Draft June 2003
RLNRV3K'VXPPHUDFDGHP\
´6XVWDLQDELOLW\0DQDJHPHQW0DUNHWLQJDQG&RQVXPSWLRQµ
Heterogeneity of Marginal Abatement Cost, and Savings
from Application of Market Based Instruments for the
Industry Sector in the MCMA
Samudra Vijay
Technology, Management and Policy Program
Engineering Systems Division,
Massachusetts Institute of Technology
Cambridge
Abstract
This paper applies equimarginal principles to estimate the abatement cost of air
pollution by the Industry sector of the Mexico City Metropolitan Area. Theory and
literature posits that the savings by applying market-based instruments are dependent on
the degree of heterogeneity of the abatement cost, and on the magnitude of transaction
cost. The analysis assumes that the intra-sectoral heterogeneity of abatement cost is lower
than inter-sector heterogeneity of abatement cost, and a sectoral approach is used. The
emissions reduction costs from 13 different industrial sectors in the MCMA have been
calculated using three different abatement policies. The abatement cost data have been
taken from the World Bank’s pollution abatement costs and expenditure (PACE) survey.
It is argued that the cost structure of the abatement activity in the Mexico is similar to
that of the US, except the difference in the labor cost. In this static analysis (for emissions
for year 1998) results indicate that in the wake of heterogeneity in abatement cost,
substantial savings can be realized by adopting market-based instruments.
Please do not cite or forward without express permission of the author.
1. Introduction
In this paper, I estimate the cost implications of uniform abatement policies, and
savings from applying equimarginal principle, in achieving different abatement levels by
the polluters in such a way that their marginal abatement cost is equal. The analysis is
presented for a static case, for emission reduction for year 1998.
First, I discuss the motivation for this paper, then I outline the air pollution
problem in the Mexico City Metropolitan Area (MCMA), and analyze role of industrial
sources of air pollution in the Mexico City’s air pollution problem. In the next section, I
identify target air pollutants from their health impact perspective. Next section lists data
sources and some problems with data. In the next section, I outline methodology, and
then I present the results of my analysis. Final section discussed caveats and assumptions,
and ways to improve the analysis. An annotated bibliography is presented at the end, and
appendix contains a graphical representation of emissions inventory of the MCMA to
provide some additional background information.
2. Motivation, and Literature Survey
We know that if there are many polluters with different marginal abatement cost,
then the polluter who can achieve emission reduction in cheapest way should be asked to
reduce its emissions first. In theory this sounds right, but in practice it poses many
problems. First, there exists asymmetry of information between the regulator and
polluter, and it is almost impossible for regulator to find out exact marginal abatement
cost for a given firm. Second, even if the regulator had that information, implementing a
policy that calls for heterogeneous abatement would ruffle quite a few socio-political and
economic feathers. Third, it would be too expensive for the regulator to determine,
2
allocate and monitor emission for each polluter. Market based instruments, such as
tradable permits allow the regulator to achieve emission reduction at lowest possible cost,
thereby maximizing social welfare, although success of tradable permits depends on
creation of a competitive and efficient market for permits. Of late, market-based
instruments to achieve environmental goals have become more acceptable. But policy
makers need to know answer to certain questions before they could adopt market based
instruments. For example, if one tries to sell marketable permits to a politician, on basis
of efficiency, the first question s/he will ask is, “OK, how much will be the cost or what
will be the savings from market based instrument?” When policy maker needs to choose
between command and control approach, and market based approach, s/he would like to
know if the trouble is worth taking, i.e., if the savings from market based approach are
worth putting in the effort to create a market for tradable permits etc.
In this paper I estimate savings from implementing a market-based approach for
abatement of NOx pollution in the MCMA industrial sector.
Policy makers, especially in developing countries, do not have enough
information about the pollution caused by the different firms, let alone the cost of
abatement by different firms. World Bank developed an industrial pollution projection
system (IPPS) that helped in developing aggregate pollution intensities by different
industrial sectors (Martin, 1993). Martin (1993) has reviewed the IPPS and suggested
how IPPS could help policy maker in prioritizing the abatement policies for emission
reduction from different sector. Hartman, Wheeler and Singh (1994) further examined
pollution abatement cost expenditure (PACE) survey data for about 20,000 industries and
calculated abatement costs for different sectors. In case no information about the
3
pollution load and abatement cost of any industries is available, aforesaid tools could be
extremely helpful in prioritizing target sectors for pollution abatement. Although the
estimates provided by the IPPS system are very aggregate, the information regarding
abatement cost can be used in conjunction with actual emission data to design appropriate
policy response using market-based or command and control instruments.
Burtraw and Cannon (2000) have explicitly incorporated heterogeneity in
abatement costs in second-best policy settings for environmental protection. They have
used a computable general equilibrium framework, and conclude that due to
heterogeneity of abatement costs, a disaggregate representation of cost results in
qualitatively different findings about cost effectiveness of different policies. They have
used different marginal abatement cost functions for different sectors of the economy,
namely, transportation, industry, electric (coal), and electric (gas). They find that the
emission tax is most preferred instrument, and choice of policy instrument depends on the
percentage reduction from the base level sought. At 4% level of reduction, tradable
permits emerge as cheapest instrument as compared to other non-revenue generating
instruments such as performance standards or technology mandate.
Carlson et al. (2000) analyze marginal cost of abatement cost of a specific
pollutant, SO2, and find that, for 678 plants in the US electricity sector, the standard
deviation is three times that of mean. However, this is not clear if these numbers will
apply to other pollutants, such as NOx, or to other sources.
Newell and Stavins (2001) focus on analysis of the relationship between the
nature and magnitude of heterogeneity and prospective cost savings. They recognize need
for large amount of data to appropriately incorporate the heterogeneity of abatement cost
4
in policy analysis, and propose a rules-of-thumb that can be employed with smaller
amount of data and can be used to conduct initial screening of policy options for
environmental problems. They emphasize on the need for policy maker to be able to
assess potential cost-savings from the market-based instruments for a particular
environmental problem. Their model indicates that the cost savings of market-based
policies relative to uniform performance standards increase in proportion to the cost
heterogeneity, which, in their model results from two sources. First, is heterogeneity in
the baseline emissions intensities, and second is heterogeneity in the slope of the cost
function.
3. Background
The Mexico City Metropolitan Area (MCMA) is considered one of the worst
polluted urban areas in the world. Pollution levels in the MCMA have been reported to
often exceed the Mexican national ambient air quality standards. The six criteria air
pollutants recognized by the MCMA’s commission for air quality in the metropolitan
(CAM), are CO, NOX, PM10, SO2, Lead and ozone. The air quality standards and peak
concentrations of the pollutants are listed in Table 1.
Primarily, emissions from transportation sectors are understood to be responsible
for the severity of air pollution in the MCMA, but emissions from industrial sources are
also significant contributor to the poor air quality in the MCMA. Industrial sources of air
pollution in the MCMA contribute about 13.6 % of the annual NOx emissions, about
26% of the anthropogenic particulate matter of size smaller than 10 microns (PM10), and
about 50% of the total SO2 emissions. Emissions of CO mainly come from the
automobiles; hydrocarbons emissions from industry are also small as compared to the
5
emissions from the household and other area sources (CAM, 2001). A chart depicting
relative contribution of different sources of air pollution in the MCMA is presented in
Appendix.
Table. 1
Standards and peak annual concentrations of criteria pollutants in the MCMA
from five measurement stations
CO
(ppm)
11
(8 hr)
NOx
(ppm)
0.21
(1 hr)
Standard
(Duration)
1988
29.5
0.327
1991
15.9
0.370
1995
14.9
0.296
1997
9.8
0.274
1999
12.1
0.216
Source: INE, 2000
MCMA = Mexico City Metropolitan Area
SO2
(ppm)
0.13
(24 hr)
O3
(ppm)
0.11
(1 hr)
PM10
(µg/m3)
150
(24 hr)
0.183
0.192
0.081
0.099
0.094
0.405
0.404
0.349
0.309
0.311
241
324
184
4. Target Air Pollutants
NOx and PM10 are the chief target emissions for devising mitigation strategies
from the industrial sector due to their relatively large contribution, and their importance
from health impacts perspective; NOx being an important ozone precursor, and PM10
being associated with mortality and morbidity. NOx is also responsible for formation of
secondary particles, which are considered responsible for increase in mortality.
Combustion of fossil fuel, primarily for generating process heat and for other industrial
applications, produces thermal NOx. Fossil fuel combustion and industrial processes
result in emissions of PM. Industry is also a major contributor (54% of total
SO2emissions) to the SO2 emissions in the valley, but the SO2 emissions are primarily
dependent on the sulfur content of the fuel, and concentration is within norms, therefore
6
analysis presented in this paper is focused on emission of NOx from the industrial
sources.
Concentration of ground-level ozone is exceeded on a frequent basis in the
MCMA, but the tropospheric ozone is not emitted by mobile or stationary sources.
Formation of ozone is a result of complex photochemistry, and is affected by the atypical
topography of the MCMA, which lends it susceptible to thermal inversions and overnight
trapping of pollutants in the valley. Although there is uncertainty about exact role of
precursors, such as NOx and hydrocarbons; NOx and hydrocarbons are understood to
play important role in formation of ozone.
Therefore, in this paper I estimate the savings in abatement cost of NOx from
industrial sources.
5. The Industrial Sector in the MCMA
There are more then 35 thousand micro, small, medium and large industrial units
operating in the MCMA, in formal and informal sector (Molina and Molina, 2002). Only
about six thousand small, medium and large, of these industrial units are required to
report their annual schedule of operation. The industrial sector in the MCMA plays a
significant role in the regional and national economy. In year 2000, the manufacturing
sector contributed about 109 million (1993 Mexican Pesos) to the regional gross product,
which is about 23% of the gross regional product (Dodder, 2002). Industrial activity is
understood to be responsible for about half of the total fuel consumption in the region
(Bazan, 2000). Annual emissions from the industry sector are listed in Table 2. The
industrial sector in the MCMA is composed of diverse industrial activities, ranging from
mining and metallurgy to textile and food processing. The metropolitan environmental
7
Table 2
Annual Emissions from Industry
NOX
CO
SO2
PM10
HC
Tonne
17448
8101
12426
2956
23932
%
8.5
0.5
54.3
14.9
5.3
Source: CAM (2001)
Commission has categorized the industrial activities in 13 different sectors. A list of the
13 industrial categories, and their annual NOx and PM emissions is given in Table 3.
6. Data
Ideally, one would like to know the exact emissions from all the sources in
question, and their marginal abatement cost curves, to apply the equimarginal principle
and determine the exact abatement responsibility for each of the sources and calculate the
total abatement cost. However, it is almost impossible to exactly determine the marginal
abatement curve for each individual firm. Therefore, I use an alternative, sectoral
approach, to estimate abatement cost.
I obtained annual emissions data for the year 1998 from a database prepared by
the Metropolitan Environmental Commission (CAM) of the MCMA. The database
includes firm level emissions of about 6200 units located in the MCMA. The database is
a compilation of mandatory filing requirement, whereby each of the industry is required
to report its annual schedule of operation (cedula de operacion), which contains data
pertaining to physical quantity of output produced by the firm, fuel consumed, and annual
8
emissions of criteria pollutants, etc. There are several problems with the database, but for
purpose of this analysis, emission data for individual firm is taken at the face value.
Next, I categorize the 6200 sources into a 13-sector classification used by the
CAM. Annual NOx and PM10 emissions from the 13 sectors for year 1998 are given in
Table 3.
Table 3
NOx and PM10 Emissions from MCMA Industry Sectors (1998)
PM10
NOx
Industry Sector
(Tonnes)
%
(Tonnes)
Non-Metalic Minerals
4570
26.4
505
%
17.4
Metalic Products
4432
25.6
175
6.0
Printing Products
145
0.8
46
1.6
Food Industry
924
5.3
515
17.8
Different Consumption Products
129
0.7
73
2.5
Petrochemical
0
0.0
0
0.0
1335
7.7
415
14.3
109
0.6
61
2.1
Woods and byproducts
1066
6.2
216
7.5
Metalic Mining Products
513
3.0
249
8.6
MediumTermLife Products
624
3.6
120
4.1
Chemical Industry
Vegetal and animal Products
Textile Industry
1316
7.6
379
13.1
Long Term Products
2128
12.3
140
4.8
Total
Source: CAM, 2001
17291
2894
Since no abatement cost data is available for the 13-sectors in the MCMA, I
obtain abatement cost data from the US industry. The World Bank sponsored an effort to
determine exact cost of abatement for different industries. A pollution abatement cost
estimation (PACE) survey instrument was designed and data for more than 20000
industries in the US was collected in a period of over 5 years. An analysis of the data on
basis of international standard industrial classification (ISIC) codes is done by Hartman
9
and Wheeler (1994). The Hartman and Wheeler abatement cost data for various
pollutants is based on 4-digit industry codes, which does not exactly match with the
Mexican industry 13-sector classification. Therefore, I studied the Mexican industry
sector activity in details, and found a closest match possible in the Hartman and Wheeler
ISIC-code based sectoral abatement cost data. The sectoral abatement cost data is
presented in the Table 4.
Table 4
Average Sectoral Abatement Cost (1999USD/Tonne)
Industry Sector
NOx
Non-Metalic Minerals
PM10
2091
25
Metalic Products
585
435
Printing Products
392
538
Food Industry
291
109
Different Consumption Products
140
48
Petrochemical
75
416
Chemical Industry
61
269
140
48
Vegetal and animal Products
Woods and byproducts
48
60
Metalic Mining Products
585
435
MediumTermLife Products
1979
474
Textile Industry
1751
503
594
806
Average
672
Std. Deviation
754
Source: Adapted from Hartman and Wheeler (1994)
321
246
Long Term Products
The abatement cost data may not be an exact fit for the Mexican industry, because
there may be country-specific structural differences in the abatement cost, such as, labor,
which may be quite different for the US and Mexico. Also, the data collected for the US
industry is not recent, and advances in science and technology of air pollution control
may have reduced the cost of pollution control. However, if we assume that the
10
abatement cost has two major components, capital and labor, and treat the abatement
technology to be uniform in Mexico and in the US, then cheap labor in Mexico is likely
to cause an upward bias in the abatement cost data from using the PACE survey data for
the US. Similarly, technical change and innovation may also cause an upward bias.
However, the abatement cost data should give us a good upper bound of the cost of
abatement, thereby, the savings achieved can be said to be minimum savings that we can
expect from adopting a policy which would enable differential abatement such that the
marginal cost of abatement for the 13 sectors is equal.
7. Methodology
For a single pollutant, say, for NOx,
Let the total abatement of the pollutant required by the regulator be: QA
Let abatement responsibility of each sector be: Qj
Let sectoral marginal abatement cost be: MCj = αj Qj
Where αj is average abatement cost of sector j for a given pollutant.
For cost effective allocation of abatement responsibility, marginal abatement costs should
be equal, therefore,
MCj = MCk
j, k, such that j  k
Total abatement constraint:
∑ Q j ≥ QA
Subject to,
Qj ≤ E j , ∀ j
Where Ej is current emission levels of sector j
Total abatement cost can be estimated as:
Qj
∑ ∫ MC dQ
j
j
0
11
I use Microsoft Excel’s optimization routine solver® to solve the aforementioned
equations, subject to the appropriate constraints, and obtain abatement level of each
industry sector using equimarginal principle. Then I calculate total abatement cost by
calculating abatement cost for each sector, depending on the level of abatement, and
summing it for all the 13 sectors.
8. Results
Since the policy maker would not exactly know how much pollution to reduce
(the efficient level of pollution), I calculated abatement cost for NOx abatement goals
ranging from 5% reduction to about 40% reduction in steps of 5% increment. I calculated
cost of abatement for the following three cases –
a. The burden of abatement is distributed in a uniform manner among all the 13
sectors, i.e, for a given abatement goal of QA, each sector is responsible for
abatement of QA/13.
b. The burden of abatement is distributed such that each of the 13 sector reduces
a given percentage of emissions, i.e., if abatement goal is 10%, then each
sector reduces 10% emissions from their current levels.
c.
The burden of abatement is distributed on basis of equimarginal principle,
i.e., each of the 13 sectors reduce emissions such that their marginal
abatement cost is equal, and total abatement constraint is satisfied.
Table 5
Total Abatement Cost( million USD1999)
% NOx
Uniform %
Uniform
Reduction Equal MAC Reduction Reduction
5
5
78
251
10
19
311
1004
15
43
700
2259
20
77
1245
4016
25
121
1945
6276
30
174
2800
9037
35
12
236
3812
12300
40
309
4979
16066
Source: Calculated by the author.
Chart 1
Level of NOx abatement and abatement cots
Abat em e nt Cost (m illion USD1999)
18000
16000
Equal Marginal Abatement Cost
14000
Equal % Abatement
12000
Equal Abatement
10000
8000
6000
4000
2000
0
5
10
15
20
25
30
35
40
Pe rce ntage Abat e m e nt (NOx)
The abatement cost for the three different cases are given in Table 5, and the
results are also presented in the Chart. Distribution of abatement level for different
industries is given in the following chart 2.
13
Chart 2
Distribution of sectoral abatement for a total 50% abatement
Distribution of Sectoral Abatement Responsibility
(50% NOx abatement from Industry)
Long Term Products
Textile Industry
MediumTermLife Products
Metalic Mining Products
Woods and byproducts
Vegetal and animal Products
Chemical Industry
Petrochemical
Different Consumption Products
Food Industry
Printing Products
Metalic Products
Non-Metalic Minerals
0
10
20
30
40
50
60
70
80
90
100
% Abatement
It is obvious that for a given level of percentage reduction in pollution, substantial
cost savings can be achieved if abatement burden is distributed among polluters on basis
of equimarginal principle, as opposed to technology or uniform emission standards. For
25% reduction in NOx, the uniform percentage abatement cost is about 1.9 billion USD
(1999), whereas cost is only 121 million USD (1999) in case of equimarginal abatement.
14
Chart 3
Abatement Cost for NOx for MCMA Industry using equimarginal principle
Abate m ent Cos t (m illion USD1999)
350
300
250
200
150
100
50
0
5
10
15
20
25
30
35
40
Percentage Abatement (NOx)
9. Discussion
There are many assumptions and simplification in the analysis, some of which
have been discussed in the previous sections, and some of them are discussed here. First,
I am assuming that the target pollutants in question, NOx and PM, are uniformly mixed
pollutants, and its spatial and temporal distribution does not make any difference to the
policy maker. This assumption is not necessarily true, as NOx reacts with hydrocarbons
at various elevations in a different manner due to different amount of sunlight present.
Moreover, certain times of the day, NOx emissions are good; in a certain proportion, it
15
actually helps in dissociating the ozone. However, this phenomenon is too complex to be
modeled and incorporated in this analysis.
I have lumped industries of different sizes, producing different output, using
different inputs, and having spatially located at different places. In its reduced form, I am
essentially analyzing 13 different (assumed) big sources within an air shed, with different
marginal abatement costs. Although data from Hartman and Wheeler supports the
assumption that heterogeneity of abatement cost within a given sector is less than that
between different sectors, it certainly cannot be uniform across different industries in a
given sector. A small difference in input such as fuel type, say, natural gas and diesel,
would make huge difference in emissions from the two sources within a given sector,
therefore the abatement cost will also be different.
Functional form of marginal abatement cost that I have used is linear, which may
not be the case in reality. I have used the average abatement cost estimates as coefficient
for my marginal abatement cost function.
Per tonne cost of emission does not reflect the actual nature of investment in the
abatement technology by a firm, which is more likely to give a step function kind of
discontinuous abatement cost curve, then a smooth quadratic curve.
Retrofitting of the ISIC codes to the Mexican sectoral data may have caused some
error. The direction of the bias introduced thus may be difficult to know. A more refined
approach could have been to assume a distribution of the industries in a given sector, and
then calculate the abatement cost distribution from the average abatement cost and
standard deviation. The abatement cost data can also be obtained from other sources, to
16
break it into capital and labor component, and see how changing the labor component
would affect abatement costs for the MCMA.
Also, I have not discussed how the tradable permits or other instruments could be
implemented to achieve the equimarginal abatement from all sources. Initially allocation
of the permits could have significant distributional impacts and may be very important
from policy perspective.
Continuous emissions monitoring of SO2 emissions from utilities played a
significant role in the success of the emissions trading program in the US. The cost of
such emissions monitoring may act as a stumbling block for policy makers interested in
adopting market-based approach in the MCMA. Therefore such a cost should also be
taken into account while calculating savings from MBIs.
Emissions from other source categories, such as transportation, and households,
and their abatement cost, have not been considered.
The analysis presented here is static, and does not take into account the time
domain. A partial equilibrium analysis can be done which could include heterogeneity of
abatement cost and dynamically calculate abatement cost for each time period.
17
References
Bazan (2002). Balancia de energia de la Zona Metropolitana del Valle Mexico. Report submitted to the
workd bank (Unpublished).
Burtraw, Dallas and Matt Cannon. 2000. Heterogeneity in Cost and Seccond-Best Policies for
Environmental Protection. (Discussion paper) 00-20. Resources for Future, Washington DC.
Carlson, Curtis, Dallas Burtraw, Maureen Cropper, and Karen L Palmer. 2000. Sulfur Dioxide Control by
Electric Utilities: What Are the Gains from trade? Journal of Political Economy.
CAM (2001). Inventario de emisiones a la atmosfera de la Zona Metropolitana del Valle Mexico, available
at www.dbf.gob.mx/secretarias/sma/index.html, under publications, D.F., Mexico.
Dodder, R. (2002) Future stories and economic growth in the Mexico City Metropolitan Area.
(Unpublished), Mexico City Project Scenario Analysis Team, Cambridge, MA
EPA (1994). Alternative Control Techniques Document: NOx Emissions from
Industrial/Commercial/Institutional (ICI) Boilers. Emission Standards Division, US EPA, Research
Triangle Park, NC.
Economics of the Environment: Selected Readings (Stavins, Robert N., Fourth Edition). (2000). New York,
NY: W. W. Norton & Company.
Hartman, R.S., Wheeler, D., Singh, M. (Working Paper). (December 1994). The cost of Air Pollution
Abatement. In http://www.worldbank.org/nipr/work_paper/1398/index.htm: World Bank.
Kolstad, C.D. (2000). Environmental Economics. New York: Oxford University Press.
Molina, L. (2002). Air Quality in Mexico Megacity: An Integrated Assessment (Luisa T. Molina; Mario J.
Molina, Science and Technology: Tools for Sustainable Development). Boston, MA: Kluwer Academic
Publishers.
Newell, R. G., & Stavins, R. N. (Working Paper). (December 2001). Cost Heterogeneity and the Potential
Savingsfrom Market-Based Policies Resources for Future, Washington, DC.
Wheeler, D. (Policy Paper). (January 1992). The economics of industrial pollution control: An international
perspective. In http://www/worldbank.org/nipr/work_paper/wheeler92/index.htm: World Bank
18
19
Appendix
Source: Connors, 2002, based on data from CAM (2001)
20