EPA Response to Commerce questions on Ozone RIA

We asked that EPA present a list of controls in the main text of the
report instead of the appendix.  Doing so would make the exposition
easier to follow.  We believe it is in EPA’s interest to make their
RIA as easy to understand as possible.

	List of controls that were used are provided in tables 3.1-3.3. Details
of the 	controls are in the appendix, so as not to overwhelm the body of
the RIA with too 	much detail.

EPA should also clarify in the text how they are using OIRA recommended
interest rates. 

	EPA is using the 3% and 7% discount rates for engineering costs as
indicated in 	OMB Circular A-4.  This is covered in section 5.1.2 with
the following text: 

The total annualized cost of control in each sector in the control
scenario is provided in Table 5.1.  These numbers reflect the
engineering costs across sectors annualized at a discount rate of 7% and
3%, consistent with the guidance provided in the Office of Management
and Budget’s (OMB) (2003) Circular A-4.  However, it is important to
note that it is not possible to estimate both 7% and 3% discount rates
for each source (see section 5.1.3). In Table 5.1, an annualized control
cost is provided to allow for comparison across sectors, and between
costs and benefits.  A 7% discount rate was used for control measures
applied to non-EGU point, area, and mobile sources.   Costs from EGU
sources, which are calculated using the IPM model and variable interest
rates, are captured in this table at an annualized 7% discount rate.  

Total annualized costs were calculated using a 3% discount rate for
controls which had a capital component and where equipment life values
were available.  In this RIA, the non-EGU point source sector was the
only sector with available data to perform a sensitivity analysis of our
annualized control costs to the choice of interest rate.  Sufficient
information on annualized capital calculations was not available for
area source and mobile controls to provide a reliable 3 percent discount
rate estimate. As such, the 3% value in figure 5.1 is representative of
the sum of the non-EGU Point Source sector at a 3% discount rate, and
the EGU, mobile, and Area Source sector at a 7% discount rate.  It is
expected that the 3% discount rate value is overestimated due to the
addition of cost sectors at a higher discount rate.   With the exception
of the 3 % Total Annualized Cost estimate on Table 5.1, cost estimates
presented throughout this and subsequent chapters are based on 7%
discount rate.

On extrapolated costs:  Regardless of where the economic impact
assessment is, it needs to incorporate whatever assessments of
extrapolated costs that EPA has estimated.  It is not appropriate to
separate out extrapolated costs from other costs in the analysis.  We
look forward to seeing the new version of this analysis before the rule
goes public.

	

	The modeled control costs and the extrapolated costs are very different
in terms of 	methodology and level of uncertainty.  The EIA has been
deleted from the RIA 	entirely.

On the size of the impact being small:  The estimated costs of the rule,
depending on what estimate you are looking at, is $7 billion. 
Economically significant rules are those whose economic effect is $100
million or more.  The impact of this rule is 70 times that threshold. 
It is one of the most expensive rules being considered this year.  The
costs should not be characterized as small.  

EPA did not address the question, “what are the total compliance costs
of this rule compared to the similar analysis done on the PM NAAQS?” 
Please include this comparison in the text to allow the compliance cost
of the present rule to be put in its proper context.

	This has been address in edits to Chapter 8: 

Tightening the ozone standards can incur significant, but uncertain,
costs 

An engineering cost comparison demonstrates that the cost of the 0.070
ppm Ozone NAAQS control strategy ($3.9 billion per year) is only
slightly higher than the Clean Air Interstate Rule ($3.6 billion per
year) and roughly one and half to just over four times higher than the
PM NAAQS 15/35 control strategy with annual engineering costs of $850
million. It should be noted that for the Ozone NAAQS $3.9 billion
represent the cost of partial attainment.  Full attainment using
extrapolation methods are expected to increase total costs
significantly.  Yet, the magnitude and distribution of costs across
sectors and areas is highly uncertain.  Our estimates of costs for a set
of modeled NOx and VOC controls comprise only a small part of the
estimated costs of full attainment.  These estimated costs for the
modeled set of controls are still uncertain, but they are based on the
best available information on control technologies, and have their basis
in real, tested technologies.  Estimating costs of full attainment
required significant extrapolation of the cost curve for known
technologies, and was based on generalized relationships between
emissions and ozone levels.  Based on air quality modeling sensitivity
analyses, there is clearly significant spatial variability in the
relationship between local and regional NOx emission reductions and
ozone levels across urban areas.  However, because we were unable to
analyze all of the urban areas that are expected to need reductions, we
used the same ratio of ozone to emissions throughout the U.S.  This
introduces significant uncertainty into the calculation of the emissions
reductions that might be needed to reach full attainment.  In addition,
because VOCs are generally much less effective than NOx in achieving
ozone reductions at key monitors (with the exception of California), we
did not use any VOC control data in the extrapolation to full
attainment.  This meant that in some areas, we assumed the need for more
expensive NOx controls than might be required if a specific area chose
to use a combination of NOx and VOC controls.  However, VOC controls
would have to be very inexpensive relative to NOx controls on a per ton
basis in order for VOC controls to be a cost-effective substitute for
NOx reductions.  Extrapolating costs by applying a cost-curve based on
known technologies also introduces uncertainties.  For some locations,
the extrapolation requires only a modest reduction beyond known
controls.  In these cases, the extrapolation is likely reasonable and
not as prone to uncertainties.  However, for areas where the bulk of air
quality improvements were derived from extrapolated emissions reductions
that go well beyond the area of the known controls, the increasing
marginal costs can suggest a cost per ton which stretches credibility. 
For example, in California, extrapolation to full attainment results in
a marginal cost for the last ton of NOx of $89,645 in Los Angeles and
$74,495 in Kern County, which are five to six times larger than the
marginal cost at the last known cost effective control.  Economic theory
would suggest that as marginal costs rise, research and development to
produce new, more cost effective technologies will also increase,
leading to a downward shift in the overall cost curve.  We did not
assume any shift in the cost curve to reflect technological innovation,
instead we provide a sensitivity analysis by showing estimates assuming
a high and low fixed cost per ton.  We are likely overstating costs in
the future when using the marginal cost and high fixed estimates.

On the EMPAX model:  Sensitivity analysis should show to what extent the
competitiveness results of the model run are driven by assumptions about
Armington and supply elasticities.  Rather than making a qualitative
statement about the relative conservative assumptions of the model, it
would be more informative for EPA to show how the results change in
response to changing key assumptions.

	The Economic Impact Analysis has been removed from the RIA, including
any 	reference to the EMPAX model.

On location issues:  EPA states that environmental regulations are not
important factors in location decisions.  Does EPA have empirical
studies that support this statement?  If not, EPA needs to run
sensitivity analysis to reflect how their assumptions of supply
elasticities drive the results of the model.

	The statement EPA is trying to make is that environmental cost is not
the only 	factor in industry location decisions.  They are one of many
important factors in 	industry location decisions.  This is in response
to comments from DOC that the 	environmental expenditures will ‘cause
industries to locate abroad’.  That	comment does not appear to
incorporate market conditions both abroad and 	domestic (labor, capital,
and perhaps foreign exchange) which may have a bigger 	influence than
environmental expenditures.  The comment also appears to assume 	that
the environmental expenditures will significantly alter the capital
labor ratio 	in the affected industries.  Other equally important
factors that need to be taken 	into account in location decisions are
tax incentives and environmental 	regulations in the relevant foreign
countries as well as overall economic and 	political stability of the
country.

Technological advance:  Despite philosophical objections, it would be
useful to run a model that assumes no technological advance to serve as
a worst case compliance cost scenario.  It will help put in context the
additional runs done for sensitivity analysis. 

	No.

On the marginal cost econometric estimations:  DOC looks forward to
receiving the data and a description of their econometric method from
EPA.

 	

Other DOC Comments:  We received no responses to the following comments:

Appendix 3, 3.2 Mobile Controls/Rules used in baseline and control
scenarios: 

3.3[SIC].2 Implement Continuous Inspection and Maintenance Using Remote
Onboard Diagnostics (OBD)

Comment: The prior section limited consideration to Class 6 and above
trucks.  For clarity, some note should made early in the first paragraph
of this section that light vehicles are the expected target of OBD for
this analysis.

	See edit in appendix 3 (page 3a-13, section 3.2.2)

3.2.4 Commuter Programs

We used the findings from a recent Best Workplaces for Commuters survey,
which was an EPA sponsored employee trip reduction program, to estimate
the potential emissions reductions from this measure.   The BWC survey
found that, on average, employees at workplaces with comprehensive
commuter programs emit 15% fewer emissions than employees at workplaces
that do not offer a comprehensive commuter program.  

Question:  Is it appropriate to assume that these results will carry
over to a larger universe of participants?  Is current participation in
BWC programs self-selected for highest possible results?  The take-up
rate should necessarily be lower as organizations become smaller and
populations become more dispersed since time costs of travel increase
and opportunities for pooling decrease. The study found that benefits
packages that don’t include financial incentives reduce emissions
about 7%.

We believe that getting 10-25% of the workforce involved in commuter
programs is realistic.

	EPA believes that the BWC survey methodology addresses these questions 
through the selection process for the cities and companies that were
included in 	the analysis.  The four metro areas selected for the study
(Denver, Washington, 	DC, San Francisco, and Houston) cover a range of
different types of commuter 	patterns that are representative of other
metropolitan areas.  In addition, the BWC 	companies that participated
in the study were randomly selected and represented a 	range of
different size and types of companies.  We agree that the participation 
rate could decline after the larger companies enroll, which is why we
limited the 	participation rate to 10-25% of the workforce.

Question:  What criteria were used to determine the likelihood of
workforce involvement?

	These rates were chosen based on the cost-effectiveness of the control
strategy 	and the need for additional reductions from the mobile source
sector.  See above 	response for more information.

Appendix 5, 5.3 Cost information for Onroad and Nonroad Mobile Sources:

Eliminating Long Duration Truck Idling, TSE

Since TSE technology can completely eliminate long duration idling at
truck spaces (i.e. a 100% fuel savings), this translates into 2,920
gallons of fuel saved per year per space. At current diesel prices
($2.90/gallon), this fuel savings translates into $8,468.

Comment:  While TSE technology can greatly improve efficiency, it does
not eliminate fuel consumption.  The electricity used to provide power
has fuel requirements and some associated emissions. 

	This is explained in our Guidance for Quantifying and Using Long
Duration 	TruckIdling Emission Reductions in State Implementation Plans
and 	Transportation Conformity:

	In the case of stationary TSE, for the purpose of this guidance, it may
be 	presumed that all emissions from power plants (including any
increase in 	demand resulting from a TSE project) will be accounted for
in projections of, 	or limits on, overall power plant  emissions in the
SIP's emission inventory. 	Therefore, related TSE emissions at power
plants should not be considered 	when quantifying the emission
reductions associated with a TSE project

Commuter Programs

We chose to apply the resulting average cost-effectiveness estimate to
one pollutant – NOx – in order to be able to compare commuter
reduction programs to other Nox reduction strategies. TRB reported the
cost-effectiveness of each measure, however, as a $/ton reduction of
both VOC and Nox by applying the total cost of the program to a 1:4
weighted sum of VOC and Nox [[total emissions reduction = (VOC * 1) +
(Nox * 4)).  There was not enough information in the TRB study to
isolate the $/ton cost-effectiveness for just Nox reductions, so we used
the combined Nox and VOC estimate.

Question:  Wouldn’t any calculation that reduces the result
necessarily make the results more accurate (such as dividing the result
in half or subtracting by the weighted value of VOC)?  Since the
cost-effectiveness could not be isolated, the cost-effectiveness of
controlling VOC could have contributed anywhere from 1 to 99 percent of
the reported result.  The combined estimate is the only one likely to be
biased.  TRB would have made no mention of VOC or reported it as such if
the results of any portion were 100% due to the cost-effectiveness of
controlling NOx.

	The 1:4 ratio of VOC:NOx used by TRB is related to the relative health
impact of 	these pollutants, and is not related to the proportion of one
pollutant or another 	that is reduced by the different control
strategies. Without knowing the actual 	emissions of each pollutant(NOx
and VOC), which could not be calculated from 	the available data in the
TRB report, we are not able to derive a better estimate 	than we have
presented in the RIA.  EPA will evaluate alternative approaches for 
addressing the VOC to NOx ratio as part of the final RIA.

