SAVINGS FROM THE APPLICATION OF TRADING AND
AVERAGING TO HEAVY DUTY ENGINE REGULATION
Prepared for:
U.S. Environmental Protection Agency
August 25, 1986
SOBOTKA & COMPANY, INC.
2501 M Street, N.W.
Suite 550
Washington, D.C. 20037
(202)887-0290
-------
TABLE OF CONTENTS
I. INTRODUCTION
A. Heavy Duty Engine Regulation and Emissions Averaging 1
B. Issues Discussed In this
Report: Trading and Averaging of NOx and PM 1
C. Methods Used to Evaluate Savings from Averaging and Trading . . 2
D. Assumptions and Limitations of the Study 3
E. Summary of the Results 6
II. CALCULATION OF COSTS
A. Summary of Types of Costs 8
B. Fuel Effects of Meeting NOx Standards 8
C. Trap Effects 11
III. SCENARIOS CONSIDERED
A. Regulatory Scenarios 15
IV. HOW AVERAGING AND TRADING GENERATE SAVINGS
A. An Example in Which Averaging" Would Yield Cost Savings ... 19
B. The Limits to Savings 20
C. Prorating to Ensure Emissions Do Not Rise 20
D. Prices of Credits and Their
Relationship to the Allocation of Savings 22
E. Illustration of Savings Relationships 23
-------
TABLE OF CONTENTS (continued)
V. RESULTS
A. Cost Comparisons by Scenario 27
B. Comparisons of Emissions Levels 33
C. Costs per Ton of Emissions Reductions 33
D. Savings per Vehicle 36
VI. SENSITIVITY ANALYSES
A. Savings Due to Averaging and Trading In
Comparison to a Baseline with Emissions Equal to the Standard . . 41
B. Reclasslflcatlon of Light-Heavy-Duty Engines 49
C. Sensitivity of Results to Assumed Functional Relationships . . 57
D. Banking 59
APPENDICES
A. Optimization
B. Background Material from ERC
C. Input Data for Computation of Savings from Regulatory Flexibility
D. Detailed Results
-------
I. INTRODUCTION
A. Heavy-Duty Engine Regulation and Emissions Averaging
On March 15, 1985 EPA promulgated regulations which significantly tighten
NOx emissions standards for heavy-duty engines (HDEs) and emissions of partic-
ulate matter (PM) for heavy-duty diesel engines (HDDEs). The NOx standard for
the 1988-90 model year HDEs is 6.0 g/BHP-hr, with a more stringent standard of
5.0 g/BHP-hr effective for 1991 and later model year engines. Model year 1988-
90 HDDEs will be required to meet a PM standard of 0.60 g/BHP-hr. In 1991, all
HDDEs except urban bus engines must meet a 0.25 g/BHP-hr standard. Urban
buses, which are excluded from averaging, are not included in this study.
Standards for NOx and PM for 1988 are intended to be met by each individual
engine family. NOx and PM limits for 1991, on the other hand, may be met by
groups of engines: some engine families may have emissions in excess of the
standards so long as the group meets the standards on average. Engines will be
considered to be part of the same group, or "averageable set," if they are
built by the same firm and fall within the same subclass. The four subclasses
envisioned are heavy-duty gasoline engines (HDGE), light-heavy-duty diesel
engines (LHDDE), medium-heavy-duty diesel engines (MHDDE), and heavy-heavy-duty
diesel engines (HHDDE). This added flexibility, termed "emissions averaging,"
is intended to allow cost reductions for the manufacturers with no compromise
in meeting the air quality goals of the regulations.
B. Issues Discussed in this Report: Trading and Averaging of NOx and PM
The purpose of this report is to quantify the implications of carrying
this concept of averaging further. It shows the annual savings that could be
realized if additional flexibility is permitted in meeting the same overall
emissions targets. The industry as a whole could be allowed to meet the stan-
dards "on average" by permitting the inter-firm trading of "credits" fo
-------
-2-
emlssions below the standards. This system, termed "emissions trading," extends
the intrafirm benefits of averaging by enlarging the averageable sets. Trading
increases the total savings while at the same time offering increased equity
across firms (since not all firms would be able to take advantage of the benefits
of averaging alone to the same degree).
This report also explores the cost-saving potential of allowing averaging
or trading across broad sets of engine types, instead of within narrow sub-
classes. Comparisons are made among three subclass assumptions. The first
assumption is that HDE emissions averaging/trading is "restricted" to the four
subclasses described above, with averaging or trading allowed within each class
but not between them. The second assumption, referred to as "partially restric-
ted" averaging or trading, involves averaging/trading among the three MODE
subclasses, but not between the two HDE classes (MODE and HDGE). The third
assumption, "unrestricted" averaging and trading, permits averaging/trading
among the three HODE subclasses and between the two HDE classes (HDDE and
HDGE).
C. Methods Used to Evaluate Savings from Averaging and Trading
Cost savings under flexible regulations—averaging or trading—are calcu-
lated by estimating the total costs of compliance with regulations traditionally
formulated on a command-and-control (every engine family must pass) basis and
comparing these costs to the costs of meeting the same emissions goal in the
most efficient way allowed by more flexible rules. The most efficient way to
meet a given emissions goal is found by applying the economic principle that
the greatest efficiency in production is reached only if the marginal costs of
production are the same for all production units. (In this case, the "product"
is emissions reduction, and the "production units" are heavy-duty engines.) In
practice, achieving equal marginal costs of pollution control among engine
classes means that some engines should emit more than average, while other
engines balance these excess emissions by emitting less than average. Engine
types that have emissions below the standard accrue credits. The number o
-------
-3-
credlts that are thus created is proportional to the number of grams per brake-
horsepower hour (g/BHP-hr) below the standard. Where averaging or trading is
permitted across engine classes/subclasses, to ensure that total (and to aid in
ensuring localized) emissions do not exceed the levels emitted if each vehicle
met the standards individually, account must be taken of differing characteris-
tics of each engine class or subclass. Hence, in such cases, emissions are not
necessarily averaged or traded on a one-for-one basis; they are "prorated"
according to sales, miles traveled per year, years of useful life, and/or power
output per mile. (An example of of the prorationing concept is presented in
Section IV.)
Marginal costs of emissions reductions are calculated based on functional
relationships (cost/emission curves) relating NOx emissions levels, PM emissions
levels, and percentage increases in fuel consumption. The functions were
developed from estimates of the technical relationships made by a specialist in
HDE emissions.^./ Individual functional relationships were developed for each
of eight engine type and usage classifications (see Appendix B for a discussion
of the classification system), including both gasoline and diesel types. An
optimizing algorithm is used to find the combination of emissions levels across
different types of vehicles that equalizes marginal costs. (See Appendix A for
an explanation of the optimization program.)
The distribution of the savings from trading for individual firms is
estimated in this study on the basis of the savings realized by different
engine type and usage classes, and individual firms' projected shares in the
sales of those classes. That is, a firm with a large share of the market for
those engine types that benefit substantially will gain relatively more.
0. Assumptions and Limitations of the Study
Costs of meeting the target levels of only the NOx and PM regulations
effective in 1991 are estimated in this study. Target levels are set at 80
JV See Section II.B. and Appendix B. Data supplied by C. Weaver of ER
-------
-4-
percent of the standards to take into account production variation in emissions
or the normal in-use deterioration of emissions performance over time, and thus
to include a. cushion to allow for engine-to-engine variability. Given this
assumption regarding maintenance of design cushions, the "effective standards"
become 4.2 g/BHP-hr for NOx and 0.22 g/BHP-hr PM. Gasoline engines are included
in NOx averaging and trading only; diesel engines are included in both NOx and
PM averaging and trading. Trucks to be sold in California, which has separate
standards, were not incorporated in this study.
The analyses in this report consider emissions from an engine only up to
the first rebuilding. Under some circumstances, allowing averaging between
engines with different likelihoods of being rebuilt could have unintended
effects on emissions. For instance, an emissions trade may allow a one g/BHP-hr
increase in NOx emissions for a LDDE-DI to be offset by a one g/BHP-hr decrease
in a LHDGE if each were assumed to travel the same number of miles before being
rebuilt and exert the same number of BHP-hrs per mile. This trade would not
result in any emissions increase if neither engine is rebuilt. If, however,
the LHDDE-DI j_s rebuilt and the LHDGE is not, the excess emissions of the
LHDDE-DI after the rebuild will not be offset and overall emissions will rise.
This study is intended to be indicative of the kinds and general magni-
tudes of percentage savings that would be seen with more flexible regulations,
but estimates of total emissions control costs, cost savings, or the distribution
of those savings over the various firms cannot be made with a great degree of
precision. The main reason for this limitation is that technological and
market forecasting for points in time as far away as 1991 is necessarily an
inexact undertaking. If sales in the industry are much greater, of course,
total costs and total savings will also be greater. Also, because the analysis
is driven by assumptions about market shares held by different types of vehicles
and different firms, the results will be sensitive to unforeseeable shifts (and
past experiences have shown such shifts to be quite common).
The recognition that exact predictions of savings are impossible led to
the decision to limit the number of factors considered to those considere
-------
-5-
most Important. Left out of the analysis are examinations of how differences
in defining useful lives of the vehicles and some of the less important catego-
ries of emissions control costs (including research, testing, and hardware
changes to meet NOx standards) affect the results.
The analysis is limited to examining the maximum savings possible due to
the added flexibility of emissions averaging and trading. The degree to which
the firms would take advantage of the program to approach the maximum savings
is not considered, though it is expected that the use of the system would be
extensive: firms have an economic incentive to trade credits to lower their
costs.
The study also considers cost savings for only one future year, meaning
that issues of spreading emissions over time in the most efficient manner
(banking) are not addressed, except in a sensitivity analysis in Chapter VI.
Banking would increase cost savings potential.
During the course of this study, every effort was made to use cost, sales,
technology, market share, and other information which was completely consistent
with EPA's Regulatory Impact Analysis (RIA) for the 1991 HDE NOx and Particulate
Emissions Standards (see Appendix C). 'However, in several areas this was not
possible because the data or other information needed for this study was not
addressed directly in the RIA. Thus, to meet the needs of this study, some
information from other sources was used and in some cases additional analyses
were necessary. This in no way detracts from the results of this study, however,
since the point here is to evaluate the incremental cost savings of different
averaging and trading programs over a given base case. As long as the input is
consistent for the base case and the averaging and trading cases, the percentage
cost savings estimated are accurate. It is also reasonable to assume that
percentage savings found here reflect those which would occur if this analysis
had been incorporated directly into EPA's RIA.
Also, it should be noted that due to the different input parameter values
used here, the baseline costs for the 1991 standards are somewhat higher tha
-------
-6-
those found in EPA's RIA. However, EPA still supports the analysis presented
in its RIA as the best estimate of the costs of compliance.
E. Summary of the Results
A summary of the results, presented in Exhibit I, shows that averaging of
NOx emissions and of PM across all HDEs (the least restrictive averaging sce-
nario) could save over $191 million per year, or 19 percent of the industry's
emissions control costs in the absence of averaging. When the industry is
divided into two subclasses, HDGEs and HDDEs, emissions averaging within those
groups saves $158 million, almost 16 percent of industry control costs. These
savings may be compared to the savings of $123 million (12% of baseline costs)
that result from averaging emissions under the current averaging regulations,
which divides engines into four subclasses (HDGE, LHDDE, MHDDE, and HHDDE).
(Estimates of cost savings were not previously calculated in the EPA FRM of
March 15, 1985.)
Trading can save an additional $107 million, or 13 percent of industry-
wide control costs, in the least restrictive scenario, over and above the
savings from averaging. A substantial portion of the savings would go to the
firms with lower gains under averaging without trading. Trading not only
results in a near-doubling of the cost savings under averaging alone but also
widens the set of firms participating in the gains from increased regulatory
flexibility. Trading thus helps to keep differential effects on the firms to a
minimum.
Trading saves less compared to averaging i.f more restrictive subclass
assumptions are made. If HDEs are divided into two subclasses, trading can
save an additional $9 million compared to averaging; if four subclasses are
assumed, trading would save $8 million. Detailed descriptions of the scenarios
and the savings are included in Section III and Appendix D.
-------
-7-
Exhibit I
EMISSIONS CONTROL COSTS
(millions of dollars per year)
Baseline - No Averaging or Trading
Four Subclass Averaging *
Two Subclass Averaging **
One Class Averaging ***
Control Costs
$ 1,009.6
$ 886.5
851.4
818.1
Savings
Four Subclass Averaging and Trading * $ 878.7
Two Subclass Averaging and Trading ** 842.4
One Class Averaging and Trading *** 710.9
vs. Baseline:
$ 123.1 12.2%
158.2 15.7
191.5 19.0
vs. Averaging:
$ 7.8 0.9%
9.0 1.1
107.2 13.1
* Restricted
** Partially restricted
*** Unrestricted
-------
-8-
II. CALCULATION OF COSTS
A. Summary of Types of Costs
The costs Imposed by the regulations that are considered in this report
fall into the following categories: 1) costs due to fuel consumption increases
resulting from NOx-reduction techniques, 2) costs due to fuel consumption in-
creases caused by PM-reducing trap oxidizers ("traps"), and 3) initial and
maintenance costs of traps. These are not the only costs imposed by the regula-
tions, but they are by far the most important, and are covered in more detail
below.
B. Fuel Effects of Meeting NOx Standards
Using conventional control technology, reducing NOx emissions from heavy-
duty truck engines* is expected to result in increases in fuel consumption. The
percentage increase will vary with the NOx emissions level, increasing sharply
as emissions per BHP-hr approach 4.0 grams. The size of the increase at any
given emissions level also varies considerably across different types of engines
and depends heavily on the control technique used. Exhibit II-l shows point
estimates of the tradeoff between NOx and fuel consumption increases at various
NOx levels (targets, not standards) as provided by ERC Inc. (EPA's RIA analysis
was not adequately detailed to be used here)V. The estimates are shown for 9
types of engines types, which are described in Appendix B. For computational
convenience, and to obtain estimates of the consumption increases at intermedi-
ate NOx levels, smooth hyperbolic functions were fit to the point estimates.
These functions are shown in Exhibit II-2.
Translating these percentage fuel consumption increases in dollar values
(assumed to be imposed on the manufacturers by market forces—though the average
]_/ Estimates for gasoline engines assume that 3-way catalysts, which have
not yet been shown to be suitable for heavy-duty use, will not be used'
-------
-9-
Exhibit II-l
POINT ESTIMATES OF NOx/FUEL CONSUMPTION TRADEOFFS
ENGINE*/USAGE TYPE
1. Light-Heavy
Indirect Injection
2. Light-Heavy
Direct Injection
3. Standard
Medium-Heavy
4. Premium
Medium-Heavy
5. Line-Haul
6. Vocational
Heavy-Heavy
7. Light-Heavy
Gasoline
8. Medium-Heavy
Gasoline
NOx TARGET (g/BHP-hr)
2.5 3.0 3.5 4.0 4.5 5.0 6.0 8.0
FUEL CONSUMPTION INCREASES
15% 8% 2% 0% 0% 0%
12% 6% 1% 0%
16% 7% 3% 0%
12% 6% 1% 0%
8% 4% 0.5% 0%
10% 5% 1% 0%
6.5% 5% 2.5% 1% 0% 0%
6.5% 5% 2.5% 1% 0% 0%
* All are diesel engines, except 7 and 8.
-------
NOx/FUEL CONSUMPTION FUNCTIONS
z
z
o
u
o:
o
u
o
x
-< m
-o i—
m
PD <-> m
oo o x
o z =r
i— co -••
i—i cz cr
o 3 -••
-O c-r
o •-«
"Z. I
ro
CD 70
TARGET G.'BHF-HR OF IJOx
-------
-11-
cost increase for all competing engines may be passed through) was accomplished
by multiplying them by EPA estimates of the discounted cost of fuel for the
engines per percent increase in fuel consumption.^/ Fuel costs were discounted
to allow for the fact that truck purchasers are less affected by costs that
will not be incurred for several years. These values are shown in Appendix C.
C. Trap Effects
Two cost impacts of using traps to control PM emissions may be identified.
First, the traps themselves are expensive to buy and to replace if and when
they wear out. Costs for initial and replacement trap oxidizer systems were
provided by EPA£/ for different engine/use types.
Second, the use of the traps increases fuel consumption by roughly 0.5 to
1 percent.2.7 This impact is translated into dollars in the same way that NOx-
related fuel consumption increases are treated.
Under the baseline scenario, all diesel engines would need traps. Under
averaging, however, only some engines would need to be fitted with traps since
those with traps will be averaged in with those without traps. To calculate
the total costs of the regulations, then, it is necessary to estimate the
fraction of engines needing traps. This depends on the efficiency of the trap,
the PM standard, and the engines' PM emissions levels without traps (called
"engine-out" emissions).
2.7 These fuel consumption functions were not used directly; EPA estimates
were used: Regulatory Impact Analysis, Oxides of* Nitrogen Pollutant Specific
Study and Summary amir Analysis of Comments, Control of Air Pollution from New'
Motor Vehicles and New Motor Vehicle Engines: Gaseous Emission Regulations for
1987 and Later Model Year Light-Duty Vehicles, and for 1988 and Later Model
Year Light-Duty Trucks and Heavy-Duty Engine s; P a rt i cu1 ateTmTssion Regulations
7or~lg8B and Later ModTPYear Heavy^Duty Diesel Engines. USEPA Uffice of Air
and Radiation, Office of Mobile Sources, March 1985.
£/ See Appendix C.
2/ See Appendix C.
-------
-12-
Traps have been assumed to be 80 percent efficient, trapping 80 percent of
engine-out PM emissions.^/ Engine-out PM emissions are expected to vary with
NOx emissions-and engine types in a manner similar to the relationship of NOx
emissions to fuel consumption increases: Exhibits II-3 and II-4 show the point
estimates provided by ERC and the curves developed by Sobotka & Co., Inc.
(SCI) for various NOx levels.
Using this information, and given a NOx and PM standard, it is straight-
forward to calculate the percentage of traps needed for a single type of engine
to meet the standard individually. For example, if a NOx standard is such that
engine-out PM is 0.5 g/BHP-hr, and the PM standard is 0.22, then (0.5-0.22)/
(80% * 0.5), or 70% of engines will need traps.£/ Not all engine types need to
meet the PM standard exactly, though, since the averageable set includes various
engines made by a firm. Firms are expected to lower their total costs by
placing the traps where they are most cost-effective: on engines where the tons
of PM removed by the traps are large compared to the costs imposed by the
traps. A simple algorithm is used in the analysis to "remove" traps on more
and more engines, starting with those on which the traps are least cost-effective
until the firm's average PM emissions just meet the standard. Appendix A
provides detail on this procedure.
Total regulatory costs for the baseline scenario are found by summing NOx
costs and adding the costs imposed by the traps to get the costs per truck.
These costs are then multiplied by the projected sales of the type of engine/use
class considered and by the share that the manufacturer is projected to have of
that engine/use class by 1991 (from EPA; see Appendix C).
}_l Regulatory Imapact Analysis, March 1985, op. cit.
£/ The RIA (p. 2-65) shows that 70% would need traps with engine-out
emissions of 0.42 to 0.54 g/BHP-hr and a target of 0.25; this is consistent
with the text.
-------
-13-
Exhiblt II-3
POINT ESTIMATES OF ENGINE-OUT PM
EMISSIONS ASSOCIATED WITH NOx EMISSIONS
ENGINE*/USAGE TYPE
1. Light-Heavy
Indirect Injection
2. Light-Heavy
Direct Injection
3. Standard
Medium-Heavy
4. Premium
Medium-Heavy
5. Line-Haul
6. Vocational
Heavy-Heavy
NOx TARGET (g/BHP-hr)
2.5 3.0 3.5 4.0 4.5 5.0 6.0
8.0
ENGINE-OUT PM EMISSIONS (g/BHP-hr)
0.60 0.52
0.46
0.45
0.65
0.75
0.58
0.45
0.54
0.50
0.60
0.44
0.37
0.40
0.34 0.30
0.44 0.40
0.32 0.28
0.28 0.25
0.30 0.27
* All are diesel engines.
-------
NOx/PM TRADEOFF FUNCTIONS
o:
I
a.
tn
a:
o
TARGET G/BHP-HR OF NOx
-------
-15-
III. SCENARIOS CONSIDERED
A. Regulatory Scenarios
Costs are computed and compared under seven scenarios: a baseline scenario
In which each Individual engine must meet the standards; three scenarios that
allow averaging; and three more that allow trading in addition to averaging.
(Trading may be thought of as averaging across manufacturers. Because more
than one manufacturer is involved, the system of accounting for emissions
increases and decreases is more complicated, but the concept is identical.
This concept is described in Section IV'.) The scenarios differ in terms of the
restrictions or boundaries of their averageable sets, i.e., in terms of the
groups of engines which must meet the standards on average.
The differences among the scenarios are illustrated in Exhibit III with a
brief explanation of each. The exhibit shows an industry comprised of two
firms, A and B, each of which markets eight families of HDEs: two HDGEs (Al,
A2, Bl, and B2); two LHDDEs (A3, A4, B3, and B4); two MHDDEs (AS, A6, B5, and
B6); and two HHDDEs (A7, A8, B7, and B8). The engine families distributed in
each box represent averageable sets: in a given scenario, emissions from one
engine family may be averaged with emissions from any engine family in the same
box, but not with emissions from engines in other boxes.
-------
-16-
Exhlblt III
REGULATORY SCENARIOS
Baseline Scenario
Firm: A
No Averaging or Trading
Subclass:
HDGE LHDOE MHDDE HHDDE
Engine Al
A2
Bl
B2
A3
A4
-B3
B4
A5
A6
B5
B6
A7
A8
B7
B8
In the baseline scenario no averaging or trading is permitted, and emissions from
each individual engine family are required to comply with standards.
Four Subclass Averaging
Firm: A
B
Subclass:
HDGE LHDDE MHDDE HHDDE
Engine "Kl
A2
Bl
B2
A3
A4
B3
B4
A5
A6
B5
B6
A7
A8
B7
B8
In this scenario, which closely resembles current regulations, the averageable
sets are large enough to include different engine families built by the same
manufacturer and falling into the same heavy-duty subclass. For instance, A3
and A4, which are both LHDDEs built by Firm A, are shown in the same box to
indicate that their emissions may be averaged together. Emissions from A3,
however, could not be averaged in with emissions from B3 (a LHDDE built by the
other firm), or from A5 (an engine built by the same firm but in a different
class).
Two Subclass Averaging
Firm: A
Subclass:
HDGE LHDDE
Engine Al
A2
Bl
B2
A3
A4
B3
B4
MHDDE
Ab
A6
B5
B6
HHDDE
A7
A8
B7
B8
This scenario widens the averageable set to allow emissions from any diesels
produced by a given firm to be averaged in with emissions from any other of
that firm's diesels. For example, engine A3 now falls in the same box, or
averageable set, with HHDDE A8.
-------
-17-
Exhibit III (continued)
One Class Averaging
Firm:
Subclass:
HDGE
A
B
Engine Al
A2
Bl
B2
LHDDE
A3
A4
B3
B4
MHDDE
A5
A6
B5
B6
HHDDE
A7
A8
B7
B8
The final averaging scenario widens the averageable set still further to include
all of a firm's engines, whether they are diesel or gasoline fueled, small or
large. (An important exception is that gasoline engines, which do not emit
significant amounts of PM, are not included in PM averaging.) In the exhibit,
all eight
other.
of A's engine families are in the same box and all of B's are in the
Four Subclass Averaging and Trading
Firm: A
Subclass:
HDGE LHDDE MHDDE HHDDE
Engine Al
A2
Bl
B2
A3
A4
B3
B4
A5
A6
B5
B6
A7
A8
B7
B8
This scenario permits averaging of emissions across firms, which is referred to
as trading. The same subclass restrictions apply in this scenario as in the
first averaging scenario, however: gasoline fueled engines are in a separate
averageable set from LHDDEs, which in turn are separate from MHDDEs, and so
forth. In the exhibit, emissions from engines A3 and A4 may now be averaged
with emissions from B3, but not with Al, Bl, or A8.
Two Subclass Averaging and Trading
Firm: A
Subclass:
HDGE LHDDE
Engine Al
A2
Bl
B2
A3
A4
B3
B4
MHDDE
A5
A6
B5
B6
HHDDE
A7
A8
B7
B8
In this scenario, the restriction on averaging different heavy-duty subclasses
of diesels is relaxed; all HDDEs produced by any firm are included in the same
averageable set. The exhibit shows all engine families in one of two boxes:
all gasoline engines are in one box, and all diesels are in the other.
-------
-18-
Exhibit III (continued)
One Class Averaging and Trading
Firm:
Subclass:
HDGE
A
B
Engine Al
A2
Bl
B2
LHDDE
A3
A4
B3
B4
MHDDE
A5
A6
B5
B6
HHDDE
A7
A8
B7
B8
The least restrictive scenario includes all HDEs in the same averageable set
(though, as with one class averaging, gasoline engines are not included in
particulate matter averaging).
-------
-19-
IV. HOW AVERAGING AND TRADING GENERATE SAVINGS
A. An Example in Which Averaging Would Yield Cost Savings
Allowing a given emissions target to be met with averaging or trading can
reduce costs because it allows emissions control activities that are relatively
inexpensive to be substituted for those that are relatively expensive. As a
simple example, consider a manufacturer whose total sales are one HHDDE and one
standard MHDDE. The total costs of meeting the same NOx standards may be
$3,000 for the HHDDE and $1,000 for the MHDDE, for a total of $4,000. The fact
that the total costs are higher for the larger truck does not by itself mean
that emissions controls are necessarily more costly per kilogram of NOx removed,
since $3000 spent cutting emissions in a large, intensively used truck might
easily reduce total emissions by just as much as $3000 spent on three smaller
trucks. The important consideration in determining if savings are possible is
not total costs but the change in total costs occurring for given changes in
emissions.
In other words, suppose that removing an additional ton of NOx by tighten-
ing the controls on the HHDDE would raise the total costs of control for this
truck to $3010, and that increasing emissions from the MHDDE by one ton would
reduce costs for that truck to $980. The same total emissions would take
place, but with one ton of emissions "transferred" from the HHDDE to the MHDDE.
By reallocating the emissions controls in this way, the manufacturer could
ensure that the same emissions would occur, but at a total cost of $3990.
In this example, we would say that the marginal costs of emission control
were $10 per additional ton for the' HHDDE and $20 per additional ton for the
MHDDE, since these are the dollar changes in total costs that result from one
ton changes in emissions. It is this difference between the marginal costs of
emissions control that allow savings to be generated: the marginal savings per
ton of emissions transferred from one truck to the other is equal to the dif-
ference between their marginal costs: $10 per ton transferred from the MHDD
-------
-20-
to the HHDDE. In terms of per engine costs, the costs for the HHDDE have risen
by $10 and the costs for the MHDDE have fallen by $20.
B. The Limits to Savings
It pays to keep transferring emissions from one engine to the other so
long as the difference between the marginal costs remains. As emissions are
reduced more and more from the HHDDE, however, the marginal cost for reducing
emissions from it will tend to rise. The opposite will happen to the marginal
cost of emissions reductions for the MHDDE as the controls are eased on it.
The gap between their marginal costs will therefore narrow, and eventually
disappear. At that point, the transferring should stop, since all of the
potential gains will have been squeezed out, and the maximum savings will have
been realized. The process of transferring to save control costs is called
"averaging" because of the requirement that per-vehicle emissions standards
still be met "on average" after the transferring. If a one ton increase is
always balanced by a one ton decrease, then average emissions for the set of
vehicles covered--the "averageable set"--are constant. The same is true with
respect to trading, i.e., the tons balance out—only in this case, the average-
able set is the overall industry.
C. Prorating to Ensure Emissions Do Not Rise
In an averaging or trading program, emissions from one engine class are
allowed to increase so long as emissions from another class are reduced. To
ensure that total emissions do not rise as a result, the program must provide
for an appropriate system for testing whether the decreased emissions from one
class are sufficient to cancel out the increased emissions from the other.
This might not be a simple matter. In the example above, emissions changes
are described in one ton increments. However, regulations are generally written
not in terms of tons of emissions over the life of a vehicle, but in terms o
-------
-21-
grams per brake horsepower-hour (g/BHP-hr). Lowering the emissions of Engine
Class B by 0.2 g/BHP-hr is not necessarily enough to compensate for an increase
in Engine Class A's emissions of the same 0.2 g/BHP-hr. The two changes will
cancel out if the same number of horsepower hours will be used by trucks using
Engine Class B as will be used by trucks using Engine Class A. This circumstance
is unlikely, and if Class A is installed in more trucks, or generates more
horsepower per mile in use, or travels more miles per year, or typically has a
longer useful life, then a given reduction in emissions by Engine Class B will
not make up for the increase from Class A. A greater number of tons of emissions
will result over the life of the engines than if each class had been held exactly
to the standard.
A partially restricted or an unrestricted averaging or trading program
could essentially ensure that emissions would not increase by prorating the
reductions and increases in certification levels to take sales, power, use pat-
terns, and length of useful lives into account. The reduction of 0.2 g/BHP-hr
for Class B in the example above should be multiplied by a factor proportional
to the total number of horsepower-hours expected to be expended by the engines
of Class B before comparing the reductions to similarly-prorated increases in
emissions of engines in Class A. The following example illustrates how this
would work.
Class A
Sales per year: 1,000
BHP-hrs/mile: 0.25
Miles per year: 20,000
Years of useful life: 9
Total BHP-hrs: 45,000,000
Total metric tons of
emissions from a change of
0.2 g/BHP-hr: 90
Total metric tons of
emissions from a change of
1.0 g/BHP-hr: 450
Class B
500
0.20
15,000
6
9,000,000
18
90
-------
-22-
Because each year the Class A fleet would generate five times the horsepower
hours of the Class B fleet, a change in Class A's certification level would have
to be balanced by an opposite change in Class B's level that was five times
greater. A 0.2 g/BHP-hr emissions increase for Class A would be fully offset
by a 1.0 g/BHP-hr decrease in emissions for Class B; then a 90 ton increase
would be offset by a 90 ton decrease.
The regulatory scenarios discussed in this paper are assumed to employ a
prorating system, identical to that described above, to ensure that total
emissions are the same with and without averaging or trading. In the operation
of the method for calculating cost savings, all changes in emissions levels are
converted into tons over the life of a year's sales fleet before being compared,
or weighted by the sales fleet's total horsepower hours before being averaged.
Traded credits are measured and priced in terms of tons of total emissions over
the life of the sales fleet.
D. Prices of Credits and Their Relationship to the Allocation of Savings
The example given above is simple partly because the whole "averageable
set" of engines is assumed to be built by the same manufacturer. Thus, the
fact that the total emissions control costs rise for one of the engines is not
important; the manufacturer is more than compensated by the lower costs of the
other. If the averageable set cuts across boundaries between firms, however,
as it does in "trading" proposals, new issues are raised. The same principles
of savings and their relationship to differences in marginal costs apply if
engines are built by different companies, but the problem of compensating the
manufacturer of the engine whose total control costs are increased arises.
This problem is taken care of by awarding credits (prorated, as described
above) for each g/BHP-hr below the standard to a manufacturer whose engines
emit less than the maximum allowed. These credits may be sold to firms in
complementary positions—those increasing emissions and reducing cpsts. The
purchasing firms would be required to show that they had obtained a credit fo
-------
-23-
each unit of emissions their products would emit over the allowed emissions for
each vehicle.
The establishment of provisions for awarding the credits and for allowing
them to substitute for emissions reductions creates a small but potentially
efficient market for trading of credits. Market forces will tend to force the
prices of the credits toward the marginal cost of generating them (and their
marginal value in substituting for emissions reductions). Assuming trades
occur when it is in the participants' economic self-interest to do so, i.e.,
they trade on the basis of their marginal costs and each manufacturer makes use
of the market to its greatest extent, then all manufacturers' products will
(after all the trades are completed) show the same marginal cost of emissions
reductions.
Market participants will benefit more from the market for credits the
greater the difference between their marginal costs and the industry-wide
marginal cost. Those with costs that are much higher than average realize
large advantages both because the average savings they realize per kilogram
traded are large and because they will find it useful to buy many kilograms'
worth of credits. Those with costs much lower than average will gain just as
much--generating and selling a larger number of credits at a large average
profit. Those with marginal costs close to the industry's marginal costs
before trading will gain little by trading, but they will not lose anything
either. This is illustrated in the following section.
E. Illustration of Sav-ings Relationships
Exhibit IV illustrates the relationship between costs relative to ipdus-
try-wide marginal costs and the cost savings available through trading. Each
of the three panels shows a marginal cost curve representative of industry-wide
marginal costs of emissions reductions (per ton), in addition to a marginal cost
curve for one of three different firms: A, B, and C. Before trading, each of
the three firms must meet the same emissions standard, and are assumed t
-------
-24-
Exhibit IV
TRADING: SAVINGS AS RELATED TO MARGINAL COSTS
$/ton
MCa -
MC,
MCa.
Marginal Cost - firm a
Marginal Cose - industry
Emissions Reduction - Firm A
$/ton
Marginal Cost - firm b
Marginal Cost - industry
Emissions Reduction - Firm B
$/ton
MC -
Marginal Cost - industry
MC,
Marginal Cost - firm c -**' . MCC
ER,,
Emissions Reduction - Firm C
-------
-25-
achieve that standard by setting emissions reductions at ERa, ER^, and ERC. At
these levels of emissions reductions, the firms' marginal costs of emissions
reductions (MCa» MC-b, and MCC) vary widely with respect to the industry average
(MCi): B's cost for removing additional tons is just slightly above the average,
but A's is much higher and C's is much lower.
If trading of emissions reduction credits is allowed, C finds it worthwhile
to increase its emissions reduction, generating credits that may be sold to A
and (to a lesser degree) to B. These credit purchases allow A and B to reduce
their emissions reductions, saving enough in costs to pay for the credits and
leave a net saving.
The savings of the various firms after trading are illustrated under the
assumption that the credit price (per ton of emissions reduction) reaches
equilibrium at the industry average marginal cost before trading. Each firm
will set its emissions reduction levels so that their marginal costs under
trading equal the price of credits. (Emissions reduction levels are shown as
ERa', ERb', and ERC'; marginal costs are MCa' etc.) Excluding its payments for
the credits, A's costs fall by the integral of its marginal costs from ERa to
ERa'—the area of the trapezoid ERaMCaMCa'ERa'. Payments for the credits equal
the price (MCi) times the quantity (ERa minus ERa'), which leaves A's net
savings as the area of the shaded triangle.
The shaded triangle representing B's savings is much smaller--the tri-
angle's base (the number of credits purchased) and height (the net savings on
the marginal ton of emissions reductions) are both proportionately smaller than
for A's. Thus, the net savings fall rapidly as a firm's marginal cost before
trading approaches the industry-wide marginal cost.
C's situation is the reverse of that of A's, but it leads to equivalent
net savings. C's costs actually rise as it increases its emissions reduction
levels, but the credits it generates in the process may be sold at a price that
yields an overall gain. This gain is shown as the shaded triangle in the
third panel of the diagram: the triangle's dimensions are again equal to the
-------
-26-
number of units of credits traded (ERC* - ERC) and the difference between
the firm's marginal cost before and after trading (MCC - MCC'). Net savings
resulting fro/n trading are, as shown by the diagrams, independent of whether a
firm has relatively high or relatively low marginal emissions control costs.
The savings are, however, much greater for firms whose costs differ sharply in
either direction from the industry as a whole.
-------
-27-
V. RESULTS
The first part of this section presents the emissions control costs for
individual firms and the industry for each of the regulatory scenarios. First,
the control costs and savings of NOx and PM regulations are compared for all
scenarios. This is followed by a presentation of the differences between the
scenarios in emissions levels, marginal costs of emissions control, and per-
engine costs and savings.
As discussed in Section III, the costs are estimated by combining the
costs of fuel consumption increases due to NOx control and traps, and the
hardware and maintenance costs of traps. The total costs are higher than would
be calculated under EPA's assumptions, due to differences in the treatment of
fuel efficiency gains over time. The savings from averaging and trading,
though, are consistent with EPA assumptions.
A. Cost Comparisons by Scenario
Exhibit V-l shows that substantial savings, both in absolute terms and as
a percent of total emissions control costs, would be realized by the industry
under the One Class Averaging scenario.^./ The industry would save almost $192
million per year, 19 percent of baseline control costs, if emissions averaging
}j The dollar figures for the savings from averaging and trading listed
for each firm do not, it should be remembered, refer only to manufacturing cost
savings to the firms. Most of the savings are fuel savings that accrue not
directly to the firms but to the purchasers of the engines. Thus, the term
"savings" is really a short-hand way to say "savings directly to the firms plus
increases in the value of the firms' products." In large measure, the increased
product values resulting from the fuel savings will be captured by the firms—
just as a firm with a less efficient product is forced to reduce its price to
stay competitive, a firm with a more efficient product is able to increase its
price without losing sales.
-------
Exhibit V-l
COST SAVINGS
ONE CLASS - AVERAGING AND TRADING
(dollars In millions)
Firm
1 Bluebird
2 Chrysler *
3 Ford
4 CumraLns
5 Caterpillar
6 Daimler-Benz
7 KHD
8 General Motors
9 Navistar
10 Hlno Motors
11 Deere
12 Mack
13 Onan
14 Perkins
15 Renault
16 Saab
17 Isuzu
18 Iveco
19 Volvo
20 White
Baseline
100Z traps
No Averaging Averaging
$0 0
3
41
382
139
8
1
201
84
T
0
91
0
0
12
1
12
5
a
a
90
06
97
44
46
31
96
94
49
90
53
41
90
97
08
33
56
27
11
SO
3
33
329
124
7
1
131
59
3
0
79
0
0
11
1
9
5
7
7
0
9
9
5
1
6
1
4
6
2
8
8
3
8
4
0
9
2
3
1
Dollar and Percentage Savings from
, Trading Averaging
Trading (vs Averaging) (vs No Averaging)
(SO
-0
2
302
114
6
1
113
56
3
0
73
0
0
10
0
9
4
6
6
0)
8
0
4
0
9
1
2
1
0
7
2
2
7
4
9
3
a
7
4
(S)
$0 0
4 7
32 0
27 1
10 1
0 7
0 1
18 2
3 5
0 3
0 1
6 6
0 1
0 1
1 1
0 1
0.6
0 4
0 7
0 7
U)
128
121
94
8
8
9
8
13
5
8
10
8
47
10
9
10
6
7
9
9
a
6
2
2
1
6
5
a
a
i
0
3
4
0
3
0
3
9
2
6
(S)
$0 0
0 0
7 1
53 4
15 3
0 8
0 2
70 6
25 4
0 3
0 1
11 8
0 1
0 1
1 5
0 1
2 4
0 4
0 9
1 0
(X)
0 0
0 0
17 4
14 0
11 0
9 /
12 3
34 9
29 9
7 4
9 9
12 8
29 1
9 9
11 8
9 9
19 5
7 2
11 2
12 /
I
ro
CO
INDUSTRY TOTAL
$1,009 6
$818 1
S/10 9 $107.1
13 1 $191
19 0
• The negative sign for Chrysler Indicates that for this firm the value of traded credits would
exceed the cost of generating them enough to more than offset the costs oi compliance with
the regulations
-------
-29-
were allowed.^./ Still further savings—over $107 million—would accompany the
introduction of trading (the One Class Averaging and Trading scenario), espe-
cially aiding-those firms with less ability to take advantage of the averaging
program. Moreover, trading "evens out" the distribution of savings that occurs
with averaging to more closely represent the original distribution of costs
incurred.
The gains from averaging would be spread quite unequally in both absolute
and percentage terms. Among domestic firms, General Motors, Navistar, and Ford
would reap the highest percentage savings. Bluebird and Chrysler, on the other
hand, would save almost nothing through averaging. However, no company loses
through averaging.
This variation in savings relates directly to the market segments in which
the different firms participate. Those with broad market coverage—General
Motors is the prime example, with LHDDEs, MHDDEs, HHDDEs, plus HDGEs—are able
to take full advantage of an averaging scheme. Those whose product lines are
much narrower—with engines that differ little in emissions characteristics—are
less fortunate: the marginal costs for emissions reductions in their different
engines are similar or identical even without averaging, so the potential to
lower costs by averaging is very limited.
Incremental gains from trading in addition to averaging are also distrib-
uted unequally: some firms show substantial gains while others are barely
better off. (Every firm is at least no worse off under trading since it can
choose not to trade credits at all, leaving it in exactly the same position
as with averaging without trading.) As mentioned above, for many firms the
distribution of large and small gains from trading counterbalances in some
important ways the distribution for" averaging by itself: a firm with one of
the largest gains under averaging, Navistar, gains only slightly more under
*/ About half of these savings result from the fact that, under averaging,
manufacturers are not forced to over-control for PM, installing expensive
traps that generate emissions reductions in excess of those required to mee
the emission standards.
-------
-30-
trading. And Chrysler, one of the firms with the smallest average gains with-
out trading, gains significantly under trading.
Again, the reason for this pattern lies in the breadth of the firms'
product lines: those firms with product lines that come close to matching the
industry's sales mix are able to bring the marginal emissions control costs for
all of their products close to the industry-wide average without trading with
other firms, and consequently are faced with very little scope for further
savings once they are allowed to trade with other firms. Conversely, a firm
like Chrysler with only one type of engine--and one whose emissions charac-
teristics differ sharply from those of the rest of the industry—finds great
opportunities to save from trading.
As shown in Exhibit V-2, when averaging is permitted within two industry
subclasses (partially restricted averaging) most firms' emissions control costs
are the same as with one-class averaging. Ford and General Motors, as the only
manufacturers of both HDDEs and HDGEs are the only ones which save less by
averaging under this subclass assumption.
For most firms, averaging and trading within two subclasses results in
significantly lower savings than those resulting from trading within one class.
Navistar and Onan are the only manufacturers with larger savings with trading
(as compared to only averaging) in this scenario than in the previous one.
However, in absolute terms the savings are slight; percentage increases are
partly due to the relative magnitudes of control costs under averaging compared
to trading in both subclass scenarios.
Exhibit V-3 shows that if (restricted) averaging, limited to four industry
subclasses, were allowed, firms manufacturing more than one type of MODE would
save less than in the two subclass scenario. The two firms saving the most
with industry-wide averaging, Navistar and General Motors, gain the least with
four subclasses. As expected, there is no change in the savings realized by
firms with only HDGE or only one type of HDDE.
-------
Exhibit V-2
COST SAVINGS
TWO SUBCLASSES - AVERAGING AND TRADING
(dollars In millions)
Dollar and Percentage Savings from
Engine Type*
C
C
C D
D
D
D
D
C D
D
D
D
D
D
D
D
D
D
D
D
D
1
2
3
4
5
6
7
8
9
10
11
Firm
Bluebird
Chrysler
Ford
Cummins
Caterpillar
Daimler-Benz
KHD
General Motors
Navistar
Hlno Motors
Deere
12 .Mack
13
U
IS
16
17
18
1'J
20
Onan
Perkins
Renault
Saab
Isuzu
Iveco
Volvo
White
Baseline
No Averaging Averaging
$0
3
41
383
139
8
1
202
84
3
0
91
0
0.
13
1
12
5
8
8
0
9
1
.0
4
5
3
0
9
5
9
5
4
9
0
1
3
6
3
1
SO
3
38
329
124
7
1
159
59
3.
0
79
0
0
11
1
9
5
/
7
0
9
8
5
1
6
1
9
6
2
8
8
3
8
4
0
9
2
3
1
Trading
Trading (vs Averaging)
SO 0
3 9
36.0
329.4
123 9
7 6
1 1
159 2
54 8
3 2
0 8
79 7
0 1
0 8
11 4
1 0
9 9
5 1
7 3
7 0
(S)
$0 0
0 0
2 7
0 2
0 3
0 0
0 0
0.6
4 7
0 0
0 0
0 0
0 2
0 0
0 0
0 0
-0 0
0 1
0 0
0.0
(Z)
0 0
0.0
7 1
0 0
0 2
0 5
0 1
0 4
8 0
1 0
0 5
0 0
71 5
0 5
0 3
0 5
1 0
0 3
0 4
Averaging
(v* No Averaging)
(S)
$0 0
SO 0
$2 3
$53 4
S15 3
SO 8
$0 2
$42 1
$25 4
SO 3
SO 1
$11 8
$0 1
$0 1
$1 5
SO 1
$2 4
$0 4
SO 9
SI 0
(X)
-0 0
0 0
5 6
14 0
11 0
9 7
12 3
20 8
29 9
7 4
9 9
12 8
29 1
9 9
11 8
9 9
19 5
7 2
11 2
12 7
TOTAL
$1.009 6
$851 4 $842 4
$9 0
11 Slbd 2
15 7
• C = HDCE. D = HDDE
-------
Exhibit V-3
COST SAVINGS
FOUR SUBCLASSES - AVERAGING AND TRADING
(dollars In millions)
Dollar and Percentage Savings from
Engine Type*
C
G
C
C
L
L M H
M H
M H
M H
L M H
L M
M
H
H
L
M
L H
N
L H
H
M H
L H
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Firm
Bluebird
Chrysler
Ford
Cummin*
Caterpillar
Daimler-Benz
KHD
General Motors
Navistar
Hlno Motors
Deere
Mack
Onan
Perkins
Renault
Saab
Isuzu
Iveco
Volvo
White
Baseline
No Averaging Averaging
SO
3
41
383
139
8
1
202
84.
3
0
91
0
0
13
1
12
5
8
a
0
9
1
0
4
5
3
0
,9
i
9
5
4
9
0
1
3
6
3
1
0 0
3 9
38 8
333 7
124 5
7 6
1 2
177 8
71 0
3 2
0 8
79 9
0 3
0 8
11 6
1.0
10.7
5 2
7 4
7 3
Trading Averaging
Trading (vs Averaging) (vs No Averaging)
0 0
3 9
36 5
330 7
124 4
7 6
1 2
176 5
69 9
3 2
0 8
79 9
0 3
0 8
11 6
1.0
10 7
5 2
7 4
7 1
(S)
$0 0
0 0
2 2
3 0
0 0
0 0
0 0
1 3
1 1
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 1
(Z)
0 0
0 0
5 7
0 9
0 0
0 0
0 0
0 7
1 6
0 0
0 0
0 0
5 0
0 0
0 1
0 0
0 0
0 1
0 0
2 0
(S)
$0 0
0 0
2 3
49 2
IS 0
0.8
U 2
24 2
13 9
0 3
0 1
11 6
0 1
0 1
1 3
0 1
1 /
0 4
0 9
0 9
(Z)
- 0 0
0 0
5 6
12 9
10 8
9 7
12 1
12 0
16 4
7 4
9 9
12 7
29 1
9 9
10 4
9 9
13 5
7 2
11 1
10 6
I
u>
ro
i
INDUSTRY TOTAL-
I.009 6
$886 S $878 7
$7 8
0 9 $123 1
12 2
* G - HDGE, L = LHDOE, M = MHDDE, H = HHDDE
-------
-33-
General Motors and Cummins, the manufacturers with the most diversity
in product line, gain the most savings from trading under the four subclass
scenario as opposed to two subclasses. White also saves more under averaging
compared to the baseline; all other firms' savings decline.
The percentage of revenues saved from trading within the entire industry
plotted as a function of the percentage saved from averaging are presented in
Exhibit V-4 (one class case) for the individual manufacturers. Here the inverse
relationship between the gains from averaging and trading for some firms is
readily apparent: Navistar, General Motors, and Isuzu all experience significant
cost savings with averaging, but little additional savings when trading is also
permitted. Ford, Chrysler, and Bluebird, however, save little or nothing by
only averaging, but significant savings accrue when trading is allowed.
B. Comparisons of Emissions Levels
This section presents the changes in the patterns of emissions that would
result from changes to more restrictive regulatory scenarios. Exhibit V-5
shows that as the subclasses become more restrictive, the range in emissions
levels across engine classes decreases. For example, there is a difference of
only 0.09 between the minimum and maximum NOx levels when emissions are averaged
in four subclasses, and a 1.73 difference when averaging within one class is
permitted.
In all scenarios, the range in both NOx and PM levels are smaller when
emissions are averaged .than when emission credits are averaged as well as
traded.
C. Costs per Ton of Emissions Reductions
In the course of solving for the most efficient emissions levels, the
optimizing model used for this analysis generates the costs of additiona
-------
Exhlhl
UJ
D
U
u.
O
C?
z
a:
u
O
or
L-
no
TKAuiNG ANu A'vtiRACiNG
% OF REVENUES SAVED, ONE CLASS
0%
D NAVISTAR
DGM
a isuzu
a CUMMINS
P MACK
D ONAN
DKHD
CATERPILLAR
n-WHCTE.VOLVO
D RENAULT
B DEERE, PERKINS, SAAB, D.BENZ
BHINO, IVE<
CJ
*.
I
D FORD
DDLU
IBIRD
SAVINGS FROM TRADING, 95 OF REVENUES
REGRESSION: R2=.64
-------
-35-
One Class
Engine Class:
LHD6E-
LHDDE-IDI-
LHDDE-DI
MHDGE-
MHDOE-NA
MHDDE-TC
HHDDE-LH
HHDDE-NLH
Exhibit V-5
EMISSIONS LEVELS *
(grams/BHP-hr)
NOx
Averaging
Trading
2,
3,
5,
25
67
06
2.21
4.80
5.17
4.90
4.97
PM
Averaging
na
0.431
0.384
na
0.108
0.251
0.246
0.085
Trading
na
0.479
0.419
na
,106
,375
,204
0,
0.
0.
0.072
Two Subclasses
Engine Class:
LHOGE
LHODE-IDI
LHDDE-OI
MHDGE
MHDOE-NA
MHDDE-TC
HHDDE-LH
HHDDE-NLH
NOx
Averaging
4.21
3.50
4.20
4.19
4.41
4.13
4.20
4.16
Trading
4.22
3.37
4.43
4.19
4.36
4.55
4.21
4.20
PM
Averaging
na
0.443
0.377
na
0.114
0.238
0.246
0.087
Trading
na
0.493
0.487
na
0.115
0.425
0.174
0.086
Four Subclasses
Engine Class:
LHDDE-DI
MHDGE
MHDDE-NA
MHDDE-TC
HHDDE-LH
HHDDE-NLH
NOx
Averaging
4.21
4.19
4.23
4.19
4.21
4.19
4.23
4.14
Trading
4.22-
3.75-
5.54
4.19-
4.23
,18
,23
4.
4,
4.14
PM
Averaging
na
0.219
0.224
na
0.122
0.273
0.286
0.087
Trading
na
0.166
0.383
na
0.119
0.274
0.286
0.087
* Figures for trading scenarios are actual emissions levels; those for aver-
aging scenarios are averages across all manufacturers, weighted by sales. The
regulations specify that no engines may exceed 6.0 g/BHP-hr of NOx or 0.60
g/BHP-hr for PM. This constraint was not binding in the scenarios examined
(i.e., even under unconstrained trading and averaging no engine would excee
these limits).
-------
-36-
emlsslons reductions for individual firms and for the industry as a whole.
Exhibit V-6 presents these marginal costs per ton of PM and NOx (emitted over
the life of a'year's fleet) for the regulatory scenarios.
This data may be compared to the costs of emissions reductions from other
sources for cost-effectiveness analyses. It is apparent that reductions in PM
are more costly than NOx emissions reductions. For all industry subclasses the
average PM control costs are greater than those for NOx.
D. Savings per Vehicle
The savings allowed by industry-wide averaging and trading varies widely
across individual firms. Exhibits V-7, V-8, and V-9 present the per-engine
costs and savings under the three regulatory scenarios. The per engine values
are, in a sense, normalized, and thus are more representative of relative
equity achieved across firms from averaging and from trading than the values
indicated in Exhibits V-l, V-2, and V-3.
-------
-37-
Exhibit V-6
MARGINAL CONTROL COSTS OF EMISSIONS CONTROL
(dollars per ton)
ONE CLASS
TWO CLASSES
SOX
PM
CAS
NOX
NOX
DIESEL
?M
AVERAGING
Finn
3laeblrd
Chrysler
Ford
Cummins
Cat erp lilac
Daimler-Benz
KHD
General Hotors
Savlstar
Si.no Motors
Oeere
Mack
Onan
Perkins
Renault
Saab
Isuzu
Iveco
Volvo
White
31.167
1.186
1.270
4.110
4,245
4,443
4,333
1,880
3.3&9
4,716
4.436
4.109
1.594
4.436
4,363
,436
.074
.753
,275
,401
S18.584
6.026
5.935
5.739
5,990
7.427
13.059
5.547
5.744
6.027
13.583
5,744
5.795
5.744
6.002
5.521
5.358
5.769
SI. 167
1.186
1.169
1.177
34,399
4.110
4.245
4.443
4.333
3,653
3,349
4,716
4,436
4.109
1.594
4,436
4.363
4.436
4.074
4,753
4.275
4.401
S12.S41
6.026
5.935
5.739
5.990
6.306
13.059
5,547
5.744
6.027
13.683
5,744
5.795
5,744
6,002
5.521
5.358
5.769
TRADING
S2.321 $6.853
SI.174 $4.009 $6.093
FOUR CLASSES
GAS LIGHT
SOX SOX
AVERAGING
Firm
Bluebird
Chrysler
Ford
Cummins
Caterpillar
Daimler-Benz
KHD
General Motors
SavLstar
Blno Motors
Deere
Mack
Onan
Perkins
Renault
Saab
Isuzu
Iveeo
Volvo
White
SI. 167
1.186
1.169 S4
4
1.177 1
1
1
1
2
4
,399
,398
.581
.581
.594
,594
.696
.406
LIGHT
PM
S12.
12.
13,
13,
13.
13.
13.
12.
641
641
682
682
683
683
268
629
MEDI'JV
sex
S4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
.436
,647
.463
.789
.615
.521
.715
.436
.436
.436
.436
.469
,753
.436
.435
MEDIUM
PM
SS
5
5
5
5
5
5
5
5
5
5
5
5
5
5
, 744
.595
.725
.750
.617
,684
.547
.744
.744
.744
,744
.720
,521
,744
,744
HEAVY
NOX
S4
34
S4
34
34
$4
$4
,109
,109
,139
.109
,109
.109
.109
HEAVY
PM
36,027
$6.027
$6,027
$6.027
S6.027
$6.027
$6,027
TRADING
$1.174 $2.494 $13.337 $4.557 $5,659 $4,109 $6.027
-------
Exhibit V-7
ONE CLASS - AVERAGING AND TRADING
PER ENGINE COSTS AND SAVINGS
Sales
13
34.016
139.009
93.817
35,259
3.023
297
329,138
79,647
1.191
331
17,441
794
331
5.265
397
7,345
1.885
2,408
3,753
Firm
1 BlOeblrd
2 Chrysler
3 Ford
4 Cummins
5 Caterpillar
6 Daimler-Benz
7 KHD
8 General Motors
9 Navistar
10 Hlno Motors
11 Deere
12 Mack
13 Onan
14 Perkins
15 Renault
16 Saab
17 Isuzu
18 Iveco
19 Volvo
20 White
I.USLS ---- -
Basel Ine
No Averaging Averaging
S22B
115
295
4,082
3,955
2,799
4,398
614
1,066
2,931
2,710
5,248
518
2,710
2,464
2,711
1.679
2.<>5u
3.436
2,160
$228
115
244
3.512
3.520
2,527
3.856
399
748
2,714
2.441
4.574
367
2,441
2,174
2.442
1,352
2,737
3,050
1,886
- - - - savings -------- .-.-
Savings from Savings from
Trading Averaging
Trading (vs Averaging) (vs No Averaging)
($64)
(25)
14
3,223
3,234
2.285
3,526
344
704
2,496
2.198
4,195
193
2.198
1.973
2,199
1.267
2,520
2,770
1,704
$292
140
230
289
286
242
330
55
44
219
243
379
174
243
201
243
85
^17
280
182
' ($0)
0
51
570
435
272
542
214
318
217
269
674
151
269
290
269
327
212
386
275
I
OJ
CO
I
-------
Exhibit V-8
TWO SUBCLASSES - AVERAGING AND TRADING
PER ENGINE COSTS AND SAVINGS
Sales
13
3d, 016
139.009
93,817
35,259
3,023
298
329,138
79,647
1,191
331
17.441
794
331
5,265
397
7,345
1.885
2.408
3,753
Firm
1 Bluebird
2 Chrysler
3 Ford
4 Cummins
5 Caterpillar
6 Daimler-Benz
7 KHD
b General Motors
9 Navistar
10 Hlno Motors
11 Deere
12 Mack
13 Onan
14 Perkins
15 Renault
16 Saab
17 Isuzu
IB Iveco
19 Volvo
20 White
Baseline
No Averaging Averaging
$223
115
295
4.082
3,955
2,799
4.398
614
1,066
2.931
2,710
5.248
518
2,710
2,464
2,711
1.679
2.950
3,436
2,160
$223
115
279
3,512
3.520
2,527
3,856
486
748
2,714
2,441
4,574
367
2,441
2.174
2.442
1,352
2.737
3,050
1.886
- - - - savings -------
Trading
Trading (vs Averaging) (vs
$223
115
259
3.511
3,513
2.514
3.854
484
688
2.687
2,429
4,572
104
2,429
2,168
2.430
1.354
2.709
3.042
1,878
$0
0
20
2
7
13
2
2
60
27
12
2
262
!2
7
12
(2)
28
8
8
Averaging
No Averaging)
$0
0
17
570
435
272
542
128
318
217
269
674
151
269
290
269
327
212
386
275
LJ
-------
Exhibit V-9
FOUK SUBCLASSES - AVERAGING AND TRADING
PER ENGINE COSTS AND SAVINGS
Sale*
13
34.016
139.009
93,817
35,259
3.023
297
329.138
79.647
1.191
331
17.4*1
794
331
5.265
397
7.343
1,885
2,408
3,753
Firm
1 Bluebird
2 Chrysler
3 Ford
4 Cummins
5 Caterpillar
6 Daimler-Benz
7 KHD
8 General Motors
9 Navistar
10 Hlno Motors
11 Deere
12 Mack
1 3 Onan
14 Perkins
IS Renault
16 Saab
17 Isuzu
18 Iveco
19 Volvo
20 White
Cons -----------
Basel Ine
No Averaging Averaging
$228
115
295
4.082
2.339
2.799
4,413
614
1.066
2.931
2,710
5.248
518
2.710
2.464
2.711
1.679
2.950
3.436
2.160
$228
115
279
3,557
2.088
2.528
3.877
540
892
2.714
2,441
4,580
367
2,441
2.208
2.442
1,451
2,737
3.054
1.932
- - - - bavlngs -------------
Trading Averaging
Trading (vs Averaging) (vs No Averaging)
$228
115
263
3.525
2,088
2,527
3,876
536
878
2.713
2.440
4.580
348
2.440
2.205
2.441
1,451
2,735
3.053
1,895
$0
0
16
32
0
1
0
4
14
1
1
0
IB
1
3
1
1
2
1
38
so-
0
17
525
251
271
536
74
174
217
269
668
151
2b9
256
269
227
212
383
228
o
I
-------
-41-
VI. SENSITIVITY ANALYSES
This chapter contains the results of four analyses of issues related to
averaging and trading. The first takes up the question of the extent to which
the savings from averaging and trading are attributable to greater efficiency
in attaining a given level of emissions reductions, or to the opportunity that
averaging/trading provides to meet the emissions standards exactly instead of
over-controlling emissions in many cases. The second examines the potential
impact of a reclassification of the smallest heavy-duty engines as light-duty
trucks, and the reduction in the scope of averaging and trading that would
result from this action. The third examines the degree to which the cost
savings results are sensitive to differences in input cost functions, i.e., in
assumptions about the relationships of NOx control levels to fuel consumption
and PM emissions. The final analysis estimates the potential of emissions
credit banking to ease the transition from the 1991 PM standards to the more
stringent 1994 standards.
A. Savings Due to Averaging and Trading in Comparison
to a Baseline with Emissions Equal to the Standard
The averaging program (and any potential trading program) is intended to
increase the efficiency of emissions control by allowing real locations of
control efforts to engines for which control is less costly, while ensuring
that overall emissions are no higher than they would be without the program.
For this report, the constraint that emissions be no higher with an averaging/
trading program has been interpreted to mean that they must be no higher, on
average, than if each vehicle met the target exactly. The costs under the
averaging scenarios are compared to a baseline in which averaging is not permit-
ted, and substantial savings are shown (see Section V-A).
Total emissions under the no-averaging or trading baseline modeled were
actually lower than the standards require (for technological reasons discussed
-------
-42-
below). Thus, under the averaging program we forego some emissions reductions
that would have been forced upon manufacturers by anticipated trap efficiency
in combination- with traditional "every engine must pass" regulatory approach.
These foregone emissions reductions are not a matter of concern because they
are not needed to meet the emissions standards established by EPA. It is
interesting to estimate, however, the degree to which the cost savings from
averaging and trading are attributable to allowing the manufacturers to avoid
over-controlling emissions versus the degree to which the savings are due to
more efficient allocations of the same emissions reductions. This estimate can
be made by comparing the costs under averaging and trading to an artificial
baseline in which over-control is avoided with no resort to averaging or trading
between different engine types.
The reason that over-control of PM emissions can be expected if averaging
jjs_ not permitted is that:
o based on the analysis in the RIA, the standards have been set at a
level sufficiently stringent that every engine would be forced to use a
trap;
o traps are expected to be so efficient that almost every engine using a
trap will have emissions well below the 0.22 g/BHP-hr level (standard
with design cushion); and
o there are no inherent cost advantages to designing traps that are less
efficient.
In quantitative terms, engine-out PM emissions are typically in the range of
0.6 g/BHP-hr (at low NOx emissions levels). The design target level unr'ap a
standard of 0.25 g/BHP-hr will be about 0.22 g/BHP-hr. This level probably
cannot be reached without a trap, and so in the absence of averaging every
engine would need a trap. With a trap, which can be expected to remove 80% of
the engine-out emissions, emissions would be only about 0.12 g/BHP-hr, well
below the target. It is the on/off, either/or nature of the trap technology
-------
-43-
that would force manufacturers to over-control their engines. Manufacturers
could try to design less efficient traps, but there appears to be little
economic incentive to do so.
For purposes of comparison, however, we are free to postulate an artificial
baseline in which each engine has "part of a trap:" the costs and effectiveness
of the trap are reduced in proportion to the need to meet the target exactly.
Costs under this artificial baseline are lower than under the actual baseline
because not every engine must pay for a "whole" trap, and emissions are exactly
at their target levels. The costs are not as low as under averaging, however,
because the distribution of traps is not optimized across different engines with
different costs of emissions reduction. The cost advantages of averaging can
therefore be split into its two components: its ability to let manufacturers
avoid over-control, and its ability to let manufacturers allocate traps opti-
mally across engine types without allowing emissions to rise.
The results for the one, two, and four subclass cases using the artificial
baseline instead of the actual baseline are presented in Exhibits VI-1, VI-2
and VI-3. While the dollar savings from trading over averaging are the same
as presented in the body of this report, the savings from averaging over the
artificial baseline are significantly lower than when the costs for averaging
were compared to the actual baseline. The reason that averaging looks less
attractive, of course, is that the artificial baseline has the savings from
reducing the number of traps already built into it.
One way to see this comparison is to examine Exhibit VI-4, which shows
four industry-wide total costs for each subclass assumption: (1) the actual
no-averaging baseline, (2) the artificial baseline, (3) the costs with aver-
aging, and (4) the costs with tradTng. In the body of the report, (3) was
compared directly to (1), and referred to as the savings from averaging. The
savings from trading, which is the difference between (4) and (3), seemed small
by comparison. As the table makes clear, though, the difference between (3)
and (1) is really composed of two components. The first component is the dif-
ference between (2) and (1), which is the savings attributable to avoiding the
-------
Exhibit VI-1
COST SAVINGS ARTIFICIAL BASELINE
ONE CLASS - AVbRAGINC AND TRADING
(dollars In millions)
Firm
Artificial Dollar and Percentage Savings from
Baseline Trading Averaging
No Averaging Averaging Trading (vs Averaging) (v» No Averaging)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
IS
16
11
IB
19
20
Bluebird
Chrysler
Ford
Cummins
Catcrpl 1 lar
Daimler-Benz
KHD
General Motors
Navistar
Hlno Motors
Deere
Mack
Onan
Perkins
Renault
Saab
IlllZII
Iveco
Volvo
Willie
SO
3
38
335
12)
7
1
1 78
71
3
0
81'
0
0
11
1
111
5
;
/
0
9
S
i
2
7
2
s
3
3
a
i
3
a
7
0
•>
2
4
J
$0
3
33
329
124
7
1
131
S9
3
0
79
0
0
11
1
9
S
7
/
0
9
9
5
1
6
1
4
6
2
a
8
]
8
4
0
•J
2
J
1
<$n
0
2
302
114
6
1
113
36
3
0
n
0
0
10
0
V
4
6
o
0)
a
0
4
0
9
1
2
1
0
7
2
2
7
4
9
)
a
7
t
($)
SO
4
32
27
10
0
0
18
3
0
0
6
0.
0
1
0
0
0
0
C)
0
7
0
1
1
7
1
2
i
3
1
6
1
1
1
1
6
4
7
7
(I
128
1 >.}
94
a
8
9
8
13
S
8
10
8
4/
10
9
10
b
7
9
•*
)
8
It
2
2
1
6
i
a
a
i
0
3
4
0
3
0
1
9
2
6
(S)
$0
0
S
6
1
0
0
47
11
0
0
0
U
0
0
0
1
0
1)
U
0
0
0
0
1
0
0
1
a
0
0
i
0
0
2
0
0
0
0
J
(I)
u a
U U
12 8
i a
0 9
0 4
0 b
26 4
Ib S
0 8
0 2
0 b
0 1
0 2
1 8
0 2
a a
u 8
0 4
. - "
INDUSTRY TOTAL
$1141 I)
SB IB 1
>/io •) $10; i
n i
i/.' 'I
a ?
-------
Exhibit VI-2
COST SAVINGS ARTIFICIAL BASELINE
TUO SUBCLASSES - AVERAGING AND TRADING
(dollars In million!)
c
c
c
c
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
II
(I
Firm
1 Bluebird
2 Chrysler
3 Ford
4 Cumins
% Tat., (.III. r
6 Dalinlet Beni
7 KHD
B General Motors
9 Navistar
10 Hlno Motors
11 Deere
12 Mack
1 3 Onan
14 Perkins
15 Renault
16 Saab
17 Isuzu
18 Ivaco
19 Vulvu
10 White
Artificial
Basel Ine
No Averaging Averaging
SO
3
38
335
12%
/
1
178
71
1
U
BO
0
0
11
1
10
5
/
•
0
9
9
5
1
/
2
5
3
3
a
3
3
a
7
0
9
2
4
1
so
3
38
329
1.-4
/
1
159
59
3
0
79
0
0
11
1
9
5
7
'
0
9
8
5
1
b
1
9
6
2
a
8
3
a
4
0
9
2
3
1
Hollar and ?<=i.
Ti adlng
Trading (v» Averaging)
SO
3
36
329
l.'l
/
1
159
54
3
0
79
0
0
II
1
9
5
}
>
0
9
0
4
V
b
1
2
B
2
0
7
1
8
4
0
9
1
1
U
(S)
SO
0
2
0
II
II
U
0
4
0
II
0
u
0
0
0
-0
0
o
II
0
0
'
2
1
U
u
6
'
u
II
0
I
0
0
0
0
1
II
u
0
•>
1
3
5
1
II
'
*
$11
0
0
6
1
U
u
18
11
0
0
U
0
0
0
0
1
0
u
1)
u
0
1
0
1
u
u
6
a
0
0
">
.0
0
1
0
u
0
u
i
(*>
o a
U 0
0 4
1 8
u t
a 4
U b
10 4
lb 5
o a
U J
0 b
0 1
0 2
i a
0 2
8 8
0 8
II •>
1 M
-p.
01
I
TOTAI
$851 4
•>H42 4
i'l 0
-------
Exhibit VI-3
COST SAVINGS. ARTIFICIAL BASELINE
FOUR SUBCLASSES - AVERAGING AND TRADING
(dollars In millions)
G
C
C
C
L
L M
M
M
H
L H
L H
H
M
L
H
L H
H
L H
H
M
L H
1
2
3
H 4
H 5
H 6
H 7
H 8
9
10
11
H 12
13
14
15
16
17
ia
H 19
20
Firm
Bluebird
Chrysler
Ford
Cuomlns
Caterpillar
Daimler-Benz
KHD
General Motors
Navistar
Ulno Motori
Deere
Hack
Onan
Perkins
Renault
Saab
Isuzu
Iveco
Volvo
UliUe
Artificial
Basel Ine
No Averaging Averaging
$0
3
38
335
125
7
1
178
71
3
0
80
0
0
11
1.
10
i
7
7
0
9
9
5
2
7
2
5
3
3
8
3
3
a
7
0
9
2
4
3
0 0
3 9
38 8
333 7
124. i
7 6
1 2
177.8
71 0
3 2
0 8
79 9
0 3
0 8
11 6
1 0
10 7
5 2
7 *
7 3
Dollar and Percentage Savings from
Trading Averaging
Trading (vs Averaging) (vs No Averaging)
0 0
3 9
36 S
330 7
12* 4
7 6
1 2
176 S
69 9
3 2
0 8
79 9
0 3
0.8
11.6
1 0
10 7
5 2
7 *
7 1
(S)
$0 0
0 0
2 2
3 0
0 0
0 0
0 0
1 3
1 1
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 1
(X)
0 0
0 0
5 7
0 9
0 0
0 0
0 0
0 7
1 6
0 0
0 0
0 0
5 0
0 0
0.1
0 0
0.0
0 1
0 0
2 0
(S)
$0 0
0 0
0 1
i a
0 7
0 0
0 0
0 7
0 3
0 0
0 0
0.4
0 0
0 0
0 0
0 0
0 2
0.0
0 0
0 0
(X)
0 0
o o-
0 4
0 5
0 6
0 4
0 4
0 4
0 4
o a
0 2
0 5
0 1
0 2
0 2
0 2
2 1
0 8
0 4
0 3
-p.
Ol
I
INDUSTRY TOTAL
$891 0
$886 5 $878.7
$7 8
0 9
$4 5
0 5
-------
ONE CLASS
TWO SUBCLASSES
-47-
Exhibit VI-4
COMPARISON OF INDUSTRY TOTAL COSTS
(dollars in mill ions)
(1)
Baseline
(2)
Artificial
100% Traps
No Averaging
$ 1,009.6
1
1
Baseline
No Averaging
$ 891.0
(l)-(3) = $ 191.5
1 1
(3)
Averaging
$ 818.1
1 1
(3)-(4
1
(4)
Trading
$ 710.9
1
) = $ 107.2
(l)-(2) = $ 118.6 (2)-(3) = $ 72.9
$ 1,009.6
$ 891.0
$ 851.4
$ 842.4
(l)-(3) = $ 158.2
I I
(3)-(4) = $ 9.0
I
(l)-(2) = $ 118.6 (2)-(3) = $ 39.6
FOUR SUBCLASSES
$ 1,009.6
I
$ 891.0
$ 886.5
$ S78.7
(l)-(3) = $ 123.1
I I
(3)-(4) = $ 7.8
(l)-(2) = $ 118.6 (2)-(3) = $ 4.5
-------
-48-
over-control of PM. The second component is the difference between (3) and (2),
which is the savings attributable to the efficiency of allowing firms to real-
locate emissions controls optimally across engine lines. This second component,
while appreciable, is much closer to the magnitude of the savings from trading
(the difference between (4) and (3)).
-------
-49-
B. Reclassification of Light-Heavy-Duty Engines
Manufacturers of some LHDEs--those for trucks in Class II-B--may choose to
reclassify their engines as light-duty truck (LOT) engines to take advantage of
test procedures that are less technically difficult or because of the availabil-
ity of off-the-shelf production technology. This reclassification could have a
significant impact on averaging and trading programs.
The extent to which this reclassification would take place is not known;
we have therefore re-estimated costs under averaging and trading for two widely
differing cases: one in which all Class II-B trucks drop out of heavy-duty
averaging and trading, and one in which only half drop out.
Estimation of Class II-B Sales and Shares
Engines for Class II-B trucks do not comprise all of an engine class in
this analysis. Instead, as defined, they represent a subset of each one of the
three light-heavy sub-classes: LHDGE, LHDDE-IDI, and LHDDE-OI. The balance of
each of these three sub-classes is composed of Class III and Class IV trucks.
Thus, to estimate how many heavy-duty trucks would remain in these classes if
100% of Class II-B trucks were reclassified as light-duty trucks, it is suf-
ficient to use the total sales and sales shares' for Class Ill's and IV's.
Based on figures published in Automotive News, we estimated that sales
of Ill's and IV's combined are in the range of 30,000 per year, with virtually
all of those sales by GM. The distribution of these sales across the three
light-heavy classes (LHDGE, LHDDE-IDI, and LHDDE-DI) is made simpler by the
fact that GM does not produce LHDDE-DIs; thus, all 30,000 should be placed into
either LHDGE or LHDDE-IDI. We have assumed that this division is the same'as
the industry-wide division between gas and diesel for this size class, yielding
20,904 LHDGEs and 9,096 LHDDE-IDIs.
-------
-50-
Once the sales and sales shares for the 100% reclassification case have
been set, the .determination of the sales and shares for the 50% case is simple:
sales and sales shares for the 50% case are unweighted averages of the sales and
shares for the "0% reclassification" (baseline) and 100% reclassification
cases. A comparison of the sales and shares for each of these three cases
are presented in Exhibit VI-5.
Results for this sensitivity case are presented in Exhibits VI-6, VI-7,
VI-8, and VI-9. The exhibits show that, for the industry as a whole, reclas-
sification would have only a moderate impact on the savings from trading and
averaging in the one-class case. Savings fall by about 15% for averaging and
for trading if half of II-Bs are reclassified, and by another 15% or so if the
other half are reclassified as well.
Under two-subclass averaging assumptions, the changes caused by reclassifi-
cation are similar to the one-class case. The savings from trading, however,
are cut dramatically by the reclassification of half of the II-Bs and are
virtually eliminated if all II-Bs are reclassified. Trading produces greater
savings when the averageable set includes engine lines and engine technologies
that are very dissimilar; apparently, subclass restrictions in conjuction with
the reclassification of II-Bs would leave very little cost-savings from heavy-
duty truck trading. (Of course, such reclassification might make trading for
light-duty trucks very appealing; an examination of this issue is beyond the
scope of this analysis.) A summary table of industry-wide comparisons across
scenarios is provided as Exhibit VI-10.
-------
VI 5
SfiiSHU SJHU.S RH MM AjanrwuH «•• iJiiir uwr unv
SALES SIMS (peuMt)
df«r- b
Ita^ ^i^* ftiiM*ii«i fiyr^i^- Rnd Qualm pillar
UU£
iHre-nn
UIIE-OI
HIDE
MIIEtM
HUE-IK
IIIIEIH
IIIIEIIH
211.668
132.811
44.278
169.112
29.818
51.114
77,405
U.7U1
0
0
0
0007
0
0
0
0
161
0
0
0
0
0
0
0
97
0
400
595
0
0
0
0
0
0
550
0
0
1 6
Ml
0
0
0
0
To 1
U.8
1J 7
13 7
ililJer-
ItaK
0
0
0
0
1 0
01
0 1
Calftldl
HI) KJUJII HfA&*r
0
0
0
0
0
02
02
02
/4 2
51 4
0
405
>> 5
a 5
11 4
II 4
U
456
0
0
17 7
0
0
IllIU
0
0
0
0
i 7
08
0
0
l^j-n* M ^ ^ Qkjf i Iliiclrui l^i "^'1* S^^> T **ft'
0
0
G
0
0
06
0
0
0
0
0
0
0
0
150
150
0
06
0
0
0
0
0
0
0
0
0
0
0
0 6
0
0
0
04
0
0
0
9 1
0
0
0
0
0
0
0
08
0
0
0
1 9
2 5
0
1 4
6 1
0
0
Ivtin
0
0
0
0
4 5
1 0
0
0
VUvo Mill*
0
0
0 '
0
0
1 4
06
06
0
0
2 5
0
0
52
0
0
755.JH1
BlufcUd Uynltf fad Orafca
Hi)
llivld.ii
Hun
Oui Itifcuu HUM.IL
laui
VUvo Uiiic
UHE 1)6. a*>
MIIE-m 70.9h5
IJIIKOI 22.119
HUE 1W.11>
HIIEtM M.HId
HIIE-n: 51. 114
IIIIKIH 77.44)5
HllLMh Id. All
0
0
0
0007
0
0
0
0
14 6
0
0
0
0
0
0
0
89
0
400
595
0
0
0
0
0
0
550
0
0
1 6
M 1
0
0
0
0
Jh 1
If. 8
117
LI 7
0
0
0
0
1 0
52
0 1
0 1
0
0
0
n
0
02
1)2
(12
to 5
546
0
Ul 5
•» 5
21 5
II 4
11 4
0
4J 6
n
0
U 7
0
0
0
0
0
0
.! 7
On
0
0
I)
0
0
0
0
06
0
0
0
U
0
0
0
0
150
15 U
U
06
0
0
0
0
0
0
0
0
0
0
0
06
0
0
0
0 4
0
0
0
9 1
0
0
u
0
0
0
u
08
0
u
0
1 8
2 %
0
1 4
6 )
0
0
0
0
0
0
4 5
1 0
0
0
0
0
0
0
0
14
06
06
0
0
2 ^
0
0
52
0
0
5/5,972
I
01
1
Typis* ^M^t fri»«J4gj Oyrilcf Fixd QJIU^JD i
HUE
uiiK-on
HUE oi
HUE
HIIEtM
MIIE-V:
IIIIEIH
IIUEMH
20.414
0
2X818
51,304
77.405
0
0
0
oou;
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
545
0
0
0
0
0
0
0
0
0
1 61
Ml
C^ittr* QuuJer*
I'l 1 1 1
Ie8
11 7
13 7
0
0
0
U
1 0
52
01
0 1
1
HI)
0
0
I)
0
0
02
0 2
02
"feSl N
HDO
luuo
0
Ul •)
m 5
21 5
11 4
11 4
Mtxjr
0,
0
n
0
11 7
0
0
lluu
0
0
0
0
2 T
08
0
0
Ufa:
0
0
0
0
0
06
0
0
fet Ci*. fed*. h.~Ji S-b U.. N~,
0
0
0
0
0
0
150
150
0
0
n
0
0
0
0
0
0
0
0
0
06
0
0
0
n
u
n
u
9 1
0
0
0
0
0
0
0
08
0
0
0
0
0
0
1 4
6 1
0
0
0
0
0
0
4 5
i a
0
0
VUvo
0
0
0
0
0
1 4
06
06
Uau
0
0
0
0
0
5 2
0
0
UI1E
IIIIE-mi L
IIIIEW
Milt Hilbiulk^yllfyr.uJbi: UH.II>
HIII.IM Hilbmll>vy(lly lluueJ hbut: - li
MIIK1L MdliiiilL^yrily blue] beu« - 1\
IIIU <*: • N
Built: - Ldimx L\k»U»
. IUJ
-------
EahlLlt VI-6
COST SAVINGS
ONE CLASS IIAIF OF Cl ASS II-b KLiIl AiilKI tlJ AS LOT
(dul lars In mil Hunt)
Finn
Baseline Dollar and Percentage Saving* IIOID
1001 traps Trading Averaging
No Averaging Averaging Trading
6
3
6
SO 01
2
28
322
124
/
1
119
51
3
0
79
0
0
11
1
9
b
7
6
D3
0
8
/
1
b
1
1
8
2
8
a
1
3
4
0
S
2
3
8
o
3
U
74
U
0
III
0
a
4
b
b
0)
H
b
3
2
1
1
0
8
0
7
;
1
7
i
9
9
8
8
I
SO
1
3d
20
7
0
U
14
1
0
u
J
0
0
0
0
u
0
0
0
0
8
b
S
9
b
1
1
0
2
1
1
1
1
9
1
6
3
S
b
(I
133
143
127
b
b
1
b
11
2
b
8
6
49
8
;
8
b
»
7
;
»
/
2
2
3
4
7
b
8
0
7
0
4
9
0
7
0
U
n
3
9
(S)
SO
0
3
iO
li
0
U
&9
17
0
0
11
0
0
1
0
1
0
0
0
0
u
b
0
3
8
>
1
4
3
1
8
1
1
4
1
/
4
9
9
(I)
1 1
U U
11 i
13 4
11 0
9 /
\J 3
a 4
2i 2
7 4
<* 9
1.' 8
29 3
9 9
1U 9
ID 0
n i
7 2
11 2
11 4
ro
i
IHUUSTKY TOTAI
$948 0
$'83 1
$t>91 ->
$92 0
11 8 $164 7
17 4
-------
Exhibit VI-7
COST SAVINGS
ONE CLASS ALL CLASS II-B RLCI.AbSIHLD AS IDT
(dollars In million*)
h Irm
Baseline Dollar and Percentage Savings iioiu
10U1 irarn Trading Averaging
No Averaging Averaging Tiadlng (vs Averaging) (vs Nu Averaging)
1 Bluebird
2 Cluy>lcr
3 toed .
4 Cuamlni
S C.. 1-rplllar
6 Daimler-Benz
7 KHD
8 General Motors
9 NavLstur
10 Hlno Motors
11 Deere
12 Ma<.k
13 On jr.
14 Perkins
15 Renault
16 Saab
17 I»uzu
IB Iveco
19 Volvo
20 Uhl la
SO 003
21
362
13V
b
1
Ibb
bi
3
0
VI
0
12
1
10
5
H
/
•
B
i
4
i
3
;
6
b
9
b
m
9
;
i
i
6
i
j
$0 003
_>j
Jib
124
7
1
10U
48
3
0
/a
0
11
i
9
5
7
u
A
8
U
1
6
1
4
8
2
8
U
*
a
4
0
1
2
3
•>
($0 00.')
-19
302
lilt
;
i
9*
46
3
0
lo
0
10
0
8
4
7
b
•
U
2
i
2
I
6
1
1
B
3
*
B
7
9
6
9
0
I
IS)
SO o
•
42 H
U 8
& 6
0 4
II 1
8 8
•i 7
0 2
0 0
3 b
•
0 0
0 7
0 1
0 &
0.3
0 4
U 4
m
179 t
1/9
4
4
&
4
8
b
i
i
4
i
6
b
b
i
S
u
•
8
4
6
7
6
1
6
1
9
4
•
9
0
9
8
1
2
u
(S)
$0 0
•
0 0
46 b
Ib 3
0 8
0 2
47 )
4 8
0 3
0 1
11 8
•
0 1
1 t
0 1
1 0
0 4
0 4
U /
(I)
1 1
•
U U
12 8
11 U
9 7
12 3
10 4
8 9
7 4
9 9
1.' B
•
9 9
9 9
9 1
9 b
7 2
11 2
•> -i
I
in
INMISTHY TlllAI
$074 7
$rin
III 6 SI U 4
u a
All
-------
Lulllbll VI -H
COST SAVINGS
TWO SUBCLASStS HALF OF CLASS) II-B KLCLASblFlKI)
(dollars In iiillllons)
C
C
C
C
D
D
0
D
0
0
D
D
0
D
0
0
0
D
U
D
D
1)
1
2
1
4
5
6
7
8
9
10
11
12
11
14
15
16
11
IB
19
70
Firm
Blueblid
Chrysler
Ford
Cumnlns
Caterpl 1 lac
Dalmler-Bcuz
KIID
Ceaeral Motors
Navistar
Hlno Motors
be ere
Mack
Onan
Perkins
Renault
Saab
Isuzu
Iveco
Volvo
Whir.
Basel Ilia
100Z craps
No Averaging Averaging
$0 001
2
12
372
139
8
1
i;a
09
3
0
91
0
0
12
1
11
4
a
7
0
4
7
4
J
1
a
i
4
9
4
I
9
a
i
i
6
3
o
SO
i.
31
322
124
7
1
. J
41
3
0
79
0
0
11
1
9
4
7
f>
0
u
i
'
i
6
1
4
a
.»
8
8
1
8
4
0
4
2
1
a
AS LOT
Trading
Tiadlng (vs Averaging)
so
-'
11
122
124
7
1
I4B
51
1
0
79
0
0
11
1
9
i
7
li
0
0
9
'
0
6
1
I
6
2
a
8
0
a
4
0
b
1
i
a
(S)
SO
0
1
0
u
0
0
u
0
0
0
0
0
0
0
n
0
0
0
0
0
0
4
0
1
0
0
2
1
0
0
0
1
0
0
0
0
0
0
0
<»:
i
u
4
0
0
0
0
IJ
0
0
0
0
73
0
0
0
U
0
0
0
Av«rag Ing
(vs No Averaging)
1
1
0
4
G
1
1
0
•
*
7
i
0
1
4
2
2
0
a
0
2
(S)
SO
0
1
50
15
0
0
10
i;
0
0
11
0
0
1
0
1
0
0
0
0
0
I
0
1
8
2
4
4
1
1
a
i
i
4
1
'
4
9
9
<.,
0 U
o o
J 5
U 4
11 0
9 7
U> 1
I/ 0
25 2
7 4
» 7
12 8
29 6
9 7
10 9
10 0
14 2
/ 2
11 i
11 4
I
en
TOTAL
SMI*)
r.si; a
u i
I il 0
14 I)
-------
VI 9
COST SAVINGS
TWO SUBCLASSES ALL CLASS II-B KECIASS1HEI) AS I Lit
(dollars In milllona)
c
c
C D
D
D
0
D
G D
D
D
D
D
0
U
D
D
D
D
D
D
1
2
3
4
S
6
7
8
9
10
11
12
11
14
li
16
17
18
19
20
E* 1 fat
9 i rio
Bluebird
Chryiler
Ford
Cunmliii
Caterpillar
Daimler- Ben*
KHD
General Molori
Navlitar
Hi no Hoion
Deere •
Mack
Oixn
Peikln*
Kenauli
Saab
Iluzu
Iveco
Volvo
Uhlte
B^ >e 11 ne
lUUt ir^pi
$0 003
21
162
119
B
1
Ibb
iJ
3
0
91
0
:2
1
10
b
a
/
•
8
b
4
b
1
7
6
5
9
b
•
9
7
1
1
6
1
^
$0 001
23
116
124
7
1
137
<.rf
3
0
79
0
11
1
9
S
7
b
•
B
0
1
b
1
b
8
2
8
8
•
B
4
0
1
2
3
5
Dollar *nJ P'l-cenrane Savings from
Hading Av«r»»ln»
$0 001
23
31b
124
/
I
HI
48
3
0
79
0
11
1
9
S
7
6
*
B
9
1
6
1
b
7
2
B
B
•
B
4
0
1
1
3
b
($)
S"
0
11
0
u
u
u
0
0
u
a
o
0
0
0
0
0
u
0
•
u
i
0
0
U
u
i
0
0
0
•
0
0
0
0
1
0
u
u>
u u
•
0 0
u o
0 0
0 U
0 0
U 0
U i
U 0
U 0
0 U
•
II U
u u
0 0
0 0
1 1
0 0
u u
(S)
SO
0
46
li
0
0
IB
4
0
0
11
0
1
0
1
0
0
u
u
•
0
i
]
8
2
1
8
3
1
B
•
I
i
1
0
4
9
;
0)
0 U
•
0 0
1.' B
11 0
9 ;
U J
11 6
B 9
7 b
9 7
\1 9
A
9 7
10 o
9 9
9 b
7 2
11 1
10 0
I
tn
LTI
TOTAL
SBob 1
$7b4
5/ttl 8
SO 1
0 U $10.' 1
11
* All en*} 1 lie & i e< ,
»i I t led
-------
Exhibit VI-10
SUMMARY OF JNDUSTKY COSTS FOK LHUt KECLASSIHLATION iCLNAHIOS
(dul l«i* til nil II tuns)
Duller *nJ Percentage Saving! I roin
1001 tr*|>i Tiadlng Averaging
No Averaging Averaging Tiadlng (v» Averaging) (v* Nu Averaging)
LDT $1.009 6 $bil 4 $812 t $« 0 11 SI48 2 1 b 7
Half II-Bi rccUsiltled o LUT 948 0 bib 0 81 i 8 22 03 Til I) U (I
All II-Bi rcclanlflcd ^s LDT 80b 3 784 1 781 8 03 0 04 102 3 lib
-------
-57-
C. Sensitivity of Results to Assumed Functional Relationships
All of the results of this report are driven by the underlying functional
relationships among NOx and PM emissions and fuel consumption. While the
functions used were developed after careful research into current and likely
future technologies, there is no guarantee that the true relationships will not
turn out to be somewhat different. There could be unexpected breakthroughs in
emissions control techniques, especially given the incentives provided by the
averaging and trading programs to produce engines capable of lower emissions
than those mandated by the standards. On the other hand, it is possible that
projections for some types of engines could turn out to be overly optimistic.
To investigate the sensitivity of the results of the analysis to moderate
shifts in the functional relationships, the potential gains from averaging and
trading were re-estimated with altered functions for one subclass. A scaling
parameter was altered in the equations relating NOx to fuel consumption and PM
emissions for all MHDDEs, increasing the fuel consumption impact and the engine-
out levels of PM associated with a given NOx target by 20%. This change was
arbitrary in the sense that there is no more reason to expect these function to
be off the true mark in one direction than the other. The shift in the functions
does increase the difference between MHDDEs and other classes, which could be
expected to make averaging and trading look more attractive. Averaging and
trading could be expected to appear less attractive if the functions had been
.shifted in the opposite direction.
The results, Exhibit VI-11, show that trading would indeed provide more
cost savings with the shifted functions, relative to the savings from trading
with the unsTiifted functions and relative to the savings from averaging In
the two subclass case, trading would save an estimated $14.2 million per year,
58% more than the $9.0 million in savings from trading estimated with the
unshifted functions. Averaging would also save more if the shifted functions
were correct, by a margin of $173.1 million per year to $158.2 million per
year. This gain represents an increase of 9.4%, significantly lower than for
trading.
-------
txl.lLU Vl-ll
COST SAVINGS
TWO SUUCI-AiSLb 201 INCREASE IN MIHUit COST JUNCTION
(dul lar» In mil I lon»)
c
c
C D
D
D
D
D
C D
D
D
D
D
D
D
D
D
D
D
D
D
1
2
3
4
i
6
7
B
9
10
11
12
13
14
1)
16
17
IB
19
20
Firm
Bluebird
Chrysler
Ford .
Cummins
Cai.. |>ll lar
Daimler-Benz
KHD
Ceneral Muloi*
Navistar
Hlno Minor*
Deere
Hack
Onan
Perkins
Renault
Saab
Isuiu
Ivcco
Volvo
White
Basel tne
100X traps
No Averaging Averaging
JO 00 J
3
41
387
IB;
22
1
262
lb/
b
2
91
0
2
36
3
2a
a
17
ji
9
1
1
9
1
a
a
j
9
6
b
4
b
b,
1
a
9
0
b
$0 003
3
38
333
169
21
1
21b
126
b
2
19
0
2
Jb
1
26
a
lb
20
9
7
4
7
2
b
4
4
6
b
a
3
b
0
0
1
b
7
4
Dollar and Percentage Savings from
Trading Averaging
Trading (vs Averaging) (v» No Averaging)
Su nui
3
36
333
169
20
1
214
121
b
2
79
U
2
31
2
2b
8
lb
19
9
0
2
2
4
6
a
4
4
4
7
1
4
S
8
4
i
b
6
<;>
SO
0
2
0
0
0
O
u
b
0
u
0
u
0
1
0
0
0
0
0
0
0
7
2
s
a
0
6
0
2
1
1
2
1
b
2
7
3
2
a
(i
\
0
7
0
0
i
0
0
4
'
4
0
66
4
4
6
2
*
1
J
)
1
0
U
1
1
a
0
j
0
•j
0
1
7
0
3
7
7
S
J
»
.*.
$0
0
2
b3
ia
0
0
4'
30
0
0
11
u
0
1
0
2
0
1
1
0
0
4
7
2
9
2
4
9
3
1
7
1
1
b
1
'
4
3
1
(It
U U
0 0
b 1
13 9
9 ;
4 1
11 ;
Id U
IV D
4 6
2 6
12 a
il 0
2 7
4 1
2 7
9 b
4 1
/ b
b U
I
LD
CO
I
TOTAL
I.JB-' 8 SI.IO'J '
S>4 2
1 ) $1/1 1
I J b
-------
-59-
D. Banking
In this section some aspects of emissions banking are discussed, and a
rough estimate of the potential savings possible using a bank is offered.
The Concept of an Emissions Bank
Under emissions banking, a firm that generates emissions credits by pro-
ducing engines that more than meet emissions standards is permitted to hold
onto them--putting them in a "bank"--and use them in later years to offset
emissions that exceed the standards. In effect, an emissions bank extends the
concept of averaging and trading over time, enlarging the averagable set to more
than one model year.
Savings from banking could be large under certain circumstances, without
compromising air quality. Concern is sometimes expressed, however, that the
use of banking could lead to emissions increases, undesirable variations in
emissions over time, or to disruptive attempts to manipulate market shares by
radical year-to-year changes in emissions characteristics. To understand the
real effects of banking, it is important to lay out the circumstances under
which emissions banks would be attractive.
Circumstances in which Emissions
Banking Would be Attractive to Manufacturers
The circumstances favoring the use of an emissions bank are:
1) if the banking rules allowed manufacturers to dispense with the "design
cushion"-- the gap between the emissions standard and the self-imposed
emissions target set so as to ensure compliance in use. In this case,
-------
-60-
after the first year of a banking regime, banked credits might take the
place of th'e design cushion, allowing manufacturers to aim directly at
the standard. This would be less costly to manufacturers over time,
but would also result in greater emissions;
2) if banking were allowed as a remedy for in-use noncompliance;
3) if the real value to consumers of vehicles with higher emissions were
expected to rise over time. Vehicles with higher emissions tend to use
less fuel, and if fuel costs rise appreciably (or if interest rates
fall appreciably, which increases the present value of fuel costs) then
consumers will be more concerned with fuel consumption and will pay a
higher premium for more efficient engines. Manufacturers could gain by
generating credits now and using them in the future, to allow them to
increase emissions and fuel efficiency when efficiency is more highly
valued;
4) if there were a perceived marketing advantage in allowing an entire
model year's output to exceed emissions standards, even at the cost of
holding an earlier year's output below the standards;
and, probably most significantly,
5) if regulations are expected to be tightened in the future. Banking
could then be used to make the adjustment to tighter regulations easier
and less costly. Manufacturers would cut emissions sooner than the
regulations require, and then ease down to the limits just after the
tighter regulations went into effect. T
-------
-61-
1) Averaging, and possibly trading, will already be permitted before bank-
ing goes into effect. These programs themselves are being structured to
force, the retention of a design cushion. A banking program could also
be structured in this way, by limiting the use of banked credits to
adjusting the PEL for an engine family, and not permitting the credits
to be used to make up for the problems of an individual engine.
2) Because of legal provisions and enforcement concerns, banking as a
remedy for in-use violations is not a serious possibility;
3) While it is possible that credit values will rise over time, it is not
likely that manufacturers will be sure enough that they will to make the
investment in generating credits attractive. Credit values could be
driven down as well as up, and so an investment in credits would be
quite risky if undertaken on the basis of expectations regarding fuel
costs or interest rates. Added to the fact that no interest would
accrue on banked credits, thus decreasing their value over time,
it would probably not seem worthwhile to accumulate credits for this
reason.
4) The idea that a firm would try to gain sales in the future by spending
banked credits is rendered dubious by the fact that sales would probably
be lost initially while the credits were being generated. In addition,
the costs of this strategy would come well before the benefits, meaning
that the costs of the capital invested in the strategy would reduce the
strategy's net value. Only if consumers behaved quite asymmetrically
with regard to increases and decreases in performance —valuing slightly
improved performance highly while being relatively indifferent to
slight deteriorations of performance—would this plan seem worthwhile.
5) This set of circumstances is the most plausible, since tightened emis-
sions standards have been promulgated. Manufacturers would have a
strong incentive to generate and bank credits before the standards were
tightened, and then use the credits in the period following the change
-------
-62-
in the regulations. This shift in emissions would not, however, cause
a worrisome spike (increase) in air pollution. Instead, it would
contribute to a smoother reduction in emissions than would take place
without banking, and would allow society to enjoy its emissions
reductions sooner than otherwise.
In sum, the only likely scenario for the use of emissions banks is one
in which the effects both for the manufacturers and society as a whole are
beneficial.
Estimating the Savings Potential of Banking
It is not possible to estimate the cost savings that would be possible
from banking without knowing the rules under which banking could take place,
and discovering the technological relationships among emissions and costs both
now and in the future. We have, nonetheless, made an estimate of the savings
that could be realized under idealized conditions to show the potential of
banking and demonstrate how a more complete analysis would be conducted. Costs
for banking with one class averaging and trading have been estimated. Since
only PM credits would be banked and manufacturers of gasoline engines would not
have PM credits, results for two subclass averaging and trading are expected to
be about the same as the one class case.
In this example, we have assumed no technological progress; in particular,
we do not assume that traps will become more efficient in 1994 than we assume
they will be in 1991. Regulations for PM are to be tightened from a standard
of 0.25 g/BHP-hr (with a target of 0.22) to a standard of 0.10 g/BHP-hr (and a
target of 0.088) in 1994. Many firms would not be able to meet this stri-nent
standard, even with averaging, so we have assumed that a non-conformance penalty
(NCP) would be charged to firms for each ton over the PM target.^.X The per-ton
}_/ The charging of a non-conformance penalty is illustrative. No NCP devel-
opment work for the 1994 standards has taken place.
-------
-63-
charge was assumed to be equal to the Industry-wide marginal cost of PM removal
calculated to prevail under the tight 1994 standards: $13,234 per ton.
Allowing banking would permit a spreading-out of the emissions tightening.
For instance, under banking there would be an early (pre-1994) reduction of
emissions from 0.22 g/BHP-hr down to 0.154 g/BHP-hr (the average of 0.22 and
0.088), a level which would generate credits. There would then be another
period of emissions at 0.154 g/BHP-hr after the regulations tightened, until
the credits were used up. Using one-class averaging assumptions, emissions
control costs under this banking scheme would be lower by about $6.5 million
dollars per year, or 0.7 percent of total emissions control costs, than they
would be without banking (see Exhibit VI-12). Banking would save more under
one-class trading assumptions than under one-class averaging: $12.7 million per
year, or 1.6 percent of emissions control costs.
The total savings from banking would depend on the number of years over
which the inter-temporal averaging would take place: the sooner the credits
started accumulating, and the longer they were used after the regulatory change,
the more would be saved. The time period can be estimated by considering the
capital cost to the manufacturers of generating a credit and holding it, with
no interest accruing, until it can be used to reduce costs. Considering the
high real cost of capital, we estimate that at most a year's worth of credits
would be banked, and so the total savings from banking in these circumstances
would be relatively small.
-------
Lxlilbll VI-12
COST SAVINGS
ONE CLASS BANKING v» NO BANKING
(Jollars In ml 11 lout)
Firm
1 Bluebird
2 Chryiler
3 Ford
4 Cummin*
5 Caterpillar
6 Daimler-Benz
7 KHD
a General Motors
9 Navlsiar
10 Hlno Motors
11 Deere
12 Hack
13 Onan
1* Perkins
16 Renault
16 Saab
17 Isuzu
IB Iveco
19 Volvo
20 Ulilto
Averaging i ran ing ---
Dollar and Percentage Dollar and Percentage
Savings trum Savlng> from
Ulllioul Avciaglng Ullh B. liking Ulll.oiil Trading Ullli Banking
Hllh Banking Banking (va Ultliuul Banking) Wltli Banking B.nking (v* Ulcliout Banking)
SO 003
3
31
312
131
8
1
14)
70
i
U
Hi
0
U
12
1
10
4
7
7
9
8
1
7
1
2
b
0
4
U
3
4
9
1
U
8
«
«
5
$0 003
3
3)
3)3
131
a
i
149
7U
3
U
8i
U
0
12
1
11
i
/
/
9
a
4
9
1
2
2
1
4
9
3
4
9
3
0
1
b
b
;
(S)
SO 0
0 0
0 0
1 3
0 1
0 1
0 0
3 7
U i
a I
0 U
0 0
0 0
0 U
0 2
U 0
0 4
U 1
0 0
0 -'
(I)
U 0
0 0
0 0
0 4
0 1
1 2
0 0
2 b
0 4
.' 9
0 U
0 0
0 U
U 0
1 6
0 0
3 6
1 b
0 0
2 b
$0 001
-0
4
321
1J
-------
Appendix A
OPTIMIZATION
This section describes the operation of the procedure (model) used to find
the lowest-cost set of emissions levels under regimes that allow averaging or
trading. It is an adaptation of a non-linear programming approach to optimiz-
ation. It finds the least-cost, optimal solution by setting all derivatives of
total emissions control cost to zero, while holding emissions constant.
The computer program takes as inputs cost and technical data on the engines
sold by firms in the industry (see Appendix B for details) and emissions targets
set by the regulations. It provides as outputs the emissions levels of each
type of engine that meet the standards at the lowest possible cost. Other
outputs are the costs per vehicle of emissions control, the total costs of the
controls for each class of engine and across engines, emissions levels, the
costs per ton of emissions reductions, and the expenditures on emissions reduc-
tion credits made for each type of engine.
Three conditions are met by the minimum cost (most efficient) set of
emissions levels. First, the total emissions equal the total emissions per-
mitted. This condition is necessary because it is not permissible to emit total
pollutant loadings at higher levels than those allowed by standards, and it is
not efficient to emit at lower levels (since emissions reductions are not
without cost). Second, the application of traps has been done so that no trap
removes less PM per dollar than it could remove if installed on any engine
without a trap. Finally, the marginal cost of removing a ton of NOx is the
same for every engine within the averageable set. .
Finding Marginal Emissions Control Costs
Finding a set of emissions levels that meets the three conditions set out
above is a straightforward, but not trivial, problem. It requires estimates o
-------
A-2
the marginal cost of NOx, which in turn requires the estimation of marginal
costs of PM removal-.
The reason the NOx estimate depends on the PM marginal cost estimate is
that these two pollutants are interrelated: decreases in NOx emissions generally
cause increases in PM emissions, which must be compensated for in some way.
Thus, the cost of NOx removal includes not only the costs of fuel consumption
increases resulting directly from NOx-control actions, but an indirect cost as
well. Removing a ton of NOx can cause an increase in PM emissions of on the
order of a tenth of a ton (for an engine without a trap). To remove this added
PM, a slightly larger proportion of engines in the averageable set must use
traps. To be efficient, the traps must be added to the engines for which the
cost per ton of PM removed is the lowest of all engines that do not already
have traps. The cost per ton of PM removed is calculated as the total trap
cost divided by the reduction in PM emissions per BHP-hr due to the trap times
the number of BHP-hrs in the engine's life. The inverse of the number of tons
removed per dollar spent on additional traps is equal to the number of dollars
per ton of PM removed—in other words, the marginal cost of PM removal. Thus,
the cost of removing a ton of NOx must include the marginal cost per ton of PM
removal times the number of tons of PM added by the removal of one ton of NOx.
Formally:
MC (NOx) = d(TC)/(NOx) = $TC/frNOx + &PM/$NOx * d(TC)/d(PM)
The functional relationships needed for the marginal cost computations are
derived by differentiating equations based on the data provided by ERC, Inc.
(see Appendix B) that relate fuel consumption increases and PM emissions to NOx
emissions levels.
The Algorithm for Finding the Most Efficient Allocation
An optimization algorithm is used to find the optimal, i.e., lowest-cost,
most efficient NOx levels and trap use patterns for each scenario. The algorith
i
-------
A-3
works by starting with each engine type within each averageable set exactly
meeting the NOx standards, and with 100% trap usage. This results in the
correct total- emissions of NOx (though not at lowest cost) and PM emissions
that are lower than necessary. The first step in finding the optimal allocation
of emissions is to calculate the cost-effectiveness of traps on each type of
engine--the dollars in trap-related costs imposed on an engine per ton of PM
removed. The program "removes" traps, starting with the least cost-effective
ones, until total PM emissions rise to the permitted level. The cost per PM
ton removed for the addition of a trap to the marginal engine type (the type
with the last trap removed by the program) equals the marginal cost of removing
a ton of PM (given the NOx emissions levels).
The next step in the optimization procedure is to calculate the marginal
costs of removing a ton of NOx from each of the engine types. As described
above, the PM increases predicted to result from the reduction of NOx by a ton
are charged to the cost of the NOx reduction at the marginal cost of PM reduction
(calculated as described above). For example, if removing a ton of NOx from
a certain engine costs $1,000 by itself (due to increased fuel consumption), and
increases PM emissions by 1/10 of a ton, and a ton of PM costs $6,000 to remove,
then the marginal cost of NOx removal for that engine is $1,000 plus $6,000/10,
or $1,600 altogether.
These marginal costs will differ for different engines, since the NOx
emissions levels have not yet been set at their optimal levels. Newton's
method is used to adjust each NOx level until the marginal cost of NOx emissions
is equalized across engine types (and for types for which some have traps and
some don't, the MC for NOx ton removal is set equal for trapped and non-trap
versions) while the average NOx emissions level is kept at the level that will
allow the standard to be met.
The process is then begun again, but with the NOx levels changed to equal
those calculated to equalize marginal costs of NOx removal. The most efficient
trap allocation scheme is again found, and the marginal cost per ton of PM
removed is again found. This time, the marginal cost of PM removal is close
-------
A-4
to its correct value because more refined NOx levels are used as inputs. Again
the NOx levels are adjusted so that the marginal costs of NOx removal are
equal. This process is repeated until no further change in NOx levels is seen
on successive iterations.
Once the iterative process is complete, the NOx levels and trap percentages
have been set so as to meet the emissions limits within each averageable set;
the traps are allocated across each averageable set so that no traps are less
cost-effective that traps would be on any engines without traps; and the marginal
costs of removing a ton of NOx (including the cost of compensating for increased
PM production by increasing the trap percentage) are equal across each average-
able set.
Total costs for each averageable set may then be found by adding the
increased fuel consumption costs for each engine type, given its optimal NOx
level, and adding the costs of the traps used.
Cost allocations within an averageable set (especially across different
manufacturers in cases in which trading is permitted) may be found by adding
to the direct costs for each engine type the cost of purchasing enough credits
to bring that engine type into compliance (with compliance defined as either
emitting few enough tons to meet standards if the standards had been expressed
on a per-truck basis, or holding enough credits to emit sufficient tons of
pollutants to make up the difference). The number of tons' worth of credits
purchased for each engine is equal to the difference between the engine's
emissions per BHP-hr and the target for the industry, times the number of
BHP-hrs expected to be generated by that engine over its life. The price per
credit is assumed to be equal to the marginal cost of generating more credits.
Because the average emissions are set it the target for the industry, net
expenditures on credits across the averageable set are zero.
-------
Appendix B
BACKGROUND MATERIAL FROM ERC
This appendix contains a detailed description provided by Energy and
Resource Consultants, Inc. (ERC) of the engine classification system used in
this report. It also provides a discussion of the development of the functions
relating emissions to each other, and to fuel consumption.
Not all of the engine families cited as typical of particular subclasses
are still in production. This fact does not change the applicability of the
functional relationships of NOx to fuel consumption or PM emissions since the
functions were based on projected technologies, rather than exclusively on
current practice.
-------
Energy and Resource Consultants, Inc.
BVBMOTJI
to: Barry Galef (SCI)
'ran: Christopher Weaver (SRC)
iubject: Classification of heavy-duty diesel engines
; have developed a classification scheme for heavy-duty engines for use in our
rock on the benefits of emissions averaging. This scheme includes six classes
if heavy-duty diesel truck engines, 1 class of diesel bus engines, and 2
lasses of heavy-duty gasoline engines, totalling nine classes in all. • The
ilasses are described below.
^
1. Light-Heavy Duty mi Engines; These are a relatively new class,
offered mainly in trucks of Classes 2b through 4 (as well as in light
trucks) . The two major examples are the GM 6.2 liter and the IH 6.9
liter engine, although Onan also produces an engine in this class.
These engines resemble passenger-car diesels in characteristics.
2. Light-Heavy Duty PI Engines; These are just beginning to cone on the
market — examples include the DI diesels used in the new Isuzu Class 3
trucks and the IVECO 8060. A number of engines of this class have been
developed in Europe and Japan, and we can expect to see more of them
over the next decade.
3. fit-andard Medium-Heavy Engine*. These engines are used in a variety of
medium-heavy applications, mostly in Class 7 and the bottom of Class 8.
Typically naturally aspirated and with less power, less durable, and
less efficient than the remaining classes. The naturally-aspirated
version of the Caterpillar 3208 is in this class, as is the IH 9.0 liter
engine.
4. Premium Medium-Heavy Engines* These engines are also used mostly in
Class 7 through the bottom of Class 8. They differ from the standard
medium-heavy engines in incorporating more "heavy-truck* features in
their design, including turbocharging and improved fuel injection
systems. The IH 466 engine is an example of this class.
5. Non Line-Haul Heavy-Heavy Engines; These engines are generally built
on the same blocks as the line-haul engines, but include changes in
calibration and accessories (such as turbcchargers) to make them more
fit for stop-and-go operation, typically in rough service such as
dump-trucks and logging trucks. Fuel-economy is less of an issue with
these applications, and high power at a broad range of RJW is more
important. Some manufacturers distinguish these as "vocational11
engines.
6. Mn«wHjiui Truck Pngtn»«. These engines are large, heavy, and highly
fuel-efficient. They are optimized for best performance in highway
cruising, generally with low rated REH. Turbcchargers and other
accessories are also optimized for best performance at cruising speeds.
-------
Energy and Resource Consultants, Inc.
Some manufacturers identify these as "economy* engines. Cumnins refers
to them as "Formula* engines.
7. BjaJUaiOfia: These are heavy-duty truck engines adapted Cor use in
transit buses. These adaptations include provision for rear-mounting,
derating to reduce smoke, and possible changes in fuel-pump calibration
to use Diesel 1 rather than 2. There are probably differences in
turbocharging as well. Examples are the Cumnins NHHTC and the various
CDA "coach" models.
8. Lighfc-Bgjwy r?uty (ilflflolinf* Enqi-neg; These basically resent? le
light-truck engines, incorporating few "heavy-duty11 features except size
and power output. They are not intended for long running at high power,
and are found mostly in Class 2B through 4. Many axe also used in light
trucks.
9. Medium-Heavy Duty p*anifpg Engine; These are true heavy-duty gasoline
engines, normally incorporating features such as heavy-duty valves,
hardened valve-seat inserts, governors, etc. They are used mostly in
trucks of Classes 5 through 7. EPA projections (with which I agree) are
that these engines will probably almost die out in the next decade,
having been replaced by diesels.
There may be some practical difficulties in distinguishing the different
diesel-engine classes from each other — especially in distinguishing between
Classes 5 and 6. If the differences in control costs for these two classes
are sufficiently low, we might want to combine them. There is also some
grading-together of Classes 4 and 6, but this should be easier to handle.
-------
Energy and Resource Consultants, Inc.
THFf TPT
u flhlnq
Indirect Injection
Typically -128 to 150 horsepower
Not rated for continuous full power
Short lifetime — about 100,000 to 150,000 miles
TVPJe
Class 2B through 4 light-heavy trucks
Heavier trucks up to 2B,000 GVW in light service
Detroit 6.2 1
IH 6.9 1
Indirect injection with prechairber optimised for low emissions
Natural aspiration or turbocharginy (no aftercooling)
Distributor injection pump with full electronic governor
Closed-loop electronic injection timing control
Ceramic monolith/additive system looks most promising.
Hay require an ignitor as well.
-------
Energy and Resource Consultants, Inc.
ahin
Direct Injection
Typically- 120 to 150 horsepower
High rated speed — 2600 to 3600 RIM
Weight range typically 600 to 1000 pounds
Not rated for continuous full power
Moderately short lifetime
Commercial trucks in Classes 2, 4, and 5.
Possibly Class 2B as well.
Light Buses such as school buses.
ISU2U 3.9 1
IVEOO 5.5 1
Cunmins 3.9 1 "B* engines
TechnolooipH
Turbocharging with jacket-water after cool ing
High-pressure in-line injection pump
Electronic governor control
Closed-loop electronic injection timing control
Electronically-modulated BGR at low NDx levels
Little information on traps in this class.
Ceramic-acnolith/additive system or catalyzed trap upstream
from turbochager look most promising.
-------
Energy and Resource Consultants, Inc.
Naturally aspirated
Typically 150 to 250 horsepower
Usually not rated for continuous full power
Moderate rated speeds — typically 2000 to 2600 RIM
Moderate lifetime — typically 250,000 miles
Weight typically 1000 to 1500 pounds
Medium-heavy trucks from 16,000 to 50,000 pounds GVW
Usually not in high-speed service (e.g. garbage trucks/
snail dump trucks, delivery trucks, school buses).
Detroit 8.2 1 MA
Caterpillar 3208 NA
IH 9.0 1
High-pressure in-line injection pump
Mechanical timing advance
Mechanical governor
Improved breathing — 4 valves/cylinder
Mechanically-modulated HER at low NDx levels
Traa
Ceramic monolith/additive system looks most promising.
-------
Energy and Resource Consultants, Inc.
BMfST1>ES
Turbocharged
Typically 170 to 250 horsepower
Usually not rated for continuous full power
Moderate rated speeds — typically 2000 to 2600 MM
Moderate lifetime — typically 250,000 milea
Weight typically 1000 to 1500 pounds
Medium-heavy duty trucks from 16,000 to 60,000 pounds GVW.
Detroit 8.2 1 turbocharged
Caterpillar 3208 turbocharged
IH DT and OTI 466
Turbocharging with jacket-water aftercooling
High-pressure in-line injection punf> with electronic governor
Closed-loop electronic injection timing control
Electronically-modulated BGR at low NDx levels
Ceramic monolith/additive, Daimler-Benz trap/additive, or catalyzed
monolith upstream from turbocharger lock most promising.
Daimler-Benz system was developed for this class.
-------
Energy and Resource Consultants, Inc.
r.TMR-HAnr
Low or very low rated speed — 1600 to 2000 RIM
Power output typically 250 to 450 horsepower
Low to moderate torque rise
Optimized for best efficiency at highway speeds and near full
load
Designed and rated for continuous operation near full power
weight typically 2000 to 3000 pounds
Very long design lifetime — 250,000 to 400,000 miles before
rebuild, and rebuildable indefinitely.
P -
Line-haul trucking
Cunmins "Fleet", "Formula", and big-cam NTC models
Detroit 6V-92 and 8V-92 (1800 and 1950 REM) models
Caterpillar 3306 and 3406 "Economy" models
Most Mack models
Emissions—
Turbocharging with cold-charge aftercooling
Ultra-high pressure in-line pump or unit injectors
Full electronic governing
Closed-loop electronic timing control
Particulate Trars
The ceramic monolith/burner system looks most promising,
because of potential for long life. The Daimler-Benz and
monolith/additive systems are also possibilities.
-------
Energy and Resource Consultants, Inc
rrJLQg Kt HEAVY-HEAVY * •VnTAfPTflMAL* f tPM T.TM^-HJUTT.
Moderate rated speed — 2200 to 2400 REM
Power output typically 250 to 550 horsepower
High torque rise and good efficiency over a vide range of
speeds.
Designed and rated for continuous operation near full power
Weight typically 2000 to 3000 pounds
Very long design lifetime — 250,000 to 400,000 miles before
rebuild, and rebuildable indefinitely.
s
On/off road operation — logging trucks, dimp trucks, trash
trucks, heavy farm trucks, other specialized applications.
Hauling heavy loads (e.g. gravel) in stop-and-go or short-haul
service.
damans Power-Torque, Twin-Turbo 475, and KT Engines
Hack 2-Valve engines
Detroit 8V-71, 6V92, and 8V-92 hlgh-RFM nodels
Caterpillar 3306, 3406, and 3408 •Vocational* ratings
go i
Turbocharging with jacket-water aftercooling (some will have
cold-charge cooling)
Ultra-high pressure in-line punp or unit injectors
Full electronic governor
Closed-loop electronic tuning control
Small amount of BGR at low NDx levels
Monolith/burner, monelicVadditive, or Daimler-Benz systems
look promising.
-------
Energy and Resource Consultants, Inc.
7; BDS
Moderate horsepower heavy-heavy engines specialized for use in
transit buses.
Transit buses
Intercity buses
Detroit 6V71 and 6V92 "Coach" Series
Cummins NUHIC
Bniaflionfl— Related Technologies Aasmned
Turbocharging with jacket-water aftercooling
Ultra-high pressure in-line punf> or unit injectors
Full electronic governor calibrated for minimal particulate
emissions (at the expense of performance)
Closed-loop electronic injection timing control
Monolith/additive system, possibly with additional igniters,
is most promising. Catalyzed monolith upstream from turbo and
Daimler-Benz system are also possibilities.
-------
Energy and Resource Consultants, Inc.
PDW GA.qnT.TTJE
Gasoline fueled
Low to moderate horsepower
Not rated for continuous high-power operation
Lack of "heavy-duty* features such as sodium-filled valves,
valv^seat inserts
Low durability — typically 100,000 miles
Basically similar to light-duty truck engines.
Light-heavy trucks in classes 2B, 3, and 4.
Class 5 trucks operating under light loads.
Tial
Ford 240, 300, 360
01 292, 307, 454
All Chrysler UDGE
Technoloj
Fuel injection
Exhaust-gas recirculation
(Possibly) Three-way catalysts at very low NDx levels
Not applicable
-------
Energy and Resource Consultants, Inc.
rTA.cs Q< MvnTrM-HPJWY nmv GASOLllE EICI1ES
niotlrvnilahina
Gasoline fueled
Typically moderate horsepower
Bated for continuous or near-continuous full-power operation
Incorporate "heavy-duty" design features such as valve-seat
inserts and sodium-filled valves
Ami
Medium-heavy trucks under moderate to severe service
A few heavy-heavy trucks in light to moderate service
Ford 370 and 429
CM 250, 366, 427 and 454
Host International Harvester HUGE
Technolo iea
Fuel injection
Exhaust-gas recirculation
Not applicable
-------
Energy and Resource Consultants, Inc
ESTIMATES OP COST FUNCTIONS BY ENGINE CLASS
Estimates of trap-oxidizer costs and efficiency the relationships between NDx
and engine-out particulars and NOx and fuel-econooy were developed separately
for each of seven classes of heavy-duty diesel engines. The NDx/fuel-econony
relationship was also estimated for two classes of heavy-duty gasoline
engines. This was necessary, since different classes of engines have
different speed and operating characteristics, and thus different emissions
patterns. The feasibility of emissions-related technologies such as
turbocharging and aftercooling also varies between engine classes. Tables
XX.1 through XX.9 list the salient characteristics of each engine class, sane
typical engine models within each class, and the emissions-related
technologies which were assumed in estimating each class1 cost functions.
Estimation of trap-oxidizer costs and efficiency relied primarily on earlier
work by one of the authors. The rationale and assumptions involved have been
documented elsewhere (Weaver, 1984b). The estimates of NOx/particulate and
NDx/fuel-econony tradeoff functions relied primarily on reports of
manufacturer's tests on development engines. These reports are mostly
confidential, and so cannot be cited directly. Analysis and application of
data from individual tests involves a great ^yl of engineering judgement, and
tradeoff curves shown are not, in general, those for any particular existing
engine. Rather, they show the emissions and fuel-econoy levels which are
estimated to be achievable by engines of each class in model year 1990.
The estimates developed represent a major improvement over the capabilities of
present-day production engines. Such estimates are necessarily speculative,
and (given the rapid progress that has been made in the last few years) they
may be overly pessimistic. On the other hand, unexpected delays in
development or the failure of some technologies to fulfill their projected
potential could reveal these estimates as having been overly optimistic
instead.
-------
Appendix C
INPUT DATA 'FOR COMPUTATION OF SAVINGS FROM REGULATORY FLEXIBILITY
Definition of Engine/Use Classes Covered In This Appendix
Class
LHDGE
LHDDE-IDI
LHDDE-DI
MHOGE
MHDDE-NA
MHDOE-TC
HHDDE-LH
HHDDE-NLH
ERC Number Definition
8
1
2
9
3
4
5
6
Light-heavy duty gasoline engine
Light-heavy duty diesel engine—indirect injection
Light-heavy duty diesel engine—direct injection
Medium-heavy duty gasoline engine
Medium-heavy duty diesel engine—standard/naturally aspirated
Medium-heavy duty diesel engine--premium/turbocharged
Heavy-heavy duty diesel engine—line haul
Heavy-heavy duty diesel engine—non-line haul/vocational
These classes are described in Appendix B.
Data Used in Costing and Optimizing Model
BHP-hrs/Truck:
LHDGE LHDDE-IDI LHDDE-DI MHDGE MHDDE-NA MHDDE-TC HHDDE-LH
78,540 86,46086,460 164,450 338,365 364,820 788,800
Source: BHP-hrs/mile (below) times miles per truck (page C-2).
HHDDE-NLH
788,800
BHP-hrs/Mile:
LHDGE LHDDE-IDI LHDDE-DI MHDGE MHDDE-NA MHDDE-TC HHDDE-LH HHDDE-NLH
0.714
0.786
0.786
1.495
1.829
1.972
2.720
2.720
Source: EEA^/ (Appendix B) provides BHP-hrs/mile for gas and diesel trucks by
MVMA weight class for 1987 & 1992. Linear interpolation was used to estimate
1991. To transform the data on MVMA classes into the ERC classes used in this
report, MHDDE-NAs were assumed to be combinations of MVMA Class 6 and 7 (weighted
by projected sales of diesels in these classes) and MHDDE-TCs were assumed to
be half of MVMA Class 6 diesel, half of MVMA Class 7 diesel, and all of MVMA
Class 8-1. HHDDE-LH and HHDDE-NLH were assumed to be MVMA Class 8-2. ' 3oth
LHDDE-IDI and LHDDE-DI were assumed to have the same BHP-hr/mile as MVMA Class
4 diesel. MHDGE was estimated as a sales-weighted average of the BHP-hr/mile of
MVMA Classes 6 and 7. LHDGE was assumed to have the same BHP-hr/mile as MVMA
Class 2B-4.
I/ Historical and Projected Emissions Conversion Fractor and Fuel Econo
for Heavy-Duty Trucks labZ-ZUUZ, Motor Vehicle Manufacturers Association of t
United States, Inc., prepared by Energy and Environmental Analysis, Inc. (EF
December 1983. /
%t
/LJ
>/7
-------
C-2
1991 Sales by Class:.
LHDGE LHDDE'-IDI LHDDE-DI MHDGE MHDDE-NA MHDDE-TC HHDDE-LH HHDDE-NLH
211,663 132,833 44,278 169,332 29,838 51,305 77,404 38,702
Source: Bob Johnson, EPA, Office of Mobile Sources, Ann Arbor, Michigan,
February 1986.
Useful Life Miles/Truck:
LHDGE LHDDE-IDI LHDDE-DI MHDGE MHDDE-NA MHDDE-TC HHDDE-LH HHDDE-NLH
11U.UUU 110,000 110,000 110,000 185,000 185,000 290,000 290,000
Source: EPA, Office of Mobile Sources, Ann Arbor, Michigan, February 1986.
Cost per One Percent Increase in Fuel Consumption:
LHDGE LHDDE-IDI LHDDE-DI MHDGE MHDDE-NA MHDDE-TC HHDDE-LH HHDDE-NLH
$0 $54 $54 $0 $259 $259 $705 $705
Source: Regulatory Impact Analysis, March 1985, op. cit., pages 2-29 and 3-42.
Fuel Consumption Impacts of Traps:
A 1.25 percent fuel consumption penalty for traps was assumed for all
heavy duty diesel engines. This represents the midpoint of the 1 to 1.5 per-
cent range based on the use of trap-oxidizer systems using ceramic monolith
substrates and fuel burners for regeneration; if the same traps were used with
an electric regeneration, it is likely the penalty would be about the same.
However, if the ceramic fiber trap is used, the fuel penalty would be somewhat
less: 0.5 to 1.0 percent.
Source: Regulatory Impact Analysis, March 1985, op. cit., page 3-86.
Discounted Costs per Trap:
LHDGE LHDDE-IDI LHDDE-DI MHDGE ' MHDDE-NA MHDDE-TC HHDDE-LH HHDDE-LH
NA $370 $370 NA $448 $448 $574 $574
The ceramic monolith/fuel burner trap oxidizer is used to determine the trap
costs for the particulate standard, due to the greater uncertainty associated
with the other designs. A 10 percent discount rate is assumed over the life of
the engine. If another trap oxidizer design were used which was more expensive,
for example, cost savings results (savings from trap avoidance) would be greater.,
Source: Regulatory Impact Analysis, March 1985, op.cit.. page 3-82.
-------
C-3
Trap Efficiency:
80% of engine-out PM emissions are removed.
Source: Regulatory Impact Analysis, March 1985, op.cit.. page 2-65.
1991 Projected sales shares by manufacturer:^/
LHDGE LHDDE-IOI LHDDE-DI MHDGE MHDDE-NA MHDDE-TC HHDDE-LH HHDDE-NLH
Bluebird
Chrysler 16.1
Ford I/ 9.7
Cummins
Caterpillar
Daimler-Benz
KHD
GM 74.2 51.4
Navistar 45.6
Hi no
Deere
Mack
Onan 0.6
Perkins
Renault 0.4
Saab
Isuzu 1.9
Iveco
Volvo
White
40.0 59.5
55.0
36.1
1.0
40.5 36.5
17.7
2.7
2.5 1.4
4.5
2.5
1.6
16.8
5.2
0.2
21.5
26.9
0.8
0.6
0.6
9.1
0.8
6.3
1.0
3.4
5.2
59.1
13.7
0.1
0.2
11.4
15.0
0.6
59.1
13.7
0.1
0.2
11.4
15.0
0.6
These are rough projections based on historical shares (from EPA sales
fractions for engine families), estimates of market penetration by new engines,
and data on sales volumes of gasoline powered trucks and imports.
These share projections are not precise enough to be considered predictions
of market shares of individual firms, e.g., Cummins, in the future. They do,
however, allow the analysis to reveal, for instance, what is likely to happen
to a "Cummins-like" firm--one with a large share concentrated in the heaviest
diesels but with some presence in other segments.
Totals may not add to 100 percent due to rounding.
Source: (see page C-4)
}_/ (see page C-4)
£/ (see page C-4)
-------
C-5
Functional Forms and Parameters:
Equations relating NOx levels to changes in fuel consumption and to PM
emissions were derived by selecting parameter values for hyperbolic functions
that defined curves closely fitting the point estimates of the relationships
made by ERC. For the relationship of NOx emissions to increases in fuel con-
sumption, the functional form used in the analysis is as shown:
% increase in fuel consumption a + b/(c + NOx)
NOx
115.9
800
4
6.15
-427.87
7100
15
15
-369.12
6200
15
12.5
where NOx is emissions of NOx measured in grams/BHP-hr, and a, b, c, and d are
parameters with values that depend on engine class:
LHDGE LHDDE-IDI LHDDE-DI MHDGE MHDDE-NA MHDDE-TC HHDDE-LH HHDDE-NLH
a -57.37 -11.6 -115.9 -57.37 2.41
b 500 24 800 500 15.3
c 6 -1.5 4 6 -2.5
d 3 1 6.15 3 -0.63
Fuel consumption functions at the level of detail shown in Exhibits II-l
and 1 1-2 were not addressed in EPA's RIA, so the ERC functions were used.
However, EPA did make point estimates of fuel consumption impacts for the 5.0
g/BHP-hr standard, and these are lower than those developed by ERC. The primary
reason for the differences lies in the area of how technological advances by 1991
are considered. EPA's analysis concluded that the next five years would bring
improvements to overcome most of any fuel consumption penalty associated with
the 1991 standards. ERC's analysis also considered technological improvement,
but evaluated fuel consumption increases as foregone gains.
While there are differences between EPA's and ERC's analyses, they do not
affect the results of this study, since a consistent set of input values was
used for all scenarios addressed here to evaluate the incremental benefits
of trading and expanded averaging.
For the relationship of NOx emissions to PM emissions, the functional form
used in the analysis is as shown:
PM
f/(g + NOx)
where NOx is emissions of NOx measured in grams/BHP-hr, and e, f, and g are
parameters with values that depend on engine class:
LHDGE LHDDE-IDI LHDDE-DI MHDGE MHDDE-NA MHDDE-TC HHDDE-LH HHDDE-NLH
e
f
9
0.407
0.134
-1.8
0.15
0.85
-1.9
0.29
0.65
-2.1
0.18
0.60
-2.1
0.16
0.60
•1.5
0.15
0.67
-1.8
Source of point estimates that were the basis of the parameter estimates: Appen
dix B and Sections II-B and INC.
-------
Appendix D
DETAILED RESULTS
This appendix presents detail on costs and emissions by firm and by subclass
for each of the scenarios.
Full data for each firm is presented separately for the three scenarios
in which averaging but not trading is allowed. (The data shown for the industry
is not the sum or average of data for the firms since it refers to the situation
under trading.) Under trading, the industry is treated as if it were combined
into a single, large firm: all data presented for the emissions, trap usage,
and per-engine costs for the industry as a whole are valid for each firm, and
all total cost figures presented for the industry become valid for individual
firms when they are adjusted for the firms' shares in the sales of the industry.
For the scenarios in which averaging is restricted to transactions among limited
sets of engine types, subtotals are shown for each subclass.
The first column identifies the engine type/use class for the row of data
in the other columns. These classes are defined in Appendix B.
The next two columns show the optimal grams of NOx' under averaging for
engines without and with traps, respectively. Emissions of NOx are kept somewhat
.higher for engines without traps to minimize PM emissions. For engines equipped
with traps, the engine-out emissions of PM are less important, because the
traps eliminate most of thenr.
The next column shows the percentage of engines that would be fitted "ith
traps. The traps are placed mostly on the larger engines to minimize the
trap cost per ton of PM removed by them.
The next two columns show the costs of emissions controls per engine. The
first, labeled "Uithout Credits," shows only the direct costs of the traps and
-------
D-2
the fuel consumption Increases at the emissions levels found to be optimal
under averaging (or trading, In the case of the industry-wide data); the second,
labeled "With Credits," includes the costs of "purchasing" credits for NOx or
PM emissions in excess of the target. Under averaging, the credits are not
really purchased, since all transactions are intra-firm. These results are
presented as though intra-firm markets for credits existed, with accounting
prices set at the marginal costs of reducing emissions.
The next two columns show the costs per engine of purchased credits for
NOx and PM, respectively. Again, these costs are internal accounting costs
only under averaging, and the total expenditures sum to zero for each firm.
The next two columns present total costs, in millions of dollars, for all
of each firm's sales in each class. The first of the two columns shows the
totals under averaging of NOx and PM without adjusting for credit transactions,
while the next column takes these transactions into account. The total costs
over all classes are shown at the bottom of the column. It is the same whether
or not credit transactions are included, since the transactions within a single
firm sum to zero.
The final column shows annual sales by class and by firm.
-------
DETAILED RESULTS: ONE CLASS AVERAGING AND TRADING
NUX U11»1UN» -
(g/BHP-hr)
FIRM ENGINE
TYPE •
No Trap
•
Trap
rn
TRAP
USAGE
(X)
COSTS PSK TKUCK -----------
Without
Credits
With
Credits
Credit Purchases.
For NOX
Foe PM
1UIAL. U»19
(millions)
Without
Credits
With
Credits
YEARLY
SALES
INDUSTRY "
LUDGE
LHDDE -
LHDDE -
HHDCE
MHDDE -
MHDDE -
HHDUE -
HHUDE -
IDI
DI
NA
TC
LH
NLH
2 25
3.67
5 06
2 21
5 0*
5 17
4 98
1 19
na
3.60
4 73
na
4 80
4 95
4 82
* 97
na
0 00
0 00
na
1 00
0 00
0 49
1 00
$407 2
169 0
189 1
857 2
2.337 0
810 8
2.403 3
3,912 6
($24.8)
193 1
517 8
(66 4)
2.644 3
2,199 7
3.884 7
4,816 6
($432 0)
(129 3)
211 0
(923 6)
571 3
1,000 3
1.568 4
1, 702 b
na
153 4
117 7
na
(264 1)
388 6
(86 9)
(79B B)
$86 2
22 5
8 4
145 2
69 7
41 6
186 0
151 4
($5 3)
25.6
22.9
(11.2)
78 9
112 9
300 7
186 4
211.668
132.833
44.278
169.332
29.838
51.304
77.405
38.703
Industry Total
$710 9 $710 9 755,361
1 Bluebird
HHDCE
4 20
na $236 2 $236 2 $00
na $0 003 $0 003
13
O
to
2 Chrysler
LHDCE
4 20
na $114 7 $114 7
$0 0
$3 9
$3 9 34.016
3 Ford
LHDCE
LHDDE - DI
MHDCE
Ford Total
4 Cu
nlns
LHDDE - DI
MHDDE - TC
HHDDE - LH
HHDDC - NLH
Cumnlns Toi .1
4 OS
6 65
4 01
4 38
4 51
4 29
4 42
na
6 07
IIJL
4 04
4 26
4 09
4 14
na $129 6 $114 1
0 42 208 5 450 2
na 273 7 234 4
U 00 $351 5 $557 2
U 00 1,519 6 2.4j? /
0 36 3.9J9 6 4,234 0
1 00 6,0*4 8 5,253 9
($15
241
(39
$63
43/
63
(210
4)
7
4)
5
7
3
0)
0
$142
460
231
(630
na
0
na
2
5
1
3)
$2
3
27
$33
$8
1
180
139
7
7
6
.9
6
1
3
4
$2
8
23
$33
$13
2
193
120
4
0
.6
9
6
0
7
2
20
17
100
139
24
45
22
.620
.711
.678
,010
.353
827
, 758
.879
$329 5 S329 5 93.817
As noted In ilie text, nic dm presented for I he Industry Is not the weighted average of the data for the
firms This It because the dat* tor the Induitrv shows the situation under trading, and the data for the
firms shows the situation under aveiaglng
-------
DETAILED RESULTS. ONE CLASS AVERAGING AND TRADING
FIRM ENGINE
NUJC HUSSIONS -
(B/BHP-hr)
TYPE • No Trap Trap
I'll
TRAP
USAGE
(X)
COSTS PER TKUbK -
' Without Ulth
Credits Credits
Credit Purchases.
For NOX For PH
(millions)
Ulchout Ulth
Credits Credits
YEARLY
SALES
5 Caterpillar
HHDDE -
HHDDE -
HHDDE -
HUDDE -
NA
TC
LH
NLH
4 45
4 45
4 22
* 35
4.30
4 20
4 02
4 06
1 00
0 00
0 44
1 00
$2.892 1
1,59? 3
4,308 8
6.347 5
$2.832 8
2.4«4 2
4,228 8
5.252 1
$147 5
379 5
(216 6)
(484 0)
($206 8)
467 4
136 6
(611 4)
Caterpillar Total
6 Daimler
HHDDE -
HHDDE -
HHDDE -
HHDDE -
-fienz
NA
TC
LH
NLH
4 40
4 36
4 13
4 24
4 26
4 11
3 92
3 94
1 00
0 60
1 00
1 00
$2.958 7
2.391 7
5.829 3
6.720 6
$2,844 6
2.442 9
4.213 7
5,239 9
$83 6
14 )
$31 2
13 7
45 6
33 b
$124 1
$0 8
6 4
(J 3
0 1
$7 6
$0 3
0 5
0 4
$1.2
$36 4
6.9
33 6
19.9
4 0
18 7
11 8
$30.5
21 0
44 8
27.8
$124 1
$0 8
6 5
0 2
0 1
$7 6
$0 2
0.6
0 3
$1-1
$13 5
16 0
11.8
25.5
19 8
27 6
17 3
10.783
8.600
10.584
5,292
35.259
284
2.6/3
44
22
3,023
99
132
66
298
157.032
68.335
68,641
10.899
11.031
8.800
4.400
o
-p.
General Hulors Total
$131 4 S131 4 329.13H
-------
DETAILED RESULTS. ONE CLASS AVERAGING AND TRADING
nux uiiaaiuna ~
(g/BHP-hr)
FIRM ENGINE TYPE * No Trap Trap
9 Navistar
LHDDE - IDI 3.59 3.47
HHDDE - NA 5 01 4 62
HHDDE - TC 5 11 4 71
Navistar Total
10 HLao Hotors
HHDDE - NA 4 33 4 20
HHDDE - TC 4 24 3 99
Hino Motors Total
11 Deere
HHDDE - TC 4.36 4 11
12 Hack
HHDDE - LH 4 29 4 09
HHDDE - NLH 4 42 4 14
Hack Total
13 Onan
LHDDE - IDI 4.30 4 15
14 Peiklns
TRAP
USAGE Without With Credit Purchases Without
(Z) Credits Credits For NOX For PM Credits
0 24 $298 1 $306 6 ($184 4) $192.9 $18 0
1 00 2.508 8 2,499 3 478 6 (488 1) 13 3
1 00 2.043 9 2.010 2 624 0 (657 6) 28 3
$59 6
1.00 $3.046 8 $2.853 7 ($5 4) ($187 8) $2 4
0 IS 2.049 2 2.436.6 10 7 376 7 08
$3 2
0 64 $2.442 6 $2.442.9 ($0 0) $03 $08
0 30 $3,814 8 $4,234 0 $104 0 $315 2 $44 4
1 00 6.092 3 5,253 8 (207 9) (630 5) 35 4
$79 8
0 66 $367 6 $367 0 $0.0 ($0.7) $0 3
(millions)
With
Credits
$18 6
13 2
27 8
$59 6
$2 3
1 0
$3 2
$0 8
$49 2
30 5
$79 8
$0.3
YEARLY
SALES
60.529
5.292
13.826
79.647
794
397
1.191
331
11.627
5.814
17.441
794
o
in
MHDDE - TC
4 36
411 0 64 $2.44.' 0 $^.442 9
SO U
$08
$0.8
$0 8
331
-------
DETAILED RESULTS. ONE CLASS AVERAGING AND TRADING
NOx EMISSIONS -
<8/BHP-hr)
FIRM ENGINE TYPE • No Trap Trip
PH COSTS PER TRUCK -
TRAP
USAGE Ulthout With
(X) Credits Credits
..---_-.. TOTAL COSTS
(millions)
Credit Purchases: Ulthout With
For NOX For PH Credits Credits
YEARLY
SALES
IS Renault
LHDDE - IDI
MHDDE - TC
Renault Total
3 30
4 39
3.25
4 1*
0 00 $272 8
0 66 2,416 5
$70 8
2,442 4
($340 t)
43 4
$138 6
(17 5)
$0 2
11.3
$11 4
$0 0
11 4
$11 4
595
4.670
$5,266
16 Saab
HIIDOE - TC
4 36
4 11
0 64 $2,442 3 $2,442 9
$0 U
$0 5
$1 0
$1 0
397
17 Isuzu
LHDDE - IDI
LIIDDE - DI
MHDDE - NA
MHDDE - TC
Isuzu Total
18 Iveco
MHDDE - NA
MHDDE - TC
Iveco Total
19 Volvo
MHDDE - TC
HHDDE - LH
HHDDE - NLH
3 35
4.39
4 50
4 52
4 32
4.23
4 43
4.21
4.33
3 30
4 06
4 35
4 28
4 19
3 98
4 18
4 00
4 04
0 00
0 00
1 00
0 80
1 00
0 01
0 29
1 00
1 00
$254
347
2,833
2,376
$3.058
1,910
$1.938
5.538
6.406
0
6
6
7
6
7
2
1
1
S97
556
2.823
2.431
$2.854
2.434
$^.441
4.231
5,255
b
1
2
7
5
a
i
9
5
(S^'JB
67
201
188
($17
45
$248
(b57
(548
3)
8
1
4
7)
3
5
8)
5)
$141
140
(211
(133
($186
478
$254
(648
(602
8
7
5)
3)
.4)
9
3
4)
2)
$0 7
0 4
1 2
7 7
$9 9
$4 1
1 0
$5 2
$3 3
2 5
1 5
$0 3
0 6
1 2
7 9
$9 9
$3.9
1 3
$5 2
$4.2
1 9
1 2
2,580
1,107
430
3.228
7,345
1,356
529
1.885
1,720
459
229
Volvo Total
$7 J
$7 3
2.408
-------
DETAILED RESULTS: ONE CLASS AVERAGING AND TRADING
NO* EMISSIONS -
(g/BHP-hr)
FIRM ENGINE TYPE • No Trap Trap
PM COSTS PER TRUCK -
TRAP
USAGE Without With
(Z) Credit! Credit*
TOTAL COSTS
(millions)
Credit Purchases. Without Ulth YEARLY
For NOX For PM Credits Credits SALES
20 White
LHDDE - DI
MUDDE - TC
White Total
* 25
4.37
3.90
* 13
0 00
0 72
$390 6
2.510 8
$553.8
2,442 7
$17 2
(7 2)
$146 0
(60 9)
$0 4
6 6
$7.1
$0 6
6 5
$7 1
1.107
2.646
3.753
* ENGINE TYPES
LIIDCE
LIIDDE
LHDDE
MHDCE
MHDDE
MHDDE
HHDDE
HHDDE
IDI
DI
NA
TC
LH
NLH
O
I
Llght-Heavy-Duty Gasoline Engine
Llglii. Heavy-Duty Diesel Engine - Indirect Injection
Llghi-Heavy-Duty Diesel Engine - Direct Injection
MedluiD-Heavy-Duty Gasoline Engine
Medlum-Heavy-Duty Diesel Engine - Naturally Aspirated
Medlum-Heavy-Duty Diesel Engine - Turbo Charged
Heavy-Heavy-Duty Diesel Engine - Line Haul
Heavy-Heavy-Quty Diesel Engine - Non-Line Haul
-------
|