United States         Off ice of           January 198]
              Environmental Protection     Planning and Management
              Agency           Washington, DC 20460
&EPA        An Analysis of Economic
              Incentives to Control
              Emissions of Nitrogen
              Oxides from Stationary

              Report of the Administrator
              of the Environmental
              Protection Agency to the
              Congress of the United
              States In compliance with
              Section 405(f) of Public Law
              95 - 95  The Clean Air Act,
              as Amended



                    Report  of  the


                  PROTECTION AGENCY

                        to  the


In compliance With Section  405(f)  of Public Law 95-95

                     January 1981


     This report to Congress was prepared under the direction of
Alex Cristofaro, of the Office of Planning and Evaluation,
United States Environmental Protection Agency.  The nucleus of the
team that performed the analysis consisted of Mr. Cristofaro;
Willard W.  Smith, Office of Planning and Evaluation; Ralph Luken,
Office of Air Noise, and Radiation; and Edward Pechan, consultant
to EPA.  Comments were also provided by John Hoffman, Steven Seidel,
Rodney Lorang, Alan Basala, Michael Shapiro, Jeff Kolb, Ed Tuerk,
Michael Levin, Eric Smith, Joan O'Callaghan, and Deb Taylor, each of
whom provided valuable criticisms and suggestions for this report.

     Contractors assisting in the preparation of the report

     MATHTECH, Inc., P.O. Box 2392, Princeton, N.J.  08540;
     Contract #EQ8AC005 (CEQ); contract amount:  $129,983;
     noncompetitively awarded; project officer:  Mr. David
     Tundermann (formerly of CEQ); and

     Energy and Environmental Analysis, Inc.  (EEA),
     1111 North 19th St., Arlington, Virginia  22209;
     Contract #EQ8AC006 (CEQ); contract amount:  $120,000;
     noncompetitively awarded; project officer:  Mr. David
     Tundermann (formerly of CEQ).

     MATHTECH and EEA collaborated on a report entitled An
Analysis of Alternative Policies for Attaining and Maintaining a
Short-Term N02 Standard.  Parts of Appendix A are directly taken
from this report, as are Figures 1 and 2.  While data gathered by
MATHTECH and EEA were used to perform the analyses appearing in
the present report to Congress, the final empirical results were
obtained by Mr. Pechan, Mr. Luken, and Mr. Cristofaro and not by
the contractors.  The role of contractor involvement in perform-
ing the analyses is further discussed at the end of Appendix A.

     Appendix B was entirely extracted from an early draft of a
report prepared by Putnam, Hayes, and Bartlett, 50 Church St.,
Cambridge, Mass.  02138, entitled Application of a Marketable
Permit System to the Control of Air Pollution (Contract #68-01-5835,
Assignment #52; dollar amount:  $29,200; competitively awarded).
The project officer for this report was Willard Smith of the Office
of Planning and Evaluation.



Conclusions and Recommendations                                 1

Section I:  Market Incentives                                   7

   Pure Emission Charges                                        8

   Standards and Charges                                       12

   Standards and Marketable Permits                            14

        Designing a Marketable Permits System                  15
        Devising an Initial Allocation of Permits              18
        Managing the Number of Permits over time               19

   EPA's Current System                                        21

        Bubble Policy                                          22
        Offset Policy                                          22
        Banking Program                                        22
        New Source Review and New Source Performance           23
        Noncompliance Penalties                                23

   Summary and General Conclusions                             23

Section II:  A NOx Case Study                                  25

     The NOx Problem                                           25

     Effects of NOx                                            26

     Ambient Air Quality Standards                             27

     An Empirical Analysis of the Advantages                   28
       of Trading NOx in Chicago

     Summary and Conclusions                                   42

Appendix A:  Data Sources and Methodology for the Chicago
               Case Study

Appendix B:  Analysis of the Operation of a Simplified Permit

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

Figure 11.
            LIST OF FIGURES

Firm Behavior in Response to an Emission Charge      10

Socially Optimal Level of Emissions                  H

A Schematic of the EPA Model                         31

Costs, Emission Reductions, and Number of Sources    32
Controlled under Alternative Regulatory Strategies

Ambient Impact of Alternative Strategies             33

Predicted Air Quality at One Location in Chicago     34
if Stationary Sources Are Not Controlled

Simulated Air Quality at One Location in Chicago     35
under SIP-without-Trading Alternative

Simulated Air Quality at One Location in Chicago     36
under a Uniform Emission Fee

Simulated Air Quality at One Location in Chicago     37
under Emission Fees by Source Type

Simulated Air Quality at One Location in Chicago     38
under Source-by-Source Emission Fee

Simulated Air Quality at One Location in Chicago     39
under SIP-with-Trading Alternative

                         EXECUTIVE SUMMARY

     Section 405(f) of the Clean Air Act Amendments of 1977 requires
the Environmental Protection Agency (EPA) to evaluate the advantages
and disadvantages of establishing a system of penalties on emissions
of nitrogen oxides  (NOx) from stationary sources and report its find-
ings to Congress.   In compliance with the Act EPA, and the Council
on Environmental Quality (CEQ) commissioned a study to: (1) compare
the costs of various emission control strategies that incorporate
various economic incentive approaches to pollution control; (2) define
the specific characteristics, including the compatibility with EPA's
existing approach to air quality management, of each economic incentive
policy; and (3) assess the usefulness of different economic incentive
policies for meeting and maintaining a short-term nitrogen dioxide
(NC>2) standard.  Short-term control for nitrogen oxides was chosen
for study because the 1977 Clean Air Act Amendments also called for
EPA to decide whether to set a short-term NC-2 national ambient air
quality standard.

     Since a contractor completed the original study in the fall of
1979, EPA has improved and updated the data-base and models used in
the contractor study.  In addition, EPA has undertaken other studies
on economic incentives.  This report presents the Agency's conclusions
on the use of market incentives to control NOX form stationary sources.


     The major conclusions of this report are:

        SOURCES.   While command-and-control strategies can be
        designed to discriminate between sources, the advantage of
        economic incentive systems, such as emission charges and
        marketable permits, is that they locate the decision for
        the installation of control equipment with those that have
        the best access to information on control costs and who have
        the greatest incentive to make economically efficient

   policies polluters have incentives to seek out  sources
   of low-cost emission reductions.  The result will  be  a
   more cost-effective pollution abatement program.   Up
   to this point, however, companies and states have  taken
   limited advantage of these policies.  One reason has  been
   the cumbersome administrative process associated with
   these policies.  EPA is currently working on streamlining
   this process to the maximum possible extent under  law.

   will submit a report to Congress under Section  405(g) of
   the 1977 Amendments in early 1981, outlining those areas
   where economic incentive approaches would result in attain-
   ment of environmental goals at less cost than under the
   existing regulatory framework and where the Act limits
   EPA's authority to implement such approaches.   Moreover,
   the best test of such new approaches would be through well
   funded pilot efforts.

        While EPA's authority to implement some types of
   market incentive approaches may not be limited  under  the
   Clean Air Act, the Agency feels that direct authority to
   implement certain approaches (where local authorities are
   willing to support such efforts) can avoid costly  litig-
   ation and delays.  Furthermore, Congressional endorsement
   of economic approaches to the regulation of air pollution
   will make such efforts more acceptable to local pollution
   control authorities.

   AMBIENT IMPACT.  The reasons for this are:

   — simplicity and ease of administration; When  emissions
   are representative of a source's contribution to an ambient
   air quality problem, a control agency does not  have to under-
   take (or supervise) elaborate ambient air quality  modeling
   to ensure that Source A can trade emission limitations with
   Source B without degrading air quality.  It needs  only to
   know both A's and B's emissions.

        — market  size;  For  some  pollutants  all  emissions  are
        thought  (or  assumed)  to contribute equally  to  the  ambient
        air quality  within an area  —  regardless of where  emitted
        within the airshed.   Emitters  of  these pollutants  could
        thus  trade emission  limitations with any source within
        the airshed  without  degrading  air quality.  The potential
        market in  emission limitations is therefore quite  large.
        However, for pollutants where  ambient impact depends not
        only  on  the  quantity of the pollutant emitted, but on the
        configuration (such  as smoke stack height)  and location of
        the source,  trading  of emission limitations must be
        restricted to sources that  can demonstrate  equivalent
        impacts  on air quality within  the same location.   This has
        the effect of decreasing  market size, which in turn de-
        creases  the  potential cost  savings that  can be realized
        from  a market-based  control approach.

        CANDIDATES.   NOX emissions, which cause  high short-term
        concentrations of NO2, do not  lend themselves as readi-
        ly to market approaches.  The  Chicago case  study showed
        that  these high  concentrations occurred  in  several very
        localized  areas.  In each such location, high concentra-
        tions were attributable to  automobile emissions and to a
        very  limited number  of stationary sources.  Thus,  a large,
        areawide market  in emission limitations  would not  pre-
        sent  itself.

              This, however,  is not  the case  for  several other
        pollutants.*  EPA's  current approach to  ozone control
        assumes  that all nonmethane hydrocarbon  emissions  contri-
        bute  equally to  ambient ozone  levels —  regardless of
        where emitted within an airshed.  Similarly, each  ton of
        "weighted" chlorofluorocarbon  emissions  is  thought to have
        the same impact  on the depletion  of  the  earth's ozone
        layer.   And  finally,  regional  emissions  of  NOX and SO2
        are believed to  cause acid  precipitation.   Thus, market
        incentive  measures could  probably play a more  important
        role  in mitigating the role of N02 in acid  precipitation
        (where all emissions  are  thought  to  contribute equally to
        the ambient  problem)  than in controlling short-term N02
        concentrations (where location and stack parameters play a
        significant  role  in  ambient concentrations).
        *While for some pollutants  it  is  likely  that  the  spatial
and temporal pattern of emissions,  as  well as  chemical  composition,
do play a role, our models are not  sophisticated enough to  account
for these variables.  Thus, for some air  pollutants  (e.g.,  ozone)
emissions are simply assumed to be  the measure of impact  on air

                          - 4 -

   TECHNOLOGIES.  Emission charge systems provide incentives
   to companies to develop better control techniques, since
   the periodic (probably annual) charge or fee billed to a
   company provides a continual incentive to further reduce
   emissions.  Under marketable permits, firms have a contin-
   uous incentive to develop innovative control technology to
   reduce their emissions so that they can sell permits on
   the market at a profit.  Since by definition one cannot
   predict how much change will come with "innovation," we
   could not quantitatively analyze these effects.  However,
   because these incentives are not present under traditional
   command-and-control regulations, it is logical that economic
   incentive approaches will better stimulate the development
   of better control techniques.

   To be efficient, a pure economic incentives approach would
   involve setting individual source-by-source fee schedules
   in which the fee for reducing emissions from level x to
   level y was set equal to the value of the additional environ-
   mental benefits of each such reduction — all the way from
   uncontrolled emission levels to zero emissions. But if the
   benefits of pollution reduction from a source are difficult
   to measure, setting correct or efficient charges is similarly
   difficult.  Therefore, unless benefits are known, fees should
   be used to drive sources to attain ambient standards, rather
   than to replace standards altogether.

   AMBIENT STANDARD?We base this conclusion on the following
   findings of our comparative analyses of these two economic

   — To implement an efficient charge system, the administer-
      ing agency must take the initiative to acquire extensive
      information about sources' 'control costs.  This is a
      difficult and expensive undertaking if costs are to be
      determined accurately.  Under a permit system, the
      agency does not need to take the initiative to acquire
      detailed source-by-source cost data.  These data are
      revealed by sources as they buy and sell permits from one
      another.  The agency may still want to perform an economic
      analysis to predict the economic consequences of its
      regulatory action.  However, this analysis would not
      likely be as detailed as that required under a charge


        — Under marketable permits, the control authority can  be
           more certain that a prespecified environmental goal  will
           be attained than under emission charges.  Under permits,
           the control authority sets the total quantity of allow-
           able pollution within an area and allows firms to re-
           distribute that quantity among themselves.  However,
           the control authority can be certain that the initial
           total quantity of pollution will not be exceeded.  Under
           charges, the control  authority sets the charge rates.
           The total quantity of pollution in an area depends upon
           how firms respond to  these rates.  Since a control
           authority cannot perfectly predict firms' responses,
           it cannot be certain  of the resulting level of pollution
           after it sets the charges.

        — A marketable permits  system self-adjusts to inflation
           and growth.  A charge system requires that the agency
           make periodic adjustments to account for these factors,
           adjustments that depend upon uncertain and perhaps
           expensive data.  Furthermore, frequent changing of
           charge  rates may undercut the credibility of a charge
        — A marketable permits  system is administratively and
           legally similar to permit programs now operated under
           state pollution control programs.  This means that it
           could be administered alongside of existing regulatory
           programs more easily  than could a charge system.  State
           and local authorities are more likely to accept a market-
           able permits system than a charge system because of  its
           similarity to the emissions offset and bubble policies
           now being implemented by the states.

        PROBLEMS.  Where regulations are already in place, market-
        able permits are less disruptive and less likely to result
        in sudden changes in the distribution of control costs
        across regulated companies and in "sunk" or worthless
        industry investments in  pollution control equipment.
        Where no regulations are in place and where emissions are
        representative of ambient impact, (e.g., chlorofluorocar-
        bons), imposing a per-emissions-unit fee on all sources
        may not only be an equitable way to stimulate pollution
        control, but result in an efficient pollution control

     The remainder of this report is divided into two sections.
Section I discusses the theory and benefits of economic approaches
to pollution control.  The potential for applying alternative
market approaches to different pollutants is discussed, along with
the relative advantages of alternative economic incentive schemes
(e.g., emission fees v. marketable permits).  EPA's progress to
date in incorporating market approaches in its policies  is also
briefly addressed.


     Section II discusses the case of a particular pollutant problem,
stationary source nitrogen oxides (NOX), and analyzes the potential
of applying economic approaches to the control of stationary source
NOx in the Chicago area.  Although we feel NOx is not a good candidate
for market approaches (insofar as its role in short-term N02 concen-
trations is concerned),  no case study can substitute for actual
field experimentation.

     Appendix A discusses the methodology of this analysis, while
Appendix B addresses some of the problems inherent in a marketable
permits approach.

                   SECTION I:  MARKET INCENTIVES

     Proponents of market incentive systems for pollution abatement
argue that such systems are more "cost-effective" in attaining
environmental goals than the traditional approach to environmental
regulation.  That is, market approaches will mitigate the effects
of air pollution  (e.g., increased respiratory disease within the pop-
ulation, acid precipitation, and visibility degradation) at a
lower cost than "traditional" government regulation.

     Under the latter approach, federal, state, and local control
agencies abated pollution by issuing regulations that limited the
total quantity of pollution a source could emit.  In the simplist
terms, these regulations might have said:  "all oil-fired industrial
boilers must meet an emission standard of x pounds per hour per
quantity of fuel burned," or "all coal-fired power plants must
meet an emission standard of y pounds per hour."  Since sources
with similar characteristics were treated alike under this approach,
it appealed to the public's sense of equity.  However, this approach
to regulation did not adequately account for two important facts
of pollution control:

     1. The costs of pollution abatement vary among sources.  It
        may cost Source A three times as much as Source B to
        achieve the same level of emission control.

     2. Sources with the same absolute quantity of emissions may
        cause different levels of harmful effects.  One source
        may have a short stack and be located in a densely populated
        area.  Its 100 tons/year of emissions will be dispersed over
        a small area, causing high ambient concentrations and
        exposing a large number of people.  Another source may
        have a tall stack and be located on the outer fringe of an
        urban area.  Its 100 tons/year of emissions may be dispersed
        over a large area, causing only low ambient concentrations
        and exposing only a small population.

      The "traditional" approach to regulation would be cost-effec-
tive only if the costs of pollution abatement and the harmful effects
of pollution did not vary from source to source.  Otherwise it would
be more desirable for the control authority to require the installa-
tion of control equipment where the greatest reduction in the harm-
ful effects 'of pollution per dollar spent on control could be achieved,

     Thus the problem for the regulatory authority is to assign
controls in a way that takes into account a source's cost of control
and the degree of its ambient impact.  While most state and local
control authorities initially adopted the "traditional" approach
to regulation, nothing in the Clean Air Act requires the states to
regulate all sources to the same extent.  Control agencies could
legally perform an air quality and pollution control cost analysis
to design regulations on a source-by-source basis, taking cost-
effectiveness into account.  While state and local authorities are
moving in this direction—especially in not requiring additional
controls if ambient air quality standards are being met in the


vicinity of a particular plant—they face several constraints in
making command-and-control regulations fully cost-effective:

     o  It is often politically necessary for state and local
        authorities to impose the same regulation on plants of the
        same type and size.  Otherwise, control authorities could
        be perceived as being unfair.

     o  Air pollution control authorities typically do not have
        sufficient resources to perform the necessary analyses
        to discriminate between sources on the basis of cost-
        effectiveness.  The development of uniform regulations
        requires much less resources on the part of the local
        control authority than does the development of source-
        by-source cost-effective regulations.

     o  Local pollution control authorities do not have direct
        access to the information necessary to design efficient
        source-by-source regulations.  For example, it would be
        extremely difficult for the regulatory body to obtain
        accurate information on all emission control opportunities
        and control cost data at sources within its jurisdiction.

     The strength of market incentive approaches relative to command-
and-control regulations is that not only are they more likely to
result in cost-effective control strategies, but they locate the
decision making responsibility for pollution control with those
who possess the information needed to make the most efficient
decision and have the incentive to do so.  Thus, the government
specifies the environmental goal but gives some flexibility to
industry in meeting that goal.

                       Pure Emission Charges


     One economic approach proposed by policy analysts to take ad-
vantage of the relationship between a source's costs of control and
its environmental impact is charges, generally emission charges.
The basic notion is that polluters "overuse" the air resource
because they do not have to pay for the privilege of using it.
Therefore, the solution is to apply a price to the use of the air
by charging polluters for their emissions.  For example,"one might
envision a charge imposed on polluters of x dollars per ton of
pollution emitted per year.  By instituting such a system, pollut-
ers have an incentive to economize on their use of the atmosphere
in the same way they make decisions to employ labor and capital

     Once an emissions charge system is in place, a polluter will
adjust the level of emissions abatement so that the cost of control-
ling the last ton of emissions is balanced by the charge that would
be paid if that one ton were emitted.  Across many polluters, this
means that the reduction in total emissions will be achieved at the
least cost.  Thus, sources whose control costs are high will control


less, while those with lower costs will control more.  In addition,
depending upon the particular characteristics of the charge scheme,
the system will provide a continuing incentive for polluters to seek
more efficient means of controlling emissions to reduce their costs.
Charges may be one of the most effective ways of stimulating innova-
tion in pollution control technology.

     The notion of an emissions charge and its effect an a firm's
behavior can be shown diagrammatically.  Figure 1 assumes that a
firm has an increasing marginal pollution control cost curve and that
before the charge system is implemented, the firm is emitting E tons
of pollution per year.  That is, the firm has incurred some control
costs to comply with current requirements.  (In the typical case
where the pollution is an unwanted by-product of no value to the com-
pany, the firm would be expected to operate uncontrolled, or at level
O1 in Figure 1.)  After a charge of f dollars per ton of emissions
is levied, the firm reduces emissions until the cost of the last unit
of control is equal to the size of the charge.  This equality occurs
at point B in Figure 1.  Thus, the firm is provided with an incentive
to reduce emissions by E minus A.  The control costs associated with
the resulting level of emissions are represented by the shaded area,
O'AB, and the charge payments that the firm would be making are re-
presented by the rectangular area ABCO and are equal to f (0 minus A).
This represents the lowest possible combination of total costs (con-
trol costs plus emission charges).

     Ideally, the charge should be set on the basis of harmful effects
(damages) caused by the source's emissions.  Specifically, the charge
should be set at the point where the marginal benefits (reduced harm-
ful effects) from pollution abatement are balanced by the marginal
costs of emissions control (point A in Figure 2).  If a lower charge
were set, the resulting level of emissions control would not be
optimal because an additional dollar spent on control would reduce
damages by more than a dollar.  On the other hand, if a higher charge
were set, the costs of additional control would exceed the benefits
from that control.


     In practice, setting charges on the basis of damages creates
formidable obstacles.  First, there is a lack of information on the
damages associated with emissions.  The most serious effects of pol-
lution often take the form of adverse health effects.  Other damages
are associated with adverse effects on materials, vegetation, visibil-
ity impairment, and so on.  Rendering these damages into quantifiable
measures is very difficult.

     A second obstacle is that the optimal charge is based upon
social damages at levels of pollution not currently experienced.  For
example', suppose a source currently emits 100 tons/year of pollution.
Quantifying the damages associated with those 100 tons/year is a
difficult task, but quantifying what the damages would be if the
source emitted 75, 50, or 25 tons/year is a problem even more fraught
with uncertainty.  Yet, this calculation would have to be made in
order to set the "optimal" fee.  Third, if each source causes different

                                   FIGURE 1

               TOTAL CONTROL COSTS
                   E         A

                                                              CHARGE PAYMENT
                                       |  (TONS PER YEAR)

                                 FIGURE 2

                                    MARGINAL COSTS

                                      OF CONTROL
                MARGINAL DAMAGES

                   OF EMISSIONS


levels of damages, optimal charges would have to be calculated on
a source-by-source basis.  The adminitration of this system would
therefore be both costly and complex, involving at least as much
governmental intrusion as the current system.


     As a more practical alternative to the theoretically ideal
charge approach just described, another technique for using economic
incentives is the standards-and-charges approach.  With this method,
the environmental authority determines a set of standards for en-
vironmental quality and attempts to achieve these standards through
imposing a system of emission charges.


    This approach circumvents the need to specifically measure the
social damages of emissions on a source-by-source basis.  Instead,
the environmental authority arrives at a set of environmental
standards that are established based on available evidence on
national or regional damages caused by a pollutant.  The data re-
quirements on the part of the control agency are therefore reduced.
The agency need only estimate the level at which the charge should
be set.  This level should be sufficiently high so that firms will
respond by reducing emissions to the levels where standards are

     While this estimation is difficult, it is generally not as
difficult as estimating and quantifying damages.  If done correctly,
charges make it possible to achieve the designated standards at the
least cost.  That is, once the environmental authority sets the
charge level(s) so that the corresponding reductions in emissions
are consistent with the prescribed standard for environmental
quality, individual responses to the charge will result in the least-
cost pattern of abatement across polluters.


     In overcoming the practical difficulty in measuring damages,
the standards-and-charges approach by necessity ignores certain
information concerning an individual source's impact.  Some of the
harmful effects of pollution the environmental authority sought to
mitigate in the setting of a national or regional air quality stand-
ard may not be accounted for in a specific situation because of such
factors as population exposure and the type of vegetation near a
source.  Thus, the charge that source pays may be more or less than
the damage to human health and welfare that its emissions create.

     A second problem relates to the fact that our current environ-
mental standards are ambient air standards.  Depending on the poll-
utant, one ton of emissions from one source may not have the same
impact on ambient air as one ton of emissions from another source.
For example, for several pollutants currently regulated by EPA —
e.g., total suspended particulates (TSP), sulfur dioxide (S02),
and nitrogen dioxide (NC>2) — the ambient air impact of an emission


point depends on meteorological conditions, stack height and dia-
meter, velocity and temperature of the emissions exiting a stack,
and local terrain.  Thus, unless- a charge system can be design-
ed to take into account each source's contribution to the ambient
air problem, emission fees will not be as cost-effective as other-
wise would be the case.  Just as firms pay different prices for
land according to its desirability, so should they have to pay dif-
ferent prices for pollutant emissions — according to their undesir-

     The implication, then, is that even under a standards-and-
charges approach, the optimal fee scale would be one in which sources
were charged different prices for their emissions according to their
ambient impact.  Thus, the air pollution control agency would have
to set a different fee for each source of air pollution.  While our
case study of nitrogen oxides (NOX) in Chicago showed that under
source-by-source charge rates, total costs of control are lower than
under any other policy alternative, a number of political and admini-
strative problems would make their implementation extremely difficult:
     o  There is the problem of treating sources that appear alike^
        differently.  Even sources with the same physical configura-
        tion would be charged different rates based on their loca-
        tion.  This may pose political problems, as allegations of
        inequitable treatment are likely to occur.

     o  The information requirements on the control agency are formid-
        able.  It must know emission rates, the relationship between
        control costs and emissions, and the relationship between
        control rates and ambient air quality.  This information is
        extremely difficult to obtain and is usually very uncertain.
        It is of course, possible to design charge systems for which
        charge rates are uniform across sources within the same
        source categories.  However, the efficiency gains of charge
        systems decrease the higher the level of aggregation.  For
        example, in our case study of Chicago, the cost of pollution
        control equipment under source category charge rates is over
        seven times higher than under source-by-source charge rates
        when standards are attained.

     o  The control authority would have to change the charge over
        time as inflation reduces the real value of the charge, as
        real prices of control equipment change, as the area moves
        into attainment, and as new information on control costs
        and ambient impacts is obtained.  Moreover, the "tinkering"
        or adjusting of charge rates necessary to achieve the
        desired results would most likely undercut the credibility
        of such a system.
1. As mentioned on page 5,it is appropriate to charge sources based
on their marginal costs of control and their contributions to ambient
air quality.  This may result in differing charges but is in accord
with the economist's definition of equity.


     In short, the difficulties in setting a reasonably efficient
standards-and-charges system are comparable to those of a pure
commandand-control approach that attempts to incorporate cost-
effectiveness into the design of individual source regulations: in
theory, all the approaches can achieve the same outcome, but in
practice one will probably have to compromise ease, certainty of
reaching environmental goals (at least in the near term), and/or
economic efficiency.  Marketable permits, combined with standards
for certain pollutants may be preferable to these other approaches.

                  Standards and Marketable Permits

     A second commonly proposed policy for dealing with air pollu-
tion is marketable pollution permits.  Under this system, the en-
vironmental control agency issues a fixed number of permits, each
entitling the holder to pollute no more than some specified quantity
per unit of time.  The two major features of the system are (1) it
relies on a market-related allocation method, and (2) after the per-
mits are initially distributed, permit holders may buy and sell them
through a regulated market.

     The marketable permits system presents each polluter with a
continual option:  reduce emissions and sell any excess permits, or
buy permits allowing pollution.  In this way, the price of the permit
in the market creates an incentive for polluters to control emissions
efficiently, just as emission charges do.  A cost-minimizing polluter
will control emissions and buy permits up to the point at which the
marginal cost of further reducing emissions equals the price of the
permit.  Thus, marketable permits provide polluters with incentives
to continually develop more effective systems of pollution control.
As better systems are developed, polluters can reduce their emissions
and sell their excess permits on the market.  Firms that develop
more effective pollution control systems are therefore rewarded by
reduced compliance costs.

     Under a marketable permits system, the control agency faces a
decision analogous to that for emission fees:  whether to issue suf-
ficient permits to equate marginal costs and marginal damages  (if
that level can be determined) or to use ambient air quality standards
as the basis for limiting the quantity of permits on the market.
Under the former approach, the information requirements on the part
of the regulatory authority are virtually identical to those when a
charge is set on emissions on the basis of damages'.  That is,  in
order to issue permits, the control authority would have to assemble
information on the damages caused by every source as well as on the
costs of pollution abatement.  As with emission charges, these infor-
mation requirements may make this alternative untractable for  some
pollutants.  Under the latter approach, damages are implicitly
estimated in the setting of uniform ambient air quality standards.
Because the national ambient standard is the same everywhere even
though individual sources cause varying amounts of damage, it  is
extremely unlikely under a standards-andpermits approach that  when
standards are attained, the marginal costs of pollution control will
be equal to the marginal benefits of pollution abatement for all
sources.  For this reason, there is a loss in economic efficiency


inherent in this approach.  That is, when the national ambient
standard is attained, it is likely that for some sources an addi-
tional dollar spent on pollution control would result in more than
a one-dollar reduction in damages, while for ^ther sources, one
less dollar spent on emissions control would result in less than
one dollar in damages.  However, this loss in economic efficiency
is offset by reduced information requirements on the part of the
control agency and by the increased certainty that the environment-
al goals will be reached within a reasonable period of time.

     The increased environmental certainty of a standards-and-
permits approach provides a major advantage over emission charges.
Under a permit system, the control authority can be more certain
that its environmental goals will be met.  With charges, the regu-
latory authority sets the charge rate.  Polluters then respond by
controlling emissions until the cost of further control exceeds
the charge rate.  Unless the regulatory authority obtains perfect
information on the abatement costs of the sources within its juris-
diction, it cannot perfectly predict how firms will respond to a
given charge rate.  Therefore, it cannot know with certainty what
the final levels of pollution within its jurisdiction will be.
While it can adjust the charge rates until the desired level of
pollution is obtained, doing so entails certain costs:  once a
firm responds to an initial charge rate by installing control
equipment, engineering constraints may make further control prohib-
itively expensive or impossible.  A firm may also complain of un-
fair treatment if changing charge rates results in a large comparative
gain or loss compared with its competitors.  Just as EPA is now often
criticized for constantly changing control requirements, industry will
not appreciate the added uncertainty if charge rates are expected to
change.  Consequently, the adjusting of charge rates may undermine
the credibility of the charge system.

     These problems are not present under permit-based approaches.
Since the control authority knows the quantity of permits it has
issued, it can be more certain of the resulting levels of pollution
within its jurisdiciton than under charges.  Of course, while there
is no need to predict how firms will respond to the permit market
in order to estimate resulting levels of pollution, the regulatory
authority will still want to predict the pollution control decisions
of firms in order to estimate the costs of its regulatory action.


     An important question is how the units of pollution in a
permit should be expressed.  Should it give its holder entitlement
to emit a pollutant, or should it entitle its holder to have an
ambient impact?  (For some pollutants, this distinction is irrelevant
because emissions are taken to be the measure of ambient impact.
This is the case for nonmethane hydrocarbons, chlorofluorocarbons,
and precursors to acid rain.)


     Ambient Permits

     Since the ultimate goal of an air pollution control strategy
 is to ensure that the air people breathe is both safe and clean,
 and since it is the ambient air that people breathe, intuitively
 the most efficient and direct approach for designing a marketable
 permits system would be one in which permits were written as entitle-
 ments* to pollute the ambient air by a given amount.  In addition,
 the pollution control authority would probably want to discriminate
 as to the location where this pollution would occur.  Ten sources
 located in one industrial park each with an entitlement to contri-
 bute 10 micrograms per cubic meter (ug/m3) of pollution to the
 ambient air, may have a different significance for human health
 and welfare than ten sources spread out over an entire region whose
 impacts do not "overlap."  It is therefore appropriate to express
 permits as entitlements to pollute the air by a certain amount at
 a certain location (or locations) for a certain period of time.

     The value of any given size of ambient permits, therefore,
 depends on where they will be used.  Just as the market for real
 estate is vastly different in different parts of an urban area, the
 market for permits could be expected to be different in different
 parts of a city or region.  One can anticipate the market to be
 "thin" in areas of the airshed where source impacts do not overlap
 (i.e., few sources are competing for the use of the air resource)
 and competitive where they do.  In the former areas, there may be
 no potential for efficiency gains through permit trading (since there
 no one to trade with), while in the latter areas there will be.
 The exact numbers of "competitive" and "noncompetitive" areas will
 depend on;

     o  The nature of the pollutant.  Some pollutants (e.g.,
        chlorofluorocarbons) have the same impact on health
        and welfare for all of the population of a region,
        regardless of where they are being emitted.  For
        these, one can anticipate a regionwide market for
        ambient air permits (it is unnecessary to even label
        "location" on the permits).  For pollutants with
        highly localized ambient impacts (e.g., short-term
        NC>2 and S02)/ localized markets for ambient permits
        would be more appropriate.

     o  The source mix and location.  For pollutants character-
        ized by localized ambient impacts, the location and
        stack characteristics of sources will affect the
        "thinness" of a market.  For example, in terms of
        annual average ambient concentrations, a source with
        a tall stack typically has a small impact over a
*"Entitlement" is used throughout this reoprt in the economist's
sense of permission to emit certain levels rather than in the
specific legal sense of rights conferred on polluters.  Similarly
in this report, the term "permit" generally refers to the unit of
emissions a source is entitled to emit, not to the legal document
authorizing him to do so.


        large area whereas a source with a short stack will
        have a large  impact in its immediate vicinity.  If
        sources are clustered together there may be more
        potential for ambient air trades than if they are
        spread apart.  Put somewhat differently, the more
        plumes from different sources overlap, the greater
        the potential for ambient trading, though the more
        difficulty in identifying each individual polluter's
        contribution  to ambient concentrations.

     o  The number of sources.  All things equal, the greater
        the number of sources, the more competitive a market
        for ambient air will be.

Emission  Permits

     Alternatively, the air pollution control agency can denominate
permits as entitlements to emit a pollutant.  One ton of NOX can be
traded against another ton, regardless of the location of the emis-
sion source.  The "thinness" of a market will be solely a function
of the number of sources present and the quantity of emissions
within an area.  While this alternative increases the efficiency of
the market, it has several disadvantages.

     First and foremost, certain trades in emissions can result in
more harmful ambient  air.  For example, a ton of emissions from a
short stack is more likely to be inhaled by humans than a ton of
emissions from a tall stack.

     A second and related problem is that by allowing unrestricted
trading in emissions, it is likely that ambient standards for some
pollutants will not be met at all locations.  Even if the control
authority severely limits the total allowable quantity of emission
permits in an area, there is no guarantee that the allowable emis-
sions will not become concentrated in one limited geographic location
within the region and thus violate the national ambient standards.

     The  pollution control authority can limit the chances of this
happening by subdividing the region into several zones, establish-
ing an emission cap for each zone, and then allowing unlimited emis-
sion trading — provided that no zone exceeds its emission limita-
tion.*  However, while this zoning will decrease the probability of
an ambient violation, it will not eliminate it, because the ambient
impact of a source is a function of its stack characteristics and
meteorology, in addition to its emissions and location.  Further-
more, some pollutants, such as short-term N02 concentrations, have
such localized impacts that emission zones would have to be small in
size and  large in number> making the administration of the system
very complex.   And finally, with more zoning comes greater political
pressure  for granting variances — further increasing the chances
of violating the standards.
*This is a variant of emission density zoning.


     Thus, if permits are expressed in emissions without respect to
ambient impact, the control authority takes only limited advantage
of the relationship between the source's marginal control cost and
its ambient impact.  The regulatory system is not as "customized" to
fit local conditions (e.g., terrain, source mix, and meteorology) as
it would be if permits were expressed in terms of ambient impacts.
Our analysis will show that this lack of "customization" would prove
extremely expensive if we were to rely on such a system to attain a
stringent short-term N02 standard in Chicago.

     However, for pollutants where emissions are the best indicator
we have of a source's ambient impact, a market in emissions is a
market in ambient air quality.  The problems of stack height and
emission density described in this section no longer apply.  Control
of acid rain may be one example of where marketable permits for emis-
sions may be a useful approach.

     Given our current knowledge of the causes of acid rain, the
most likely control strategy to achieve this goal would be to reduce
regionwide emissions of SC>2 and NOX (the suspected precursors to acid
rain).  Thus, the control agency may have two goals:  to attain the
local ambient standards and to reduce acid rain regionwide.  This
implies that the control agency desires to ration two commodities
— ambient air and regional emissions — and that two separate, but
interrelated, markets could be created for each.  To operate such a
system, the control agency would have to allocate the number of
ambient air and emissions permits consistent with attaining both of
its goals.  All trades would be ton for ton, but the control agency
would also model the effects of the proposed trade on ambient air
quality and approve the trade only if the resulting air quality is
as good as or better than that existing before the transaction.


     We see, then, that it is possible in theory to design market-
able permits systems to achieve various goals related to air quality.
Probably the most significant problem facing the control authority
is determining how to initially allocate the permits.  Should it give
permits away for free, or should it require sources to bid for them?

Allocation Via Bidding Mechanisms

   Most economists have argued for the bidding approach, maintaining
that a bidding approach will ensure that those who have the best use
for permits will get them — since these people will be willing to
bid the highest.  Some practical problems of using a bidding system
(especially to allocate permits of a pollutant now covered by a
national ambient air quality standard) are the following:

   o  It may be politically difficult to limit certain firms'
      pollution levels or use of the air resource much below cur-
      rent levels.  This is the principal reason why the emission
      limitations contained in many State Implementation Plans
      now fall short of attaining standards.  This problem would
      still persist under a bidding scheme.


   o  An auction is likely to generate large amounts of government
      revenues, the size of which are likely to cause political
      problems"!Under a bidding system, sources will have to pay
      the government for their permits.  Even if the control
      agency gives away some quantity of permits for free (e.g.,*
      the government might give all existing sources permits for
      75% of their current emissions) and auctions the rest, these
      transfer payments could be extremely large.*

   o  It is difficult to ensure that an auction will not result in
      localized violation of ambient standards.  To ensure that
      standards are attained in all areas within a region, the con-
      trol agency would have to know something about the air quality
      in the vicinity of a bidder in order to decide whether to
      grant the bidder an additional permit.  In other words, for
      some pollutants, obtaining a permit would have to depend not
      only on the bid price, but on the special circumstances of
      the bidder (the air quality in his location and his stack

   o  How should we treat sources that do not obtain permits in the
      auction?  Should they be shut down? Should they have to pay
      some kind of a fee?  If so, how much?

   o  Gaining the support of industrialists and environmentalists
      may be difficult.Auctions in the air quality field are an
      untried regulatory approach.  Environmentalists may believe
      it unethical to "auction off" air quality.  Businessmen may
      feel it is not in their interest to participate.  Gaining
      public support for the concept may be a difficult task.

Nonmarket Allocation Methods

     Alternatively, the control authority could give away permits
for free.  Formulae could be used to allocate the quantity based
on historical emission levels, a percentage reduction requirements,
etc.  One possibility would be to grant permits to sources allowing
emissions up to levels consistent with current SIP regulations.  After
this allocation, sources could trade emission limitations subject
to certain rules to ensure that air quality standards are attained.

     If the quantity of permits initially allocated is so large so
that ambient air standards will not be attained (e.g., because of
* A source will bid for a permit from the government up to its marginal
cost of control or up to the asking price of permits on the market  (in
other words, no source will bid $25 for a permit from the government
to emit a pound of a pollutant if it only costs $20 to control that
pound or if it can buy a permit from another source for $22).  However,
the marginal cost of control at levels of emissions consistent with
achieving ambient standards is typically significant, implying that
these transfer payments could be extremely high.


political opposition of vested interests) the control agency can
reduce the number of permits initially allocated in several ways,

     o   retiring the permits of firms that shut down or cur-
         tail operation.  This reduces the supply of permits
         on the market but has the undesirable effect of giv-
         ing economically inefficient firms an incentive to
         keep operating in order to preserve their air resource
         "assets" for later sale on the market.

     o   expropriating permits by forcing firms to install addi-
         tional control equipment.  Although air pollution coiji-
         trol agencies can do this by issuing new regulations,
         it causes great uncertainty for firms holding permits.
         A permit may be a valuable asset one day and not the
         next.  In turn, uncertainty will discourage trading of

     o   limiting the time a permit entitles its holder to pol-
         lute the air.  For example, a permit may allow its
         holder to emit 10 pounds of NOX per hour for a period
         of five years, after which time the permits will be
         reallocated.  Although this approach eliminates some
         uncertainty since emitters know that they hold their
         assets for specified periods of time, they are still
         unsure of their ability to obtain permits in the future.

     o   setting limits on the consumption of permits by new
         sources through the establishment of performance
         standards.  At some point in time existing capital
         equipment will wear out and be replaced by new capital
         equipment.  The control agency can take advantage of
         this turnover in capital stock to reduce the allocation
         of emission permits in an area by (1) retiring the per-
         mits of sources when they shut down or are replaced and
         (2) forcing the sources that replace them to employ high
         levels of pollution control by issuing performance
         standards for new sources.  Although over time this
         approach reduces the total regional quantity of permits,
         its main disadvantage is that it imposes additional
         costs on new investment.

     o   "taxing" trades in permits.  Following this principle
         each trade will have to improve rather than maintain air
         quality.  For example, a new source emitting 10 tons per
         year may have to obtain permits for 15 tons — 10 tons
         for its own consumption and 5 tons to be "given" to the
         control authority.  This scheme also has its costs.
         Because trades are implicitly taxed (and the revenues
         used to retire permits), the total number of trades will
         be reduced, as will the (previously discussed) economic
         benefits from trading.  However, if implemented as
         described here, both new and existing sources will bear
         these costs.


                        EPA's Current System

     Much of EPA's current approach to air quality management contains
characteristics of standards-and-permits and air quality permits
(without standards) systems, with control of existing sources resembl-
ing the former, and control of new sources, the latter.  As noted
earlier, under a marketable ambient permits system, permits would be
expressed as entitlements to pollute the air by a certain amount at a
certain location (or locations) for a certain period of time.  This is
essentially how permits are now written under EPA's New Source Review
Program.  The applicant applies for a permit to emit at a particular
site and conducts dispersion modeling to calculate its ambient impact.
The permit itself contains an emission limitation.  Since the permit
is site- and source-specific, limits emissions, and is issued in
conjunction with air quality modeling, it can be thought of as an
entitlement to pollute the ambient air.

     However, EPA has only issued New Source Review permits to a limit-
ed number of sources — typically large emitters constructed after
1975.  Other sources face limitations on their consumption of the air
resource either through state permit programs (where these exist) or
through regulations issued under State Implementation Plans (SIPs).
To the extent that states rely on air quality modeling in designing
their SIPs, these limitations implicitly control the sources's ability
to pollute the ambient air.  From the economist's point of view they
are quite similar to ambient air permits.

     Through its bubble, offset, and banking policies, EPA currently
allows some trading of these entitlements to pollute the ambient air
in a way that resembles a standards-and-permits approach.

     We stated earlier that one function of a pollution control
authority in devising a marketable permits system is to establish an
initial allocation of permits consistent with achieving its air
quality goals.  The regulatory mechanism EPA currently uses to ful-
fill this task is the State Implementation Plan.

     Under the Clean Air Act, states must submit to EPA a set of regu-
lations requiring a quantity of emission reductions from existing
sources sufficient to attain the ambient air standards.  Thus, while
the initial allocation is accomplished through a nonmarket mechanism,
trading of permits determines the final allocation.

     Under this approach, states implicitly award a certain quantity
of permits to existing sources free -of charge.  This quantity, however,
may be less than a source's current level of emissions.  A source is
then faced with three options:

     o  "buy" (or, at least, obtain) extra permits for its
        emissions via the bubble policy.  The policy's rules —
        specifically that of ambient air quality equivalence —
        have the effect of making the transaction one involving
        permits to pollute the ambient air by a given amount.


     o  control its emissions so that they are in line with its
        SIP limitation.

     o  face possible enforcement action.


     The bubble policy prescribes the rules by which existing sources
may adjust or trade their individual SIP emission limitations designed
to attain the ambient air quality standards.  The basic idea of the
bubble policy is to encourage industry to develop the most cost-
effective mix of controls sufficient to attain air quality standards.
While we will not describe bubble trading rules further here, they
contain certain conditions traders must meet to ensure that they do
not exceed the current allocation of air quality permits.


     Under EPA's offset policy, new sources that wish to locate in a
nonattainment area must offset their emissions (and ambient impact)
by reducing the emissions of existing sources.  In other words,
before they may operate, new sources must buy permits or "offsets"
for their emissions from existing sources.  These permits must be
sufficient in number to ensure, on balance, a net benefit in the
ambient air quality.  Hence, they are implicitly ambient air permits.

     This policy was applicable during the period January 16-July 1,
1979 (the later date being the deadline for states to submit their
SIPs, which would contain their own rules for new sources) and still
applies to newly discovered nonattainment areas (until states develop
a SIP).  However, most states have adopted some version of the offset
policy in their own SIPs.  In many cases, these plans lay out not only
the rules by which new sources must acquire offsets from existing
sources, but also certain measures that reduce the total number of per-
mits.  One such measure is the requirement that new sources more than
offset their emissions.  Several states have adopted this approach,
and offset ratios range from 1:1 to 2:1 (existing permits secured for
new permits).


     EPA is developing regulations to permit banking of emission reduc-
tions.   When a source is certified by the state authority as having
reduced its emissions beyond the level it actually needs to comply
with its requirements (so that it holds more permits than it needs),
we say that it has "banked" emission reductions.  These "emission
reduction credits" are the permits that are available for sale or
trade.   Again, certain aspects of EPA's regulation will deal with the
administrative process by which a control agency "certifies" that a
source has permits available for sale, while others deal with how to
reduce the allocation of permits.  Under the latter, rules will be
developed for retiring permits when a source shuts down.  Under the
former, rules will ensure that after a transaction, sources will not
"consume" more air than the amount for which they have permits.



     Section 173 of the Clean Air Act specifies that all new major
sources and major modifications to existing sources locating in a
nonattainment area must employ control equipment that will result in
the "lowest achievable emission rate"(LAER) contained in any State
Implementation  Plan or achieved in practice (whichever is more strin-
gent).  Section 111 calls on the Administrator of EPA to develop
standards of performance for all new and modified sources falling
within certain  source categories — irrespective of source location.
Both of these sections have the effect of limiting the consumption of
permits by new  and modified sources.  As discussed earlier, this will
reduce the total regional quantity of permits over time as plants turn


     If sources do not comply with the limitation facing them under
the SIP,  in addition to litigation, they are required under Section
120 of the Act  to pay a penalty equal to the economic value of the
delay  in  compliance.  Sources complying with the emission limitation
pay no penalty  at all.  This scheme has some similarity to a standards-
and-charges approach:

     o  Under NCPs, penalties are zero if a source complies with
        the emission limitation mandated by the SIP.  The SIP it-
        self is designed to attain ambient standards.  Under a
        standards-and-charges approach, the optimal charge on emis-
        sions is also zero if a source's emissions are consistent
        with the standard.

     o  Under NCPs, penalties are significant if a source does not
        comply, j Similarly, under a pure standards-and-charges
        approach, if a source exceeds the prescribed standard,
        charges on its emissions should be applied sufficient to
        induce  the source owner to comply with the limitation.
        Whil£ under NCPs, penalties are not explicitly linked to
        the source's emissions, they provide an incentive to com-
        ply somewhat similar to the incentive from emission charges
        under the standards-and-charges approach.  That is, there
        is no incentive to exceed emission limitations under either

                   Summary and General Conclusions

     Based upon the discussion in Section I, we conclude the following:

     o  The main advantages of emission charges and marketable permits
        approaches over "traditional" regulation  (in which all sources
        in the  same category face the same emission limitation) are
        that the former locate the decision for installing controls
        with those who have the best information on control costs and
        have the greatest incentives for making cost-effective decisions,
        Futhermore, to the extent that "traditional" regulation does not


discriminate among individual source impacts in determining the
need for control, market approaches are more likely to result
in cost-effective outcomes.

With its offset and bubble policies, EPA has moved in the
direction of incorporating market incentives into its air

A system of marketable permits is generally preferable over
a system of emission charges.  The reasons for this are:

— marketable permits are more like our existing system of
   regulations.  The transition to and administration of
   such a system is likely to be smoother than for emission
   charges.  Moreover, sources have a vested interest in
   the current permit system, which could be jeopardized if
   EPA adopts a new charge approach.

— attainment of standards is more certain under marketable
   permits.  Under permits, the control authority "fixes"
   the total quantity of allowable emissions in an area.
   No matter how much firms trade permits this total quantity
   cannot be exceeded.  Under emission charges, the control
   authority sets the charge rates.  The total quantity of
   emissions in an area depends on how firms respond to
   these rates.  Since the control authority cannot be sure
   of firms' responses, it cannot be certain of the resulting
   level of total areawide emissions.  Thus, marketable per-
   mits offer more certainty regarding future air quality than
   do emission charges.

An exception to this rule may be the use of emission charges
to control currently unregulated pollutants where sources do
not have a vested interest in a permit system of regulation.

A market in permits is likely to function best for pollutants
for which a source's emissions are representative of its ambient
impact.  For these pollutants trading rules need not take source
location or stack parameters into account.

Unless damages from pollution are known with relative certainty,
market incentive systems should be used in conjunction with,
rather than to replace, ambient standards.  Otherwise control
authorities face formidable obstacles in setting charge rates
or in limiting permit quantities for individual sources.

Economic approaches to regulation provide continual incentives
to sources to use innovative control methods.  Under charges,
reducing emissions lowers charge payments; under marketable
permits, reducing emissions frees additional permits for sale
on the market.  In both cases, the development of innovative
control technology reduces compliance costs.

                   SECTION II:  A NOx CASE STUDY

     While in the previous section, the theory of market approaches
was discussed, this section addresses the application of market ap-
proaches to the control of a particular pollutant — nitrogen oxides
(NOx).  This section begins with a characterization of the NOx problem
and concludes with a case study of market approaches to attain a short-
term nitrogen dioxide (N02) standard.

                          The NOx Problem

     Each year more than 20 million metric tons of nitrogen oxides
(NOX) are released into the atomosphere as a result of fuel-burning
activities in the United States.  About half of the NOX emissions
come from automobiles, and the other half from stationary sources,
such as furnaces and boilers:
                                                     industrial 12%



     By 1985 we estimate that stationary sources will account for
70% of man-made NOx emissions.  There are two major reasons for
this trend:  (1) the growing tendency towards using coal for elec-
trical generating stations and industrial boilers (because there is
more nitrogen in coal than in most other fuels, burning coal produces
more NOx than burning oil or gas); and (2) reduction in automobile
NOx emissions (older, more polluting cars will be replaced by newer,
cleaner models).

     Nitric oxide is formed by two chemical processes'that occur during
combustion.  In one process, the heat of combustion causes the oxida-
tion of nitrogen in the air; in the other, nitrogen is oxidized in
the fuel itself.  The former process results in thermal NOx, and the
latter, in fuel NOx.  The formation of thermal NOx greatly depends
upon the amount of heat generated during 'combustion and can be con-
trolled by reducing combustion temperatures.  On the other hand,
the formation of fuel NOx primarily depends upon the amount of
oxygen available during combustion.

     The ultimate goal of processes for controlling NOx emissions is
either to convert the harmful oxidized forms of nitrogen (NO, N02)
to molecular nitrogen (a colorless and odorless gas that constitutes
78 percent of the atmosphere by volume) or to prevent the oxidation
of nitrogen in the first place.  Oxidation of nitrogen may be prevent-
ed by controlling the amount of oxygen (for fuel NOx) or heat (for
thermal NOx) present during combustion.  For example, one method
of reducing NOx formation involves "staging" the combustion process.
In the initial stage of combustion, the air supplied to the burners
of a boiler is less than the amount needed to completely burn the
fuel.  During this stage, fuel-bound nitrogen is released but cannot
be oxidized, so it forms stable molecules of harmless molecular
nitrogen.  In a second stage of combustion, more air is added to the
fuel-air mixture, and the combustion is completed.

     Converting NOx to molecular nitrogen is accomplished by treating
the exhaust gases that are released from burners.  One such process
currently in the experimental stage is selective catalytic reduction,
in which process ammonia is added to the flue gas before the gas
passes over a catalyst.   The catalyst enables the ammonia to react
with NOx, converting it to molecular nitrogen and water.  This
system, although extremely expensive, promises to remove as much as
90 percent of the NOx from flue gases.

                           Effects of NOx

     Exposure to NOx increases the risks of acute respiratory disease
and susceptibility to chronic respiratory infection.  Nitrogen dioxide
(N02)  damages the heart, lungs, liver, and kidneys and, at high concen-
trations,  can be fatal.   At lower levels of 25 to 100 parts per
million (ppm),  N02 can cause acute bronchitis and pneumonia, while
occasional exposure to low levels of N02 can irritate the eyes and skin.


     Probably the most significant effect of NOx is the formation of
photochemical oxidants, commonly known as smog.  NOx reacts with
other pollutants  (mainly hydrocarbons) in the presence of sunlight to
form ozone, which is the main constituent of smog.  Breathing smog
irritates the lungs and can seriously aggravate asthma and other
respiratory diseases.

     NOX emissions contribute to acid precipitation.  They can be con-
verted  to nitric acid, which may fall to the ground in the form of
rain and snow.  Acid rainfall in New York's Adirondack Mountains has
reduced several species of fish in at least 90 lakes.  Furthermore,
the proportion of this acidity that is due to nitric acid has been
increasing in recent years -- from 15 percent in 1964-65 to 30 percent
in 1973-74.

                    Ambient Air Quality Standards

     The current primary standard for nitrogen oxides (as expressed in
terms of NO^) is an annual average of 100 micrograms per cubic meter
ug/m3.  When EPA  issued this standard in 1971, we believed this level
would protect human health with a "margin of safety."  The actual
health  effects impact level for an annual period was thought to be 150

     EPA is currently evaluating more recent epidemiological work and
is considering setting a short-term standard for NC^.  Since some
studies have shown acute health effects from exposure to NOx emissions,
EPA is  considering setting an hourly standard.  The levels under con-
sideration range  from 250 to 1000 ug/m3.  if the standard is set
close to 1000 ug/m3, most areas would be in compliance, even without
placing additional controls on point sources, and there would be little
need to be concerned over implementing market incentive approaches.
If the  standard is set close to 250 ug/m3, there would be a nonattain-
ment problem of the same magnitude as those for oxidants and total
suspended particulates.  The incremental requirement for this more
stringent standard would be the retrofitting of controls on a large
number  of stationary sources.  In this case, economic incentive.
approaches merit careful consideration because of their potential to
reduce  control costs.

     These economic incentive approaches face a major obstacle in
ensuring attainment of a short-term N02 standard:  elevated concen-
trations of NO2 during short averaging periods are characterized by
extremely localized impacts.  That is, a violation of a short-term NO2
standard may occur at one location, while concentrations only short
distances away may be well below the standard.  The implication is
that markets in short-term ambient N©2 concentrations would be numer-
ous in  number and small in size.

     If EPA promulgates a short-term standard, existing State Imple-
mentation Plans (SIPs) have to be revised to accommodate it.  These
revisions if necessary would involve state regulations prescribing
emissions limitations for specific source categories and timetables
for compliance.  Under the Clean Air Act, states are allowed nine


raonths after a standard is issued to develop their plans, which must
then demonstrate attainment within three years.  Until the area attains
the standards, new sources locating in the area would be required to
control emissions to levels reflecting the lowest achievable emissions
rate (LAER) achieved in practice for that source category, or the most
stringent emission limitation contained in any SIP (whichever is more

              An Empirical Analysis of the Advantabes of
                        Trading NOx in Chicago

     Whether EPA continues to rely on marketable permits in ambient
air through mechanisms like the bubble and offset policies, or whether
the agency adopts some other scheme, a control strategy that discrimi-
nates between sources on the basis of control costs and ambient impact
is more cost-effective than one that does not.  But how much more
cost-effective is it?

     To answer this question, we analyzed different control strategies
for attaining a hypothetical short-term N02 standard of 250 ug/m3
(hourly average) in the Chicago region.  We chose the Chicago area
because it is one of the few regions now in violation of the annual
average N02 standard and because it has one of the best inventories
of NOx emissions now available.  The inventory is based on a 1975
survey of stationary sources.  The survey was projected to 1984, which
was considered to be the earliest date for implementing a short-term
NC>2 standard.

     For our analysis we used an EPA ambient air quality/control cost
model for the Chicago area.*  Figure 3 shows five inputs and three out-
puts of the model in general terms.  The inputs consist of:

     o  estimates of emissions from point and nonpoint sources

     o  cost and level of control data for point sources

     o  meteorological data for the region

     o  the short-term ambient standard to be met

     o  the control strategy and/or policy to be followed

     The model processes these inputs in two modes:  (1) by solving a
mathematical programming problem, it can find the least-cost strategy
for controlling emissions; and (2) it can simulate a wide variety of
alternatives to the least-cost strategy, such as those considered in
this report.
   See E. Pechan and S. Strathmeyer, "A Computer-Assisted System for
   Developing Regional Cost-Effective Air Pollution Control Strategies
   (AIRMOD)," Energy and Environmental Analysis, July 1980 (Draft).


     For each strategy analyzed, the model produces three estimates
that serve as a basis for comparison:  (1) estimates of the annual
cost of controlling emissions;  (2) estimates of the ambient concentra-
tions that will be experienced at different locations under each
strategy; and (3) estimates of the total emissions that will remain
after the strategy's implementation.

     Our base case for making cost and air quality comparisons was
"no control."  If stationary sources of NOx were not controlled in
Chicago by 1984, approximately 36 locations in the region would have
hourly concentrations in excess of 250 ug/m3.

     The results of this analysis are presented in Figures 4-11.
Figure 4 shows the costs, emission reductions, and number of sources
controlled under each policy alternative.  Figure 5 summarizes the
ambient concentrations we would expect to find at the locations we
specified in the analysis under the alternative policies.  In order
for the reader to better visualize these ambient impacts, Figures 6
through 11 show NC>2 "isopleths" at one five-kilometer square location
in the Chicago region under the various policy options.  These
"isopleths" show the levels of ambient NC>2 concentrations in map

     The policy scenarios and the results of the analysis are sum-
marized below:

     o  SIP Control Strategy without Trading — We applied a
        uniform level of emission control across similar cate-
        gories of sources that are major polluters of the ambient
        air.  These controls were sufficient to attain the ambient
        standards.  This scenario simulated the "traditional"
        regulatory approach to pollution control.  Thus, all sources
        in three of the nine source categories present in Chicago
        had to install the highest level of emission control.  The
        technology required — flue gas treatment by dry selective
        catalytic reduction — is projected to be available by 1985.
        These three source categories are industrial coal-fired boil-
        ers, industrial oil- and gas-fired boilers, and industrial
        process units.  Uniform emission levels for these three
        source categories represent an attempt to circumvent the
        need for source-by-source emission limitations.  As a
        result of these emission limits all locations are brought
        into compliance with the short-term NC>2 standard.  We esti-
        mate the annual control costs resulting from this strategy
        to be $130 million for the Chicago area.

     o  Emission Fees — We set a charge on emissions to control
        pollution and examined three distinct systems of emission

        1.   a uniform charge system, in which a single charge rate
            just sufficient to attain the ambient standards is
            levied on all sources.  Total control costs for attaining
            standards under this alternative are $305 million, while
            charge payments would amount to $414 million.  The charge

rate itself would have to be at least $15,800/yr/lb/hr
(per year per pound per hour).   The "double burden"
i.e., the charge payment required in addition to that
spent on the purchase of equipment to control emissions—
is enormous, amounting to 136%  of the costs of
the pollution control equipment itself.   Note, however,
that while the total costs of this system far exceed
those of other alternatives, areawide point source
emissions would be reduced by 84% — 4 times greater
than the SIP alternative.  Yet, as Figure 5 shows,
these additional emission reductions result in only
limited improvements in ambient air quality.

a source-category charge system, in which different
charges are levied on different source categories (e.g.,
all oil-fired industrial boilers face the same charge
rate, but this rate differs from that levied on coal-
fired power plants).  This alternative would attain
standards at a total cost (pollution control equipment
plus charge payments) to polluters of $152 million, with
$66 million resulting from control costs, and $89 million
from charge payments.  The source category charge rate
schedule would be:

$15,300/yr/lb/hr         industrial oil-fired boilers
$15,800/yr/lb/hr         industrial coal-fired boilers
$ 3,500/yr/lb/hr         industrial process units
$ 0                      Bll other source categories
The "double burden" under this system is similar to that
under a uniform charge rate — with charge payments
amounting to 135% of the costs of control.  If this
double burden could be removed from this system, it
would be a relatively attractive alternative, since the
costs of pollution control equipment equal $66 million
— almost half of what they would be under a SIP-no
trading alternative.  One possible way to lessen this
burden would be to charge sources only for "excess"
emissions over some specified level—e.g., the charge
rate could be $0/yr/lb/hr for emissions up to 75%
of base-year emissions, and then $x/yr/lb/hr thereafter.

a source-by-source charge rate, in which we assigned a
different charge rate for each source.  When the fee
system is "custom fit" and each source faces an emis-
sion charge designed specifically for it, total costs
for attaining standards are extremely small — amount-
ing to $14.7 million.  The most important conclusion
from this result is that policies that discriminate in
their treatment of individual sources on the basis of
costs and ambient impacts can produce tremendous cost
savings.  For example, under a uniform charge rate

                                              FIGURE 3

                                A SCHEMATIC OF THE EPA MODEL
                                        /       \
                             AMBIENT NO2

                                                Figure 4
               Costsf Emissions Reductions, and Number of Sources Controlled under Alternative
Regulatory Strategies

Control Number of Sources
SIP - no
Emission Fees
Uniform Rate
Source Category
Charge Rates
SIP with




Areawide point source
Emission Rate Reductions





Control Charge
Costs Payments
(106 $/yr) (10* $/yr)

305 414

66 89

9 4

>13 -
Total C
to Sour
(106 $/;




1.   Excluding administrative costs.
2.   A range is given because the outcome is a function of who trades with whom.

                                Figure  5
             Ambient  Impact of Alternative  Strategies  in  Chicago

Range of
tions (ug/m3)

0 -
101 -

151 -
201 -
251 -
> 300



Number of
(base case)


locations with ambient

SIP No Uniform
Trades Charge

11 10
14 17

46 43
16 17
0 0
0 0
concentrat ions
Emission Fees
Charge by

falling within

Source Charge


SIP with

2 i
* Represents case where total  costs are $13 million  (see  text)

                   Figure  6
Predicted Air Quality  in 1984 at One Location in
    Chicago if Stationary Sources Are Not Controlled
        + denotes  location of sources

                  Figure  7
Simulated Air Quality at One Location in
   Chicago under SIP-without-Trading
       + denotes location of sources

                 Figure 8
Simulated Air Quality at One Location in
   Chicago Under a Uniform Bnission Fee'
       + denotes location of sources

                  Figure  9
Simulated Air Quality at One Location in
   Chicago Under Bnission Fees by Source
       + denotes location of  sources

                Figure  10
Simulated Air Quality at One Location in
   Chicago Under Source-by-Source Emission

       + denotes location of sources

                   Figure  11
Simulated Air Quality at One  Location in
   Chicago lander SIP-vath-Trading Altern-
       + denotes location of  sources


rate itself would have to be at least $15,800/yr/lb/hr
(per year per pound per hour).   The "double burden"
i.e., the charge payment required in addition to that
spent on the purchase of equipment to control emissions—
is enormous, amounting to 136%  of the costs of
the pollution control equipment itself.   Note, however,
that while the total costs of this system far exceed
those of other alternatives, areawide point source
emissions would be reduced by 84% — 4 times greater
than the SIP alternative.  Yet, as Figure 5 shows,
these additional emission reductions result in only
limited improvements in ambient .air quality.

a source-category charge system, in which different
charges are levied on different source categories (e.g.,
all oil-fired industrial boilers face the same charge
rate, but this rate differs from that levied on coal-
fired power plants).  This alternative would attain
standards at a total cost (pollution control equipment
plus charge payments) to polluters of $152 million, with
$66 million resulting from control costs, and $89 million
from charge payments.  The source category charge rate
schedule would be:

$15,300/yr/lb/hr         industrial oil-fired boilers
$15,800/yr/lb/hr         industrial coal-fired boilers
$ 3,500/yr/lb/hr         industrial process units
$ o                      all other source categories
The "double burden" under this system is similar to that
under a uniform charge rate — with charge payments
amounting to 135% of the costs of control.  If this
double burden could be removed from this system, it
would be a relatively attractive alternative, since the
costs of pollution control equipment equal $66 million
— almost half of what they would be under a SIP-no
trading alternative.  One possible way to lessen this
burden would be to charge sources only for "excess"
emissions over some specified level—e.g., the charge
rate could be $0/yr/lb/hr for emissions up to 75%
of base-year emissions, and then $x/yr/lb/hr thereafter.

a source-by-source charge rate, in which we assigned a
different charge rate for each source.  When the fee
system is "custom fit" and each source faces an emis-
sion charge designed specifically for it, total costs
for attaining standards are extremely small — amount-
ing to $14.7 million.  The most important conclusion
from this result is that policies that discriminate in
their treatment of individual sources on the basis of
costs and ambient impacts can produce tremendous cost
savings.  For example, under a uniform charge rate


       total costs are $719 million, or roughly 49 times greater
       than when charge rates are designed to fit the special
       circumstances of each source (33 times greater when one
       considers only the cost of pollution control equipment).
       However, this alternative still poses the implementation
       problems previously discussed.   Fortunately, the economic
       efficiency of a source-by-source charge rate is not due
       to any characteristic of emission charges per se, but to
       the policy of discriminating among individual sources in
       determining how to regulate or influence emissions levels.

o  SIP Control Strategy with Trading of Permits — We followed
   the initial allocation of permits allowed under the SIP
   described earlier, but allowed trading of permits to
   determine the final allocation.  Under this alternative,
   total costs of attaining standards range from $13 million to
   $130 million.  As explained in detail in Appendix B, it
   is difficult to be more precise in this estimate without
   knowing who trades with whom.  For example, if those with
   the highest cost of abatement trade with those with the
   lowest cost — and make no profit in doing so — total
   costs could be as low as $13 million.  However, there is
   no a priori reason why one should expect this outcome.

   For example, a source with a high control cost of
   $10,000/yr/lb/hr, certainly has an incentive to trade
   with a source with costs of $2,000/yr/lb/hr.  All that
   can be said is that the price of a permit will range
   from $2,000 to $10,000.  The actual price of a permit
   will be a function of who is the better negotiator and
   how well the market works.  Because there are only a limited
   number of sources with which to trade and sources possess
   imperfect information about other sources' control costs, the
   outcome of a trading system will depend on which firms are
   the best strategic bidders.  Total costs are likely to vary
   significantly, depending upon which sources are involved in
   the transactions.

   Certainly total costs to sources cannot exceed $130 million
   —since this figure represents the pretrading allocation
   of permits.  And we know that total costs could never
   be any less than $13 million — since this is the absolute
   least-cost way of meeting both the emission allocation
   mandated by the initial permit allocation under the SIP
   and the 250 ug/m^ standard.  The actual total cost to
   sources of a trading alternative will fall somewhere in
   between.  Exactly where is a function of:

       o the transaction costs in the market due to its adminis-
         tration.  If the control authority requires a long time
         to approve trades or, similarly, if traders must under-
         take extensive analysis before a trade is approved,
         fewer trades will take place, making the final total
         cost figure more likely to fall near $130 million than
         $13 million.


            o the information shared by sources.   The more sources
              are conscious of the economic gains of trading,  the
              more likely trading is to take place.   However,  it is
              difficult to state whether shared knowledge of control-
              cost curves across sources will be  beneficial for
              permit trading.  Instead it may result in more firms
              "holding out" — trying to obtain the  highest possible
              price for their permits.

            o strategic behavior.  Again, the outcome of the market
              will depend on the relative skills  of  the negotiators

     Thus, permit trading can reduce the total costs of attaining air
standards by a wide margin, with the actual cost  savings dependent upon
the degree to which the control authority and sources mitigate the
obstacles described above.

                      Summary and Conclusions

     Based upon the Chicago case study,  we conclude  the following:

       o  The control equipment necessary for attaining of a
          short-term NO2 standard of 250 ug/m3 in the Chicago area
          is not likely to be available until the mid 1980s.

       o  Short-term NOX concentrations in Chicago are characterized
          by very localized impacts.  Permit markets for this pollu-
          tant would be extremely thin.

       o  The economic gains from permit trading  are a function of
          the order in which trades occur.  Because  of the limited
          number of trades in the market, it would be overoptimistic
          to assume that a market in permits would result in the
          least cost pattern of pollution abatement.

       o  Discriminating among individual sources in determining how
          to regulate or influence emission levels results in cost
          savings.  For example, costs of pollution  control equipment
          under source-by-source emission charges were seven times
          less than under source category charges, and thirty-three
          times less than under uniform emission  fees.  This is
          not to imply that source-by-source emission fees should
          be adopted by EPA.  As discussed in Section I, this alter-
          native poses serious implementation problems.  However,
          approaches that do discriminate between sources should
          be given serious consideration.


                        FOR THE CHICAGO CASE STUDY



      A.   Source Inventories

      B.   Control Technologies and Costs

           1.  Sources and Methods of Control

           2.  Cost and Effectiveness of Control Technologies
                 for Stationary  Sources  Used in the Analysis

      C.   Limitations


      A.   Approach

      B.   Meteorological Conditions

      C.   Selection of Receptor Sites

      D.   Cases Used in the Analysis

      E.   Limitations


                        LIST OF TABLES
A.I   Steps in Applying AIRMOD to the Chicago NOX Problem

A. 2   NOX Control Technologies and Costs for Boilers and
        Process Heaters

A.3   NOX Control Technologies and Costs for Gas Turbines,
        Internal Combustion Engines, and Nitric Acid Plants

A.4   Highest and Second Highest One-Hour N©2 Levels in

A.5   Meteorological Conditions Used in the Air Quality
        Assessment Model

A.6   Comparison of Point and Area Source Contributions
        Under Different Meteorological Conditions

                        FOR THE CHICAGO CASE STUDY

      In this Appendix, additional information is presented
concerning the methods and data used in developing the results
of the analyses in the body of the report.

      The analyses presented in the body of the report were
performed using the AIRMOD computer programs (developed by
Pechan and Strathmeyer 1980).  The AIRMOD system is a series
of linked computer programs, operated in stages, which can
be used to simulate a wide variety of air pollution control
strategies on a regional basis.  The steps involved in using
AIRMOD for the Chicago NOX study are shown in Table A.I.

      The remainder of this Appendix is divided into three
sections.  The first section discusses the initial two items
from Table A.I -.- the pollution source inventories and the
control cost information.  The second section discusses the
ambient air quality modeling performed, while the third section
discusses the simulation of the "least cost" solution and
other policies.

     A.  Source Inventories

     Information was developed for the following three
categories of NOX sources in the Chicago region:

      o    Large stationary sources emitting at least 10
           pounds per hour of NOX (termed point sources).

      o    Small stationary sources emitting less than 10
           pounds per hour of NOX, e.g., residential furnaces
           (termed area sources).

      o    Mobile sources — primarily automobiles.

Data was collected for sources as of 1975.

Point Sources

      For point sources, data were collected for each "point"
of emissions for electric utility, industrial, and major

TABLE A.I — Steps in Applying AIRMOD to the Chicago NOy
      o    Develop Pollution Source Inventories

      o    Develop NOX Emission Reduction and Cost Information

      o    Perform Air Quality Modeling

      o    Prepare a Least-Cost Solution
      o     Identify  Possible Control  Policies and Simulate
             Their Application

                                — A 9 —
                                 c\ £,

commercial and institutional sources.  Examples of points are:

      o    a fossil fueled boiler

      o    an industrial process heater

      o    a gas turbine unit

Data  collected for each point source included:

      o    location (in the UTM system to the nearest tenth
           of a kilometer)

      o    level of NOX emissions and associated activity
           level (capacity of unit, hours per year of operation)

      o    characteristics of the emission point (stack height
           and diameter, exhaust gas temperature and volume)

      The set of 1975 point sources consisted of 472 separate
emission points.  Where the required information was not available,
default values typical of normal operating practice were utilized.
For area and mobile sources, data were developed into a set of 634
cells which covered the entire Chicago region.  Each cell consisted
of a  square five kilometers on a side.  For each cell, separate data
were  collected for area and mobile sources.

      Since 1984 is the earliest possible date for implementation
of a  short-term NOX control strategy, it was necessary to update
the 1975 source inventory to represent expected 1984 conditions.
For point sources, growth information was derived at a two-digit
Standard Industrial Classification code level for the region, and
the applicable growth parameters were applied to uncontrolled NOX
emissions.  Expected increases in NOX emissions were split into
two categories — (1) increases in activity and resulting NOX
emissions from existing sources and (2) NOX emissions resulting from
entirely new sources.  Half of the total increase was allocated to
existing sources (with the increase for each specific source based
on its industry type) while the remaining half was allocated to
typical new sources (based on "cloning" a sample of existing sources)
These new sources were assumed to locate with equal probability at
nine  industrial parks in the Chicago region.

      Growth in area sources was assumed to occur equally throughout
the region, since data were not available to divide growth in more
detail for these small sources.  For mobile sources, it was assumed
that no increase in vehicle miles travelled (VMT) would occur.*
The effect of reductions in automobile emissions due to Federal
tailpipe regulations resulted in an assumed decrease of about 20%
in regional mobile source emissions.
*The Chicago State Implementation Plan (SIP) assumes that
regional VMT will decrease by 20%.


       B.   Control Technologies and Costs

            1.  Sources and Methods of Control

       The major categories of stationary source NOX emissions
 identified by EPA are:

       o    utility and industrial boilers

       o    large internal combustion engines

       o    industrial process heaters

       o    industrial process emissions

       o    incinerators*

       NOX emissions from the first three categories above
 result from fossil-fuel combustion — either from oxidation
 of nitrogen in the fuel or from oxidation of nitrogen  in the
 air used to support combustion.  The various existing  and  new
 approaches for the control of NOX emissions  from combustion
 sources can be categorized as follows:

       o    Combustion Process Modification

       o    Fuel Modification

       o    Removal of NOX From Stack Gases [Flue Gas Treatment

       o    Alternative Combustion Processes

       Combustion  Process  Modification.   This  technique consists
 of altering operating conditions  of  existing  systems or altering
 the design of  new units to minimize  NOX  emissions.  The underlying
 principle  in the  combustion modification is  the  reduction of  oxygen
 concentration  in  the  combustion zone and the  reduction of residence
 time at  high temperatures above 3600°F.   The  most  effective combus-
 tion modification techniques  include:   low excess  air  firing  (LEA)/
 off-stoichiometric combustion  through biased  firing, overfire air,
 flue gas recirculation, low air preheat,  and  water  injection.
* Municipal and industrial incineration of waste is a significant
source of NOX in many areas.  There are no anticipated control
alternatives for NOX from these sources (Anderson, et al., 1979)
other than their phase-out and replacement by other disposal methods,
such as sanitary landfill.  There are no incineration sources con-
tributing to elevated ambient NOX levels in the present analysis.


      Excess air is an operationally controllable parameter in
most combustion processes.  The reduction of oxygen in the com-
bustor air results in excess carbon monoxide (CO) when burning gas
or excess smoke when burning oil.  NOX emissions can be reduced
by lowering excess air to the level just above that of the onset
of excess CO or smoke.  Reducing excess air from 25 to 15 percent
normally results in lowering NOX by an equivalent percentage
with an additional benefit of reducing fuel requirements by one
percent.  Equipment for modifying the existing combustion system
for LEA consists of improved oxygen measuring systems and controls.
The costs of such equipment are relatively independent of size.
The annual operating and maintenance costs are normally offset by
the fuel savings.

      Flame temperature has the most dramatic impact on the NOX
formation rate and is therefore the most important combustion
modification parameter.  Flame temperature can be reduced by:
shifting the air-fuel ratio from the stoichiometric value in the
primary flame zone; diluting the incoming air with recirculated
flue gas; reducing the temperature of the incoming air; or inject-
ing a coolant, such as water or steam.  Each of these modifica-
tions has been found to reduce NOX from combustion systems.
Their applicability depends upon the combustion parameters of each
furnace.  NOX reductions of up to 60 percent can be achieved
with the application of a combination of these modifications to
existing systems.

     New equipment being designed incorporates one or more of the
above modifications to comply with the NSPS for coal-fired boilers
of 0.7 Ib. NO2/106 Btu.  Further improvements in the burner
designs are expected to continue with NOX emission reductions
(over the present uncontrolled level), reaching 65 percent by the

      Fuel Modification.  Modification of fuel is another method
used to reduce NOX emissions.  Thermal NOX emissions are decreased
by using fuels that burn at lower temperatures.  The three most
popular techniques are fuel switching, fuel additives, and fuel

      Solid fuels are considered to have more nitrogen than liquid
fuels; gaseous fuels have virtually no nitrogen.  Despite the
current cost-effectiveness of cleaner fuels in the control-of NOX,
switching to them cannot be considered a long-range strategy because
of their projected shortage in the future.  With regard to using
fuel additives,  current evidence indicates that the emission
reductions are quite limited and that the strategy is not likely
to be cost-effective.   At the same time, current technology for
fuel denitrification generally involves fuel pretreatment to remove
other pollutants.  For example, marginal reductions in the nitrogen


content of coal and oil have been achieved from desulfurization.
However, this technique is not economically attractive if used
solely for the purpose of NOX control.  Thus, if fuel denitrifica-
tion is to prove cost-effective, it will be on the basis of simul-
taneous sulfur and nitrogen removal.

     Removal of NOX From Stack Gases  [Flue Gas Treatment (FGT)].
In the wake of recently increased emphasis on the control of NOX
from stationary sources, development activity for flue gas treat-
ment of NOX has accelerated.  Several FGT processes are presently
in use (or in development stages) in Japan where N02 standards
are more stringent.  Specific processes include selective catalytic
reduction  (SCR), selective noncatalytic reduction, and wet scrub-
bing.  The dry systems, such as SCR using ammonia, show the most
promise for clean  inlet streams.  The system works well with gas-
or oil-fired combustion processes in which the flue gas to be
scrubbed has low SO2 and dust loadings.  NOX removal efficiencies
of 90 percent are  possible.  Selective noncatalytic reduction
using ammonia is being tried on a system in Japan, and efficiencies
of up to 70 percent have been achieved.  Wet systems for the removal
of NOX from flue gas involve oxidation of NO to NC>2 with ozone
or other oxidizing agents.  The costs of oxidizing agents and/or
energy input for generation of ozone added to the FGT costs seem
to render the wet  simultaneous SOX/NOX processes impractical for
commercial use.  Other combinations presently being tried for the
simultaneous removal of SOX/NOX are wet S0x/dry NOX and dry
S0x/wet NOX.

     The energy impact of the wet NOX scrubbing system is dis-
couraging.  For a  utility boiler, the energy consumption by the wet
NOX scrubbers could be as high as 10 percent of the total plant
©utput.  But considering the trend toward increasing use of coal
in the future, the wet NOX removal processes, even though expensive,
show promise because of the simultaneous SOX/NOX removal potential.

      Alternative  Combustion Processes.  Fluidized bed combustion
(FBC) and catalytic combustion are the major alternative combustion
processes presently being explored.  FBC appears to be competitive
with control of conventional combustion methods from a NOX control
standpoint.  However, the versatility of the FBC concept to a wide
variety of equipment applications to conventional boilers has yet
to be demonstrated.

      Catalytic combustion of clean fuels has been shown to limit
NOX emissions to 5 ppm, or a greater than 99 percent reduction
over noncatalytic  combustion units.  Research is presently continu-
ing to demonstrate the technical feasibility of applying catalytic
combustion to a variety of systems.


     Nitric acid manufacturing constitutes a major noncombustion
source of NOX emissions.  The available control techniques include
catalytic reduction, extended absorption, wet scrubbing, molecular
sieve absorption, and flaring.  Catalytic reduction is currently
the most widely used NOX control technique for nitric acid manufactur-
ing plants.  Reductions of up to 90 percent have been achieved.

           2.  Costs and Effectiveness of Control Techniques
                 for Stationary Sources Used in the Analysis

      This section summarizes the costs and effectiveness of control
techniques described earlier for the various stationary sources of
NOX.  Combustion processes in which the combustion parameters are
similar have been grouped together and have been assumed to require
similar control techniques.

      Tables A.2 and A.3 contain summaries of the control techniques
applicable to each source category, including their effectiveness,
availability, and annual costs.  The cost estimates for a control
technique are for an average size source and are assumed to be linear
with source size.  In the analysis, information on source size is used
to estimate capital cost, while information on source utilization (hours
per year of operation) is used to estimate operation and maintenance
costs.  It is important to stress that most of the data in the
analysis are based on pilot projects, with limited field experience.
Thus, the cost-effectiveness of these techniques could vary by a wide
margin when applied to individual sources.  The annual costs in the
tables consist of the initial investment annualized at a rate of
16 percent plus annual operating and maintenance costs.*

      Utility Boilers.  Combustion process modification (CPM) and
flue gas treatment (FGT) have been considered to be the most
effective options for the control of NOX emissions from existing
utility boilers.  Combustion process modification techniques have
been found to be less effective on coal-fired utility boilers than
on either gas- or oil-fired units.  Table A.2 shows the effective-
ness of CPM techniques on coal-, gas-, and oil-fired utility
boilers.  NOX emission reductions of up to 90 percent from multiple
boilers may be necessary if a stringent short-term NC>2 standard
is promulgated in the future.  To achieve this high degree of
control, FGT in conjunction with CPM may be necessary.
*  The 16 percent capital recovery factor assumes 100 percent debt
financing, a 10 percent interest rate, and a 10-year recovery
period.  The actual life of most NOX controls (i.e., those based
on combustion modification techniques) would equal the remaining
life of the sources controlled.

                      TABIE A.2 — NOy Control Technologies and Costs for Boilers and Process Heaters
                                                                     Annual Cost for Typical Facility
                                                                         (Dollars Per Billion BTU)
              Level of     Coal-    Oil- and
                 NOX       Fired   Gas-Fired
 Technology   Reduction   Utility   Utility
Availability  Achievable  Boilers   Boilers
 Fired     Oil- and
Indus-   Gas-Fired   Industrial
trial    Industrial   Process
Boilers   Boilers     Heaters
Low Excess Air  (LEA)                Present       10-25%       2.00       .09       2.90      2.60

IEA Plus Off-Stoichiometric
  Combustion  (OSC)                  Present       16-40%       5.00      2.80       6.60      6.00

LEA Plus Flue Gas Recirculation
LEA Plus OSC Plus Ammonia
Low NOX Burner*
Dry Selective Catalytic
Reduction (Flue Gas
Treatment )
N/A - Technology not applied to this source category.
Source:  Anderson et al., 1979.

           TABIE A.3 — NOX Control Technologies and Costs for Gas Turbines, Internal
                           Combustion Engines, and Nitric Acid Plants
Annual Cost for Typical Facility
Water Injection
Fine Tuning and Chang-
ing Air-to-Fuel Ratio
Level of
Gas Combustion
Turbines! Engines1
46.00 N/A
N/A 20.00
Dry Selective Catalytic
  Reduction (Flue Gas
Treatment )
Chilled Absorption
N/A - Technology not applied to this source category.

^'•Dollars per billion ETU

2Dollars per 1000 tons production

Source:  Anderson et al., 1979.

While dry selective catalytic reduction has been demonstrated on
gas- and oil-fired units with low SC>2 in the inlet stream, the
flue gas must be desulfurized before dry SCR can be used for coal-
fired units.  Nevertheless, recent reports indicate that applica-
tions of dry SCR with special catalysts on coal-fired units are
under investigation.  In addition, as noted above, several simul-
taneous SOX/NOX removal processes are currently under study,
although their energy consumption may make them impractical.
Finally, it should be noted that several low NOX burners are
presently under development for coal-fired utility boilers.  These
burners would also be suitable for retrofit application.  Prelimin-
ary cost projections are under $l/kW for new units and about $2/kW
for retrofits and reductions in emissions from 45 to 60 percent
are expected.

      Industrial Boilers.  Combustion process modification tech-
niques for the control of NOX emissions from industrial boilers
are effective in the range of 16 to 40 percent depending upon the
fuel burned and the type of boiler.  Reducing excess oxygen has
been found to be more effective in reducing emissions in coal-fired
boilers than in either oil- or gas-fired boilers.  Retrofitting
boilers with advanced-design burners could boost the NOX emissions
control up to 40 percent.  There are presently no data available on
the costs of flue gas treatment on small industrial boilers.

     Gas Turbines.  Reductions of up to 60 percent can be realized
with the application of water injection on existing gas turbines.
New equipment with advanced design is expected to result in NOX
emission reductions of up to 75 percent at an annual cost of $4-$10/kW.

      Internal Combustion Engines.  NOX emissions from internal
combustion(1C) engines can be reduced up to 30 percent by changing
the air-to-fuel ratio.  New 1C engines scheduled for 1983 are
expected to have NOX emissions reduced by 50 percent.

      Industrial Process Furnaces.  Uncontrolled emissions from
industrial process furnaces, such as those associated with metal-
lurgical, cement, and glass manufacturing, vary depending on the
excess air level per unit of fuel burned, though at the minimum
permissible level of excess air, emissions per unit of fuel are
approximately the same.  NOX reductions with LEA for different
furnaces investigated are in the range of 20 to 35 percent.  Further-
more, the cost of implementing low excess air modification is
independent of furnace size and does not vary significantly among
various processes.  Adoption of advanced process technologies and
burner designs is estimated to reduce the NOX emissions from
these furnaces by about 70 percent in the 1980s at a cost of
approximately $5/kW.


      Nitric Acid Plants.  Nitric acid manufacturing and other
related processes using nitric acid as feedstock account for most
of the noncombustion-generated NOx.  Stack gas treatment employing
the method listed in Table A.3 can reduce NOX emissions from this
source category up to 90 percent.

      C.   Limitations

      The source information used in the analysis was derived
from previous studies and is subject to the limitations discussed
in them.  Despite rather extensive checking, it is likely that
at least some information in the data base is inaccurate to some
extent.  The simulation of emissions growth from 1975 to 1984
included in the analysis is somewhat speculative.  It is interest-
ing to note, however, that aggregate results from cases assuming
no growth in point source emissions did not differ significantly
from the growth case actually used in the analysis.

      Cost data were developed using a model plant approach which
developed costs based on typical facilities.  Costs for specific
facilities could be expected to vary significantly in either
direction from the estimates used.


      A.   Approach

      The development of cost-effective pollution control strategies
requires not only information on the actual levels of pollution occur-
ring in the ambient air but also information on the sources which
contribute to the elevated pollution levels.

     Even if ambient data alone were enough for developing appropriate
NOX control strategies, it is clear that the ambient N02 data available
for the Chicago AQCR are inadequate for even adequately characterizing
the short-term ambient N02 problem.  Currently there are only four
continuous N02-NOX monitors in the Chicago AQCR, far below the number
needed to adequately cover an area approximately 150 km by 80 km.
Nevertheless, the four sites are representative of severely impacted
locations in the region.  The peak one-hour N02 levels observed at
these sites are among the highest concentrations observed in any urban
area of the United States.  Table A.4 shows the highest and second
highest yearly one-hour NO2 levels measured at these sites during

      In order to further evaluate the extent of the short-term N02
problem in the Chicago AQCR, as well as to gain information from which
abatement strategies could be developed, it was necessary to develop an
air quality assessment model.  For this purpose, a multiple point and
area source model known as RAM was used for assessing short-term NOX
concentrations.*  RAM is an EPA-approved, Gaussian steady-state model
capable of predicting short-term (averaging times ranging from an hour
to a day) ambient concentrations of relatively stable pollutants from
multiple point and area sources.  The model requires hourly meteoro-
logical data on wind direction, wind speed, stability class, and mixing
height.  In addition, the data required on the point sources consist
of source coordinates, emission rate, physical stack height, stack-
gas volume flow, and stack-gas temperature.  Area sources are specified
in terms of the southwest corner coordinates ^of the area source grid,
the total grid cell area, the total cell emis'sion rate, and the
effective area source height.
* See Turner and Novak (1978).

   TABIE A.4 — Highest and Second-Highest One-Hour N00  Levels  (g/m3) in Chicago

1-hr. NO?


1-hr. NO2_

1-hr. NO?


1-hr. NO?

1-hr. NO?


1-hr. NO?

a/ Questionable data.

N/A - Not available.

Source:  Anderson et al., 1979.


      As noted, RAM is designed for modeling nonreactive pollutants.
However, nitrogen dioxide (NC>2) is primarily a secondary pollutant
formed by oxidation of nitric oxide (NO).  The initial NO concentra-
tion in the exhaust gases, the plume diffusion and travel time,
and the ambient concentration of photochemical oxidants and reactive
hydrocarbons are all important factors that affect the conversion
of NO to NO2-  Therefore, a dynamic model of N02 formation from
point source emissions of NOX was developed for use in conjunction
with RAM to predict the NOX concentrations at a receptor due to
point sources and to translate them into N02 concentrations.

      With regard to area sources, a fixed N02/NOX ratio for each
period of the day based on the observed data at the continuous
monitoring sites was used to translate ambient NOX concentrations
into the corresponding NO2 levels.

      B.   Meteorological Conditions

      Meteorological conditions play a major role in the short-
term buildup of pollutants from both point and area sources.
High ground level concentrations from point sources are normally
caused by fumigation (inversion breakup) or plume downwash where
the plume intersects the ground quickly.  The meteorological para-
meters that characterize these conditions are an unstable atmos-
phere (stability class of B or C) and moderate-to-high wind speeds.
High ground level impacts from vehicles and other area sources
occur when the atmosphere is quite stable (stability class D or
E), wind speeds are low, and mixing heights are low.  Thus, the
meteorological conditions that maximize the impact of point
sources are at the opposite extreme from, those that maximize the
impact of ground level area sources; hence, it is difficult to
estimate their simultaneous maximum impact.  To circumvent this
problem, several sets of meteorological conditions were used to
simulate worst-case conditions for point and area sources together.

      The actual meteorological data for the Chicago AQCR used in
the analysis were extrapolated from the mixing height and wind
speed data collected by the National Weather Service at its
Greenbay, Peoria, Flint, and Dayton stations.*  Typical meteorolog-
ical conditions as well as those that correspond to stagnant and
unstable atmospheres in the Chicago AQCR are shown in Table A.5.
*  See Holzworth (1972, 1974, 1978).

 TABIE A5 — Meteorological Conditions Used in the Air Quality Assessment Model
o Wind Speed
o Mixing Height
o Stability Class
o Wind Direction
o Anbient Temperature
Worst Case - Point Source
Season and Hour of the Day
Summer Mid-Morning

4.0 meters/sec.
650 meters

Summer Afternoon

6.5 meters/sec.
1600 meters

Winter Mid-Morning

6.0 meters/sec.
450 meters
20 °F


o  Wind Speed
o  Mixing Height
o  Stability Class
o , Ambient Temperature

Worst Case - Area Source

o  Wind Speed
o  Mixing Height
o  Stability Class
o  Ambient Temperature

Predominant Wind
2.5 meters/sec.
300 meters
0.5 meters/sec.
300 meters
4.5 meters/sec.
800 meters
                    2.5 meters/sec.
                    800 meters
                         80 °F
                                       2.50 meters/sec.
                                       300 meters
                                           20 °F
                   1.5 meters/sec.
                   200 meters
Source:  Holzworth (1972, 1974, 1978).


      C.   Selection of Receptor Sites

      The RAM model operates by simulating the ambient air quality
levels which would be experienced if there were a set of continuous
monitors located at various points throughout the region.  These
simulated monitors are termed receptors.  The receptors which are
likely to experience elevated NO2 concentrations are of prime
interest.  Their locations are hard to determine since the concen-
tration at any point is an extremely complex function of meteorolog-
ical parameters, source locations and strengths, source overlaps,
etc.  One possible solution is to develop a matrix of receptors
which blankets the entire region.  The level of coverage is then
only limited by available computer resources.  Another solution is
available by using a capability of the RAM model in which the
model selects likely "hot spots" based on the characteristics of
major emission sources.  While the second method is excellent in
identifying points of maximal impact for individual point sources,
its selections do not fully identify areas in which many sources
overlap nor consider the effect of the NO to NO2 transformations.

      The problem of selecting critical receptors is exacerbated
by the short averaging time being simulated, since each source's
maximal plume impact is rather small.

      To assist in developing a viable approach to select receptors
to include in the analyses, several series of preliminary analyses
were made.  These analyses revealed the infeasibility of attempting
to fully blanket the entire region with receptors.  It was decided
to base the final analyses on 200 high-impact receptors selected
by RAM, along with approximately 400 other receptors which were
selected to blanket the region.

      D.   Cases Used in the Analysis

      Several preliminary runs of the model were then made to
determine the combinations of hourly emissions and meteorological
conditions that would result in the highest short-term NO2 concen-
trations as well as to see if the model results were in line with
the limited monitoring data.

      High hourly NO2 values were estimated at several receptors
for both sets of worst case conditions depicted in Table A.5.
However, these conditions are distinct in that they select for
point source impacts on the one hand and area source impacts on
the other, with limited interaction between these source types
under either set.  To test for the possibility that the highest
hourly N02 levels could result from area and point source interac-
tion, an intermediate set of meteorological conditions was specified;

      o    Wind Speed:  1.5 meters per second;

      o    Stability Class:  C; and

      o    Mixing Height:  300 meters.


The NC>2 emissions used with these conditions correspond to the 1975
summer mid-morning levels.  Since these conditions indeed led to the
highest estimated ambient NC>2 levels (see Table A.6), they were used
in the final analysis.

      The results indicated the presence of two distinct types of
point sources:  (1) large plants with tall stacks such as power plants,
and (2) plants with a large number of smaller sources with short stacks
such as steel mills and refineries.  The diffusion characteristics of
emissions from the second category were found to be similar to those
of the area-source emissions.  During meteorological conditions that
cause high concentrations from both the point sources with short stacks
and the area sources, point sources such as power plants caused low
concentrations due to their high effective stack heights.  This is
illustrated by the results presented in Table A.6, which compares the
relative point- and area-source contributions to ambient N02 concentra-
tions under three different meteorological conditions.  It should be
noted that while some extreme meteorological conditions could result
in point sources with tall stacks having greater individual impacts,
these same conditions tend to minimize the impact of other source
category emissions with the result being that overall concentrations
are relatively low.  Thus, the point sources with short, multiple
stacks and the area sources seem to be the dominant cause of high
short-term N02 concentrations in Chicago.

      E.   Limitations

      Previous work on characterizing the N02 problem from both
point and area sources was valuable in developing this relatively
sophisticated approach for estimating short-term NC>2 concentrations.
Nevertheless, there remain important limitations in the analysis that
are associated with the underlying assumptions.  These assumptions
include the following:

      1.   The meteorological conditions are assumed to hold for
           a period of one hour and are assumed to be uniform over
           the whole AQCR.  Uniformity is not likely to be very real-
           istic because of the variations in the topography across
           the region.  In addition, center-city meteorological condi-
           tions are considerably altered by the "built" environment.
           However, a "microscale" analysis seemed unwarranted in
           view of the objectives of this study and the prohibitive
           resources that it would have required.  Thus, the estimates
           of the NC"2 concentrations may have been biased due to
           these factors.

      2.   Mobile-source emissions are uniformly distributed through-
           out each grid cell.  This results in a loss of spatial
           resolution that may reduce the impact of these emissions
           on air quality.  The uniformity assumption is again question-
           able since ambient levels of the mobile-source pollutants
           (HC, CO,  NOX) tend to peak near roadways and normally
           diffuse within relatively short distances.

         TABLE  A. 6 — Comparison of Point and Area Source Contributions Under Different
Meteorological Conditions

Meteorological Conditions

Number of Receptors
above 200 ug/m3*
Average for all
Receptors above
200 ug/m3*
Five Highest
Receptors :
Percentage of Fewer
plants with Signi-
ficant Contribu-
Worst Case - Point Source
Total Point Area

277 165 111

509 428 81
589 409 81
348 209 139
348 225 123
342 219 123

Worst Case - Area Source
Total Point Area

341 142 199

568 549 19
479 279 200
472 272 200
472 272 200
472 272 200

Total Point

316 142

603 493
602 434
600 407
598 430
553 383




*  The 200  ug/m3 level of concentration is for a one-hour averaging period.

**- Defined  as effective stack height less than mixing height.


           Thus, NC>2 levels, though considerably smoothed due
           to the slow conversion rate of NO to NC>2/ are still
           significantly higher in the vicinity of roadways.
           TO circumvent this problem, some version of a high-
           way model would be needed to estimate the impact
           of mobile sources more accurately.

      3.   As noted above, the NOX concentrations due to the
           point sources were transformed to NC>2 concentra-
           tions by use of a dynamic plume-specific approach
           incorporated into RAM.  However, the translation of
           NOX emissions from area sources to NOo concentrations
           was accomplished by using a fixed N02/NOX factor of
           0.5.  This factor was derived from hourly NO2 and NOX
           data for the summer morning period recorded at the
           continuous monitoring stations in the Chicago AQCR.
           In general, the daily N02/NOX ratios for the mid-
           morning period ranged from 0.4 to 0.6.  Nevertheless,
           it is probable that on some days, the monitors were
           influenced by area sources to a greater degree than on
           others.  Furthermore, the factor of 0.5 was applied
           to all area sources.  While it is true that NOX emis-
           sions from area sources are mixed with the atmosphere
           homogeneously within relatively short distances of
           the point of discharge, concentrations in the immedi-
           ate vicinity of discharge points may differ.  Notwith-
           standing, it seemed reasonable to assume that for a
           given period of the day, the ratio of N02 to NOX
           concentrations across area sources would be similar.
           Yet, this assumption should be considered only as a
           crude approximation to the actual conversion rate.

      4.   The conversion rates used in the analyses -- variable
           for point sources and fixed for area and mobile sources
           — do not exactly replicate the photochemical type
           reactions which influence the NO to N02 conversions.
           The details of the chemical transformations are not
           well understood and it is likely that approximations
           similar to those used in the present analysis will
           continue to be required.

      An example of difficulties associated with estimation of NO
to NO2 conversion rates is illustrated by a recent study performed
for EPA's Industrial Energy Research Laboratory (IERL) (Radian
1980) which used an alternative method which predicted NOX concen-
trations of over twice those of the present study.*  While the air
modeling results used in the present study are only approximations,
these results are in close agreement with the limited hourly monitor-
ing data available.  The IERL results, on the other hand, are con-
siderably higher than those from the limited available monitoring
*  If the IERL study is correct, it would most probably be
impossible to meet a 250 ug/m3 hourly NO, standard even with
technologies expected to be available later in this decade.



      Operation of the RAM model provides information on the expected
worst-case ambient air quality impacts as well as detailed infor-
mation on the sources which contribute to the elevated NOX concen-
trations.  This information can be used to simulate a variety of
alternative control strategies of two major types — those which
are sensitive to each source's contribution to elevated pollution
concentrations as well as those which are insensitive.

     The various economic incentive policies simulated by AIRMOD
are derived from the solution to a mathematical programming problem
which is most simply expressed as identification of that source-
specific control strategy which meets the selected ambient air
quality standard throughout the region with the lowest possible
annualized expense for pollution control measures.

      The method of achieving this least-cost solution is based
on an algorithm presented by Miller and Lent (1978).  The method
operates by repeatedly selecting the most cost-effective control
option among all sources in the region until the simulated standard
is achieved.  The cost-effectiveness criterion is computed for
each source as the incremental annual cost of implementing the
next most stringent control alternative, divided by the sum of the
expected ambient air quality improvement at each violating receptor,
multiplied by the amount each receptor is in violation at that
operation.  The most cost-effective option is then selected.

     The various strategies presented in the body of this report
are derived from the least-cost strategy as computed above.  The
derivation of each of these is presented below.

      o    SIP Control Strategy Without Trading - The most
           stringent control alternative for any source in
           each of the nine source categories required in
           the least cost strategy to meet the ambient
           standard is assumed to be the control which would
           be applied uniformly by the SIP to each source
           within these categories.

      o    Emission Charges - In the emission charge strategies,
           the analysis consists of two steps.  First, the
           sources are divided into the groupings as speci-
           fied by the particular charge approach — e.g., for
           the uniform charge, all sources are treated together,
           while for the source-by-source fee, each source
           is treated separately.  Second, for each group,
           a charge is set at the minimum amount per quantity
           of emissions which is required to ensure that
           the ambient standard will be achieved by the sources
           acting in their economic self-interest to minimize the
           sum of fees plus annual control costs.


      o    SIP Control Strategy With Trading of Permits - In this
           case, the number of permits is defined as the emissions
           required under the SIP control strategy without trading.
           Since the operation of the trading market cannot be
           specified exactly, a range of possibilities must be
           developed.  The most expensive case in terms of annual
           regional control costs is if no trading occurs — exactly
           the SIP without trades.  The least-cost case would be
           no profit trading, that is, each source is willing to
           trade if it can cover its costs.  This latter case is
           equivalent to a new least-cost problem with a dual set
           of contraints which (1) meet the ambient standard, and
           (2) limit emissions to the total permitted under the
           SIP no-trading case.

Comparison to Previous Analysis

     Results presented in this report for the simulation of an
hourly 250 ug/m3 NC^ standard in Chicago differ in several
significant respects from those previously provided in a contractor
study and in the 1979 CEQ Annual Report.  The present results both
correct errors in the previous work as well as use more realistic
and detailed assumptions.  The major differences are outlined

     (1) Correction of errors in the inventory of point sources.
         In the original analysis, the single source contrib-
         uting most to elevated ambient NC>2 concentrations was
         an incinerator.* Because no control measures are available
         or are expected" to be available for this type of source,
         the contractor study assumed that it would be shut down and
         its  emissions reduced to zero.  It was further assumed
         that this shutdown would occur at no cost.  Contact with
         the source revealed that its emissions were drastically
         overstated in the contractor study.  Corrected results
         showed an emission rate of less than 10 pounds per hour
         and no contributions to elevated concentrations.
     *This source alone would cause violation of an hourly N02
standard of 500 ug/m3.

     (2) Correction of errors in the least-cost algorithm.  The
         least-cost method in both the contractor study and the
         present analysis is based on the solution to a mathemat-
         ical programming problem using a gradient solution
         method.  However, the version of the solution algorithm
         upon which the contractor results were based was defective
         in several respects.  The net result of these defects was
         to overestimate the cost of achieving a least cost

     (3) Growth is now treated explicitly and consistently in the
         analysis.  In the contractor report, the hypothetical
         future NC>2 standard was imposed on an inventory represent-
         ing 1975 emissions for point sources combined with 1984
         emissions for automobile sources.*  The adjustment made
         by the contractor to account for reducing tailpipe emis-
         sions was developed on an ad hoc basis and applied to the
         sum of mobile and nonmobile source emissions rather than
         separately for these two categories.  The present analyses
         corrects these problem in the following ways:

            o    emission growth for point sources is treated
                 explicitly — both "in-place" growth at existing
                 sources and new sources based on "clones" of
                 existing sources.

            o    area sources are now divided into "mobile" and
                 "other" sources.  Adjustments due to the effects
                 of federal automobile standards are made directly
                 to the mobil source emissions.

     (4) A "de minimus" criterion was incorporated into the analysis.
         In its report, the contractor claimed that only sources
         with NOX emissions greater than 10 Ibs/hr were required
         to install controls.  In the analysis itself, however,
         this "de minimus" criterion was not included.  Thus, the
         costs of some control strategies were overstated.  The
         analysis was corrected using a 10 Ib/hr cutoff.

     (5) Reporting capabilities have been significantly enhanced.
         Results which simulate various policies can now be easily
         displayed in the form of contour isopleth maps.  Thus,
         the differential effects of air quality occuring under
         different policies can now be directly observed. In addition,
         more meaningful and useable printed reports are also
     *The authors were forced to make some adjustment to the mobile
source component (which is decreasing due to federal tailpipe
emission limits and fleet aging) since a 250 ug/m^ standard cannot
be achieved without reducing emissions from the 1975 vehicle fleet.

                      REFERNCES FOR APPENDIX A

Anderson, R. J., Reid, R. o., and Seskin, G. P., "An Analysis of
      Alternative Policies for Attaining and Maintaining a
      Short-Term NO2 Standard,"  Mathtech, Inc., 1979.

Blanks, J. P., et. al., "Investigation of NC>2/NOX Ratios in
      Point Source Plumes," 1980.

Eppright, B. R., et. al., "Impact of Point Source Control
      Strategies on N02 Levels," Radian Corporation," 1978.

Holzworth, G. C., "Daily Mixing Height and Wind Speed Data
      .at National Weather Service Sites at Greenbay, Peoria,
      Flint, and Dayton," EPA (unpublished), 1978.

Holzworth, G. C., "Mixing Height, Wind Speeds and Potential
      for Urban Air Pollution Throughout the Contiguous
      United States," EPA, AP-101, 1972.

Miller, C. and J. Lent,,"A Heuristic,Integer Programming
      Approach to Estimating the Cost of Air Quality Standards,"
      Energy and Environmental Analysis, Inc., 1978.

Pechan, E., and S. Stratmeyer, "A Computer-Assisted System for
      Developing Regional Cost-Effective Air Pollution Control
      Strategies (AIRMOD)," Energy and Environmental Analysis, Inc.,
      july, 1980 (Draft).

Turner, D. B., and J. H. Novak, "User's Guide for RAM," Volume I,
      Algorithm Description and Use, EPA-600/8-78-016a, 1978.

                      SIMPLIFIED PERMIT SYSTEM

     This appendix models a simplified marketable permit system.
Throughout, it is assumed that the air pollution control authority
can attain  its air quality goals simply by rolling back the emis-
sions within the area.  Source location, stack parameters, and
meteorology are ignored in this control strategy.  This approach
is generally accepted for certain pollutants — nonmethane hydro-
carbons, chlorofluorocarbons, and acid rain precursors (i.e.,
cases where total emission loadings are thought to be responsible
for the air quality problems).  The model does not emulate the
intricacies of control strategies for attaining ambient TSP, S02/
or N02 standards — where source location, stack parameters, and
meteorology play important roles.

     Trading in a marketable permit system would occur when it is
in the economic interest of a firm to do so.  This interest can be
explained by the decision diagram shown below.  The diagram repre-
sents a situation in which a firm was allocated fewer permits than
its current emission level.  It must now make a decision whether
to install pollution control (PC) equipment or to buy permits to be
in compliance.  A firm may choose not to comply and take the risk
of being caught.  However, for the purposes of this analysis, we
are assuming that firms will comply.

                           	   Buy permits  	
                                  Install PC equipment
The economics of the situation can be represented by the following

                            	Buy permits PR X (Qp -

                                  Install PC equipment
                                  and sell excess permits
                                                  - PS (OA - QPC)
The variables in the diagram are defined as follows:

PB - weighted average buying price
QC - current emission level
   - allocated number of permits
    - net present value of total pollution control costs
Ps - weighted average selling price
QpC ~ emission level after installation of PC equipment


One alternative is to buy enough additional permits to equal current
emission levels.  The cost of this alternative is represented by:

PB x (Qc " QA) = cost of buying permits
The quantity (Qc - QA) is the number of permits that must be
bought.  It is the difference between the firm's current emission
level, Qc, and the number of permits that were allocated to it, QA.
The other alternative involves installing pollution control equip-
ment and selling any excess permits.

The following equation represents this alternative:

                       - PS X (QA - QPC)
The quantity (QA - Qpc) is the number of excess permits that
may be sold.  The sale price (PS) times this quantity equals the
revenue from the sale of the excess permits.

          Before looking at an actual example, it is useful to
understand a number of practical aspects to this equation.   First,
there are a number of areas where uncertainties can affect  the
economics of the decision.

     o    Periodic revisions of air quality standards will  cause
          uncertainty in the value of the permit, and affect busi-
          ness decisions.  Also, changes in the standards may
          result in more controls being required.  To guard against
          this contingency, a firm may decide to hold additional
          permits over and above its current level of emissions.

     o    The estimate of the cost of PG equipment will have some
          uncertainty in it.  This cost equals the net present
          value of all of the future cash flows associated  with
          the procurement, installation land operation of the PC
          equipment.  There are uncertainties with the capita}.
          investment needed for equipment installation.  In addi-
          tion, there are longer-term uncertainties associated
          with projecting operating and maintenance costs of the
          equipment.  In general, firms will tend to cover  the
          risks caused by these uncertainties by adding safety
          factors into the cost estimates.

     o    The estimate of the effectiveness of the PC equipment
          will also have some uncertainty.  As a result, a  firm
          will probably not commit to the sale of all its excess
          permits, but keep some in reserve in case the effective-
          ness estimates are incorrect.


The effect of these uncertainties will be illustrated in the follow-
ing example.  In order to illustrate the individual firm's decision
environment, the parameters of a typical plant are shown below.

                      Carbon Steel Plant

Qc - current level of emissions (TSP) - 102 Ibs/hr
QpC - emissions after additional PC equipment is installed - 31 Ibs/hr
CPC ~ net Present value of pollution control costs - $14,100,000
QA - quantity of permits allocated - 61 Ibs/hr permits

              Buy permits    PB X (102 Ibs/hr - 61 Ibs/hr)

              Install PC
              equipment	 $14,100,000 - PS (61 Ibs/hr -
                                               31 Ibs/hr)

     This decision then is to either buy 41 Ibs/hr permits or
install PC equipment and sell as many as 30 permits.  One way to
look at the problem would be to consider the "indifference" price
- the price that would result when the decision maker is indifferent
about buying permits or installing PC equipment.  In this case
the indifference price would be $199,000 per permit.2  If the
price of permits, therefore, is below $199,000, the firm will
choose to buy permits.  However, when the price is above $199,000,
the firm will choose to install control equipment and sell the
excess permits.

     The above calculations assume the parameter values for emission
control efficiency and control costs are known.

     If, for example, it is assumed that only 90% of the planned
emission Deduction was achieved and the PC costs could be 20%
higher than the estimate, the indifference price would now be
$264,000 per permit.3

     One can see the substantial increase in the indifference
price.  The implications fbr a marketable permits system is that
uncertainty will increase the asking price for permits.  This may
result in a potential reduction of the number of permits actually
exchanged between sources.
     3-Annual capital amortization, maintenance and operating cost
discounted at 10% in perpet'uity.
     2This price is calculated by setting the equation for the cost
of buying permits equal to the equation for the cost of installing
PC equipment and solving for the price.
     3The calculation  is as follows:
         41 P = (1.2)  X ($14,100,000)-P(61-38)
         P = $264,000


     Another  important factor to consider is the timing of the
cash flows.   If our example firm were to buy permits, it would
have to  initially spend possibly as much as $10,824,000.1 If
the firm were to install PC equipment, it would receive at least
$6 million for the sale of permits and take on a negative cash
flow liability with a net present value of $14,100,000.2 From a
cash flow standpoint, installation of PC equipment may look attrac-
tive to those firms that expect to sell permits as a result of the
PC expenditures.  In return, these firms would receive a large
positive cash flow initially, with negative cash flows in later
years.  The following section of this report considers a situation
involving a number of sources that would be included in a trading

     Exhibit  Bl presents a simplified air quality control region
with nine sources.  For each source, the current level of emissions
is presented  along with the emissions that would be controlled
with the installation of pollution control equipment.  Unit costs
and total costs are also presented.

     For analysis purposes, it is assumed that the following number
of permits are allocated free to each source in proportion to
current emission levels.

               Source                        Permits

                  1                             10
                  2                             10
                  3                             10
                  4                             15
                  5                             15
                  6                             20
                  7                             20
                  8                             25
                  9                             !3C>

                Total                          155

     An overall regional emission reduction of 31 units (186 current
units minus 155 units) will be required.  Each source will have to
either buy additional permits or install pollution control (PC)
equipment (and possibly sell permits).  Exhibit B2 illustrates the
initial decision environment along with the indifference price,
i.e., that permit price at which a source is indifferent as to
whether it buys permits or installs PC equipment.
           estimate is based on purchasing 41 permits at $264,000
per permit.
     2This estimate is based on selling 23 permits at $264,000 per

                          EXHIBIT Bl
              Simple Air Quality Control Region
           Current      Potential        Unit        Total
          Level of      Emission        Control      Control
Source    Emissions     Reductions       Costs        Costs

  1           12            4             $10          $40
  2           12            3               8           24
  3           12            5              15           75
  4           18            5               7           35
  5           18            7               4           28
  6           24            4              16           64
  7           24            5              10           50
  8           30            7               4           28
  9           _36           _2              _!           H

             186           47                         $407


     For the initial portion of this analysis, the following assump-
tions are made:

          permits are initially allocated free

          complete certainty regarding
               - PC costs
               - PC effectiveness

          no time lag is assumed between selling the permit and
          installing PC equipment

          only sources can buy and sell permits

          one binding standard is Assumed

     .    all sources generate equivalent emissions on a
          continuous basis

     As stated earlier, one objective of a marketable permits
system is to achieve economic efficiency in pollution control
costs.  In this simple model the least-cost solution would
initially involve PC equipment on sources 8, 5, 4, 2, 9 and 1 for
a total cost of $218.  This would result in controlling 33 emis-
sion units — two more than necessary.  This can be compared to a
straight rollback on all sources where the cost would be $407 and
the region overcontrolled by 16 emission units.

     The question is whether the least-cost solution (or any pro-
gress toward the least-cost solution) would be obtained if marketable
permits were allocated and traded.  Two approaches are considered.
The first involves trades at indifference prices.  The second involves
trades where permit prices reflect a built-in profit factor.  Several
trading scenarios are examined.  Cases 1, 2 and 3 contrast different
trading sequences.  These three cases illustrate that there is not a
unique solution to matching buyers and sellers.  The costs in Case 2
are equal to the least cost solution.  Cases 1 and 3 are alternative
solutions which result in higher cost.  Case 3 in particular could be
interpreted as a worst-case, first-come-first-served matching of buyers
and sellers.  Case 4 shows the impact on net costs of varying the per-
mit prices.  Case 5 shows the effect of trading at a profit as opposed
to trading at the indifference price.  Following a brief description
on each case the income transfer for all cases will be compared with
the least-cost solution.

                          EXHIBIT B2


                                                      Indifference  Price

                	buy  2
      	   $40 + sell  2           $10

      	buy 2

    \	   $24 + sell  1           $  8

      	buy 2

    %	   $75 + sell  3           $15

                                        buy 3

    "X	$35 + sell  2           $  7

                                        buy 2

    \	   $28 + sell  4           $  4

      	   buy 4

        	   $64                    $16

        	buy 4


Case 1 - Trade at indifference price, lowest-price source sells to
         highest-price source at the highest's price.

Selling Firm   # of Permits Sold   Buying Firm   Transaction Price

    5                  46            $16/unit
    8                  23            $15/unit
    4                  21            $10/unit

In addition, firms 2, 7 and 9 will have to install PC equipment.

     The total cost of PC control in this case equals $228.^  The
region would be overcontrolled by three units (total emissions after
control equal 152 units, but only 155 units were required).

Case 2 - Same as Case 1, but different matching of buyers and sellers.

Selling Firm   # of Permits Sold   Buying Firm   Transaction Price

    5                  46            $16/unit
    8                  23            $15/unit
    4                  27            $10/unit
    2                  17            $10/unit
    9                  17            $10/unit

In addition, source 1 will have to install PC equipment.

In this case the total cost of PC equals $218, the least-cost solu-
tion.  However, the difference between this case and the least-cost
solution is that under a permits system, the sources that do not
apply PC equipment compensate those sources that do apply controls.

Case 3 - Trade at closest price (e.g., $16/unit trade with $15/unit),
         sale at highest indifference price.

Selling Firm   # of Permits Sold   Buying Firm   Transaction Price

    3                  36             $!6/unit
    7                  16             $!6/unit
    9                  11             $10/unit
    2                  11             $10/unit
    8                  2                4             $ 7/unit
    5                  1                4             $ 7/unit

Total cost equals $268 and results in emission reductions of 34 units.

1 See Exhibit B3 for the cost to each source.


As can be seen, the total PC cost far exceeds the least-cost solution -
principally because high-cost PC equipment is being installed.  This
situation could arise if trading occurred on a first-come-first-served
basis.  However, efficiency gains will be obtained any time trades
Case 4 - Same as Case 1, but at a transaction price midway between
         buyer and seller.

Selling Firm   # of Permits Sold   Buying Firm   Transaction Price

    5                  4                6            $10.00/unit
    8                  23             $9.50/unit
    4                  21             $8.50 unit

The net pollution control costs are $228, the same as in Case 1, but
the income transfers are not as large.
Case 5 - Trading at a profit, i.e., sell permits at 20% higher, buy
         permits at 20% discount, sell at highest price.

Selling Firm   # of Permits Sold   Buying Firm   Transaction Price

    5                  46             $12.80/unit
    8                  23             $12.00/unit
The following firms must also install PC equipment:  1, 2, 4, 7, and 9.
Total cost is $269 and emission reduction would equal 39 units.  It can
be seen that this case reduces market activity in comparison to other
cases.  It also forces more PC investment than needed.

     Exhibit B3 shows the net cost for each source under the various
cases.  As can be seen from the exhibit, under marketable permit
systems, sources that have lower costs of pollution control are com-
pensated by sources that have higher pollution control costs.

                          EXHIBIT B3


$ 40
$ 40

Case 1
$ 20

Case 2
$ 40

Case 3
$ 20

Case 4
$ 17

Case 5
$ 40
Reduction  47
     The following observations and conclusions from analysis of the
five cases are apparent:

1.   Marketable permit solutions are not unique.  The least-cost
     solution is not always attained.  Costs may vary signif-
     icantly depending on which sources are involved in the
     transactions.  The final distribution of permits often
     depends on the order in which bids to buy and sell permits
     are transacted.

2.   Operation of a trading exchange to appropriately match buyers
     and sellers of permits is needed to obtain maximum cost reduc-
     tion.  One method to efficiently match buyers and sellers
     would be to allow permit trades to be executed only on specfic
     dates.  Prior to these dates, sources would provide the
     exchange with their prices to buy or sell permits.  On the
     given day, the exchange would execute all trades.  This would
     reduce the probability of high-cost sources buying permits for
     other high (but slightly lower) cost sources rather than from low-
     cost sources.

3.   In the case of a total market failure (i.e., no permit trades
     occur), the marketable permits approach would result in the
     rollback solution.  As long as some trades occur, a marketable
     permits approach will result in pollution control cost reductions,

1. Report No.
3. Recipient's Accession No.
4. Title and Subtitle
                     An Analysis of Economic Incentives to
        Control Emissions of Nitrogen  Oxides from
        Stationary Sources
                                                  5. Report Date
                                                    January, 1981
7. Author(s)
            Report of the EPA Administrator to Congress
                                                  8. Performing Organization Rept.
9. Performing Organization Name and Address
         Office of Planning and Management
         USEPA  ,
         401 M St.,  S.W.
                   m,  D.C. 20460	
                                                  10. Project/Task/Work Unit No.
                                                  11. Contract/Grant No.
12. Sponsoring Organization Name and Address
         Environmental Protection Agency
         401 M St., SW
         Washington,  D.C. 20460
                                                  13. Type of Report & Period
                                                  Report to Congress
 15. Supplementary Notes
 16. Abstracts
       Study analyzes alternative economic schemes for controlling airborne
 pollutants and reccnmends  under what conditions each scheme  is appropriate.
 Study also estimates the cost of attaining  a hypothetical short-term N02
 standard  in the Chicago region under various policy alternatives.
 17. Key Words and Document Analysis.  17a. Descriptors

    Ambient  standards                    Nitrogen oxides

    Economic incentives

    Marketable Permits

    Emission Fees

I7b. Identifiers/Open-Ended Terms

    see above
17c. COSATI Field/Group
18. Availability Statement

  on request
                                      19.. Security Class (This
                                      20. Security Class (This
           21. No  of Pages
           22. Price
                                THIS FORM MAY BE REPRODUCED
                                                                                   USCOMM-DC 82«S-PjM

    INSTRUCTIONS  FOR  COMPLETING FORM  NTIS-35         (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-73-87 and FAA-RD-73-09.

    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, subordinate subtitle to the
       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. F.ach 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, date published).

   6.  Performing Organization Code.  Leave blank.

   7.  Author(s).   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 Mailing 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 in-
       dexes such as  Government Reports Index (GRI).

   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 Name and  Mailing Address.  Include  zip code. Cite main sponsors.

   13.  Type of Report and Period Covered.  State  interim, final, etc., and, if applicable,  inclusive dates.

   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
       Cite availability of related parts, volumes, phases, etc. with report number.
   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 1964  COSATI Subject Category  List.
       Since the majority of documents are multidisciplinary in nature, the primary Field/Group assignment(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 public  releasability, for example "Release unlimited",  or limitation  for reasons other
       than security.  Cite any  availability to the public,  other than NTIS, with address,  order  number and price, if known.

   19 & 20. Security Classification.  Do not submit classified reports to the National Technical  Information Service.

   21.  Number of  Pages.  Insert the total number of pages, including introductory pages, but excluding distribution list, if any.

   22.  NTIS  Price. Leave blank.
FORM NTIS-3S (REV. 10-73)                                                                                  USCOMM-DC 826S-P7/I

                                        * U.S. GOVERNMENT PRINTING OFFICE : 1981 -341-085/4637