
1.000
1.000-
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
5
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
Precipitates Number
678
. .970
.993
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1. 000.
.943
.990
.997
1.009
1.000
1,003
1.000
l.OOD
1.003
1.000
1.000
.893
.947
.990
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
9
.813
.913
.987
.997
1.000
1.000
1.000
1.000
1.000
1.000
1.000
10
.790
.907
.977
.997
1.000
1.000
1.000
1.000
1.000
1.000
1.000
11
.767
.873
.953
.987
.997
1.000
1.000
1.000
1.000
1.000
.1.000
12 -
.557
.617
.723
.780
.833
.880
.913
.917
.537
..943
.960
13
.260
.233
.217
.200
.187
.173
.157
.140
.130
.110
.097
14
.113.
.087
.057
.013
.003
.010
.007
.003
.000
.000
.000
-------
TABLE A.4
COMPLIANCE TEST FAILURE PROBABILITIES
(R • 1.1V, Compliance Test Standard •» .04)
Sample *Precipitator Number
Size '
1 2 3 4 5 6 7 8 9 10 11 12 13 14
3 1.000 1.000 1.000 1.000 .993 .920 .883 .797 .737 .690 .660 .380 .157 .043
5 1.000 1.000 1.000 1.000 1.000 .983 .953 .877 .820 .810 .743 .450 .100 .033
10 1.000 1.000 1.000 1.000 1.000 1.000 .990 .977 .937 .903 .860 .493 .083 .003
15 1.000 1.000 1.000 1.000 1.000 1.000 1.000 .993 .960 .977 .930 .493 .037 .000
20 1.000 1.000 ' 1.000 1.000 1.000 1.000 1.000 1.000 .990 .983 .957 .567 .017 .000
25 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 .987 .997 .983 .573 .020 .000
30 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 .990 .573 .013 .000
35 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 .997 .993 .603 .007 .000
40 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 .997 .60,7 .003 .000
45 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 .613 .000 .000
50 1.000 1.000 1.000 1.000 1;000 1.000 1.000 1.000 1.000 1.000 1.000 .597 .000 .000
-------
00
TABLE A. 5
COMPLIANCE TEST FAILURE PROBABILITIES
(R - 1.15V, Compliance Test Standard - .1)
Sample
Sice
3
5
10
15
20
25
30
35
40
45
50
1
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
2
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
3
.997
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
4
.987
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
5
.973
.993
'1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
Precipitator
6 7
.823
.910
.983
.993
.997
1.000
1.000
1.000
1.000
1.000
1.000
.767
.830
.930
.960
.987
.993
.990
.993
.997
l.COO
l.COO
Number
8
.633
.703
.840
.890
.937
.950
.960
.970
.993
.993
.990
9
.527
.593
.670
.747
.773
.800
.823
.820
.850
.873
.877
10
.483
.550
.620.
.650
.683
.717
.730
.760
.790
.790
.803
11
.410
.4,73
.463
.460
.467
.470
.487
.503
.510
.507
.513
12
.207
.160
.090
.060
.033
.023
.017
.010
.010-
.003
.001
13
.047
.017
.000
.000
.000
.000
.000 .
.000
.000
.000
.000
14
.000
.003
.000
.000
.000
.000
.000
.000
.000
.000
.000
-------
AO
TABLE A.6
COMPLIANCE TEST FAILURE PROBABILITIES
(R • 1.1V, Compliance Test Standard • .1)
Sample
Size
3
5
10
15
20
25
30
35
40
45
50
1
1.000
1.000
1.000
1.000
1.000
1.000
1.000
^.000
1.000
1.000
1.000
»
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
3
.997
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
4
.970
1.000
1.000
1.000
1.000
1.000
1.000
1:000
1.000
1.000
1.000
5
.917
.977
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
Precipitator
6 7
.710
.823
.920
.967
.983
1.000
1.000
1.000
1.000
1.000
1.000
.590
.647
.743
.790
.860
.383
.913
.927
.')30
.953
.960
Number
8
.510
.553
.620
.660
.697
.747
.740
.727
.770
.810
.780
9
.393
.433
.380
.410
.400
.417
.387
.377
.390
.400
.370
10
.350
.370
.333
.303
.287
.283
.273
.293
.303
.250
.253
11
.287
.317
.200
.140
.103
.110
.110
.no
.087
.070
.067
12
.100
.050
.020
.000
.000
.003
.000
.000
' .000
. .000
.000
13
.020
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
14
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
-------
TABLE A.7
PRECIPITATOR CHARACTERISTICS
(Inflexible Technology)
Precipitator
1
.2
3
4
5
6
7
8
9
10
11
12
13
14
z
2.756
2.998
3.289
3.654
4.134
4.836
5.347
5.598
5.974
6.094
6.437
7.444
8.861
9.859
First Day*
Efficiency
(%)
91.96
93.37
94.72
96
97.19
98.3
98.8
98.98
99.21
99.27
99.41
99.69
99.86
99.92
Last Day*
Efficiency
(%)
82.12
84.39
86.74
89.14
91.59
94.15
95.47
96 .
96.67
96.86
97 .33
98.33
99,1
99.41
Average*
Efficiency
(%)
87.04
88.88
90.73
92.57
94.39
96.23
97.14
97.49
97.94
98.07
98.37
99.01
99.48
99.67
% of Time
in Violation of
Standard*
(%)
100
100
100
100
100
100
86
78
65
61
48 ,
0
0
0
*During base load years, normal load conditions
100
-------
TABLE A.8
PRECIPITATOR CHARACTERISTICS
(Flexible Technology)
z*
Frecipitator (IV)
1
2
3
4
5
6
7
8
9
10
11
» •
12
13
*4
2.756
2.998
3.289
3.654
4.134
4.836
5.347
5.598
5.974
6.094
6.437
7.444
8.861
9.859
First Day Last Day
Efficiency* Efficiency*
(%) (%)
91.96
93.37
94.72
96.00
97.19
98.30
98.80
98.98
99.21
99.27
99.41
99.69
99.86
99.92
89.47
91.16
92.80
94.40
95.95
97.43
98.14
98.40
98.72
98.81
99.03
99.46
99.75
99.85
Average % of Time in
Efficiency* Violation of Standard*
(%) (%)
90.72
92.27
93.76
95.20
96.57
97.87
98.47
98.69
98.97
99.04
99.22
99.58
99.81
99.89
100
100
100
100
100
100
29
0
0
0
0
0
0
0
*During base load years, normal load conditions
101
-------
TABLE A.9 .
CHARACTERISTICS OF COST PARAMETER DISTRIBUTIONS*
Cost
Parameter**
a
AC
B
b0
bl
C0
cl
HR
mo
ml
MC
r
XI
X2
Yl
Y2
Y3
Low
Value
80
.5
.002
70
.5
7.2
.08
8,800
1,817
.000027
5,000
.05
6,000
500
125
.0005
1
Modal
Value
140
1
.004
180
4.7
7.6
.12
9,400
2,317
.000033
10,000
.09
12,000
1,000
165
.001
1.2
High
Value
180
3
.008
290
8.9
8.0
.16
10,000
2,817
.000039
30,000
.13
24,000
2,000
205
.004
2
Distribution
Type
7
3
6
2
2
8
8
5
8
8
6
8
3
3
5
3
6
* For parameters measured in dollars, low, modal, and high values are in
1967 dollars.
** Characteristics for bg» b]_, CQ, c^, mQ, and m^ are based upon regression
analysis (Watson (1970)). Characteristics for <*, AC, 3, HR, MC, r, XI,
X2, and Yl are representative of known estimates. Characteristics for
Y2 and Y3 are based upon engineering judgment.
102
-------
TABLE A.10
FIXED "COSTING" PARAMETERS*
f
ht =<
7,440 yrs.
5,200 yrs.
2,160 yrs.
880 yrs.
1-12
13-17
18-25
26-30
s**** =
Ah
CF
ZFFA
.5%
6%
.9
.0085
*» n
kt = <
.889 yrs.
.8988 yrs.
.9634 yrs.
.9851 yrs.
\
kt*** = l for all
1-12
13-17
18-25
26-30
•
years
Fan Efficiency = .6
Pressure Drop = .5
?T— R = *
EC =• 8500
n =30
* Values are representative of known estimates.
** These values are appropriate for inflexible precipitator technology.
They reflect negative exponential failure of discharge electrodes.
*** This value is appropriate far flexible precipitator technology. It
indicates full utilization of power input to discharge electrodes.
**** This value, representative of a low sulfur western coal and a desul-
furized eastern coal, satisfies new source performance standards for
sulfur dioxide emissions.
103
-------
TABLE A.11
REPRESENTATIVE FLUE GAS VOLUMES AND PRECIPITATOR
SECTIONALIZATION FOR DIFFERENT SIZED POWER PLANTS
Plant
Size
(MW)
25
200
800
1,300
V*
(1000fs of Actual Cubic
Feet per Minute)
| 131
706
2,173
3,221
EBS
(Electrical
Bus Sections)
4
6
18
36
* Estimated using V - 9.03 MW811((T + 460)/760)1'21 (Watson (1970) p. 80).
It is assumed that flue gas temperature (T) is 340°F.
104 *U.S. GOVERNMENT PRINTING OFFICE: 1974 546-317/294 1-3
-------
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
EPA-600/5-73-014
3. Recipient's Accession No.
4. Title and Subtitle
"Enforcement Economics in Air'Pollution Control"
5* Report Date
December 1973
6.
7. Autboi(s)
Paul B. Downing and William D. Watson, Jr.
&• Performing Organization Kept.
No.
9. Performing Organization Name and Address
Washington Environmental Research Center
Implementation Research Division
Environmental Protection Agency
Washington, DC 20460
10. Project/Task/Work Unit No.
1HA094
11. Contract/Grant No.
12. Sponsoring Organization Name and Address
(same as 9.)
13. Type of Report & Period
Covered
Final Report
14.
15. Supplementary Notes
16. Abstracts The effects of alternative enforcement strategies on the pollution control
activities of the firm are investigated. There are a number of tradeoffs available to
a firm including delay and non-compliance which allow it to minimize expected pollution
control costs. These are identified within the context of a generalized behavioral
model for the firm and an empirical study is undertaken to determine their importance.
In a simulation of current enforcement of the federal new source particulate mattei
discharge standard for coal-fired power plants it is found that cost-minimizing power
plants will install relatively costly pollution control technology and frequently vio-
late federal fly ash standards. Two alternative enforcement strategies for overcoming
these shortcomines, namely compliance tests in combination with emission taxes and
emission taxes alone, are analyzed.
In the case of the federal fly ash discharge standard for coal-fired power plants
it is tentatively concluded that emission tax enforcement would probably result in an
approximate minimization of the sum of firm and enforcement agency resource costs. The
17. Key Words and Document Analysis. 17a. Descriptors general applicability of this result to Other
Economic Analysis enforcement problems is discussed.
Policy Tradeoffs
Pollution Control
Enforcement Policy
Cost-Effectiveness
Monte Carlo Simulation
17b. Identifiers/Open-Ended Terms
17c. COSATI Field/Group Q503
18. Availability Statement
Release Unlimited
19.. Security Class (This
Report)
.SSIFI
UtyCLAS
curity Cli
20. Security Class
Page
UNCLASSIFIED
21. No. of Pages
104
22. Price
FORM NTIS-35 (REV. 3-72)
USCOMM-DC 14992-P72
-------
INSTRUCTIONS FOR COMPLETING FORM NTIS-35 (10-70) (Bibliographic Data Sheet based on COSATI
Guidelines to Format Standards for Scientific and Technical Reports Prepared by or for the Federal Government
PB-180 600).
1. Report Number. Each individually bound report shall carry a unique alphanumeric designation selected by the performing
organization or provided by the sponsoring organization. Use uppercase letters and Arabic numerals only. Examples
FASEB-NS-87 and FAA-RD-68-09.
2. Leave blank.
3. Recipient's Accession Number. . Reserved for use by each report recipient.
4. Title and Subtitle. Title should indicate clearly and briefly the subject coverage of the report, and be displayed promi-
nently. Set subtitle, if used, in smaller type or otherwise subordinate it to main title. When a report is prepared in more
than one volume, repeat the primary title, add volume number and include subtitle for the specific volume.
5. Report Dote. Each report shall carry a date indicating at least month and year. Indicate the basis on which it was selected
(e.g., date of issue, date of approval, date of preparation.
6. Performing Organization Code. Leave blank.
7. Author<«). Give name(s) in conventional order (e.g., John R. Doe, or J.Robert Doe). List author's affiliation if it differs
from the performing organization.
8. Performing Organization Report Number. Insert if performing organization wishes to assign this number.
9. Performing Organization Name and Address. Give name, street, city, state, and zip code. List no more than two levels of
an organizational hierarchy. Display the name of the organization exactly as it should appear in Government indexes such
as USGRDR-I.
10. Project/Task/Work Unit Number. Use the project, task and work unit numbers under which the report was prepared.
11. Contract/Grant Number. Insert contract or grant number under which report was prepared.
12. Sponsoring Agency Nome and Address. Include zip code.
13. Type of Report and Period Covered. Indicate interim, final, etc., and, if applicable, dates covered.
14. Sponsoring Agency Code. Leave blank.
15. Supplementary Notes. Enter information not included elsewhere but useful, such as: Prepared in cooperation with . . .
Translation of ... Presented at conference of ... To be published in ... Supersedes . . . Supplements . . .
16. Abstract. Include a brief (200 words or less) factual summary of the most significant information contained in the report.
If the report contains a significant bibliography or literature survey, mention it here.
17. Key Words and Document Analysis, (a). Descriptors. Select from the Thesaurus of Engineering and Scientific Terms the
proper authorized terms that identify the major concept of the research and are sufficiently specific and precise to be used
as index entries for cataloging.
(b). Identifiers and Open-Ended Terms. Use identifiers for project names, code names, equipment designators, etc. Use
open-ended terms written in descriptor form for those subjects for which no descriptor exists.
(c). COSATI Field/Group. Field and Group assignments are to be taken from the 1965 COSATI Subject Category List.
Since the majority of documents are multidisciplinary in nature, the primary Field/Group.as signment(s) will be the specific
discipline, area of human endeavor, or type of physical object. The application(s) will be cross-referenced with secondary
Field/Group assignments that will follow the primary posting(s).
18. Distribution Statement. Denote releasability to the public or limitation for reasons other than security for example "Re-
lease unlimited". Cite any availability to the public, with address and price.
19 & 20. Security Classification. Do not submit classified reports to the National Technical
21. Number of Pages. Insert the total number of pages, including this one and unnumbered pages, but excluding distribution
list, if any.
22. Price. Insert the price set by the National Technical Information Service or the Government Printing Office, if known.
ORM NT.S-S8 (REV. 3-72, iJsCOMM^C .4**P7!
------- |