								4/8/2009 

Technical Memo:  Comparison of Analytical Tools

Used to Evaluate the Cement NESHAP Proposal

For the proposed Portland cement NESHAP, the EPA utilizes three models
to evaluate the impacts of the regulation on the industry and the
economy.  The models include:  the Control Cost Model from the
Engineering Analysis, the Economic Impact Analysis model, and the
Industrial Sector Integrated Solutions (ISIS) model.  

Typically in a regulatory analysis, EPA determines the regulatory
options suitable to meet statutory obligations under the Clean Air Act. 
Based on the stringency of those options, EPA then determines the
control technologies and monitoring requirements that may be selected to
comply with the regulation.  This is conducted in an Engineering
Analysis.  The selected control technologies and monitoring requirements
are then evaluated in a cost model to determine the total annualized
control costs.  The annualized control costs serve as inputs to an
Economic Impact Analysis model that evaluates the impacts of those costs
on the industry and society as a whole.  

For the Portland cement NESHAP, EPA also employs the ISIS model which
conducts both the engineering cost analysis and the economic analysis in
a single modeling system.  The ISIS Model is a dynamic and integrated
model that simulates potential decisions made in the cement industry to
meet an environmental policy under a regulatory scenario.  ISIS
simultaneously estimates 1) optimal industry operation to meet the
demand and emission reduction requirements, 2) the suite of control
technologies needed to meet the emission limit, 3) the engineering cost
of controls, and 4) economic response of the policy, in an iterative
loop until the system achieves the optimal solution.  The objective
function in ISIS minimizes the cost of production and any emission
control requirements over the time horizon of interest, while meeting
projected demand for applicable commodities and emissions constraints. 
ISIS has been designed to analyze the impacts of control strategies upon
the cement manufacturing industry and has the capability to estimate the
change in cost of production within the cement industry resulting from
implementation of a regulation, determining control technology
installations as well as estimating price and output changes based on
the emission limits determined by the regulatory option under
consideration.  While the Engineering Analysis and Economic Impact
Analysis models evaluate a snapshot of implementation of the proposed
rule in a given year (i.e., 2005), ISIS evaluates impacts of compliance
dynamically over time (i.e., 2013-2018).  

The costs of the Engineering Analysis are estimated based on the type of
control device that is assumed necessary for a source to comply with the
proposed emission standards. Based on a kiln’s baseline emissions of
mercury, Total Hydrocarbon, Hydrogen Chloride and Particulate Matter,
and the removal efficiency necessary to comply with the proposed
emission limit for each Hazardous Air Pollutant (HAP), an appropriate
control device, or control devices, is identified. In assigning control
devices to each kiln where more than one control device would reduce
emissions of a particular HAP below the limit, it is assumed that the
least costly control would be installed. However, in some instances, a
more expensive technology is considered appropriate because the selected
control reduced emissions of multiple pollutants. For many kilns, this
analysis assumes that multiple controls will have to be added because
more than one control will be needed to control all HAP.  This approach
and the resulting cost estimates are considered conservative in that it
is not possible to consider what other alternatives, in lieu of multiple
add-on controls, may be available to cement plants to comply with the
proposed emission limits. Because individual cement manufacturing plants
may implement process or other changes to reduce emissions, the costs of
the proposed amendments likely overstate the costs that actually will be
incurred by individual cement plants.  

The Cement Economic Impact Analysis (EIA) model uses a single-period
static partial-equilibrium model to compare a pre-policy cement market
baseline with expected post-policy outcomes in cement markets. This
model was used in previous EPA analyses of the Portland cement industry
(EPA, 1998  XE “U.S. EPA, 1998”  ; EPA, 1999b  XE “U.S. EPA,
1999b”  ).  The time horizon for the analysis is for the intermediate
run when producers have some constraints on their flexibility to adjust
factors of production. This time horizon allows us to capture important
transitory impacts of the program on existing producers. The model uses
traditional engineering costs analysis as “exogenous” inputs (i.e.
determined outside of the economic model) and computes the associated
economic impacts of the proposed regulation.

This memo provides a comparison of parameters in these models to provide
an evaluation of how the differences between the models may impact the
results presented in the supporting documentation to the regulation.   

Engineering Analysis

In table 1, we present the comparison of the Engineering Analysis and
the Cost Module of ISIS.  With regard to data inputs, both models used
the same control technologies inputs such as control efficiency, and
installation and operation cost, as well as used the same 2005 database
of kilns.  A key difference regarding the two models is that while the
Engineering Analysis assumes the existing kiln population in 2005 to be
constant for its analysis, ISIS takes into account change in operation,
e.g., closure, expansion, retirements, and new units that operate more
efficiently and thus emit lower levels of pollution.  Another difference
between these two models resides in the way the total cost of controls
is calculated.  While, the Engineering Analysis identify the specific
controls required to meet the proposed emissions limits for each kiln,
and then developed capital and annualized control costs, ISIS uses a
dynamic optimization approach to select control technologies to meet the
emission limits at each kiln in a manner that considers cost,
co-pollutant reductions, and economics of operation, simultaneously. 
For example, ISIS allows for optimization at the kiln and plant level. 
If one plant operates two kilns, ISIS might determine that it is more
cost effective to decrease production at one kiln to a level that meets
the emission limit, and increase production at the second kiln while
applying control technology.  In this example, the Engineering Analysis
would apply controls at both kilns, while ISIS applies controls at only
one kiln.  Overall, because of these differences, and the iterative
approach applied in ISIS, we would expect the total control costs from
ISIS to be lower than that of the Engineering Analysis.

Table 1: Comparison of the Analytical Models: Engineering Cost Model
Dimension

Parameter	Engineering Analysis	ISIS

Kiln Population

  -Existing

  -New	

Kilns in existence in 2005

No retirements, expansions or replacements

Calculates a 5 year estimate based on historical production and growth. 
Assumes each new kiln capacity is 1.2 million tons. Total new kilns is
20.	

Kilns expected to exist in 2013 based on reported closures and
expansions between 2005 and 2013

Model allows for retirements, expansions, and replacements, when
optimizing production costs in a given year.

New capacity is estimated based on industry growth and operational
economics.  Model determines total new kilns.  New kiln capacity is 1.2
million tons each. 

Control technology application	Determined by engineering judgment
assuming operation of the kiln constant at a 2005 levels	Endogenously
determined in the model such that total production cost to meet the
demand and emission reduction requirement is minimized. 

Cost of Control Technology	Based on empirical cost data, EPA Control
Cost Manual, published data, Manufacturer Data and emission test data
when available	Based on empirical cost data, EPA Control Cost Manual,
published data, Manufacturer Data and emission test data when available

Total Control Cost	Summation of capital and annualized costs for each
control technology applied to the kilns comes from EPA reports such as
EPA Air Pollution Control Cost Manual, Alternative Control Technology
(ACT) reports, and others

Costs for existing sources are estimated separately from new sources.

Total Control Costs are summed and amortized using a 7% discount rate
Uses same technology costs as the Engineering Analysis for calculating
total control costs. However, the control cost also takes into account
the operation of the kiln. Further, the model allows for fuel-switching,
which could affect flue gas volume and control costs.

Costs on new and existing sources are estimated simultaneously.

Total Control Costs are summed and the net present value is calculated
using a 7% discount rate

Major vs. Area Source Kilns	Known area sources not subject to emission
limits are factored into engineering cost estimates	Will allow
unaffected area sources to increase production to meet demand.

Has the capability to apply differential regulatory requirements on
major point and area sources which can affect supply curve in a given
market. 

Emission reductions	Baseline emissions at each kiln/facility are
multiplied by the control efficiency measures for each pollutant reduced
by the selected control technology.

Emission reductions are based of emissions in 2005 and do not reflect
capacity changes that occurred after 2005.	Begins with baseline
emissions in 2005 and considers growth and capacity changes out to the
year of promulgation of the rule (2013)

Estimates ex-ante and ex post emissions of pollutants under
consideration. 

Evaluates collateral impacts, such as reductions (or increases) in
criteria pollutants or greenhouse gas emissions as a result of
policy-induced control applications.



Economic Impact Analysis

In table 2, we present the comparison of the Cement EIA Model and the
Economic Module of the ISIS Model. With regard to the characterization
of the cement industry, both models treat the cement industry as
regional in nature due to the costs of transporting the commodity across
long distances (i.e., cost prohibitive conditions creates regional
markets), with each kiln modeled in one of 20 cement markets. A key
difference between the models is that the EIA models cement regions as
independent markets, treating them as isolated regions, while ISIS
allows for inter-market regional trade with cost of transportation
dependant on distance which serves to limit the amount of trading
between markets.  The EIA’s model of cement regions and independent
markets is consistent with review of recent economic literature on
cement market behavior undertaken prior to the running of this model for
this proposed regulation and documented in the RIA.  The inter-market
regional trading that takes place within ISIS accounts for the
overlapping and interdependent market behavior reported by the Portland
Cement Association (PCA) (APCA, 1997).  At the same time, the
inter-regional trading characteristic of ISIS offers flexibility that
can result in lower cement prices when compared with the EIA analysis. 
Thus, the EIA model characterizes the cement industry as an oligopoly
while the ISIS model characterizes the industry as perfectly competitive
with restrictions on the level of trade via the transportation cost
function.  Because the oligopoly structure limits production to smaller
markets, cement prices are projected to be higher in the EIA model than
in ISIS where firms take the market price as given.  

Regarding imports, both models characterize imports as a perfect
substitute for domestically produced cement, but while the EIA model
treats imports as elastic, ISIS treats imports as perfectly elastic but
with constraints on the maximum quantity of imports in a given region. 
The difference in the treatment of imports has an indeterminate impact
on prices in ISIS when compared with the EIA cement prices.  

Table 2: Comparison of Analytical Models:  Economic Model Dimension

Characteristic	Cement EIA Model	ISIS Model

Direct  Compliance Costs

(before market adjustments) 	Assigned to kiln based on Engineering
Analysis	Choice of control technology determined within the model 

Markets 	20 markets	20 markets

Cement Supply & Demand	Single period 

Baseline year: 2005

USGS Data

Supply & Demand adjust with cement price changes	Multi-period 

Baseline Year:2005 

Model Horizon: 2013-2018

 

USGS combined with PCA future year projections

Demand adjusts with cement price changes

Imports	Perfect substitution

USGS custom district data

Elastic = 2.4.

	Perfect substitution 

USGS custom district data

Perfectly elastic constrained at baseline of historical levels for 2005.

(i.e., capped at 27% of total demand)

Seller behavior	Cournot multi-firm oligopoly 

With a small number of firms in each region, firm’s recognize their
production decision influences the market price

Seller behavior implies there is a pre-existing distortion in the market

(e.g. price is not equal to marginal costs)

	Perfect competition with limitations due to transportation costs

 

Firms take the market price as given

Assumes no pre-existing distortion in the market

(e.g. price is equal marginal costs)



Energy Use	No Fuel Switching	Fuel Switching is permitted, which gives
additional flexibility to a firm to meet the emission requirements

Closures	No closures modeled.  If profits are negative, the plant is
considered to be in danger of closing.	A kiln is retired if in the
beginning of a given year and throughout the policy horizon capacity
utilization rate is zero

Regional Market Interactions	Each regional market is independent of the
other with no cross boundary trades	Allows for inter-market trading. 
Transportation matrix includes a step function of increasing cost, with
cost constant in each step, as distance increases. 

Total Regulatory Cost	Measure =

Social cost

 Social cost is estimated based on traditional social welfare economics
(changes in producer and consumer surplus) in reaction to the direct
compliance costs.	Measure = Direct Compliance Costs  + total cost to
meet the demand (cost of production, imports and inter-market
transportation)





Further details on how economic impacts are estimated by these models
can be found in the Regulatory Impact Analysis (RIA) and the ISIS
Technical Support Document (TSD) for the proposed NESHAP.   In addition,
EPA conducted a peer review of ISIS, which can be found in the docket to
the Cement Proposal. 

Appendix A

PARTIAL EQUILIBRIUM ANALYSIS:  COMPARIson oF MODEL FEATURES

EPA considered the proposed rule’s economic effects using partial
equilibrium analysis; the approach is highlighted as an important
assessment tool in EPA’s Guidelines for Preparing Economic Analyses
(EPA, 2000). However, partial equilibrium models can be implemented in a
variety of ways. As a result, it is important to discuss key model
features when presenting a range of estimated economic effects.   

Since the Economic Impact Model and the ISIS model put partial
equilibrium analysis into practice in different ways, this appendix
provides additional information about two model features that can help
key stakeholders better understand and interpret the analysis. First, we
discuss how the two approaches define a cement company’s “best”
production decision given the market definition and the existing number
of plants producing cement in these markets. Second, we discuss
differences in the ways the approaches allow regional markets to
interact.

A.1	Cement Production Decision and Marginal Analysis

Both economic models assume that the cement company acts in the best
interest of its shareholders and maximizes profits. When deciding
whether to make another ton of cement, the company considers the
production effect on profits by comparing the current market price of
cement and the marginal cost; if price is above marginal cost, producing
and selling the extra ton of cement increases profit. The company
continues to produce additional cement until the profit from producing
an extra ton of cement is zero (price equals marginal cost) or capacity
constraints are reached. The ISIS model uses this decision rule and is
consistent with the assumption of pure competition. 

Now consider an alternative view used in the Economic Impact Model that
examines the possible implications of a different market structure; one
that has few companies but sells similar or identical products. In this
case, the cement company continues to consider the production effect
described above but adds another dimension to the decision-making
process; it considers the market price effect that is associated with
producing an additional ton of cement. Given the small number of cement
producers, adding an extra ton of cement to the market may lower the
market price of cement and reduce the profits on all the other cement
sold. If the price effect is large enough, it may be more profitable for
the company to reduce production below the levels implied by pure
competition. As the number of producers expands, the cement company
would become less concerned about the market price effect, and the
production decision becomes more consistent with decisions made in pure
competition.  

A.1.1	Market-Level Effects 

What are the consequences of these differences as they relate to
measures of the projected market-level price and quantity changes? To
provide some intuition about the size of the differences, we present a
simple market analysis with two kilns operated by different companies.
Assume the marginal cost of cement production for each plant is $98 and
$100 per metric ton. The market demand for cement is met by both kilns;
the least expensive kiln operates at full capacity, and the highest cost
kiln operates at 85% capacity. The baseline cement price is equivalent
to the high-cost kiln ($100).  

Three examples show the mechanics of how the pure competition model
projects market price changes in response to the regulation. The market
price changes are driven by underlying differences in the baseline kiln
production costs and the distribution of incremental compliance costs
across kilns.

Example 1: Both kilns face an additional $1 per ton in compliance costs.
The marginal cost of production for each kiln is now $99 and $101. In
equilibrium, the new market price is $101 (1% increase). Assuming a
demand elasticity of −1, the quantity demanded with regulation falls
by 1%. 

Example 2: Next, only the least-cost kiln faces an additional $1 per ton
in compliance costs. The marginal cost of production for each kiln is
now $99 and $100. In equilibrium, there is no change in the market price
($100), and there is no change in the quantity demanded.

Example 3: Now the least-cost kiln faces an additional $3 per ton in
compliance costs. The marginal cost of production for each kiln is now
$101 and $100. In equilibrium, the new market price is $101 (1%
increase). Assuming a demand elasticity of −1, the quantity demanded
with regulation falls by 1%. Note the similarity with the market
outcomes in Example 1.  

Now consider the same market in the Economic Impact Model. As discussed
earlier, the company considers the market price effect of the production
decision. For example, when making the decision to raise production by
one ton of cement, the change in marginal revenue (dMRi) must equal the
change in the marginal cost of producing cement:

ΔMarginal Revenue = ΔMarginal Cost

where the change in marginal revenue depends on the market price effects
brought on by producing an additional ton of cement. To estimate price
and quantity changes in this framework, we used the model equations
described in Appendix C of the RIA.

Example 1: Both kilns face an additional $1 per ton in compliance costs.
In equilibrium, the new market price is $102 (2% increase). Assuming a
demand elasticity of −1, the quantity demanded with regulation falls
by 2%.  

Example 2: Next, only the least-cost kiln faces an additional $1 per ton
in compliance costs. In equilibrium, the new market price is $101 (1%
increase). Assuming a demand elasticity of −1, the quantity demanded
with regulation falls by 1%.  

Example 3: The least-cost kiln faces an additional $3 per ton in
compliance costs. In equilibrium, the new market price is $103 (3%
increase). Assuming a demand elasticity of −1, the quantity demanded
with regulation falls by 3%.  

Table A-1 compares the projected market-level changes in these examples.
As expected, the projected relative price and quantity changes are
higher under the multi-firm Cournot oligopoly model.   

Table A-1.	Market-Level Effects Comparisons

Type	Baseline Marginal Cost ($/metric ton)	Incremental Compliance Cost
($/metric ton)	Price Change 

(Percent)	Quantity Change

−1.0%	−2.0%

Kiln 2	$100	$1





Example 2

Kiln 1	$98	$1	0.0%	1.0%	0.0%	−1.0%

Kiln 2	$100	$0





Example 3

Kiln 1	$98	$3	1.0%	3.0%	−1.0%	−3.0%

Kiln 2	$100	$0







A.2	Cement Transportation and Trade between U.S. Cement Regions

 

5

Q

h

h

♀怀愀Ĥ摧璋

h

h

h‹t

h

 h‹t

h

\

$

\

愀Ĥ摧ẦÍ

\

愀Ĥ摧ẦÍ

\

\

愀Ĥ摧ẦÍ

\

愀Ĥ摧ẦÍ

\

愀Ĥ摧ẦÍ

\

h‹t

h‹t

愀Ĥ摧ẦÍ

\

\

 h·

 hf

摧绡 Ò欀๤

摧绡 D欀ꁤ

h	b

h	b

 h

Impact Model and the ISIS model use the same 20 regional cement markets,
the Economic Impact Model assumes that transportation costs between
regions are high enough that interregional trade is unlikely to occur,
at least in the short run. In contrast, the ISIS model adds some
additional flexibility in the model that allows cement plants to ship
cement to other markets if the additional benefits of selling cement to
these markets outweigh the costs of transporting the cement outside the
regional market. Details of transportation cost assumptions of this ISIS
module are described in the ISIS technical documentation. 

A.2.1	Market-Level Effects 

What are the consequences of the trade assumptions as they relate to
measures of the projected market-level price and quantity changes?
Allowing trade in the Economic Impact Model would expand the cement
market definitions and increase the number of producers. As discussed
above, as the number of producers in a market increases, the production
decision becomes more consistent with decisions made in pure
competition. The extent to which trade across regions is feasible would
be driven by the assumptions about transportation costs. The examples in
Table A-1 show that increasing the number of producers participating in
the market could moderate projected price increases and subsequent
consumption changes.  

 ISIS characterization of seller behavior takes into account PCA
comments that producers “rather than restricting

output to control price they seek to produce whatever amount of cement
their customers demand at the prevailing

price levels, arranging for imported cement when domestic production
capacity is insufficient to meet demand”

(APCA, 1997).

 This model is formally known as a multi-firm Cournot Oligopoly model.

 The equation is formally discussed in Appendix A and Appendix B of the
RIA.

 PAGE   

 PAGE   12 

