The Effects of New-Vehicle Price Changes
on New- and Used-Vehicle Markets
and Scrappage: Peer Review and
Response to Reviewer Comments
gPk United States
Environmental Protection
^1 Agency
-------
The Effects of New-Vehicle Price Changes
on New- and Used-Vehicle Markets
and Scrappage: Peer Review and
Response to Reviewer Comments
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
Prepared for EPA by
RTI International
EPA Contract No. EP-C-16-021
Work Assignment No. 4-28
This technical report does not necessarily represent final EPA decisions
or positions. It is intended to present technical analysis of issues using
data that are currently available. The purpose in the release of such
reports is to facilitate the exchange of technical information and to
inform the public of technical developments.
and
ICF International
EPA Contract EP-C-16-020
Work Assignment 4-04
NOTICE
&EPA
United States
Environmental Protection
Agency
EPA-420-R-21-020
August 2021
-------
August 2021
MEMORANDUM
SUBJECT: Peer Review for RTI Report, "The Effects of New-Vehicle Price Changes on
New- and Used-Vehicle Markets and Scrappage"
In March 2021, EPA contracted with ICF to conduct a peer review of a study conducted by RTI
International. The draft study, titled "The Effects of New Vehicle Price Changes on New and
Used Vehicle Markets and Scrappage," conducted a literature review and developed a method to
examine the effects of changes in the prices of new vehicles on new and used vehicle sales and
vehicle scrappage.
The peer reviewers selected by ICF were Drs. Ashley Langer of the University of Arizona,
Benjamin Leard of the University of Tennessee, and James Sallee of the University of California
at Berkeley. EPA would like to extend its appreciation to all four reviewers for their efforts in
evaluating this survey. The reviewers brought useful and distinctive views in response to the
charge questions.
The first section of this document contains the final RTI response to the peer reviewers'
comments. The second section provides the peer review report conducted by ICF. It documents
the peer review process, provides both a summary of the peer review comments and the detailed
responses, the peer reviewers' curriculum vitae, conflicts of interest declarations, and notes from
the peer reviewer mid-review meeting.
CONTENTS
I. Response to External Peer Review Comments on The Effects of New-Vehicle Price Changes
on New- and Used-Vehicle Markets and Scrappage, July 30, 2021, from RTI (Mark Jacobsen
and Robert Beach).
1. Comments by Dr. Ashley Langer
2. Comments by Dr. Benjamin Leard
3. Comments by Dr. James Sallee
II. Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle
Markets and Scrappage, June 11, 2021, from ICF
1. Introduction
2. Peer Review Process
3. Responses to Charge Questions
4. Resumes of Selected Reviewers
5. Conflicts of Interest Declarations
6. Peer Reviewer Mid-Review Meeting Notes
-------
I
Memorandum
INTERNATIONAL
To:
Gloria Helfand, U.S. Environmental Protection Agency
From:
Mark Jacobsen and Robert Beach
Date:
July 30, 2021
Subject: Response to External Peer Review Comments on The Effects of New-Vehicle Price
Changes on New- and Used-Vehicle Markets and Scrappage
The report "The Effects of New-Vehicle Price Changes on New- and Used-Vehicle Markets and
Scrappage" was submitted for independent, external peer review in April 2021. The external peer
reviewers provided their independent responses to EPA's charge questions. This memorandum documents
our responses to the comments provided by three expert reviewers.
We present the individual reviewer comments (verbatim) in response to the 6 peer review charge
questions in the following tables, organized by reviewer, along with our responses. In some cases,
reviewers provided comments in addition to their direct responses to the charge questions, each of which
are also addressed within this memorandum. We have responded to all the detailed comments
individually and indicate how the draft report was revised to response to peer reviewer comments in all
cases where revisions were made. The peer review report also included a summary of the review
comments, but we did not include separate responses to the summary comments because we responded to
each of the detailed comments upon which the summary comments are based.
Peer Review Charge Questions
1. Does the Presentation Describe the Data and Methods Sufficiently?
2. Does the Report Miss Relevant Literature; Are the Interpretations of the Elasticities Defensible?
3. Are the Data and Assumptions Appropriate and Objectively Chosen?
4. Are the Methods and Procedures Employed Technically Appropriate and Reasonable?
5. Does the Modeling Analysis Appear to Produce Results Consistent with the Assumptions and
Data?
6. Are the Results Sensitive to the Data and Assumptions Used in Model Development; Are There
Alternative Assumptions?
Expert Peer Reviewers
An EPA contractor identified and selected three reviewers who met the technical selection criteria
provided by EPA and who had no conflict of interest in performing this review:
Ashley Langer, Ph.D.
Assistant Professor, Department of Economics, University of Arizona
Rii International is a trade name of Research I riangie institute. Rli and the Rii iogo are U.S. registered trademarks of Research i riangie institute.
-------
Faculty Research Fellow of the National Bureau of Economic Research
Benjamin Leard, Ph.D.
Assistant Professor, Department of Agricultural and Resource Economics, University of Tennessee
Faculty Fellow, Howard Baker, Jr. Center for Public Policy, University of Tennessee
James Sallee, Ph.D.
Associate Professor, Department of Agricultural and Resource Economics, University of California,
Berkeley
We thank the reviewers for their very detailed and conscientious expert review of our report. Their
queries and recommendations have enabled us to improve the report by correcting oversights, adding new
material, and improving clarity.
Comments by Dr. Ashley T,anger
C IIARdi:
Ql ESTION
COMMENTS
RESPONSE
1. Does the
presentation describe
the data and methods
sufficiently to allow
the reader to form a
general view of the
quality and validity of
the analysis approach?
Yes. This is the best government
report I've ever read. The model is
laid out clearly and is in line with the
state-of-the-art research in the best
economics journals. The authors are
extremely clear about where each
parameter is coming from and why
they made the modeling assumptions
that they did. I feel like after reading
this report I could, with relatively
little additional thought, actually code
up the dynamic simulation in the
paper and test alternative models.
That is a feat in a technical report like
this.
We appreciate this positive feedback
on the clarity of the report.
2. Does the report
miss relevant
literature in its
review? Are the
interpretations of the
elasticities in the
literature review, and
the estimates of
elasticities used in the
model stemming from
I think that the authors overall have
gone well beyond just conducting a
literature review. They have
calculated elasticities using important
estimates from the literature and in
some cases have literally gone back
to the data and code from the papers
and calculated elasticities that are not
recoverable from only the published
version. They have also used good
judgement to choose which
elasticities from the literature are
Thank you for the suggestion of
additional papers to incorporate in the
discussion. We focused specifically
on U.S. papers in our assessment of
the elasticities available from the
existing literature because we felt
those were most relevant for
characterization of the U.S. vehicle
market. However, we agree it makes
sense to incorporate the suggested
literature to help place our model in
the context of dynamic vehicle
-------
the literature review,
defensible?
more likely to be applicable to their
model and give the reader solid
reasoning for why they made the
choices they did.
That said, I do think that there are a
couple of papers that should probably
be at least discussed. Schiraldi (2011)
and Adda and Cooper (2000) both
model dynamic adoption of new and
used cars. They aren't based in the
U.S., but they may be the closest
models to the ones the authors are
after here. Putting their model in the
context of the dynamic car
purchasing literature seems
important.
purchase literature. We added
footnote 2 referring to and citing
these papers.
3. Are the data and
assumptions
appropriate for the
analysis conducted
and objectively
chosen? If not, do you
know of other data or
proposed alternative
assumptions that
might be used in this
analysis?
I think that the data and assumptions
of the model are appropriate and
reasonable. The authors have
provided extensive sensitivity testing
and their take-aways from the model
appear to be extremely generalizable.
We appreciate this positive feedback
on the data and assumptions used for
this study.
4. Are the methods
and procedures
employed technically
appropriate and
reasonable? Please
distinguish between
cases involving
reasonable
disagreement in
methods as opposed to
cases where you
conclude that current
methods involve
specific technical
errors.
The methods and procedures are
technically appropriate and
reasonable. In particular, because the
authors are not attempting to defend a
specific point estimate, but are more
attempting to lay out intuition for
how policy affects equilibrium new
and used vehicle adoption and
pricing, I don't have any real
critiques of the approach. The authors
are clear about the assumptions of the
model and the ways that these
assumptions could be relaxed in
future work.
We appreciate this positive feedback
on the methods and procedures used
for this study.
5. Does the modeling
analysis appear to
The results are consistent with the
assumptions and data used in model
We appreciate this positive feedback
on the modeling analysis.
-------
produce results
consistent with the
assumptions and data
used for model
development? Do the
results presented by
RTI follow from the
data and assumptions
used in the analysis?
development. Beyond following from
the data, the results provide intuition
for how the results would change
with alternative modeling
assumptions and the authors are very
clear about which assumptions are
most critical for changing the results.
6. In what ways are
the results sensitive to
the data and
assumptions used in
model development?
Are there alternative
assumptions and data
that the researchers
should consider
providing improved
analysis?
The authors are very careful to keep
the analysis to the adoption and
pricing of a single representative
vehicle of each age. I do worry
slightly that others who read the
report may want to use these results
to draw conclusions about the effect
of things like fuel economy policy on
the overall fuel economy of the
vehicle fleet over time. I think it
might make sense for the authors to
add some discussion of the
complications that within-vintage
vehicle heterogeneity is likely to add
in the real world. In particular, a
policy like a CAFE standard, which
penalizes fuel inefficient vehicles
while subsidizing fuel efficient
vehicles, will have different long-run
effects on the age distribution of fuel
inefficient and efficient vehicles. I
see this not as something that the
authors should do to improve this
analysis: adding heterogeneity is very
complicated as the authors point out.
But I do think that it would be helpful
for the authors to be fairly clear about
the limits of the analysis they have
conducted by explicitly saying that
policies that do not uniformly affect
the new vehicle fleet will have
complicated effects on the long-run
age distribution of vehicles.
We agree with the reviewer's
suggestion to add more discussion of
the potential implications of
distributional effects beyond the
single representative vehicle used for
each vintage in the report. We added
discussion of these points and related
caveats associated with the findings
implied by the current analysis in
paragraph 2 of Section 6.2 and in the
new Section 10.4.
ADDITIONAL OVERALL COMMENTS PROVIDED
(NOT CIIARCE Ql ES 11ON-SPIX 1 I K ):
-------
Overall, this is the best report of this type I have ever read. The
analysis provides both intuition and a concrete path forward
for future analysis. The model rests on reasonable assumptions
and is very clear about how things might change under
alternative assumptions. The calibration is reasonable and
makes choices about which values from the literature to use
that are based on the quality of the studies rather than
weighting all previous work equally. The authors conduct
extensive sensitivity analysis that allows the reader to fully
understand which results are robust to alternative assumptions
and which depend on the parameterization. Congratulations on
great work!
We appreciate this positive feedback
on the analysis, model, and report.
ADDITIONAL COMMENTS BY SPECIFIC REPORT
CHAPTER:
Dr. Langer also commented on specific sections of text
throughout the report.
3.1, Bullet 4 ("Production of vehicles..."), which discusses
long-run changes in vehicle models: Cite Knittel and coauthors
here on long-run model changes? Was that paper ever
published?
We thank the reviewer for reminding
us of this work. We added the
citation to Blonigen, Knittel, and
Soderbery (2017) on long-run
competition and entry.
3.1.1.1, Paragraph 3 ("In the literature estimating..."), which
discusses new-vehicle demand systems and the distinction
between purchase price and net ownership costs: This
paragraph was helpful for explaining what you are trying to get
at.
We appreciate the positive feedback
on clarity of this text.
3.1.1.2, Paragraph 2 ("Much of the literature..."), which
discuss the distinction between average own-price elasticity
among individual models of new vehicles and aggregate own-
price elasticity: But it's also important to recognize that the
identifying variation behind these estimates may not be well
set up to get at these aggregate elasticities.
We added discussion of this caveat
on p. 3-4.
3.1.1.3, Paragraph 1, in reference to the sentence: "The
derivatives of demand here reflect a world where (at least from
the consumer's perspective) only the price of the vehicle has
increased:" Single vehicle?
We clarified this reference on p. 3-4.
The point was that they reflected only
changes in price, not vehicle quality.
3.1.1.3, Paragraph 2 ("More often..."), which discusses
derivatives of demand and consumer willingness to pay for
vehicle attributes/regulations that change vehicle attributes:
It's probably worth being clear here that these derivatives are
at least theoretically possible to get from the models, if the
models include the right vehicle attributes.
We added clarification to this point
starting at the top of p. 3-5.
-------
3.1.1.3, Paragraph 3 (starting at "A consumer might..."),
which discusses the effects of fuel economy regulations on
consumer demand and explains that consumers may perceive
fuel economy-related gas savings differently. This doesn't
need to be phrased as heterogeneity in beliefs. It could just be
that some consumers driver more and others drive less so
increases in fuel economy have different effects for different
consumers.
We expanded on our discussion to
incorporate the reviewer's point
about differences in driving patterns
on p. 3-5.
3.1.1.3, Paragraph 4, in reference to the sentence: "We will
refer to the combined change in perceived cost at the time of
purchase as a 'generalized cost:'" Shouldn't this be something
like the compensating variation? Is it really "perceived cost"?
and it's clearly heterogeneous, so it's likely the mean change
in perceived cost?
We expanded on our discussion on
p.3-5 to clarify the meaning of
"generalized cost" as used in this
report.
3.1.2, Paragraph 1, in reference to the sentence: "If assuming
perfect competition, supply will in fact be perfectly elastic:"
But we wouldn't do this in the car industry.
We expanded on our discussion of
supply in the U.S. vehicle market on
p. 3-6.
3.1.2, Paragraph 1, in reference to the sentence: "However, the
presence of a used market interacts with the equilibrium
system and the new-vehicle demand elasticity is no longer
sufficient to measure the effect of a policy, even when new
vehicle supply is perfectly elastic:" Related to the above, I
would be stronger here that supply isn't perfectly elastic, even
in the long run.
We expanded on our discussion of
supply in the U.S. vehicle market to
incorporate the reviewer's point
about the vehicle industry not being
perfectly competitive on p. 3-6.
3.1.2, Paragraph 2, in reference to the sentence: "When a
particular used model's price rises, more of that model become
available (e.g., because scrap dealers, insurance companies,
and mechanics decide to repair and sell more of them as
vehicles instead of as scrap metal):" Owners don't also
respond? Trade in vehicles rather than keeping them as an
extra car for a child or relative?
We added a reference to potential
owner response on p. 3-6.
3.1.2, Paragraph 3, which discusses scrappage functions. In
response to the phrase "vehicle scrappage depends only on
own price of vehicles:" It should depend on the price of other
vehicles to the extent that parts for repairs are interchangeable
and supply curves for those parts slope up.
We added a footnote identifying the
possibility that scrappage could
depend on the prices of other vehicles
on p. 3-6, though we expect this
effect to be small and maintain this
assumption.
3.1.2, Paragraph 4, in reference to the sentence: "Empirical
estimates will typically be presented as a derivative or
elasticity of this function, rather than of the density of
underlying shocks that determines the function:" Derivative of
the density, no?
We clarified to refer to the derivative
of the density on p. 3-7.
-------
3.1.3, Paragraph 1, which discusses how the presence of
equilibrium effects in observational data may lead to biased
estimates:" Would this be clearer if it were framed as omitted
variables bias? So there are omitted variables that are
correlated with both the likelihood of scrappage and the cost of
repair?
Equilibrium effects can create bias in
a number of ways, so we kept the
more general text. However, we
agree omitted variables are certainly
one useful way to view the source of
bias and have added footnote 9 in
Section 3.1.3 to this effect.
3.1.3, Paragraph 3, which discusses the effects of new-vehicle
prices on scrap rates of used vehicles, in reference to the
phrase: "a downward shock to the scrap rate will create more
competition for new versions of a vehicle (because there are
now more used ones entering the market):" Remaining in the
market, not entering the market.
We corrected the text on p. 3-8.
3.1.3, Paragraph 3, which discusses the effects of new-vehicle
prices on scrap rates of used vehicles, in reference to the
phrase: "Estimating the reduced form successfully would
require good quasi-experimental variation in new-vehicle
prices, but the likely confounders and long time series required
have meant that the literature has been unable to find many
suitable settings:" And is this even the right elasticity? As the
authors say above, this is only for a price change for a single
vehicle, and the cross-price elasticities are likely to matter
because used car prices are determined in equilibrium.
We added text on p. 3-8 to
incorporate the reviewer's point.
3.2.1, Paragraph 3, in reference to the question: "How do the
effects on the composition of the vehicle inventory relate to the
initial policy goals?" It seems like somewhere here there also
needs to be recognition that it's not just one "substitution"
effect to used cars but that it will depend a lot on which
vehicles the new car buyers are substituting toward, and how
substitution is happening within the used car market (so 20
year old vehicles may still be scrapped at high rates, but there
may be much less scrappage of 10 year old vehicles and this
changes the policy impacts).
We have added additional questions
to our list starting at the bottom of p.
3-9 in response to the reviewer's
comment and a footnote referencing
our assumption that there are
differences in substitutability
between vehicles that are farther
apart in age.
3.2.1, Paragraph 3, in reference to the question: "For the group
of consumers that substitutes from new to used, how much
does this substitution drive up equilibrium used-vehicle
prices?" And WHICH used vehicle prices?
We expanded on this point about
differing impacts on different vintage
used vehicles in the text.
3.2.2, Paragraph 6, which starts with "A dynamic model..." in
reference to the sentence "Assuming the policy shock is long-
lived the effects on used-vehicle prices and scrappage evolve,
likely strengthening, over time." Why? Not sure it's not true,
but not completely obvious why it is. Couldn't producers
evolve/adapt to weaken the effects?
We have added text on the
mechanical nature of the
strengthening: shortages in the used
market become more severe with
each new vintage directly affected by
the policy. However, as pointed out
by the reviewer, the size of the direct
-------
effects could also diminish if
technology allows easier compliance
over time. This is also of potential
dynamic interest, and we added it in
footnote 13.
4.1, Paragraph 3, in reference to the sentence "The fact that the
elasticities differ in this way is entirely consistent with
economic theory; more disaggregated choice sets (in terms of
attributes, time of purchase, etc.) mean that the best substitutes
for any given good within the choice set will be more similar
to it (e.g., more substitution would be expected between sedan
models or between makes of sedan than between new and used
vehicles or between a personal passenger vehicle and
alternative means of transportation)." This is not exactly how I
would think about the issue. Substitution happens for many
reasons. In particular, since people are heterogeneous, having
more products means that there are products that are similar to
a vehicle on different dimensions, which means that more
heterogeneous people can be on the margin between buying
this vehicle and another.
We added a sentence to reflect this
point on p. 4-2.
4.2, Table 4-2: List of Papers Included and Parameter
Estimates Provided. Busse, Knittel, and Zettlemeyer doesn't
have any relevant elasticities? I guess everything that they're
looking at is responses to gas prices rather than responses to
price changes, but isn't that central to their generalized price
measure?
We were not able to calculate the
relevant elasticities we were focused
on for this review. As the reviewer
notes, they focused on responses to
changes in gas prices and did not
provide enough information to
estimate demand elasticities in
response to changes in vehicle prices.
5.1, in reference to this sentence about Table 5-1: "The
elasticities in the first panel hold constant most other aspects of
the equilibrium system, including substitution possibilities to
used vehicles." What does this mean, exactly? Are you just
saying that the attributes of used vehicles are held constant?
Not sure what a substitution "possibility" means.
Clarified in the text on p. 5-2 that we
were referring specifically to their
holding used-vehicle prices constant.
5.1.1, which discusses aggregate own-price elasticity of
demand for new vehicles with respect to the price of new
vehicles: I feel like somewhere in here there needs to be more
discussion of the fact that this is not necessarily all substitution
to used vehicles but also includes substitution to fewer total
vehicles (either not traveling at all, using a used vehicle more,
or switching to bike/transit/etc.). But maybe I missed that
elsewhere?
We do have some discussion of
substitution to the outside good
elsewhere in the text but added a
sentence making this point at the
beginning of Section 5.1 on p. 5-2.
5.1.1, Paragraph 1, in reference to the sentence: "In addition,
we made a simple assumption regarding average model-level
Yes, the reviewer is correct. We
clarified in the text on p. 5-3 that we
-------
elasticities and average substitution to the outside good:" This
is a little confusing, but I think is saying that everything should
technically be share-weighted averages instead of simple
averages, right?
would ideally calculate share-
weighted rather than simple averages.
5.1.1, Paragraph 4, which starts with "Barry, Levinson..."
noted that it should say "(2004)" after Pakes in the last
sentence.
We added the year to this reference
on p. 5-4.
5.1.3, Paragraph 3, in reference to the first sentence: "As
expected given the relationship between the new- and used-
vehicle markets, the available elasticities of new-vehicle
demand when used-vehicle prices can adjust (-0.18, -0.36)
indicate less responsiveness." What timeframe should we think
of these elasticities as being over? In BLP, it's very clearly a
year. But here are they longer-term? Presumably not super
long-term though?
We added text indicating that these
elasticities are representative of the
medium term, allowing for new and
used markets to adjust to a new
equilibrium.
5.2.2, Paragraph 1, in reference to a sentence about estimates
of own-price elasticity for used vehicles, "Their estimate of
own-price elasticity for used vehicles ranged from -0.75 to
-1.93 with a mean of -1.23." Aggregate own price elasticity?
We added a sentence on p. 5-8
clarifying that this estimate is not
directly comparable to the aggregate
elasticities for new vehicles because
it is an average of model-level values
rather than an aggregate elasticity.
5.2.2, Paragraph 2, in reference to the sentence "Bento et al.
(2009) also reported model-level price elasticities among used
vehicles: the average vehicle-level elasticity among all used
vehicles was -0.54." I guess I'm a bit confused about what a
used vehicle own-price elasticity is given that there are
consumers on both the supply and demand sides. Given that
there are lots of substitutes I would expect an own-price
elasticity of a used vehicle to be well above 1 in magnitude.
But if the seller is basically going to sell the car one way or
another regardless, then does that pull down the elasticity? So
if prices go up, the supply of used vehicles goes up and the
demand goes down so the number of vehicles TRANSACTED
doesn't change? Is there a way to provide some clarity on this?
We thank the reviewer for raising this
point. In our representative agent
model, we are considering total
demand (irrespective of how many
transactions would be needed to
reshuffle to a particular level of
demand). We have clarified this point
on p. 5-8.
5.2.4, but seems to be in reference to section 5.2 as a whole:
Does Schiraldi provide an estimate that is useful? Adda and
Cooper? They're not on US data, but it still seems like they're
informative.
We incorporated only U.S. data in
this study as most representative of
market conditions in the U.S., so we
did not use non-U.S. estimates. In
addition, these papers do not present
estimates of our elasticities of
interest.
5.3.1, which discusses the elasticity of aggregate scrappage
with respect to the average price of used vehicles, in reference
Moved sentence referenced in the
comment below into the first
-------
to the sentence: "In Hahn's (1995) Table 2, he relates an
exogenous "bounty" (subsidy to scrappage) to changes in the
absolute number of vehicles scrapped relative to a baseline."
Timeframe seems important to mention here. Short-run, I
assume?
paragraph on Section 5.3 on p. 5-9
noting that the studies using bounty
programs are focused on short-run
responses, which may differ
substantially from the long-run
response. Also mentioned that Hahn
is capturing short-run response on p.
5-10.
5.3.1, Paragraph 3, in reference to the sentence: An important
caveat of studies looking at a bounty for scrappage lies in the
temporary nature of the program: it may be that a short-lived,
salient subsidy will quickly harvest a stock of "almost-
scrapped" vehicles, overstating the elasticity. Yes, maybe just
move this up front in the discussion?
Moved sentence into the first
paragraph of Section 5.3 on p. 5-9 to
make this point sooner and clarify the
implications of studying scrappage
through the simulation of bounty
programs.
5.4, Table 5-4. "Time Frame of Data Used" is hard to read.
This text looks the same as for the
other tables in our version, but we
will check readability in the final
report.
Section 6 (overarching statement about the section): I get that
this is sort of an obnoxious comment, but I could have used
more of a road-map up front on this section. For instance, it
wasn't completely clear up front that the price vector is sort of
secondary and just guarantees equilibrium in the system of
equations. It's obvious after reading through, but being a little
bit clearer about how this section would be laid out would
have been helpful.
We have added a paragraph on p. 6-1
outlining the content of Section 6 and
think it was especially useful to give
people a road map about the two
parallel price vectors (which indeed
are equivalent in the long-run section
but enter importantly later when we
think about dynamics and
expectations).
6.1, in reference to "q° number of vehicles of age a
demanded" (as well as the notation for number of vehicles of
age a supplied). Demanded in transactions or overall? Ditto
with supplied.
These are aggregate quantities, and
we added that to the variable
description on p. 6-1 to clarify.
6.2, Paragraph 1, in response to the sentence "The elasticities
for individual vehicles measured by this approach do not
provide clear signals about an elasticity of demand for new
vehicles in the aggregate, because the outside good for
individual vehicles includes other new vehicle models; thus,
demand for an individual model is likely to be much more
elastic than demand for a generic new vehicle." I get what
you're saying here, but wouldn't it be clearer to just say "the
alternative choice for each individual vehicle includes other
new vehicle models and the outside good"?
We incorporated the reviewer's
comment into the discussion on p. 6-
2.
6.2, Paragraph 2, which starts with "We assumed..." and
discusses assumptions about the demand system and the
We appreciate the positive feedback.
-------
electrification of the vehicle market. Good discussion in this
whole paragraph.
6.4, Paragraph 5, which starts with "Importantly, supply and
demand..." Related to an earlier comment: this paragraph is
super helpful for understanding how everything is coming
together and more of this earlier in the section would be
helpful to the reader.
Thank you for this useful suggestion:
we now also raise the connection
between price vectors on p. 6-1 in the
introduction to Section 6.
6.4.1, Paragraph 3, in reference to the sentence "Because
scrappage occurs when repair costs exceed the value of the
vehicle the form of the scrap function (constant elasticity)
determines the form of the repair cost density." Isn't this
backwards? The repair cost density is the primitive and the
scrap function follows from it? I appreciate that you are
putting an assumption on the scrap function rather than the
repair cost density though.
Agreed, the density seems more like
the fundamental and we have updated
the language to reflect that on p. 6-9.
In practice, an assumption on either
this density or the scrap function
fully specifies the other one.
6.6, Paragraph 6 about f>a (Share of aggregate spending on
each vehicle age), in reference to the sentence "Expenditure
shares come from combining the age profile above with data
on the cost of vehicles of different ages." I'm fairly confident
that you explained this somewhere and I'm just missing it, but
why isn't \beta_a just calculated in the model from the
equilibrium prices and quantities? Why does it need to be
treated as an exogenous parameter? It seems very odd given
the simulation exercise?
As pointed out by the reviewer, the
baseline values are calibrated
internally using equilibrium prices
and quantities. They are exogenous
from the perspective of policy (since
underlying consumer preferences
need to be the same in the baseline
and policy world), and we have
clarified these points in the text.
6.6, Figure 6-2: Baseline Price Profile by Age. I think that the
"baseline" here just means that you use it for the expenditure
shares and then update it in the simulation, yes? A clearer title
would be helpful.
We think the use of "baseline" here
and elsewhere is clear (i.e., simulated
patterns over time based on historical
data in the absence of a shock),
though we added "vehicle" to the title
and added a sentence on p. 6-12,
further clarifying what is meant by
this baseline price path.
7.1, Paragraph 2, which discusses Table 7-1: Demonstration of
Channels of Adjustment in Quantities and Prices When
Generalized Cost of New Vehicles Rises by 1%, in reference
to the sentence "The 1% increase in generalized cost measures
the strength of the policy, hence the term "policy elasticity."
Impact? I'm not really sure what "strength" means here.
We revised the text on p. 7-1 to
clarify what was meant here. The
idea is that the level of the
generalized cost represents the
magnitude of the policy. The policy
elasticities are then capturing the
responsiveness of the vehicle market
relative to the magnitude of a policy
shock.
7.1, Table 7-1, in reference to the Cross-Price New/Used
column: Maybe I missed this earlier, but this is the elasticity to
Yes, this reflects substitution to all
used vehicles (summed over ages),
-------
all ages of used vehicles? I get that this might make sense
since new buyers switch to newer used vehicles and those
buyers switch to older used vehicles and it all works its way
through the system, but being explicit about this would
probably be helpful.
and we clarify this in the text. We
also added text to footnote 26,
directing readers to Appendix C
where we provide more detail on the
role of age difference in substitution.
7.2, Paragraph 5, in reference to the first sentence: "To
complete the setting, note that the demand system contains 30
ages, but so far we have only specified three demand
elasticities." Ok, this makes sense re: my above comment. A
quick sentence about this earlier where you talk about
assumptions would be helpful.
Related to the comment above, we
added a sentence describing the
assumption of an aggregate used
vehicle market at the beginning of
Section 7.1 on p. 7-1, as well as the
footnote on Table 7-1 to help clarify.
7.4, Paragraph 3, which describes Table 7-2: Policy Elasticities
Corresponding to Selected Demand and Scrappage Elasticities,
in reference to the sentence: "The first panel (rows A through
E) holds the scrappage elasticity fixed at a value of-0.7
(Jacobsen and van Benthem [2015] and also close to the
median of values in Table 5-3)." Rows A through E: Not
labeled.
We were not sure what you meant by
those rows not being labeled, though
this comment and the one below may
have referred to our lack of showing
these as panels despite referring to
the first and second panels in the text.
We have corrected the text on p. 7-5
to refer to everything by the row
labels A through I.
7.4, Paragraph 3, which describes Table 7-2: Policy Elasticities
Corresponding to Selected Demand and Scrappage Elasticities,
in reference to the sentence: "The second panel (rows F
through I) explores changes in this elasticity, spanning most of
the range in the literature with values of-0.2 to -1.2.)." No
second panel.
See response to the comment above.
7.4, Paragraph 4, in reference to the sentences: "Scenarios A
through C explore an increase in substitution to the outside
good. As we saw in Table 7-1, this is one of the most
important elasticities in determining the effect of policy on
new-vehicle sales." It's probably worth being really clear early
in the report that this is critical and perhaps the worst-
identified parameter in the modern discrete choice literature.
We added sentences making this
point in the Executive Summary and
the Introduction.
7.4, Paragraph 11, which starts with "Figure 7-1 displays..." in
reference to the sentence "When the scrap elasticity is -0.7 on
the horizontal axis, the values on the Y axis reflect the policy
elasticities as shown in Table 7-2 for Scenario D." Instead of
Y, vertical. Or X instead of horizontal [for parallel sentence
structure].
We revised the text on p. 7-8 to refer
to horizontal and vertical axes for
consistency.
8.3, Paragraph 3, referring to the sentence "For instance, there
is some empirical evidence of increased buying in advance of
regulatory changes going into effect, once those changes have
Agreed. We added citations to a
couple of papers examining
anticipatory responses to
transportation policy on p. 8-3.
-------
been announced." Do you want cites on these types of
statements?
9.1, Paragraph 2, referring to the sentence "This is a small-
scale version of changes in real estate prices that can occur
even when the announced change is decades away." Exactly
what Holland, Mansur, and Yates get for electric vehicle
mandates in their recent AEJ.
We added a citation to Holland,
Mansur, and Yate and a couple of
sentences on this paper starting at the
bottom of p. 9-1.
9.1, Paragraph 2, referring to the word "sales" in the last
sentence, "Preannouncement has no effect on the long-run
outcome: sales converge to the 0.25% decline in a little over 5
years." New vehicle sales. Just being clear that the changes in
the used vehicle market take longer to get worked out (I
assume).
We revised text on p. 9-2 to clarify
that this is referring specifically to
new-vehicle sales.
Comments by Dr. Benjamin Leard
c MAkci:
Q1 ESTION
COMMENTS
RESPONSE (INITIAL DRAET)
1. Does the
presentation
describe the data
and methods
sufficiently to
allow the reader
to form a general
view of the
quality and
validity of the
analysis
approach?
In general, yes. The document could cite a
few citations that have either been updated
or published.
Citation 1: Dou and Linn (2018), which
was a working paper, is now a published
paper appearing in the Journal of
Environmental Economics and
Management. The citation is
Dou, X. and J. Linn (2020). How do US
passenger vehicle fuel economy standards
affect new vehicle purchases? Journal of
Environmental Economics and
Management. 102: 1-21.
Citation 2: Leard (2021) is an updated
version of Leard (2020). The current
version is available at
httos: //media, rff. ors/documents/WP 19-
01 rev 2021 .Ddf. The market Drice
elasticity of demand estimate from the most
recent version is -0.37.
We updated these citations
throughout the text of the report
and in the references. We revised
the text to incorporate the latest
elasticity estimate from Leard
(2021) in the text.
2. Does the report
miss relevant
literature in its
review? Are the
Yes, the report misses several relevant
citations.
Our literature searches included
grey literature but did not capture
the Stock et al. comment on the
proposed SAFE rule or the Center
-------
interpretations of
the elasticities in
the literature
review, and the
estimates of
elasticities used in
the model
stemming from
the literature
review,
defensible?
Citation 1: James A. Stock et al. Comment
on Proposed SAFE Rule, EPA-HQ-OAR-
2018-0283-6220 (October 26, 2018).
The Stock et al. comment on the proposed
SAFE Rule does not report elasticities, but
new vehicle market price elasticities in the
range of -0.03 to -0.09 can be calculated
from their results. The Stock et al. comment
also includes an important correction of a
spreadsheet error in the Notice of Proposed
Rulemaking, resulting in a revised elasticity
estimate for that model of -0.07.
It is worth mentioning how this citation is
different from other papers such as Leard
(2021). My understanding of the paper is
that it has a model that accounts for general
equilibrium effects. It might be useful to
include a new column in Table 5-1 that has
a binary yes-no variable "Includes general
equilibrium effects" since the point of the
current document is to account for these
effects.
Citation 2: The report. Sean P. McAlinden
et al., The Potential Effects of the 2017-
2025 EPA/NHTSA GHG/Fuel Economy
Mandates of the US Economy, Center for
Automotive Research, 27 (Sept. 2016),
httDs://\\ww.careroiiD.ore/\\D-
content/uDloads/2017/02/The-Potential-
Effects-of-the-2017 2025 -EPANHTS A-
for Automotive Research report.
We thank the reviewer for
identifying those. We reviewed
these references and have added the
McAlinden et al. (2016) report to
our summary tables and counts that
we include in the report. We have
also added a citation to the Stock et
al. (2018) comment but did not
include that reference in our
summary tables or reported range
because it did not meet our criteria
for inclusion in the primary set of
studies. While the authors are very
highly qualified transportation
researchers, their comment is
neither peer-reviewed nor grey
literature in the form of a working
paper series, report issued by a
university or independent research
institute, or otherwise available as a
publicly available output published
by an independent research
institution, which were the criteria
for our searches.
GHGFuel-Economv-Mandates-on-the-US-
Economv.Ddf
This is a report that estimates the long-run
market-price elasticity of demand for new
vehicles to be -0.61.
The interpretation of one of the citations is
incorrect:
- Fischer, Harrington, and Parry (2007):
This study finds an implied new vehicle
market price elasticity of demand equal to -
1 by reporting separate elasticities for cars
and light trucks. This ignores substitution
between the two vehicle categories. This
Fischer, Harrington, and Parry
(2007) detail the elasticities in
Appendix B. For the set of
elasticities that adjusts for changes
in the used market, they find -0.79,
-0.85, and -0.36 for cars, trucks,
-------
paper also reports a long-run elasticity for
the combined market for new motor
vehicles, finding it to be -0.36. This value is
reported in the bottom panel of Table 5-1.1
am puzzled why the -1 value is reported in
the top panel when the combined effect is
what is relevant here.
and combined, respectively. We
report only the combined value of
-0.36 in the lower panel. For
elasticities looking at new vehicles
without adjustments for the used
market, they report (from GM
source data) values of-2.25, -0.97,
and -1.0 for cars, trucks, and
combined, respectively. We report
only the combined value of-1.0 in
the upper panel.
3. Are the data
and assumptions
appropriate for
the analysis
conducted and
objectively
chosen? If not, do
you know of other
data or proposed
alternative
assumptions that
might be used in
this analysis?
Certain papers that report new vehicle
market price elasticities are, in my opinion,
inappropriate to cite and use to construct a
range of elasticities. The following papers
cited in Table 5-1 are inappropriate:
Berry et al. (2004): This study assumes a
new vehicle market price elasticity of
demand for new vehicles equal to -0.4 (in
addition to model where they assume it is
equal to -1). This elasticity is not estimated
in this paper. Therefore, it is inappropriate
to cite.
Fischer, Harrington, and Parry (2007): This
study finds an implied new vehicle market
price elasticity of demand based on model
simulations, not estimated from data.
Generally, these types of "calibration"
results should not be included with other
studies that estimate the elasticity.
BLP (1995), Goldberg (1998), Bento et al.
(2009), Knittel and Metaxoglou (2014),
Dou and Linn (2020): My issue with citing
these papers is that none of them focus on
properly estimating the market price
elasticity of demand for new vehicles. None
of them exploit proper statistical variation
necessary to identify this parameter. Take,
for example, BLP (1995). This study uses a
panel of aggregate sales and price data for
vehicle models sold during 1971-1990. The
study is almost exclusively focused on
obtaining unbiased estimates of the model-
specific own-price elasticity of demand
We thank the reviewer for careful
review and feedback on the papers
included. We agree that there are
caveats to many of the studies
being used, including the fact that
some were not focused on
estimation of the market price
elasticity of demand. Nonetheless,
those are the elasticities that are
implied by the universe of available
studies. Thus, we feel that it is
reasonable to include estimates
generated from simulation models
(which are denoted as being
calculated from calibrated models
in the table), as well as those
calculated from studies of vehicle
demand providing sufficient
information to calculate the
elasticities implied by those
models. While Berry et al. (2004)
did not estimate the aggregate price
elasticity of demand directly, they
estimated a parameter representing
price responsiveness in their central
case that can be used to calculate
the market elasticity. An important
implication of our review of the
literature is that the availability of
appropriate elasticity estimates is
quite limited (which we mention in
the overview of the literature at the
beginning of Section 5). While
these available estimates help set
our ranges for the cases presented
-------
(e.g., Toyota Camry). One small part of the
paper discusses the implied market-price
elasticity of demand. As discussed in Leard
(2021), this elasticity is a function of the
own-price elasticity and the propensity of
new vehicle buyers to leave the new vehicle
market. It is not clear how BLP (1995)
identify this propensity (they have no
discussion of this).
BLP (1995) should instead be cited when
discussing the model level own-price
elasticity of demand, since this is the focus
of the paper.
The other papers - Goldberg (1998), Bento
et al. (2009), Knittel and Metaxoglou
(2014), Dou and Linn (2020) - also have
alternative focuses and do not discuss much
at all how the market price elasticity of
demand is identified from the data.
Therefore, these papers should not be cited
to construct a plausible range of values for
the market price elasticity of demand. A
paper like Bento et al. (2009), for example,
which focuses on the link between vehicle
choice and VMT choice, should be cited
when constructing a plausible range for the
VMT elasticity.
using our simulation model, the
model is flexible and allows for the
use of alternative assumptions.
4. Are the
methods and
procedures
employed
technically
appropriate and
reasonable?
Please distinguish
between cases
involving
reasonable
disagreement in
methods as
opposed to cases
where you
conclude that
current methods
I have a few concerns with the methods.
Concern 1: The modeling of vehicle
ownership costs (r) and prices (p) is unclear
to me.
Why is it necessary to model vehicle
ownership costs in the first place? Why not
just model vehicle prices?
The discussion in 6.4.1 seems to be
motivated from this distinction. But it is not
clear why the model cannot be simplified to
only have vehicle prices or vehicle prices
plus fuel costs.
Ownership costs appear in the vehicle
demand function equation (6 - 1). The
authors motivate using ownership costs
with the following:
(1) This comment was quite helpful
in clarifying the way we describe
depreciation and survival. We now
make the link clearer in the first
paragraph of Section 6 and the
(new) third paragraph of Section
6.4.1.
Specifically, modeling a fixed path
of depreciation (i.e., ownership cost
being a fixed percentage of asset
value at any given age) and
modeling a fixed survival path are
closely connected. Consider instead
what would happen if survival of
vehicles became shorter but
depreciation were a fixed fraction
-------
involve specific
technical errors.
"Modeling ownership cost as the variable
of interest to the consumer follows the
approach in Bento et al. (2009) and is
important when allowing choices to be
made between new and used vehicles."
Bento et al. (2009) define ownership costs
as the sum of fuel costs and a rental rate,
where the rental rate is defined as the sum
of
Foregone return of the real value of the car,
which is proportional to the used vehicle
price.
Depreciation, which is typically
proportional to the used vehicle price (e.g.,
a typical used vehicle depreciation rate is
often cited to be 15 to 20% of the value of a
used vehicle).
Insurance and registration, which are
generally tiny relative to the first two
components.
Given these features, it seems
straightforward to model used car demand
as a function of vehicle prices and fuel
costs only. This would simplify the model
set up and it would be consistent with how
supply is defined, which is a function of
used vehicle prices.
Concern 2: It is unclear how a vehicle is
defined in the model. Is a vehicle defined
by age only? So the model has 31 vehicles:
new, 1 year old, 2 years old, ..., 30 years
old. This should be more explicitly stated in
the first paragraph of Section 6.2.
Concern 3: Section 6.3.1 New Vehicles
states that the price of a new vehicle is
taken as given. Is it equivalent to say that
the new vehicle market is perfectly
competitive? This is in contrast to most of
of asset values: vehicle owners
would suffer unexpected losses
because they would still be
depreciating vehicles that are no
longer operational. Or, if vehicle
survival became longer, as in many
of the cases we examined, there
would be windfall gains on vehicles
that are fully depreciated but still
operational. By recomputing
depreciation together with survival
we are able to capture equilibrium
changes. When the up-front cost of
new vehicles increases, the choices
owners make to extend survival are
linked to (and, in fact, are implicitly
motivated by) spreading out the
higher vehicle cost over more
years.
More subtly, consider the intuition
in line 1 of Table 7-2. If we made
depreciation a fixed fraction of
asset value, then demand for new
vehicles would fall by 1% (because
ownership cost would just be a
constant fraction of the now-higher
new-vehicle price). This would not
be consistent with a zero elasticity
to the outside good because new
and used car quantities are also
linked in that scenario; depreciation
needs to adjust to produce an
equilibrium.
(2) Yes, the reviewer is correct that
a vehicle is defined only by vintage
in this aggregate model. We
clarified this and added more detail
on the limitations of this
aggregation in the new second
paragraph of Section 6.2.
(3) Consumers taking prices as
fixed is not equivalent to assuming
the new vehicle market is
competitive because the price
could, and likely does, include a
-------
the vehicle market literature (Bento et al.
2009 for example), which models the new
vehicle market is imperfectly competitive.
It would be useful to include more details
on this assumption, why it was adopted, etc.
markup. To the extent negotiating
power at a dealership might change,
we did abstract from that. We have
now clarified Section 6.3.1 to this
effect and explain more clearly the
role of imperfect competition in
new vehicle supply (paragraph 2 of
Section 6.3.1). The extent a policy
might be expected to increase or
decrease the degree of competition
would affect the prices faced by the
consumer. In the present setting,
this would appear as an input in the
sense that generalized cost (i.e.,
total price paid by the consumer,
net of attribute changes) would
need to be adjusted to account for
policy-induced changes to market
power. We do not have a strong
prior on if policy is likely to
increase or decrease market power,
so we suggest a neutral assumption
that markups remain fixed.
Concern 4: In Section 6.6, the authors
mention that they explore a range of the
demand elasticities theta_aa and theta_Na.
It would be useful to provide a table of
elasticities assumed. I realize this table is
presented in Table 7-2, but it would be
constructive to include a table in Section
6.6 with discussion defending the choice of
elasticities.
(4) We have now added to Section
6.6 a summary and discussion of
the elasticities we chose (we
attempted to span most of the range
in the literature table). This version
of the report also includes a new
case, exploring a very small new-
vehicle demand elasticity (-0.1) in
Appendix C. This expanded limit
reflects, for example, a value closer
to the Stock et al. (2018) comment
on the SAFE rule.
Concern 5: The authors assume that the
own-price elasticities for used vehicles is
identical to the new-vehicle elasticity. This
is problematic because of how households
sort into different vehicle ages. New
vehicles are more frequently bought by
wealthy households, and a lot of literature
has shown that wealthy households tend to
be less responsive to price than average or
low income households. For this reason, I
(5) Footnote 21 provides some
detail on this question, noting that
used-vehicle owners may face more
limited substitution possibilities to
the extent they may be considering
public transit, or not traveling, as a
substitute. We also fully agree with
the possibility in the reviewer's
comment that income or other
effects could instead make
-------
would expect the own price elasticities to
be larger in magnitude for older vehicles.
elasticities higher among used-
vehicle owners. Appendix Table C-
1 examines this possibility: in row
(3) used vehicles have double the
own-price elasticity as new
vehicles. We have expanded the
text in Appendix C to discuss the
rationale mentioned by the
reviewer.
Concern 6: At the end of Section 7.2, the
authors state that "Cross-price demand
elasticities between individual ages are set
to fall at 8% per year of difference in age."
This seems reasonable, but it is not clear
where 8% comes from. Is this based on
results from a demand system such as
Bento et al. (2009)?
(6) There is very little evidence in
the literature about the degree of
substitutability across ages. We are
glad to hear that the reviewer finds
the 8% assumption to be
reasonable, but we have also added
text emphasizing that for this
parameter exploring multiple
values is especially important. We
now do this in Section 7.2 and in
Appendix C, where we have added
cases ranging from very
concentrated substitution all the
way to perfectly flat. The overall
policy elasticities are not very
sensitive to this parameter.
Intuitively, it does not appear to
matter much if substitution is direct
(e.g., from age 5 vehicles directly
to ages 6 through 30, which is what
can happen when the profile of
substitution is very flat) or through
an indirect cascade (e.g.,
substitution from age 5 to 6, which
induces others to substitute from 6
to 7, 7 to 8, and so on, which is
what happens when the elasticity
falls off sharply with age
difference).
5. Does the
modeling analysis
appear to produce
results consistent
with the
assumptions and
data used for
Yes. It is clear that the policy elasticities
are smaller in magnitude than the new
vehicle demand elasticities, which makes
sense given the equilibrium response.
We agree with the reviewer that the
relative magnitude of the
elasticities is reasonable.
-------
model
development? Do
the results
presented by RTI
follow from the
data and
assumptions used
in the analysis?
6. In what ways
are the results
sensitive to the
data and
assumptions used
in model
development? Are
there alternative
assumptions and
data that the
researchers should
consider
providing
improved
analysis?
Given my comments above, I suggest
applying a different range for the market
price elasticity of demand for new vehicles.
I would also consider adding greater detail
to the definition of a vehicle. It seems to me
that part of the value of this exercise is to
simulate the effects of various policies (for
example, CAFE standards). The current
structure of the model has a highly
aggregated vehicle definition that seems
lacking for performing simulations of an
actual policy. At a minimum, it would be
useful to provide a discussion on how the
model could practically be expanded, and
the potential calibration and computational
challenges involved with such a task.
As discussed in responses to
comments above, we added the
McAlinden et al. (2016) study to
our primary set of elasticities from
the literature and updated text to the
latest version of Leard (2021)
(changing from -0.40 to -0.37),
but found the Stock et al. (2018)
comment to be outside our criteria
for inclusion and continue to
believe that the other papers used to
develop the range should be
retained as representative of the
available literature that can be used
to estimate the market price
elasticity of demand for new
vehicles. As noted above, certainly
caveats are associated with a
number of the available estimates,
but we continue to feel that the
range specified is appropriate.
Along those lines, were an analyst
to prefer to focus on alternative
elasticity values, they could
emphasize specific scenarios in
Table 7-2 that align with their
preferred specification. In addition,
the simulation model was
developed with the flexibility to
readily use alternative parameters.
Adding more detail to the definition
of a vehicle is outside the scope of
the current study, which is focused
on representation of vehicle
markets at an aggregate level.
Outputs from this model can be
linked to more detailed engineering
models to capture additional detail
-------
within those models. Nonetheless,
we agree with the reviewer that it
would be valuable to expand the
model presented in this report in the
future (e.g., adding electric vehicles
or other advanced fuel vehicle
technologies and otherwise
capturing within-vintage vehicle
differences and characteristics
likely to be differentially affected
by transportation policy).
ADDITIONAL OVERALL COMMENTS PROVIDED
(NOT CIIARCE Ql ES 11ON-SPIX 1 I K ):
ADDITIONAL COMMENTS liV SPECIFIC REPORT
CHAPTER:
Comments by Dr. James Sallee
7. CIIARdE
Ol 1 S I ION
S COMMENTS
RESPONSE (INITIAL
DRAIT)
1. Does the
presentation describe
the data and methods
sufficiently to allow
the reader to form a
general view of the
quality and validity
of the analysis
approach?
The report is clear. Data and
assumptions are documented well.
Assumptions and modeling choices
are explained thoroughly. Moreover,
the results of the analysis are well
explained, and every effort is taken to
provide basic economic intuition that
gives context to the quantitative
model output.
As a result, readers should be fully
able to evaluate the merits of the
model and understand both its
contributions and its limitations from
the information contained in the
report.
The report lays out the main analysis
in a sequence of steps that ensure
clarity by adding complexities and
features one at a time (e.g., section
7.1). The report also has many
sections that seek to consolidate
results and illustrate the underlying
We appreciate this positive feedback
on the quality and clarity of the
report.
-------
intuition of the results (e.g., the
summary in section 7.6).
Altogether, the report explains a
complex economic model in a concise
and effective way. This is not easy to
do, and the authors of the report
should be commended for the quality
of the exposition.
2. Does the report
miss relevant
literature in its
review? Are the
interpretations of the
elasticities in the
literature review, and
the estimates of
elasticities used in
the model stemming
from the literature
review, defensible?
The report does an admirable job of
surveying the literature and putting
the information available from past
research into context.
I am not aware of other studies that
should have been included in the
literature review. As the report notes
at several points, there are many
studies of the automobile market, but
attention to the elasticities pivotal for
the current analysis are rarely the
focus and are often not even
calculated. Thus, the report's finding
that few existing papers can be used
to calculate the key elasticities
conforms with my understanding of
the literature.
All interpretations of the elasticities
offered in the report appear
appropriate. The use of elasticities
from the literature to inform the
model is certainly defensible.
The report explains clearly why the
existing literature offers limited
guidance on key values. It therefore
emphasizes strongly the need for
sensitivity analysis, which is then
provided.
This is the appropriate stance to take.
If anything, I would recommend even
more caution in relying on the
existing literature (and hence elevate
even further the critical role of
sensitivity analysis) for two reasons,
We appreciate the positive feedback
on the literature review and the
constructive suggestions for further
enhancement. We have further
expanded our discussion of the
caveats associated with the
parameters available from the
literature starting on p. 5-1 (including
addition of the excellent points raised
by the reviewer regarding study age
and consumer income and changes in
embedded technology). We agree
with the reviewer that many of the
study authors were not necessarily
focusing on estimation of the
elasticities of interest for this study
and may therefore not have conducted
a thorough evaluation of the implied
elasticity. Nonetheless, those are the
elasticities implied by the demand
systems in the literature, and we
believe it is reasonable to use those as
part of the assessment of available
elasticities. As noted by the reviewer,
standard errors are not necessarily
available from each study and are not
necessarily calculated with consistent
methods.
-------
both of which are noted in the report
itself, but I note here for emphasis.
First, many of the studies are more
than a decade old. Mechanically,
vehicles are becoming more durable
and lasting longer (this is noted in
section 2). At the same time,
embedded technology is also
changing rapidly, which might make
newer and older vehicles less close
substitutes. Both factors imply that
empirical relationships observed in
the automobile more than a decade in
the past may be misleading indicators
of the vehicle market of the future.
Moreover, one might suspect that the
elasticity of demand for new vehicles
shrinks as consumers get richer. If
true, older studies might exaggerate
the new vehicle price elasticity.
(Some interaction between income
and price sensitivity is often estimated
in prior studies. Perhaps there is some
information in those that would
suggest how large might be any trend
in the elasticity driven by economic
growth. The report does not mention
this.)
Second, a substantial share of the
available studies do not focus on the
elasticity parameters that are of
interest here. In several cases the
authors of the report had to construct
an estimate that was not even reported
in the original research. This is
relevant because researchers have
discretion in how to construct and
estimate these models. When a
parameter (like the substitution to the
outside good) is viewed as a
byproduct of estimation, the original
researchers may not have exerted
maximal effort in ensuring that the
estimation was reliable.
-------
The report does not discuss standard
errors in the original studies. Instead it
uses the point estimates from prior
studies to form a qualitative sense of
the range of estimates across papers.
This is a defensible approach,
especially because the standard errors
are likely calculated with different
methods across studies and are thus
not always comparable. Even so,
where available, it might be useful to
have standards errors from the
original studies included in the review
to more fully characterize uncertainty.
We agree with the reviewer that it
would be useful to incorporate
standard errors. However, as was
noted in the introduction to Section 5
of the report in the paragraph
beginning with "Unfortunately,
available information from these
papers is generally insufficient
there is generally not enough
information available from the papers
to even identify or calculate a
standard error, let alone compare
them on a consistent basis. Thus, we
have continued to focus on point
estimates for the purposes of this
report. The model does include
flexibility to conduct additional
sensitivity analyses.
3. Are the data and
assumptions
appropriate for the
analysis conducted
and objectively
chosen? If not, do
you know of other
data or proposed
alternative
assumptions that
might be used in this
analysis?
The data and key model parameters
used in the baseline analysis and the
various scenarios appear appropriate.
The data sources cited are all reliable
and would be considered standard,
objective sources. I do not know of
any alternative data sources that I
believe to be superior.
Likewise, the parameter assumptions
are appropriate and appear consistent
with an objective analysis.
(In answering this question, I interpret
"assumptions" to refer specifically to
parameters used in the model, rather
than assumptions about the structure
of the model or analysis, which I
interpret as falling under the
definition of "methods and
procedures" addressed in the next
question. Essentially, this answer is
about section 6.6 of the report.)
The baseline scrappage scale factors
(ba in the notation of the model) are
said to simply be "calculated using
baseline vehicle survival
probabilities." The relevant paragraph
(p. 6-10) adequately explains the
source of data, but I was uncertain of
We appreciate this positive feedback
on the data and assumptions used in
this report.
The scale factor in the scrappage
equation ba is indeed a simple
algebraic calculation such that
baseline prices exactly reproduce
baseline scrap rates. We added more
detail on the calculation of baseline
scrappage scale factors in paragraph 4
of Section 6.6 as suggested by the
reviewer.
-------
whether there was a degree of
modeler discretion here, or if each ba
is simply an algebraic calculation
based on equation 6-5, where both sa
and pa are raw data inputs. This is a
minor detail that could be easily
clarified.
4. Are the methods
and procedures
employed technically
appropriate and
reasonable? Please
distinguish between
cases involving
reasonable
disagreement in
methods as opposed
to cases where you
conclude that current
methods involve
specific technical
errors.
The methods and procedures
developed and deployed in the report
are technically appropriate and
reasonable.
The model is complex: it deals with
thirty vehicle ages while
incorporating forward looking
expectations. To maintain tractability,
it must make simplifications—there is
a single composite vehicle model;
consumer heterogeneity is not
explicated; forward looking beliefs
are fully rational; new vehicle supply
is perfectly elastic; the model
abstracts from the choice of miles
traveled; and, there are no income
effects.
Those assumptions are justified and
reasonable. They serve the interest of
transparency and clarity, and they are
all appealing as a benchmark case.
The report is transparent and clear
about assumptions and offers caveats
where appropriate. For instance, the
analysis is built on a model of vehicle
purchase, rather than vehicle use. As
noted in the report, it is difficult to see
exactly how incorporating changes in
mileage in response to vehicle prices
might play out in the full equilibrium.
This decision could perhaps be added
to the model at some future date, but
even then, the key parameters (e.g.,
how vehicle price affects mileage,
presumably through a change in
depreciation) are not well estimated in
the literature and thus would require
We appreciate the reviewer's
comments generally agreeing that the
methods and procedures we used in
this report are appropriate and
reasonable. As noted by the reviewer,
it may be possible to further expand
the model in the future to add vehicle-
miles traveled, but we agree that the
necessary parameters for defining
vehicle use relationships are even less
available from the literature than the
vehicle demand elasticities reviewed
in the current study. The reviewer
raises a good point about potential
changes in substitutability between
new and used vehicles related to
systematic changes in vehicle
characteristics overtime. We added
discussion of this consideration in
footnote 6.
In addition, we incorporated
discussion of the implications of
imperfect competition for
interpretation of the model and its
inputs in paragraph 2 of Section 6.3.1.
The degree to which costs are passed
through (and/or subjected to markups)
can be embodied in the generalized
cost input to the model. We have
updated Section 6.3 to relax our
assumptions on the supply side. The
model here focuses on demand, so to
the extent there is information on
competition changes, those could also
be reflected in generalized cost. From
the perspective of the (representative)
consumer, the price of new vehicles is
given throughout.
Finally, we expanded on the
discussion of assumed substitution
patterns among vehicles of different
ages in Section 7.2 and especially in
-------
additional conjecture. As such, a
model focused on ownership seems
appropriate.
The report offers a generalized
interpretation of a cost shock to new
vehicles. It suggests that a regulatory
change that creates a combination of
changes in price and attributes can be
interpreted as a tax equal to the "net
utility cost." This is sensible and
appealing as a modeling approach. (It
is also very important because new
vehicle regulations create just this sort
of collection of changes
simultaneously.)
That said, a minor caveat would seem
to be that a change in other attributes
(as opposed to just a change in price)
might change the substitutability
between new and used vehicles; that
is, a regulation that changes the
attributes of new vehicles might
change the theta terms in the demand
system. This is likely to have a
minimal impact in the current setup.
But, it could become a more
important issue in the future if the
model is used to contemplate a
transition to electric vehicles or
vehicles featuring different levels of
automation, in which case attributes
affected by the policy could radically
change the substitution between new
and used vehicles.
The report assumes full pass through
of cost shocks to consumers: because
new vehicle supply is assumed to be
perfectly elastic, all regulatory
burdens are borne by consumers. The
report argues that this is a sensible
interpretation for the long run, and it
is indeed the most appealing
benchmark. That said, I am not
certain that we would expect full pass
through of costs in a market with
Appendix C. Appendix C also now
includes several new cases where we
explore a wide range of assumptions,
including a very fast rate of falloff
that effectively constrains all
substitution to within a few years.
Rapid falloff further reduces the
policy elasticity below the demand
elasticity, as might be expected given
that it makes substituting away from
new vehicles more difficult in the
long run. Appendix C also explores a
case with entirely flat substitution.
This is perhaps less realistic but
provides a useful extreme bound on
the substitution pattern.
-------
imperfect competition. It is perhaps
useful then to contemplate how the
results would be different if part of a
regulatory cost increase is borne by
new vehicle sellers, and part is passed
on to consumers. Is it possible to
simply interpret the cost shock in the
model as the portion passed on to
consumers?
The only method and procedure in the
modeling analysis that I had trouble
understanding was the assumption
about how the demand system was
assumed to be structured across used
vehicles of different ages. The text
describes the assumptions deployed in
an intuitive (and appealing) way, but
it does not write out algebraic
explanations. Furthermore, in some
places, the report offers as intuition
that older vehicles are less close
substitutes for newer vehicles, which
is meant to rationalize several
patterns. I struggled to follow this
reasoning because I could equally
imagine a world in which there was
only substitution between adjacent
ages (e.g., the only cross-price
elasticity for a 3-year old car is
between 2- and 4-year old cars), but
nevertheless a price shock to new
vehicles cascades through the entire
distribution of ages.
5. Does the modeling
analysis appear to
produce results
consistent with the
assumptions and data
used for model
development? Do the
results presented by
RTI follow from the
data and assumptions
used in the analysis?
The modeling analysis produces
results that appear to follow logically
from the assumptions and data. As a
peer reviewer, I did not attempt to
replicate any of the findings or inspect
the model code that executes the
simulations (nor was I given the
material to do so). Instead, my
judgment that the results reported
follow from the data, model, and
assumptions is based on the fact that
the pattern of results are consistent
We appreciate the positive feedback
on the reasonableness of our modeling
results.
-------
with the economic theory that the
analysis purports to embody. The
qualitative results that emerge in the
baseline model, and the difference
across the nine scenarios described in
section 7.4 (as well as the additional
versions in the appendix), conform
with theoretical expectations. These
results, and the economic reasoning
that aligns with them, are described
clearly in the report. I found that
discussion sound and convincing.
6. In what ways are
the results sensitive
to the data and
assumptions used in
model development?
Are there alternative
assumptions and data
that the researchers
should consider
providing improved
analysis?
The report explores the sensitivity of
key results by generating simulations
based on a range of parametric
assumptions that span the range of
elasticities from the literature. While
it is always possible to perform more
sensitivity analysis (and future users
will apparently be able to do so on
their own, which is a great feature of
the project), the current report does a
nice job overall of showing the likely
range of key results.
Quantitative values of course vary
across scenarios, but several key
findings, including the policy
elasticity being significantly smaller
than the new vehicle elasticity and the
overall fleet size shrinking in response
to a new vehicle price increase,
appear to be robust to a range of
inputs and assumptions. This is
reassuring and seems to suggest that
key qualitative insights that can
inform policy are very likely to hold.
Section 7.4 presents the core
sensitivity analysis by running a
variety of scenarios. This section
emphasizes variation in three
parameters: the new vehicle elasticity,
overall substitution to the outside
good, and the scrap elasticity. But, it
seems there is a fourth crucial
elasticity, which is the substitutability
As suggested by the reviewer, we
added more discussion of sensitivity
analysis at several points in the report.
We now look at different values for
substitutability between different
vehicle vintages in Appendix C. In
addition, we added a new section
focused on model and parameter
uncertainties associated with this
model in Section 9.4. Finally, we
expanded Appendix C to also include
output from both alternative survival
data sources noted in the main text
(Leard, and Jacobsen and van
Benthem).
-------
between vehicles of different ages.
The appendix provides a couple of
alternative model runs that seem to
address this point. The discussion
around these scenarios is quite
concise, and as a result I am not fully
certain I understand how to interpret
those results. The appendix results
seem to suggest that key model results
are not sensitive to big changes in
these assumptions.
The report does not feature standard
errors. Instead of focusing on standard
errors, the report focuses on
alternative scenarios. This is a
sensible approach, but it does open
the possibility that some users may
underestimate (or overestimate)
uncertainty. To help future users, it
might be useful to include a
discussion about uncertainty in one
section, starting with a bullet point of
sources of uncertainty in the results,
with some guidance as to what the
authors believe are the biggest
sources. These would include both
uncertainty about key parameters,
which might be amenable to some
statistical characterization, and model
uncertainties, which might be harder
to quantify.
At a few points, the report simply
states that the model results are robust
to differences in data inputs. This
includes reference to vehicle market
data used in a paper by Leard and one
by Jacobsen and van Benthem, rather
than the AEO report that is the main
data source. I fully believe this is true,
but it might be useful to report those
robustness checks somewhere.
ID ADDITIONAL ()\ LRAI.I. ( ()\I\IL\ I S
PROVIDLD (\()T( IIAR(.L ()l LSTIOV
SPL( ILK )
-------
The analysis described in this report is a significant step
forward and represents a valuable tool for improving
regulatory analysis of the car market.
Attempts to incorporate interactions between new vehicle
regulations and the used vehicle market that are based on
statistical relationships between new vehicle prices and the
used and new vehicle markets studied separately are
inadequate grounds for policy analysis and highly likely to be
misleading. In essence, this model studies long-run general
equilibrium interactions, something that is notoriously
difficult to study using observed market data. Thus, an
understanding of how interactions between new and used
vehicle markets dampen, amplify or propagate the effects of
new vehicle regulations requires an integrated model that
provides internally consistent effects constrained by economic
theory. The analysis in this report does exactly that, which is
why it is so valuable. Specifically, this new tool should foster
improved analysis of the ways in which used vehicle market
interactions may alter the net impact of new vehicle
regulations on consumers and the environment.
The model described in the report identifies key mechanisms
that govern the relationship between new vehicle costs and the
used vehicle market. There is substantial uncertainty about
key parameters, so care should be taken in interpreting the
quantitative output of the model (this is also true, of course, of
other models that underly regulatory analysis of automobiles),
but the analysis here reveals that certain facts are robust (e.g.,
the policy elasticity is smaller than the new vehicle demand
elasticity). Moreover, it provides a tool that can be used to
assess the plausibility of different outcomes (e.g., the total
fleet size can increase in response to a new vehicle cost
increase only under implausibility limited substitution to the
outside good).
A central contribution of the report is to highlight the
difference between a new vehicle demand elasticity and what
the report refers to as the policy elasticity, the latter of which
takes into account the feedback of used vehicle prices into the
demand for new vehicles following a cost shock to new
vehicles. The analysis demonstrates that the policy elasticity is
consistently smaller than the demand elasticity, and that the
magnitude of the gap between the two is of importance. This
is a critical insight, and it conforms with economic theory.
The report explains this relationship effectively, and it
We appreciate the positive comments
on the valuable contributions of this
report. As suggested by the reviewer,
we have added discussion of future
research and analyses that we feel are
likely to provide the most valuable
new information for continuing to
refine the use of this model for future
policy analyses in anew Section 10.4.
-------
provides a tool that users can customize to perform sensitivity
tests and additional analyses.
This tool might become even more useful in the future if
subsequent studies provide new information about the key
parameters that serve as inputs to the model. Having run many
iterations of the model (presumably many more than are
presented in the report), the authors likely have insights about
the most important parameter choices and modeling decisions.
A possible addition to the report would be a prioritized list
that emphasizes where additional estimates from future
research, or additional sensitivity analysis from future users,
would provide the most valuable new information
ADDITIONAL COMMENTS BY SPECIFIC REPORT
CHAPTER:
Included here are a series of comments, many of which are
quite minor expositional notes, ordered with respect to each
section of the report.
Section 3
This section provides a valuable discussion of the key
economic forces that link the new and used vehicle markets
and the role of scrappage. I suspect that it would be possible
to illustrate these relationships further by using a pair of
simple supply and demand diagrams to illustrate the steady-
state (long-run equilibrium) effects.
Initially, a new vehicle cost increase shifts up the supply curve
in the new vehicle market. The magnitude of the impact
depends on the elasticity of demand. The change in the new
vehicle market impacts used vehicle supply (shifted down
because there are fewer new cars to become used) and used
vehicle demand (more so as buyers view used and new as
close substitutes), resulting in a new price and quantity. The
size of this change depends on the scrap elasticity, which
determines the slope of used vehicle supply. The resulting
change in used vehicle price then shifts the new vehicle
demand curve, resulting in a new quantity.
The role that each of the key elasticities places could be
labeled in a heuristic diagram of these two markets.
Analytical expressions that show the simple relationship
between key elasticities would also be helpful. This is not
feasible in the full model with 30 vintages of used vehicles,
but a future modeling exercise that reduced the used vehicle
market to a single composite might make it possible to
As suggested by the reviewer, we
have incorporated a diagram
illustrating the supply and demand
relationships in the new- and used-
vehicle markets. This appears as
Figure 3-1, and we provide text
explaining the figure in Section 3.2.1.
Also as suggested, we can now
describe how the slopes of lines in the
figures, and therefore the relative size
of effects, connect to the core model
parameters. Deriving analytical
expressions in a two-age world is, we
think, best left for a future exercise.
We now discuss this is Section 3.2.
We think a two-age model would
likely not directly inform the results
of the present study because the
relationships in the more realistic
multi-age model here allow many
more interactions.
-------
characterize the key comparative statics, showing how they
depend on the handful of key elasticities.
Section 5
The literature synthesis is effective and clear overall.
The report offers a rich discussion of some of the studies, but
in a few cases the results of prior studies are reported without
a lot of commentary about their reliability or applicability to
the purpose at hand.
For example, my impression is that authors of papers in the
tradition of Berry, Levinson and Pakes (1995) are relatively
skeptical of the reliability of their own estimates of the
elasticity of substitution to the outside good, relative to the
other estimates in those systems. I gather that the magnitude
of this elasticity can scale with assumptions made by the
researcher about what fraction of consumers are shopping for
a car in a given period, something which is nebulous and often
chosen without hard data. This is not to suggest that there are
better estimates available, but rather to emphasize our relative
ignorance about the market level elasticities that are central to
the analysis here.
In some cases, the authors of the report had to do additional
calculations to extract an elasticity estimate from a prior
paper. To the extent possible, it would be useful to have those
steps explicated somewhere, for future reference. (For
example, I'm not sure that someone else can replicate the
author's interpretation of the elasticity in the Hahn study, or
replicate the new calculations from Bento et al. 2009.)
Based on the reviewers' comments,
we have incorporated additional text
on the reliability and applicability of
available elasticity estimates,
particularly caveats associated with
many of these estimates, starting on p.
5-1. In addition, we added footnote 17
indicating that the underlying
spreadsheets used to calculate
elasticity parameters for some studies
are available from the report authors
upon request.
Section 6
It might be helpful to include units (for the prices) in Figure 6-
2.
The demand system has (A+2)(A+l)/2 substitution
elasticities, where A is 30 in the present analysis. This is a
large number. The calibration feeds in basically three
elasticities. All of the rest are filled in via assumption about
the structure of cross-price elasticities through the age
structure. This assumption (described in sections 7-2 and 7-3,
in particular footnote 21) appears reasonable and is an
appealing solution to the problem that no estimates of this age
substitution pattern exists. But, I was not confident that I
could fully reconstruct the age structure of substitution from
the description in the text. A small, helpful step would be to
add algebraic expressions detailing this procedure and
We added units for the prices in
Figure 6-2.
We added algebraic expressions
detailing the calculation of the
substitution elasticities (and
discussion of methods for conducting
additional sensitivity analyses related
to substitution between different
vehicle vintages) in Appendix C
(footnote 42).
-------
commenting further on how interested users might conduct
additional sensitivity analysis (noting that some is provided in
the appendix).
Section 7
Table 7-1 (and several related tables) do not label the main
results as the "policy elasticity." This term is emphasized in
the text, but it is not called out in the tables. Adding it might
be helpful.
It might be helpful to also include some measure of price
changes (for used cars this may be an index across ages) in
Table 7-2. The price changes are not the object of interest, but
they are a key channel, so including them might help the
reader.
Figure 7-1 should include somewhere (title or note) that the
new vehicle elasticity is -0.8 because the main point of the
graph is that the policy elasticity is substantially smaller. This
is not apparent without referring back to a prior table.
We made edits to Tables 7-1 and 7-2
and Figure 7-1 to more clearly
identify the policy elasticities, based
on the review comments. We note that
the relevant price changes for a
selection of used vehicles are reported
in Table 7-3.
Section 10
Scrappage plays a key role in the analysis, but scrappage can
mean either that a vehicle is exported from the United States
or that it is fully decommissioned. The paper does not mention
this distinction. It is standard practice for regulatory analysis
to focus on emissions in the US fleet, and as such it is logical
to ignore this distinction in the present analysis. The
distinction is, however, relevant to the true greenhouse gas
emissions implications of new standards. This might be a
subject of future work.
The model built here is free of transaction costs. In reality,
transaction costs associated with buying and selling vehicles
are substantial. This is unlikely to matter for analysis of long-
run effects, but it could have an impact on transition
dynamics. Modeling transaction costs is certainly beyond the
scope of the current report, but it might be useful to speculate
as to whether transaction costs would amplify or dampen the
main effects during a transition.
As surmised by the reviewer, we did
not distinguish between vehicles
being exported or decommissioned.
We agree with the reviewer that this
could be valuable to explore in future
work and now mention this
distinction in our discussion of
potential future research in a new
Section 10.4.
We agree with the reviewer that
modeling of transaction costs is
beyond the scope of the current report
but could have effects on transition
dynamics for certain scenarios,
though we think even these are likely
to be small. We added discussion of
potential implications of incorporating
transaction costs in the new Section
10.4.
Appendices
It might be helpful to add further labels to Figure D-l, which
is showing two completely different model runs with separate
assumptions about expectations, rather than two sets of results
that come out of the same model run.
We added labels and a note to Figure
D-l to further clarify the contents.
-------
Some of the figures in appendix D label time just as "year"
instead of the more helpful "years since policy effective date."
We updated the labels to consistently
indicate "years since policy effective
date," as appropriate.
-------
\l/
'ICF
Peer Review of "The
Effects of New Vehicle
Price Changes on New
and Used Vehicle
Markets and Scrappage"
RTI Report 0215574.004.028
Submitted to:
U.S. Environmental Protection Agency
2000 Traverwood Drive
Ann Arbor, Ml 48130
Submitted by:
ICF
100 Cambridgepark Drive, Suite 501
Cambridge, MA 02140
EPA Contract No. EP-C-16-020
Work Assignment No. 4-04
June 11, 2021
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Table of Contents
I. Introduction 3
II. Peer Review Process ................................................................................................................... 4
1. Selecting Reviewers 4
2. Administering the Review and Receiving Comments 5
3. Difficulties Encountered 6
III. Responses to Charge Questions ................................................................................................ 6
1. Comment Overview and Summary 6
1.1 Does the Presentation Describe the Data and Methods Sufficiently? 7
1.2 Does the Report Miss Relevant Literature; Are the Interpretations of the Elasticities
Defensible? 7
1.3 Are the Data and Assumptions Appropriate and Objectively Chosen? 8
1.4 Are the Methods and Procedures Employed Technically Appropriate and Reasonable?. 8
1.5 Does the Modeling Analysis Appear to Produce Results Consistent with the Assumptions
and Data? 9
1.6 Are the Results Sensitive to the Data and Assumptions Used in Model Development; Are
There Alternative Assumptions? 9
2. Comments by Reviewer 10
2.1 Comments by Ashley Langer 10
2.2 Comments by Benjamin Leard 18
2.3 Comments by James Sallee 23
Appendix A. Resumes of Selected Reviewers ......................................................................... 33
Appendix B. Conflicts of Interest Declarations 49
Appendix C. Peer Reviewer Mid-Review Meeting Notes ......................................................... 53
2
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
I. Introduction
The U.S. Environmental Protection Agency (EPA) sets emission standards for new light-duty
vehicles. Such standards are expected to increase the up-front costs of new vehicles, and thus
to affect new-vehicle sales as well as the market for used vehicles by altering substitution
between new and used vehicles. Changes in the valuation of used vehicles are, in turn, likely to
affect vehicle scrappage rates by reducing the number of old vehicles that are retired each year.
Consumer response to changes in vehicle prices induced by regulatory policy can therefore
have important implications for the total size and average age of the U.S. vehicle inventory,
influencing net impacts of regulatory actions on greenhouse gases, criteria pollutants, vehicle
safety, and other outcomes. Being able to estimate the effect of regulations on the new and
used vehicle markets will enhance EPA's ability to examine the economic and environmental
effects of new light-duty vehicle emission standards.
The RTI report: 0215574.004.028, "The Effects of New Vehicle Price Changes on New and
Used Vehicle Markets and Scrappage" (referred to as the Report) describes the development of
methods to estimate these impacts. The methods involve identifying a relationship between
changes in new-vehicle costs and new and used vehicle fleet sizes using a simulation model
parameterized with literature estimates of various market and consumer response elasticities.
EPA's guidelines specify that all highly significant scientific and technical work products shall
undergo independent peer review according to specific agency protocols. This process is
designed to ensure the use of the highest quality science in its predictive assessments and to
assure stakeholders that each analysis/study has been conducted in a rigorous, appropriate,
and defensible way. Therefore, EPA submitted the Report for external peer review, seeking
independent expert opinion on the methodologies employed and analyses presented in the
Report. ICF facilitated this peer review process. This memorandum contains a summary of the
peer review results as well as documentation of the process.
The peer review was conducted from March to June 2021 in accordance with the current
version of EPA's Peer Review Handbook,1 At the conclusion of the review process, ICF
collected all unedited peer reviewers' comments and provided them to EPA. This Technical
Report contains a brief summary of the reviewers' comments to EPA's charge questions, along
with the unedited answers presented by each peer reviewer.
Supporting documentation collected from the reviewers, including their curriculum vitae (CV)
and conflict of interest (COI) statements, is also provided.
1 U.S. Environmental Protection Agency, Peer Review Handbook, 4th Edition, October 2015. Prepared for the U.S.
EPA by Members of the Peer Review Advisory Group, for EPA's Science Policy Council, EPA/100/B-15/001.
Available at https://www.epa.gov/osa/peer-review-handbook-4th-edition-2015. including OMB's Information Quality
Bulletin for Peer Review (Handbook, Appendix B) provisions for the conduct of peer reviews across federal
agencies.
3
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
The following materials are included in this Task 3 Technical Report.
1. Description of the Peer Review Process (Section II)
2. Reviewer Responses to Charge Questions (Section III)
3. Reviewer Supporting Documentation (Appendix A and Appendix B)
4. Notes from mid-review meetings with EPA, ICF, and the contracted peer reviewers
(Appendix C).
II. Peer Review Process
ICF conducted the peer review in three stages. First, ICF identified a qualified set of reviewers;
second, ICF contracted with the selected peer reviewers and conducted the review; then, ICF
collected reviewers' feedback on the RTI Report. ICF documented the peer review process,
consisting of this Technical Report, to submit the assembled information from the peer
reviewers to EPA. Ultimately, EPA will convey results of the peer review process to the authors
of the RTI Report, who will respond to the comments received. The following sections provide
detail on these steps.
1. Select i vers
ICF first identified a pool of independent subject matter experts from which to select three
qualified candidates to form a review panel. Qualifications included two technical
considerations. First, candidates could not have actual or apparent conflict(s) of interest or lack
of impartiality that would preclude an independent review. Secondly, it was necessary that the
combined expertise of the candidates covered the two focus fields of this analysis:
1. Simulation modeling of light duty vehicle price changes and consumer behavior
response.
2. Fuel economy policies of light duty vehicles.
ICF identified fourteen potential reviewers for the report based on a combination of individuals
originally suggested by EPA and identified through ICF's research. ICF then contacted each
candidate by e-mail to assess reviewer's expertise in the field, ability to perform the work during
the period of performance, and any association with the topic that would preclude an
independent review (note that at this stage, the Report was not shared with the potential peer
reviewers). ICF also collected a CV or resume from each candidate that expressed availability
and interest in participating in the review panel.
Based on all the information gathered during this initial phase, ICF selected three reviewers.
While all candidates were highly qualified to act as peer reviewers, ICF sought to select
candidates whose combined expertise would cover all technical aspects of the Report while
bringing diverse and complementary perspective to the peer review process. ICF suggested the
4
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
following reviewers in the April 12, 2021 Peer Review Selection Memo to EPA2, here listed in
alphabetical order
1. Ashley Langer
Assistant Professor, Department of Economics
Faculty Research Fellow of the National Bureau of Economic Research
University of Arizona, Tucson
McClelland Hall 401
Tucson, AZ 85721
alanqer@email.arizona.edu
520-621-0117
2. Benjamin Leard
Assistant Professor, Department of Agricultural and Resource Economics
Faculty Fellow, Howard Baker Jr. Center for Public Policy
University of Tennessee, Knoxville
Morgan Hall 31 OA
Knoxville, TN 37996
bleard@utk.edu
865-974-5005
3. James Sallee
Associate Professor, Department of Agricultural and Resource Economics
University of California, Berkeley
207 Giannini Hall
Berkeley, CA 94720
sallee@berkelev.edu
510-643-5133
ICF anticipated that this selected group of reviewers would provide extensive and
complementary expertise to conduct the peer review, as documented by ICF in the Peer Review
Selection Memo. EPA concurred with ICF choices and approved all selected reviewers.
2. Acl it ifi intents
ICF composed and delivered a charge letter to the three selected reviewers to send along with
the RTI Report "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets
and Scrappage". The charge letter included EPA's charge questions to the reviewers,
instructions on how to complete the review, a timeline of when comments were due to ICF, and
a conflict of interest (COI) form to be returned to ICF along with the comments. ICF sent these
materials to each individual reviewer on May 3, 2021.
2 Draft peer reviewer selection memo, REVISED (Task 1) Contract EP-C-16-020, Work Assignment No. 4-04: Peer
Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage", to
Elizabeth Miller and Dana Jackson, US EPA OTAQ, from: Paola Massoli, and Sarah Lettes, ICF.
5
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
EPA then arranged and hosted a mid-review teleconference on May 6, 2021 with the selected
peer reviewers and ICF. The goal of the meeting was to introduce the peer reviewers to the
EPA staff and address early questions or concerns. The 1-hour meeting included an overview of
the review process, background information on the Report, and a discussion on technical and
practical aspects. ICF's notes from this meeting are included as Appendix C.
ICF requested that the peer reviewers provide comments to the charge questions within two
weeks. All peer reviewer comments and completed COI forms were received by May 25, 2021.
ICF compiled all unedited peer review comments, charge letter information, and attachments
into a peer review report. ICF organized all comments into tables so that the individual
comments could be easily grouped and compared for review purposes. ICF prepared and
submitted a Draft Peer Review Summary Report that assembled the unedited reviewer
comments for EPA review, and delivered the draft report to EPA on June 2, 2021. EPA provided
ICF with additional comments and edits, and a revised, final report was delivered to EPA on
June 11, 2021.
3. Difficulties Encountered
No significant difficulties were encountered while performing this review.
III. Responses to Charge Questions
Section 1 presents an overview of the peer reviewers' comments received on the six charge
questions. This overview is followed by Section 2, which provides the direct, unedited peer
reviewer responses to each of the charge questions. The unedited responses by reviewer
appear in a table format. In those tables, the left column lists the EPA's charge question, and
the right column provides the reviewer's comments.
1. Co mm iew
The following section summarizes the peer reviewers' comments to the charge questions. The
questions have been abbreviated for easier presentation. These summaries do not rewrite the
responses or supersede the unedited comments presented in Section 2.
Overall, all three peer reviewers agree that the modeling analysis presented in the Report
produces results consistent with the assumptions and data used for model development. This
said, each reviewer had comments and constructive suggestions to improve content or to clarify
the literature cited and the methodology used.
Both Dr. Langer and Dr. Sallee also provided extensive additional comments beyond those
requested by the six prescribed charge questions. Those are not summarized here but are
presented in their entirety in Section 2. In addition, Dr. Langer provided direct edits to the Report
8
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
draft document, which was shared with EPA; those comments are included in the Additional
Comments section of the table.
1.1 Does the Presentation Describe the Data and Methods Sufficiently?
The reviewers agreed that the report presented the data and methods sufficiently. In fact, Dr.
Langer called it the "best government report" that she has read.
Both Dr. Langer and Dr. Sallee noted that the explanations of each assumption, parameter, and
component of the model were clear, and that the document was organized in a way that
facilitated the understanding of such a complex model.
Dr. Sallee appreciated that the report explained the economic intuition underlying each
component of the quantitative model. Dr. Langer noted that the report was so complete that she
would have been able to code the simulation to test alternative models. Dr. Leard agreed that
the report described the data and methods sufficiently. He pointed two citations that should be
revisited to reflect recently updated or published papers and provided the updated citations.
1.2 Does the Report Miss Relevant Literature; Are the Interpretations of
the Elasticities Defensible?
The reviewers provided mixed responses to this charge question.
Both Dr. Langer and Dr. Sallee praised the authors of the Report for having extensively
reviewed and utilized information available in the literature to extract or calculate elasticities
include in their model, while using good judgment on which values to include or exclude based
on relevance and data quality. Dr. Langer suggested the citation of two additional studies that
model the dynamic adoption of new and used cars and should be included in the report.
Dr. Sallee commented that the report included all the relevant studies of which he was aware of,
but also recognized that there are only a few studies in the literature that provide suitable
elasticity parameters to characterize demand for new vehicles and other consumer behaviors.
He pointed out that many of these studies are dated, and that current substitutability of new and
old vehicles may have changed due to changes in embedded technology (i.e., new vehicles are
more durable and last longer). He also added that the elasticity of demand for new vehicles is
likely smaller as consumers get wealthier, so older studies might exaggerate new-vehicle price
elasticity.
Finally, he noted that a substantial share of the available studies does not focus on the elasticity
parameters that are of interest in the Report. Hence, while recognizing the overall approach and
interpretations of elasticities from the literature as defensible, he suggested adding a sensitivity
analysis that incorporates standard errors from the original literature studies. This exercise
would provide additional context on the uncertainty of the elasticity parameters used in the
model.
7
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Dr. Leard noted that the Report misses several relevant citations, while pointing out that the
interpretation of one of the citations is incorrect and suggested an alternative take. Regarding
the studies that Dr. Leard suggested adding, he pointed out that each of these sources would
provide additional context on elasticities.
1.3 Are the Data and Assumptions Appropriate and Objectively Chosen?
The reviewers provided mixed responses to this charge question.
Dr. Langer and Dr. Sallee both found that the data, key model parameters and parameter
assumptions were appropriate, reasonable, based on reliable data sources, and consistent with
an objective analysis. However, Dr. Sallee posed a question on the baseline scrappage scale
factors for which he suggested a minor clarification.
Conversely, Dr. Leard found several citations to be inappropriate. He expressed concerns
regarding the use of certain papers to construct a range of elasticities, offered some
suggestions on how those papers should be cited in other sections of the Report and shared
alternative papers that could be cited instead. The concern around using certain elasticities is
consistent with Dr. Sallee's remark on the need to address the uncertainty of the elasticity
parameters used in the model (see Dr. Sallee's response to charge question 2). If the authors of
the Report accept Dr. Leard's comments on this charge question, the estimates of the market
price elasticity of demand for new vehicles might have to be revised.
1.4 Are the Methods and Procedures Employed Technically Appropriate
and Reasonable?
Each of the peer reviewers expressed different views regarding methods and procedures
employed in the Report.
Dr. Langer had no objections and felt that the authors are clear about the assumptions of the
model.
Dr. Leard raised six specific concerns with the method. Some concerns (Concerns 2, 3, 4, 6)
appear to be solvable by providing additional clarifications, while others (Concerns 1 and 5)
might require a more in-depth analysis of methods and assumptions. A brief summary of
Concerns 1 and 5 is provided here. Concern 1 questions the modeling of vehicle ownership
costs, and suggests a simplified approach to model used-car demand by considering vehicle
prices or vehicle prices plus fuel costs only. This would simplify the model set up and it would be
consistent with how supply is defined, which is a function of used vehicle prices. Concern 5 is
related to the assumptions of price elasticities for used and new vehicles and points out that
there has to be a distinction based on households' income. Data show that new vehicles are
more frequently bought by wealthy households, and that wealthy households tend to be less
responsive to price than low-income ones. For this reason, Dr. Leard expects the own-price
elasticities to be larger in magnitude for older vehicles.
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
In addition, Dr. Leard's Concern 3 suggests that the price of a new vehicle should not be taken
as given, and that it would be useful to add more details on why the new vehicle market was
assumed to be perfectly competitive.
Dr. Salle provided an extensive answer to this charge question. He recognized that the model is
complex and that the Report must make assumptions, including focusing the model on vehicle
ownership (this is in contrast with Dr. Leard's view on including vehicle ownership; see Concern
1 in the response to charge question 4). However, Dr. Sallee expressed a concern regarding
'the assumption about how the demand system was assumed to be structured across used
vehicles of different ages' which appears to be related some of Dr. Leard's concerns around the
treatment of elasticities associated with vehicle age(s).
Finally, Dr. Sallee incorporated a qualitative analysis of the model applicability to future
scenarios. For example, if the model were to be used to contemplate a transition to electric
vehicles or vehicles featuring different levels of automation, attributes affected by the policy
could radically change the substitution between new and used vehicles.
1.5 Does the Modeling Analysis Appear to Produce Results Consistent
with the Assumptions and Data?
Regarding the results produced by RTI's modeling analysis, all three reviewers agreed that the
modeling results are consistent with the assumptions and data used in model development, and
that such results appear logical. As specified by Dr. Sallee, the qualitative results that emerge in
the baseline model conform with theoretical expectations. Dr. Leard pointed out that it makes
sense that the policy elasticities are smaller in magnitude than the new vehicle demand
elasticities.
1.6 Are the Results Sensitive to the Data and Assumptions Used in Model
Development; Are There Alternative Assumptions?
All three reviewers recognized the overall qualitative and quantitative value of the Report but
had several recommendations in response to this charge question. Dr. Langer suggested
adding a discussion to clarify the limitations of the model in predicting or representing the effect
of policies that do not uniformly affect the new vehicle fleet. Dr. Sallee suggested that the
authors explore the role of an additional elasticity - the substitutability between vehicles of
different ages - in section 7.4, which presents the core sensitivity analysis. He also notes that
while it is acceptable that the Report does not feature standard errors, it would be beneficial to
add a discussion covering uncertainty about key parameters and overall model uncertainty.
Finally, Dr. Leard encouraged applying a different range for the market price elasticity of
demand for new vehicles. ICF noted that this might require a re-run of the simulations. Dr. Leard
also recommended adding greater detail to the definition of a vehicle to make the model more
suitable to performing simulations of an actual policy.
9
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
2. Comm< ewer
AS i
CHARGE QUESTION
COMMENTS
1. Does the presentation describe
the data and methods sufficiently
to allow the reader to form a
general view of the quality and
validity of the analysis approach?
Yes. This is the best government report I've ever read.
The model is laid out clearly and is in line with the state-
of-the-art research in the best economics journals. The
authors are extremely clear about where each
parameter is coming from and why they made the
modeling assumptions that they did. I feel like after
reading this report I could, with relatively little additional
thought, actually code up the dynamic simulation in the
paper and test alternative models. That is a feat in a
technical report like this.
2. Does the report miss relevant
literature in its review? Are the
interpretations of the elasticities in
the literature review, and the
estimates of elasticities used in
the model stemming from the
literature review, defensible?
I think that the authors overall have gone well beyond
just conducting a literature review. They have calculated
elasticities using important estimates from the literature
and in some cases have literally gone back to the data
and code from the papers and calculated elasticities that
are not recoverable from only the published version.
They have also used good judgement to choose which
elasticities from the literature are more likely to be
applicable to their model and give the reader solid
reasoning for why they made the choices they did.
That said, I do think that there are a couple of papers
that should probably be at least discussed. Schiraldi
(2011) and Adda and Cooper (2000) both model
dynamic adoption of new and used cars. They aren't
based in the U.S., but they may be the closest models to
the ones the authors are after here. Putting their model
in the context of the dynamic car purchasing literature
seems important.
3. Are the data and assumptions
appropriate for the analysis
conducted and objectively
chosen? If not, do you know of
other data or proposed alternative
assumptions that might be used in
this analysis?
I think that the data and assumptions of the model are
appropriate and reasonable. The authors have provided
extensive sensitivity testing and their take-aways from
the model appear to be extremely generalizable.
10
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
4. Are the methods and
procedures employed technically
appropriate and reasonable?
Please distinguish between cases
involving reasonable
disagreement in methods as
opposed to cases where you
conclude that current methods
involve specific technical errors.
The methods and procedures are technically appropriate
and reasonable. In particular, because the authors are
not attempting to defend a specific point estimate, but
are more attempting to lay out intuition for how policy
affects equilibrium new and used vehicle adoption and
pricing, I don't have any real critiques of the approach.
The authors are clear about the assumptions of the
model and the ways that these assumptions could be
relaxed in future work.
5. Does the modeling analysis
appear to produce results
consistent with the assumptions
and data used for model
development? Do the results
presented by RTI follow from the
data and assumptions used in the
analysis?
The results are consistent with the assumptions and
data used in model development. Beyond following from
the data, the results provide intuition for how the results
would change with alternative modeling assumptions
and the authors are very clear about which assumptions
are most critical for changing the results.
6. In what ways are the results
sensitive to the data and
assumptions used in model
development? Are there
alternative assumptions and data
that the researchers should
consider providing improved
analysis?
The authors are very careful to keep the analysis to the
adoption and pricing of a single representative vehicle of
each age. I do worry slightly that others who read the
report may want to use these results to draw
conclusions about the effect of things like fuel economy
policy on the overall fuel economy of the vehicle fleet
over time. I think it might make sense for the authors to
add some discussion of the complications that within-
vintage vehicle heterogeneity is likely to add in the real
world. In particular, a policy like a CAFE standard, which
penalizes fuel inefficient vehicles while subsidizing fuel
efficient vehicles, will have different long-run effects on
the age distribution of fuel inefficient and efficient
vehicles. I see this not as something that the authors
should do to improve this analysis: adding heterogeneity
is very complicated as the authors point out. But I do
think that it would be helpful for the authors to be fairly
clear about the limits of the analysis they have
conducted by explicitly saying that policies that do not
uniformly affect the new vehicle fleet will have
complicated effects on the long-run age distribution of
vehicles.
11
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
ADDITIONAL OVERALL COMMENTS PROVIDED (NOT CHARGE QUESTION-SPECIFIC):
Overall, this is the best report of this type I have ever read. The analysis provides both
intuition and a concrete path forward for future analysis. The model rests on reasonable
assumptions and is very clear about how things might change under alternative assumptions.
The calibration is reasonable and makes choices about which values from the literature to
use that are based on the quality of the studies rather than weighting all previous work
equally. The authors conduct extensive sensitivity analysis that allows the reader to fully
understand which results are robust to alternative assumptions and which depend on the
parameterization. Congratulations on great work!
ADDITIONAL COMMENTS BY SPECIFIC REPORT CHAPTER:
Dr. Langer also commented on specific sections of text throughout the report:
3.1, Bullet 4 ("Production of vehicles..."), which discusses long-run changes in vehicle
models: Cite Knittel and coauthors here on long-run model changes? Was that paper ever
published?
3.1.1.1, Paragraph 3 ("In the literature estimating..."), which discusses new-vehicle demand
systems and the distinction between purchase price and net ownership costs: This paragraph
was helpful for explaining what you are trying to get at.
3.1.1.2, Paragraph 2 ("Much of the literature..."), which discuss the distinction between
average own-price elasticity among individual models of new vehicles and aggregate own-
price elasticity: But it's also important to recognize that the identifying variation behind these
estimates may not be well set up to get at these aggregate elasticities.
3.1.1.3, Paragraph 1, in reference to the sentence: "The derivatives of demand here reflect a
world where (at least from the consumer's perspective) only the price of the vehicle has
increased:" Single vehicle?
3.1.1.3, Paragraph 2 ("More often..."), which discusses derivatives of demand and consumer
willingness to pay for vehicle attributes/regulations that change vehicle attributes-. It's probably
worth being clear here that these derivatives are at least theoretically possible to get from the
models, if the models include the right vehicle attributes.
3.1.1.3, Paragraph 3 (starting at "A consumer might..."), which discusses the effects of fuel
economy regulations on consumer demand and explains that consumers may perceive fuel
economy-related gas savings differently. This doesn't need to be phrased as heterogeneity in
beliefs. It could just be that some consumers driver more and others drive less so increases
in fuel economy have different effects for different consumers.
3.1.1.3, Paragraph 4, in reference to the sentence: "We will refer to the combined change in
perceived cost at the time of purchase as a 'generalized cost:'" Shouldn't this be something
12
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
like the compensating variation? Is it really "perceived cost"? and it's clearly heterogeneous,
so it's likely the mean change in perceived cost?
3.1.2, Paragraph 1, in reference to the sentence: "If assuming perfect competition, supply will
in fact be perfectly elastic:" But we wouldn't do this in the car industry.
3.1.2, Paragraph 1, in reference to the sentence: "However, the presence of a used market
interacts with the equilibrium system and the new-vehicle demand elasticity is no longer
sufficient to measure the effect of a policy, even when new vehicle supply is perfectly elastic:"
Related to the above, I would be stronger here that supply isn't perfectly elastic, even in the
long run.
3.1.2, Paragraph 2, in reference to the sentence: "When a particular used model's price rises,
more of that model become available (e.g., because scrap dealers, insurance companies,
and mechanics decide to repair and sell more of them as vehicles instead of as scrap metal):"
Owners don't also respond? Trade in vehicles rather than keeping them as an extra car for a
child or relative?
3.1.2, Paragraph 3, which discusses scrappage functions. In response to the phrase "vehicle
scrappage depends only on own price of vehicles:" It should depend on the price of other
vehicles to the extent that parts for repairs are interchangeable and supply curves for those
parts slope up.
3.1.2, Paragraph 4, in reference to the sentence: "Empirical estimates will typically be
presented as a derivative or elasticity of this function, rather than of the density of underlying
shocks that determines the function:" Derivative of the density, no?
3.1.3, Paragraph 1, which discusses how the presence of equilibrium effects in observational
data may lead to biased estimates:" Would this be clearer if it were framed as omitted
variables bias? So there are omitted variables that are correlated with both the likelihood of
scrappage and the cost of repair?
3.1.3, Paragraph 3, which discusses the effects of new-vehicle prices on scrap rates of used
vehicles, in reference to the phrase: "a downward shock to the scrap rate will create more
competition for new versions of a vehicle (because there are now more used ones entering
the market):" Remaining in the market, not entering the market.
3.1.3, Paragraph 3, which discusses the effects of new-vehicle prices on scrap rates of used
vehicles, in reference to the phrase: "Estimating the reduced form successfully would require
good quasi-experimental variation in new-vehicle prices, but the likely confounders and long
time series required have meant that the literature has been unable to find many suitable
settings:" And is this even the right elasticity? As the authors say above, this is only for a price
change for a single vehicle, and the cross-price elasticities are likely to matter because used
car prices are determined in equilibrium.
3.2.1, Paragraph 3, in reference to the question: "How do the effects on the composition of
the vehicle inventory relate to the initial policy goals?" It seems like somewhere here there
also needs to be recognition that it's not just one "substitution" effect to used cars but that it
will depend a lot on which vehicles the new car buyers are substituting toward, and how
substitution is happening within the used car market (so 20 year old vehicles may still be
13
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
scrapped at high rates, but there may be much less scrappage of 10 year old vehicles and
this changes the policy impacts).
3.2.1, Paragraph 3, in reference to the question: "For the group of consumers that substitutes
from new to used, how much does this substitution drive up equilibrium used-vehicle prices?"
And WHICH used vehicle prices?
3.2.2, Paragraph 6, which starts with "A dynamic model..."in reference to the sentence
"Assuming the policy shock is long-lived the effects on used-vehicle prices and scrappage
evolve, likely strengthening, over time." Why? Not sure it's not true, but not completely
obvious why it is. Couldn't producers evolve/adapt to weaken the effects?
4.1, Paragraph 3, in reference to the sentence "The fact that the elasticities differ in this way
is entirely consistent with economic theory; more disaggregated choice sets (in terms of
attributes, time of purchase, etc.) mean that the best substitutes for any given good within the
choice set will be more similar to it (e.g., more substitution would be expected between sedan
models or between makes of sedan than between new and used vehicles or between a
personal passenger vehicle and alternative means of transportation)." This is not exactly how
I would think about the issue. Substitution happens for many reasons. In particular, since
people are heterogeneous, having more products means that there are products that are
similar to a vehicle on different dimensions, which means that more heterogeneous people
can be on the margin between buying this vehicle and another.
4.2, Table 4-2: List of Papers Included and Parameter Estimates Provided. Busse, Knittel,
and Zettlemeyer doesn't have any relevant elasticities? I guess everything that they're looking
at is responses to gas prices rather than responses to price changes, but isn't that central to
their generalized price measure?
5.1, in reference to this sentence about Table 5-1: "The elasticities in the first panel hold
constant most other aspects of the equilibrium system, including substitution possibilities to
used vehicles." What does this mean, exactly? Are you just saying that the attributes of used
vehicles are held constant? Not sure what a substitution "possibility" means.
5.1.1, which discusses aggregate own-price elasticity of demand for new vehicles with
respect to the price of new vehicles: I feel like somewhere in here there needs to be more
discussion of the fact that this is not necessarily all substitution to used vehicles but also
includes substitution to fewer total vehicles (either not traveling at all, using a used vehicle
more, or switching to bike/transit/etc.). But maybe I missed that elsewhere?
5.1.1, Paragraph 1, in reference to the sentence: "In addition, we made a simple assumption
regarding average model-level elasticities and average substitution to the outside good;" This
is a little confusing, but I think is saying that everything should technically be share-weighted
averages instead of simple averages, right?
5.1.1, Paragraph 4, which starts with "Barry, Levinson..." noted that it should say "(2004)"
after Pakes in the last sentence.
5.1.3, Paragraph 3, in reference to the first sentence: "As expected given the relationship
between the new- and used-vehicle markets, the available elasticities of new-vehicle demand
when used-vehicle prices can adjust (-0.18, -0.36) indicate less responsiveness." What
14
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
timeframe should we think of these elasticities as being over? In BLP, it's very clearly a year.
But here are they longer-term? Presumably not super long-term though?
5.2.2, Paragraph 1, in reference to a sentence about estimates of own-price elasticity for used
vehicles, "Their estimate of own-price elasticity for used vehicles ranged from -0.75 to -1.93
with a mean of-1.23." Aggregate own price elasticity?
5.2.2, Paragraph 2, in reference to the sentence "Bento et al. (2009) also reported model-
level price elasticities among used vehicles: the average vehicle-level elasticity among all
used vehicles was -0.54." I guess I'm a bit confused about what a used vehicle own-price
elasticity is given that there are consumers on both the supply and demand sides. Given that
there are lots of substitutes I would expect an own-price elasticity of a used vehicle to be well
above 1 in magnitude. But if the seller is basically going to sell the car one way or another
regardless, then does that pull down the elasticity? So if prices go up, the supply of used
vehicles goes up and the demand goes down so the number of vehicles TRANSACTED
doesn't change? Is there a way to provide some clarity on this?
5.2.4, but seems to be in reference to section 5.2 as a whole: Does Schiraldi provide an
estimate that is useful? Adda and Cooper? They're not on US data, but it still seems like
they're informative.
5.3.1, which discusses the elasticity of aggregate scrappage with respect to the average price
of used vehicles, in reference to the sentence: "In Hahn's (1995) Table 2, he relates an
exogenous "bounty" (subsidy to scrappage) to changes in the absolute number of vehicles
scrapped relative to a baseline."Timeframe seems important to mention here. Short-run, I
assume?
5.3.1, Paragraph 3, in reference to the sentence: An important caveat of studies looking at a
bounty for scrappage lies in the temporary nature of the program: it may be that a short-lived,
salient subsidy will quickly harvest a stock of "almost-scrapped" vehicles, overstating the
elasticity. Yes, maybe just move this up front in the discussion?
5.4, Table 5-4. "Time Frame of Data Used" is hard to read.
Section 6 (overarching statement about the section): I get that this is sort of an obnoxious
comment, but I could have used more of a road-map up front on this section. For instance, it
wasn't completely clear up front that the price vector is sort of secondary and just guarantees
equilibrium in the system of equations. It's obvious after reading through, but being a little bit
clearer about how this section would be laid out would have been helpful.
6.1, in reference to "q° number of vehicles of age a demanded" (as well as the notation for
number of vehicles of age a supplied). Demanded in transactions or overall? Ditto with
supplied.
6.2, Paragraph 1, in response to the sentence "The elasticities for individual vehicles
measured by this approach do not provide clear signals about an elasticity of demand for new
vehicles in the aggregate, because the outside good for individual vehicles includes other
new vehicle models; thus, demand for an individual model is likely to be much more elastic
than demand for a generic new vehicle." I get what you're saying here, but wouldn't it be
clearer to just say "the alternative choice for each individual vehicle includes other new
vehicle models and the outside good"?
15
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
6.2, Paragraph 2, which starts with "We assumed..." and discusses assumptions about the
demand system and the electrification of the vehicle market. Good discussion in this whole
paragraph.
6.4, Paragraph 5, which starts with "Importantly, supply and demand..." Related to an earlier
comment: this paragraph is super helpful for understanding how everything is coming
together and more of this earlier in the section would be helpful to the reader.
6.4.1, Paragraph 3, in reference to the sentence "Because scrappage occurs when repair
costs exceed the value of the vehicle the form of the scrap function (constant elasticity)
determines the form of the repair cost density." Isn't this backwards? The repair cost density
is the primitive and the scrap function follows from it? I appreciate that you are putting an
assumption on the scrap function rather than the repair cost density though.
6.6, Paragraph 6 about fia (Share of aggregate spending on each vehicle age), in reference
to the sentence "Expenditure shares come from combining the age profile above with data on
the cost of vehicles of different ages." I'm fairly confident that you explained this somewhere
and I'm just missing it, but why isn't \beta_a just calculated in the model from the equilibrium
prices and quantities? Why does it need to be treated as an exogenous parameter? It seems
very odd given the simulation exercise?
6.6, Figure 6-2: Baseline Price Profile by Age. I think that the "baseline" here just means that
you use it for the expenditure shares and then update it in the simulation, yes? A clearer title
would be helpful.
7.1, Paragraph 2, which discusses Table 7-1: Demonstration of Channels of Adjustment in
Quantities and Prices When Generalized Cost of New Vehicles Rises by 1%, in reference to
the sentence "The 1% increase in generalized cost measures the strength of the policy,
hence the term "policy elasticity." Impact? I'm not really sure what "strength" means here.
7.1, Table 7-1, in reference to the Cross-Price New/Used column: Maybe I missed this earlier,
but this is the elasticity to all ages of used vehicles? I get that this might make sense since
new buyers switch to newer used vehicles and those buyers switch to older used vehicles
and it all works its way through the system, but being explicit about this would probably be
helpful.
7.2, Paragraph 5, in reference to the first sentence: "To complete the setting, note that the
demand system contains 30 ages, but so far we have only specified three demand
elasticities." Ok, this makes sense re: my above comment. A quick sentence about this earlier
where you talk about assumptions would be helpful.
7.4, Paragraph 3, which describes Table 7-2: Policy Elasticities Corresponding to Selected
Demand and Scrappage Elasticities, in reference to the sentence: "The first panel (rows A
through E) holds the scrappage elasticity fixed at a value of -0.7 (Jacobsen and van Benthem
[2015] and also close to the median of values in Table 5-3)." Rows A through E: Not labeled.
7.4, Paragraph 3, which describes Table 7-2: Policy Elasticities Corresponding to Selected
Demand and Scrappage Elasticities, in reference to the sentence: "The second panel (rows F
through I) explores changes in this elasticity, spanning most of the range in the literature with
values of-0.2 to -1.2.)." No second panel.
18
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
7.4, Paragraph 4, in reference to the sentences: "Scenarios A through C explore an increase
in substitution to the outside good. As we saw in Table 7-1, this is one of the most important
elasticities in determining the effect of policy on new-vehicle sales." It's probably worth being
really clear early in the report that this is critical and perhaps the worst-identified parameter in
the modern discrete choice literature.
7.4, Paragraph 11, which starts with "Figure 7-1 displays..."in reference to the sentence
"When the scrap elasticity is -0.7 on the horizontal axis, the values on the Y axis reflect the
policy elasticities as shown in Table 7-2 for Scenario D." Instead of Y, vertical. Or X instead of
horizontal [for parallel sentence structure],
8.3, Paragraph 3, referring to the sentence "For instance, there is some empirical evidence of
increased buying in advance of regulatory changes going into effect, once those changes
have been announced." Do you want cites on these types of statements?
9.1, Paragraph 2, referring to the sentence "This is a small-scale version of changes in real
estate prices that can occur even when the announced change is decades away." Exactly
what Holland, Mansur, and Yates get for electric vehicle mandates in their recent AEJ.
9.1, Paragraph 2, referring to the word "sales" in the last sentence, "Preannouncement has no
effect on the long-run outcome: sales converge to the 0.25% decline in a little over 5years."
New vehicle sales. Just being clear that the changes in the used vehicle market take longer to
get worked out (I assume).
17
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
ijamin Leard
CHARGE QUESTION
COMMENTS
1. Does the presentation
describe the data and methods
sufficiently to allow the reader to
form a general view of the
quality and validity of the
analysis approach?
In general, yes. The document could cite a few citations
that have either been updated or published.
Citation 1: Dou and Linn (2018), which was a working
paper, is now a published paper appearing in the Journal
of Environmental Economics and Management. The
citation is
Dou, X. and J. Linn (2020). How do US passenger vehicle
fuel economy standards affect new vehicle purchases?
Journal of Environmental Economics and Management.
102: 1-21.
Citation 2: Leard (2021) is an updated version of Leard
(2020). The current version is available at
httDs://media.rff.ora/documents/WP 19-01 rev 2021.odf.
The market price elasticity of demand estimate from the
most recent version is -0.37.
2. Does the report miss relevant
literature in its review? Are the
interpretations of the elasticities
in the literature review, and the
estimates of elasticities used in
the model stemming from the
literature review, defensible?
Yes, the report misses several relevant citations.
Citation 1: James A. Stock et al. Comment on Proposed
SAFE Rule, EPA-HQ-OAR-2018-0283-6220 (October 26,
2018).
The Stock et al. comment on the proposed SAFE Rule
does not report elasticities, but new vehicle market price
elasticities in the range of -0.03 to -0.09 can be calculated
from their results. The Stock et al. comment also includes
an important correction of a spreadsheet error in the
Notice of Proposed Rulemaking, resulting in a revised
elasticity estimate for that model of -0.07.
It is worth mentioning how this citation is different from
other papers such as Leard (2021). My understanding of
the paper is that it has a model that accounts for general
equilibrium effects. It might be useful to include a new
column in Table 5-1 that has a binary yes-no variable
"Includes general equilibrium effects" since the point of the
current document is to account for these effects.
Citation 2: The report. Sean P. McAlinden et al., The
Potential Effects of the 2017-2025 EPA/NHTSA GHG/Fuel
18
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Economy Mandates of the US Economy, Center for
Automotive Research, 27 (Sept. 2016),
httDs://www.cararouD.ora/wo-
content/uDloads/2017/02/The-Potential-Effects-of-the-
2017 2025-EPANHTSA-GHGFuel-Economv-Mandates-
on-the-US-Economv.odf
This is a report that estimates the long-run market-price
elasticity of demand for new vehicles to be -0.61.
The interpretation of one of the citations is incorrect:
- Fischer, Harrington, and Parry (2007): This study finds
an implied new vehicle market price elasticity of demand
equal to -1 by reporting separate elasticities for cars and
light trucks. This ignores substitution between the two
vehicle categories. This paper also reports a long-run
elasticity for the combined market for new motor vehicles,
finding it to be -0.36. This value is reported in the bottom
panel of Table 5-1. I am puzzled why the -1 value is
reported in the top panel when the combined effect is what
is relevant here.
3. Are the data and assumptions
appropriate for the analysis
conducted and objectively
chosen? If not, do you know of
other data or proposed
alternative assumptions that
might be used in this analysis?
Certain papers that report new vehicle market price
elasticities are, in my opinion, inappropriate to cite and use
to construct a range of elasticities. The following papers
cited in Table 5-1 are inappropriate:
Berry et al. (2004): This study assumes a new vehicle
market price elasticity of demand for new vehicles equal to
-0.4 (in addition to model where they assume it is equal to
-1). This elasticity is not estimated in this paper. Therefore,
it is inappropriate to cite.
Fischer, Harrington, and Parry (2007): This study finds an
implied new vehicle market price elasticity of demand
based on model simulations, not estimated from data.
Generally, these types of "calibration" results should not
be included with other studies that estimate the elasticity.
BLP (1995), Goldberg (1998), Bento etal. (2009), Knittel
and Metaxoglou (2014), Dou and Linn (2020): My issue
with citing these papers is that none of them focus on
properly estimating the market price elasticity of demand
for new vehicles. None of them exploit proper statistical
variation necessary to identify this parameter. Take, for
19
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
example, BLP (1995). This study uses a panel of
aggregate sales and price data for vehicle models sold
during 1971-1990. The study is almost exclusively focused
on obtaining unbiased estimates of the model-specific
own-price elasticity of demand (e.g., Toyota Camry). One
small part of the paper discusses the implied market-price
elasticity of demand. As discussed in Leard (2021), this
elasticity is a function of the own-price elasticity and the
propensity of new vehicle buyers to leave the new vehicle
market. It is not clear how BLP (1995) identify this
propensity (they have no discussion of this).
BLP (1995) should instead be cited when discussing the
model level own-price elasticity of demand, since this is
the focus of the paper.
The other papers - Goldberg (1998), Bento et al. (2009),
Knittel and Metaxoglou (2014), Dou and Linn (2020) - also
have alternative focuses and do not discuss much at all
how the market price elasticity of demand is identified from
the data. Therefore, these papers should not be cited to
construct a plausible range of values for the market price
elasticity of demand. A paper like Bento et al. (2009), for
example, which focuses on the link between vehicle
choice and VMT choice, should be cited when
constructing a plausible range for the VMT elasticity.
4. Are the methods and
procedures employed
technically appropriate and
reasonable? Please distinguish
between cases involving
reasonable disagreement in
methods as opposed to cases
where you conclude that current
methods involve specific
technical errors.
I have a few concerns with the methods.
Concern 1: The modeling of vehicle ownership costs (r)
and prices (p) is unclear to me.
Why is it necessary to model vehicle ownership costs in
the first place? Why not just model vehicle prices?
The discussion in 6.4.1 seems to be motivated from this
distinction. But it is not clear why the model cannot be
simplified to only have vehicle prices or vehicle prices plus
fuel costs.
Ownership costs appear in the vehicle demand function
equation (6 - 1). The authors motivate using ownership
costs with the following:
"Modeling ownership cost as the variable of interest to the
consumer follows the approach in Bento et al. (2009) and
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
is important when allowing choices to be made between
new and used vehicles."
Bento et al. (2009) define ownership costs as the sum of
fuel costs and a rental rate, where the rental rate is
defined as the sum of
Foregone return of the real value of the car, which
is proportional to the used vehicle price.
Depreciation, which is typically proportional to the
used vehicle price (e.g., a typical used vehicle
depreciation rate is often cited to be 15 to 20% of
the value of a used vehicle).
Insurance and registration, which are generally tiny
relative to the first two components.
Given these features, it seems straightforward to model
used car demand as a function of vehicle prices and fuel
costs only. This would simplify the model set up and it
would be consistent with how supply is defined, which is a
function of used vehicle prices.
Concern 2: It is unclear how a vehicle is defined in the
model. Is a vehicle defined by age only? So the model has
31 vehicles: new, 1 year old, 2 years old, ..., 30 years old.
This should be more explicitly stated in the first paragraph
of Section 6.2.
Concern 3: Section 6.3.1 New Vehicles states that the
price of a new vehicle is taken as given. Is it equivalent to
say that the new vehicle market is perfectly competitive?
This is in contrast to most of the vehicle market literature
(Bento et al. 2009 for example), which models the new
vehicle market is imperfectly competitive. It would be
useful to include more details on this assumption, why it
was adopted, etc.
Concern 4: In Section 6.6, the authors mention that they
explore a range of the demand elasticities theta_aa and
theta_Na. It would be useful to provide a table of
elasticities assumed. I realize this table is presented in
Table 7-2, but it would be constructive to include a table in
Section 6.6 with discussion defending the choice of
elasticities.
21
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Concern 5: The authors assume that the own-price
elasticities for used vehicles is identical to the new-vehicle
elasticity. This is problematic because of how households
sort into different vehicle ages. New vehicles are more
frequently bought by wealthy households, and a lot of
literature has shown that wealthy households tend to be
less responsive to price than average or low income
households. For this reason, 1 would expect the own price
elasticities to be larger in magnitude for older vehicles.
Concern 6: At the end of Section 7.2, the authors state
that "Cross-price demand elasticities between individual
ages are set to fall at 8% per year of difference in age."
This seems reasonable, but it is not clear where 8%
comes from. Is this based on results from a demand
system such as Bento et al. (2009)?
5. Does the modeling analysis
appear to produce results
consistent with the assumptions
and data used for model
development? Do the results
presented by RTI follow from the
data and assumptions used in
the analysis?
Yes. It is clear that the policy elasticities are smaller in
magnitude than the new vehicle demand elasticities, which
makes sense given the equilibrium response.
6. In what ways are the results
sensitive to the data and
assumptions used in model
development? Are there
alternative assumptions and
data that the researchers should
consider providing improved
analysis?
Given my comments above, 1 suggest applying a different
range for the market price elasticity of demand for new
vehicles. 1 would also consider adding greater detail to the
definition of a vehicle. It seems to me that part of the value
of this exercise is to simulate the effects of various policies
(for example, CAFE standards). The current structure of
the model has a highly aggregated vehicle definition that
seems lacking for performing simulations of an actual
policy. At a minimum, it would be useful to provide a
discussion on how the model could practically be
expanded, and the potential calibration and computational
challenges involved with such a task.
ADDITIONAL OVERALL COMMENTS PROVIDED (NOT CHARGE QUESTION-SPECIFIC):
ADDITIONAL COMMENTS BY SPECIFIC REPORT CHAPTER:
t-y
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
2.3 Comments by Dr. James Sallee
CHARGE QUESTION
COMMENTS
1. Does the presentation
describe the data and methods
sufficiently to allow the reader to
form a general view of the quality
and validity of the analysis
approach?
The report is clear. Data and assumptions are
documented well. Assumptions and modeling choices are
explained thoroughly. Moreover, the results of the
analysis are well explained, and every effort is taken to
provide basic economic intuition that gives context to the
quantitative model output.
As a result, readers should be fully able to evaluate the
merits of the model and understand both its contributions
and its limitations from the information contained in the
report.
The report lays out the main analysis in a sequence of
steps that ensure clarity by adding complexities and
features one at a time (e.g., section 7.1). The report also
has many sections that seek to consolidate results and
illustrate the underlying intuition of the results (e.g., the
summary in section 7.6).
Altogether, the report explains a complex economic
model in a concise and effective way. This is not easy to
do, and the authors of the report should be commended
for the quality of the exposition.
2. Does the report miss relevant
literature in its review? Are the
interpretations of the elasticities
in the literature review, and the
estimates of elasticities used in
the model stemming from the
literature review, defensible?
The report does an admirable job of surveying the
literature and putting the information available from past
research into context.
I am not aware of other studies that should have been
included in the literature review. As the report notes at
several points, there are many studies of the automobile
market, but attention to the elasticities pivotal for the
current analysis are rarely the focus and are often not
even calculated. Thus, the report's finding that few
existing papers can be used to calculate the key
elasticities conforms with my understanding of the
literature.
All interpretations of the elasticities offered in the report
appear appropriate. The use of elasticities from the
literature to inform the model is certainly defensible.
23
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
The report explains clearly why the existing literature
offers limited guidance on key values. It therefore
emphasizes strongly the need for sensitivity analysis,
which is then provided.
This is the appropriate stance to take. If anything, I would
recommend even more caution in relying on the existing
literature (and hence elevate even further the critical role
of sensitivity analysis) for two reasons, both of which are
noted in the report itself, but I note here for emphasis.
First, many of the studies are more than a decade old.
Mechanically, vehicles are becoming more durable and
lasting longer (this is noted in section 2). At the same
time, embedded technology is also changing rapidly,
which might make newer and older vehicles less close
substitutes. Both factors imply that empirical relationships
observed in the automobile more than a decade in the
past may be misleading indicators of the vehicle market of
the future.
Moreover, one might suspect that the elasticity of demand
for new vehicles shrinks as consumers get richer. If true,
older studies might exaggerate the new vehicle price
elasticity. (Some interaction between income and price
sensitivity is often estimated in prior studies. Perhaps
there is some information in those that would suggest
how large might be any trend in the elasticity driven by
economic growth. The report does not mention this.)
Second, a substantial share of the available studies do
not focus on the elasticity parameters that are of interest
here. In several cases the authors of the report had to
construct an estimate that was not even reported in the
original research. This is relevant because researchers
have discretion in how to construct and estimate these
models. When a parameter (like the substitution to the
outside good) is viewed as a byproduct of estimation, the
original researchers may not have exerted maximal effort
in ensuring that the estimation was reliable.
The report does not discuss standard errors in the original
studies. Instead it uses the point estimates from prior
studies to form a qualitative sense of the range of
24
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
estimates across papers. This is a defensible approach,
especially because the standard errors are likely
calculated with different methods across studies and are
thus not always comparable. Even so, where available, it
might be useful to have standards errors from the original
studies included in the review to more fully characterize
uncertainty.
3. Are the data and assumptions
appropriate for the analysis
conducted and objectively
chosen? If not, do you know of
other data or proposed
alternative assumptions that
might be used in this analysis?
The data and key model parameters used in the baseline
analysis and the various scenarios appear appropriate.
The data sources cited are all reliable and would be
considered standard, objective sources. 1 do not know of
any alternative data sources that 1 believe to be superior.
Likewise, the parameter assumptions are appropriate and
appear consistent with an objective analysis.
(In answering this question, 1 interpret "assumptions" to
refer specifically to parameters used in the model, rather
than assumptions about the structure of the model or
analysis, which 1 interpret as falling under the definition of
"methods and procedures" addressed in the next
question. Essentially, this answer is about section 6.6 of
the report.)
The baseline scrappage scale factors (ba in the notation
of the model) are said to simply be "calculated using
baseline vehicle survival probabilities." The relevant
paragraph (p. 6-10) adequately explains the source of
data, but I was uncertain of whether there was a degree
of modeler discretion here, or if each ba is simply an
algebraic calculation based on equation 6-5, where both
sa and pa are raw data inputs. This is a minor detail that
could be easily clarified.
4. Are the methods and
procedures employed technically
appropriate and reasonable?
Please distinguish between
cases involving reasonable
disagreement in methods as
opposed to cases where you
conclude that current methods
involve specific technical errors.
The methods and procedures developed and deployed in
the report are technically appropriate and reasonable.
The model is complex: it deals with thirty vehicle ages
while incorporating forward looking expectations. To
maintain tractability, it must make simplifications—there is
a single composite vehicle model; consumer
heterogeneity is not explicated; forward looking beliefs
are fully rational; new vehicle supply is perfectly elastic;
the model abstracts from the choice of miles traveled;
and, there are no income effects.
25
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Those assumptions are justified and reasonable. They
serve the interest of transparency and clarity, and they
are all appealing as a benchmark case.
The report is transparent and clear about assumptions
and offers caveats where appropriate. For instance, the
analysis is built on a model of vehicle purchase, rather
than vehicle use. As noted in the report, it is difficult to
see exactly how incorporating changes in mileage in
response to vehicle prices might play out in the full
equilibrium. This decision could perhaps be added to the
model at some future date, but even then, the key
parameters (e.g., how vehicle price affects mileage,
presumably through a change in depreciation) are not
well estimated in the literature and thus would require
additional conjecture. As such, a model focused on
ownership seems appropriate.
The report offers a generalized interpretation of a cost
shock to new vehicles. It suggests that a regulatory
change that creates a combination of changes in price
and attributes can be interpreted as a tax equal to the "net
utility cost." This is sensible and appealing as a modeling
approach. (It is also very important because new vehicle
regulations create just this sort of collection of changes
simultaneously.)
That said, a minor caveat would seem to be that a change
in other attributes (as opposed to just a change in price)
might change the substitutability between new and used
vehicles; that is, a regulation that changes the attributes
of new vehicles might change the theta terms in the
demand system. This is likely to have a minimal impact in
the current setup. But, it could become a more important
issue in the future if the model is used to contemplate a
transition to electric vehicles or vehicles featuring different
levels of automation, in which case attributes affected by
the policy could radically change the substitution between
new and used vehicles.
The report assumes full pass through of cost shocks to
consumers: because new vehicle supply is assumed to
be perfectly elastic, all regulatory burdens are borne by
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
consumers. The report argues that this is a sensible
interpretation for the long run, and it is indeed the most
appealing benchmark. That said, 1 am not certain that we
would expect full pass through of costs in a market with
imperfect competition. It is perhaps useful then to
contemplate how the results would be different if part of a
regulatory cost increase is borne by new vehicle sellers,
and part is passed on to consumers. Is it possible to
simply interpret the cost shock in the model as the portion
passed on to consumers?
The only method and procedure in the modeling analysis
that 1 had trouble understanding was the assumption
about how the demand system was assumed to be
structured across used vehicles of different ages. The text
describes the assumptions deployed in an intuitive (and
appealing) way, but it does not write out algebraic
explanations. Furthermore, in some places, the report
offers as intuition that older vehicles are less close
substitutes for newer vehicles, which is meant to
rationalize several patterns. 1 struggled to follow this
reasoning because 1 could equally imagine a world in
which there was only substitution between adjacent ages
(e.g., the only cross-price elasticity for a 3-year old car is
between 2- and 4-year old cars), but nevertheless a price
shock to new vehicles cascades through the entire
distribution of ages.
5. Does the modeling analysis
appear to produce results
consistent with the assumptions
and data used for model
development? Do the results
presented by RTI follow from the
data and assumptions used in
the analysis?
The modeling analysis produces results that appear to
follow logically from the assumptions and data. As a peer
reviewer, 1 did not attempt to replicate any of the findings
or inspect the model code that executes the simulations
(nor was 1 given the material to do so). Instead, my
judgment that the results reported follow from the data,
model, and assumptions is based on the fact that the
pattern of results are consistent with the economic theory
that the analysis purports to embody. The qualitative
results that emerge in the baseline model, and the
difference across the nine scenarios described in section
7.4 (as well as the additional versions in the appendix),
conform with theoretical expectations. These results, and
the economic reasoning that aligns with them, are
described clearly in the report. 1 found that discussion
sound and convincing.
6. In what ways are the results
sensitive to the data and
The report explores the sensitivity of key results by
generating simulations based on a range of parametric
27
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
assumptions used in model
development? Are there
alternative assumptions and data
that the researchers should
consider providing improved
analysis?
assumptions that span the range of elasticities from the
literature. While it is always possible to perform more
sensitivity analysis (and future users will apparently be
able to do so on their own, which is a great feature of the
project), the current report does a nice job overall of
showing the likely range of key results.
Quantitative values of course vary across scenarios, but
several key findings, including the policy elasticity being
significantly smaller than the new vehicle elasticity and
the overall fleet size shrinking in response to a new
vehicle price increase, appear to be robust to a range of
inputs and assumptions. This is reassuring and seems to
suggest that key qualitative insights that can inform policy
are very likely to hold.
Section 7.4 presents the core sensitivity analysis by
running a variety of scenarios. This section emphasizes
variation in three parameters: the new vehicle elasticity,
overall substitution to the outside good, and the scrap
elasticity. But, it seems there is a fourth crucial elasticity,
which is the substitutability between vehicles of different
ages. The appendix provides a couple of alternative
model runs that seem to address this point. The
discussion around these scenarios is quite concise, and
as a result I am not fully certain I understand how to
interpret those results. The appendix results seem to
suggest that key model results are not sensitive to big
changes in these assumptions.
The report does not feature standard errors. Instead of
focusing on standard errors, the report focuses on
alternative scenarios. This is a sensible approach, but it
does open the possibility that some users may
underestimate (or overestimate) uncertainty. To help
future users, it might be useful to include a discussion
about uncertainty in one section, starting with a bullet
point of sources of uncertainty in the results, with some
guidance as to what the authors believe are the biggest
sources. These would include both uncertainty about key
parameters, which might be amenable to some statistical
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
characterization, and model uncertainties, which might be
harder to quantify.
At a few points, the report simply states that the model
results are robust to differences in data inputs. This
includes reference to vehicle market data used in a paper
by Leard and one by Jacobsen and van Benthem, rather
than the AEO report that is the main data source. I fully
believe this is true, but it might be useful to report those
robustness checks somewhere.
ADDITIONAL OVERALL COMMENTS PROVIDED (NOT CHARGE QUESTION-SPECIFIC):
The analysis described in this report is a significant step forward and represents a valuable
tool for improving regulatory analysis of the car market.
Attempts to incorporate interactions between new vehicle regulations and the used vehicle
market that are based on statistical relationships between new vehicle prices and the used
and new vehicle markets studied separately are inadequate grounds for policy analysis and
highly likely to be misleading. In essence, this model studies long-run general equilibrium
interactions, something that is notoriously difficult to study using observed market data. Thus,
an understanding of how interactions between new and used vehicle markets dampen,
amplify or propagate the effects of new vehicle regulations requires an integrated model that
provides internally consistent effects constrained by economic theory. The analysis in this
report does exactly that, which is why it is so valuable. Specifically, this new tool should foster
improved analysis of the ways in which used vehicle market interactions may alter the net
impact of new vehicle regulations on consumers and the environment.
The model described in the report identifies key mechanisms that govern the relationship
between new vehicle costs and the used vehicle market. There is substantial uncertainty
about key parameters, so care should be taken in interpreting the quantitative output of the
model (this is also true, of course, of other models that underly regulatory analysis of
automobiles), but the analysis here reveals that certain facts are robust (e.g., the policy
elasticity is smaller than the new vehicle demand elasticity). Moreover, it provides a tool that
can be used to assess the plausibility of different outcomes (e.g., the total fleet size can
increase in response to a new vehicle cost increase only under implausibility limited
substitution to the outside good).
A central contribution of the report is to highlight the difference between a new vehicle
demand elasticity and what the report refers to as the policy elasticity, the latter of which
takes into account the feedback of used vehicle prices into the demand for new vehicles
following a cost shock to new vehicles. The analysis demonstrates that the policy elasticity is
consistently smaller than the demand elasticity, and that the magnitude of the gap between
the two is of importance. This is a critical insight, and it conforms with economic theory. The
report explains this relationship effectively, and it provides a tool that users can customize to
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
perform sensitivity tests and additional analyses.
This tool might become even more useful in the future if subsequent studies provide new
information about the key parameters that serve as inputs to the model. Having run many
iterations of the model (presumably many more than are presented in the report), the authors
likely have insights about the most important parameter choices and modeling decisions. A
possible addition to the report would be a prioritized list that emphasizes where additional
estimates from future research, or additional sensitivity analysis from future users, would
provide the most valuable new information
ADDITIONAL COMMENTS BY SPECIFIC REPORT CHAPTER:
Included here are a series of comments, many of which are quite minor expositional notes,
ordered with respect to each section of the report.
Section 3
This section provides a valuable discussion of the key economic forces that link the new and
used vehicle markets and the role of scrappage. I suspect that it would be possible to
illustrate these relationships further by using a pair of simple supply and demand diagrams to
illustrate the steady-state (long-run equilibrium) effects.
Initially, a new vehicle cost increase shifts up the supply curve in the new vehicle market. The
magnitude of the impact depends on the elasticity of demand. The change in the new vehicle
market impacts used vehicle supply (shifted down because there are fewer new cars to
become used) and used vehicle demand (more so as buyers view used and new as close
substitutes), resulting in a new price and quantity. The size of this change depends on the
scrap elasticity, which determines the slope of used vehicle supply. The resulting change in
used vehicle price then shifts the new vehicle demand curve, resulting in a new quantity.
The role that each of the key elasticities places could be labeled in a heuristic diagram of
these two markets.
Analytical expressions that show the simple relationship between key elasticities would also
be helpful. This is not feasible in the full model with 30 vintages of used vehicles, but a future
modeling exercise that reduced the used vehicle market to a single composite might make it
possible to characterize the key comparative statics, showing how they depend on the
handful of key elasticities.
Section 5
The literature synthesis is effective and clear overall.
The report offers a rich discussion of some of the studies, but in a few cases the results of
prior studies are reported without a lot of commentary about their reliability or applicability to
the purpose at hand.
For example, my impression is that authors of papers in the tradition of Berry, Levinson and
Pakes (1995) are relatively skeptical of the reliability of their own estimates of the elasticity of
substitution to the outside good, relative to the other estimates in those systems. I gather that
the magnitude of this elasticity can scale with assumptions made by the researcher about
what fraction of consumers are shopping for a car in a given period, something which is
nebulous and often chosen without hard data. This is not to suggest that there are better
30
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
estimates available, but rather to emphasize our relative ignorance about the market level
elasticities that are central to the analysis here.
In some cases, the authors of the report had to do additional calculations to extract an
elasticity estimate from a prior paper. To the extent possible, it would be useful to have those
steps explicated somewhere, for future reference. (For example, I'm not sure that someone
else can replicate the author's interpretation of the elasticity in the Hahn study, or replicate
the new calculations from Bento et al. 2009.)
Section 6
It might be helpful to include units (for the prices) in Figure 6-2.
The demand system has (A+2)(A+1)/2 substitution elasticities, where A is 30 in the present
analysis. This is a large number. The calibration feeds in basically three elasticities. All of the
rest are filled in via assumption about the structure of cross-price elasticities through the age
structure. This assumption (described in sections 7-2 and 7-3, in particular footnote 21)
appears reasonable and is an appealing solution to the problem that no estimates of this age
substitution pattern exists. But, I was not confident that I could fully reconstruct the age
structure of substitution from the description in the text. A small, helpful step would be to add
algebraic expressions detailing this procedure and commenting further on how interested
users might conduct additional sensitivity analysis (noting that some is provided in the
appendix).
Section 7
Table 7-1 (and several related tables) do not label the main results as the "policy elasticity."
This term is emphasized in the text, but it is not called out in the tables. Adding it might be
helpful.
It might be helpful to also include some measure of price changes (for used cars this may be
an index across ages) in Table 7-2. The price changes are not the object of interest, but they
are a key channel, so including them might help the reader.
Figure 7-1 should include somewhere (title or note) that the new vehicle elasticity is -0.8
because the main point of the graph is that the policy elasticity is substantially smaller. This is
not apparent without referring back to a prior table.
Section 10
Scrappage plays a key role in the analysis, but scrappage can mean either that a vehicle is
exported from the United States or that it is fully decommissioned. The paper does not
mention this distinction. It is standard practice for regulatory analysis to focus on emissions in
the US fleet, and as such it is logical to ignore this distinction in the present analysis. The
distinction is, however, relevant to the true greenhouse gas emissions implications of new
standards. This might be a subject of future work.
The model built here is free of transaction costs. In reality, transaction costs associated with
buying and selling vehicles are substantial. This is unlikely to matter for analysis of long-run
effects, but it could have an impact on transition dynamics. Modeling transaction costs is
certainly beyond the scope of the current report, but it might be useful to speculate as to
whether transaction costs would amplify or dampen the main effects during a transition.
31
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Appendices
It might be helpful to add further labels to Figure D-1, which is showing two completely
different model runs with separate assumptions about expectations, rather than two sets of
results that come out of the same model run.
Some of the figures in appendix D label time just as "year" instead of the more helpful "years
since policy effective date."
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Appendix A: Resumes of Selected Reviewers
1. Ashley Langer
Ashley Lanqer
http://www.ashlevlanaer.com
alanger@email.arizona.edu
1130 E Helen St
McClelland Hall 401
Tucson, AZ 85721
Office: (520) 621-0117
Appointments:
• Assistant Professor, Department of Economics, University of Arizona, 2013-present
• Faculty Research Fellow, National Bureau of Economic Research, 2020-present
• Adjunct Research Scientist, Ford School of Public Policy, University of Michigan, 2013-
2016
• Visiting Scholar, Energy Policy Institute of Chicago, University of Chicago, Fall 2013
• Visiting Assistant Professor, Department of Economics, University of Arizona, 2012-
2013Assistant Professor, Ford School of Public Policy and Department of Economics,
University of Michigan, 2010- 2013
Degrees:
Ph.D. University of California, Berkeley, Economics, 2010
B.A. Northwestern University, Mathematical Methods in the Social Sciences and
Economics, 2002
Publications:
"Escalation of Scrutiny: The Gains from Dynamic Enforcement of Environmental Regulations"
(with Wesley Blundell and Gautam Gowrisankaran) American Economic Review, 110(8),
August 2020, 2558-85
"From Gallons to Miles: A Disaggregate Analysis of Automobile Travel and Externality Taxes"
(with Vikram Maheshri and Clifford Winston) Journal of Public Economics, 152, August
2017
"The Intergenerational Transmission of Automobile Brand Preferences" (with Soren Anderson,
Ryan Kellogg, and James Sallee) Journal of Industrial Economics, 63(4), December 2015
"Automakers' Short-Run Responses to Changing Gasoline Prices"
(with Nathan Miller), Review of Economics and Statistics 95(4), October 2013
"Toward A Comprehensive Assessment of Road Pricing Accounting for Land Use"
(with Clifford Winston) Brookings-Wharton Papers on Urban Affairs, 2008
33
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
"The Effect of Government Highway Spending on Road Users' Congestion Costs"
(with Clifford Winston) Journal of Urban Economics 60(3), 2006
Working Papers:
"Designing Dynamic Subsidies to Spur Adoption of New Technologies" (with Derek Lemoine)
Revisions requested at JAERE
"What Were the Odds? Estimating the Market's Probability of Uncertain Events" (with Derek
Lemoine)
"Fueling Alternatives: Evidence from Naturalistic Driving Data" (with Jackson Dorsey and
Shaun McRae)
"Energy Transitions in Regulated Markets" (with Gautam Gowrisankaran and Mar Reguant)
Teaching:
• Empirical Environmental Economics, University of Arizona (PhD)
• Environmental Economics, University of Arizona (Undergraduate)
• Honors Principles of Economics, University of Arizona (Undergraduate)
• Government Regulation of Industry and the Environment, University of Michigan (Masters
in Public Policy)
Energy and Environmental Policy, University of Michigan (Undergraduate)
• Microeconomics B: Cost-Benefit Analysis in Depth, University of Michigan (Masters in
Public Policy)
• Economics of Discrimination Reading Group
Fellowships and Awards:
2019 Sloan Foundation Grant for "Economics of Innovation in the Energy Sector"
(Wth David Popp and NBER,
$399,895)
Eller Dean's Research Award for Assistant Professors
2018 NBER Economics of Energy Use in Transportation Grant (Wth Shaun
McRae, $15,000)
Kalt Prize for best graduate student placement in the Eller Business School
(Jackson Dorsey, Kelley Business School at Indiana University)
2017 Center for Management Innovations in Health Care (Wth Derek Lemoine,
$3,900)
Eller Small Grant (Wth Derek Lemoine, $2,500)
2016 Institute of the Environment Program Development Grant (With Derek
Lemoine, $15,000)
2015-2016 Eller Small Grant (Wth Derek Lemoine, $2,500)
2014-2016 University of Arizona Renewable Energy Network Grant (Wth Derek
Lemoine, $15,000 each year)
2013-2014 University of Arizona Renewable Energy Network Grant ($15,000)
2011,2012 Faculty Teaching Honor Roll for Microeconomics B (twice), Energy Policy,
and Government Regulation of Industry and the Environment
2009-2010 Association of American University Women Dissertation Fellowship
2008 U.C. Berkeley Dean's Normative Time Fellowship
34
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
2007-2008 Outstanding Graduate Student Instructor Award
Research Presentations:
2021: Allied Social Science Association Conference, University of Georgia, University of
Albany (scheduled), ZEW Mannheim Energy Conference (keynote, scheduled)
2020: Allied Social Science Association Conference, University of Arizona Quant Law
Conference, Booth, Wharton, ETH/ZEW/CMCC/Grenoble Ecole de Management
2019: NBER Future of Energy Use in Transportation, Yale University Economics/SOM,
University of Chicago (EPIC), NBER EEE Summer Institute, University of California
Berkeley, University of Virginia Law School, National University Singapore, Jinan
University, Triangle Resource and Environmental Economics Seminar, University
of Calgary, University of California, San Diego
2018: Arizona State University, NBER Future of Energy Use in Transportation, University
of Nevada, Reno, Cowles Foundation Conference on Structural Microeconomics,
NBER Industrial Organization Summer Institute, NBER Environmental and Energy
Economics Summer Institute, University of Maryland Environmental Tax
Workshop, Georgetown University, Santa Barbara Occasional Workshop in
Environmental Economics, Yale University School of Forestry and Environmental
Studies
2017: Stanford Institute for Theoretical Economics Conference, NBER Environmental and
Energy Economics Summer Institute, MIT CEEPR Workshop
2016: POWER Conference, HOC Conference, AERE Annual Conference, Heartland
Conference, University of Michigan Conference on Transportation, Energy,
Economics and the Environment
2014: University of Colorado Boulder Environmental and Resource Economics
Workshop, University of California, Davis, University of Michigan Conference on
Transportation, Energy, Economics and the Environment
2013: Northeast Workshop on Energy Policy and Environmental Economics, American
Environmental and Resource Economists (AERE) Annual Conference, NBER
Summer Institute in Industrial Organization, Quantitative Marketing and Economics
Conference, University of Chicago (Harris)
2012: American Economic Association Annual Conference, University of Arizona,
University of British Columbia,
Copenhagen Business School
2011: International Industrial Organization Conference, Midwest Bioenergy Conference,
University of Chicago (Harris School), UC Berkeley (Energy Institute), Carnegie
Mellon University (Heinz School)
2010: Arizona State University, Columbia University, Michigan State University, Ohio
State University, Stanford University, Tufts University, University of British
Columbia, University of Maryland (Agricultural and Resource Economics),
University of Michigan (Economics, Erb Institute, and Public Policy),
University of Toronto, University of Wsconsin (Economics and Agricultural and
Applied Econ.)
2009: Collegio Carlo Alberto (MOOD Conference), San Francisco Federal Reserve,
University of California, Berkeley (Economics, Business, and Public Policy)
35
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Service:
2021- Editorial Council, Journal of Environmental Economics and Management
2021 Presentation to Congressional Budget Office ("Recent Research on the
Economics of Discrimination")
2015-2020 Program Committee for the Summer Conference of the Association of
Environmental and Resource Economists
Graduate Advising (all UA):
EK Green (chair), Paul Fisher (chair), Wei Zhou, Chase Eck (chair), Wendan Zhang
(chair), Timothy Roberson (chair), Yujia Peng, Phuong Ho (co-chair), Arundhati Tillu,
Jackson Dorsey (chair), Sangllk Nam, Keith Meyers, Wesley Blundell (co-chair), Anatolii
Kokoza, Kyle Wlson, Ahmad Mohassel (chair), Cong Liu, Charles He, Alex Hollingsworth,
Michael Matheis, Soudeh Mirghasemi, Hoa Nguyen, Thiagarajah Subramaniam, Leila
Asgari
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
2. Benjamin Leard
Contact Information:
Department of Agricultural and Resource Economics
Howard H. Baker Jr. Center for Public Policy
University of Tennessee - Knoxville
Phone: (865) 974-5005
Email: bleard@utk.edu
Website: www.beniaminleard.com
Office: Morgan Hall 31 OA
Academic Positions and Professional Affiliations
Assistant Professor, Department of Agricultural and Resource Economics, University of
Tennessee, 2020-present
Faculty Fellow, Howard Baker Jr. Center for Public Policy, 2020-present
University Fellow, Resources for the Future, 2020-present
Fellow, Resources for the Future, 2014-2020
Education
Cornell University, Applied Economics and Management, Ph.D. 2014
James Madison University, Economics and Mathematics, B.S. 2008
Refereed Publications
How Much Do Consumers Value Fuel Economy and Performance? Evidence from Technology
Adoption (with Joshua Linn and Christy Zhou), forthcoming, Review of Economics and
Statistics.
Pushing New Technology into the Market: California's Zero Emissions Vehicle Mandate (with
Virginia McConnell). Review of Environmental Economics and Policy, 2021, 15(1): 169-179.
What Does an Electric Vehicle Replace? (with Jianwei Xing and Shanjun Li). Journal of
Environmental Economics and Management, 2021, 107: 1-33.
The Effect of Fuel Price Changes on Fleet Demand for New Vehicle Fuel Economy (with
Virginia McConnell and Christy Zhou). Journal of Industrial Economics, 2019, 67(1): 127-159.
37
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Explaining the Evolution of Vehicle Miles Traveled in the United States (with Joshua Linn and
Clayton Munnings). Energy Journal, 2019, 40(1): 25-54.
Voluntary Exposure Benefits and the Costs of Climate Change (with Kevin Roth). Journal of the
Association of Environmental and Resource Economists, 2019, 6(1): 151-185.
Flawed Analyses of U.S. Auto Fuel Economy Standards (with Antonio M. Bento, Kenneth
Gillingham, Mark R. Jacobsen, Christopher R. Knittel, Joshua Linn, Virginia McConnell, David
Rapson, James M. Sallee, Arthur A. van Benthem, and Kate S. Whitefoot). Science, 2018,
362(6419): 1119-1121.
Consumer Inattention and the Demand for Vehicle Fuel Cost Savings. Journal of Choice
Modeling, 2018, 29: 1-16.
Fuelling Behaviour Change. Nature Energy News & Views, 2018, 3: 541-542.
How Do Low Gas Prices Affect Costs and Benefits of US New Vehicle Fuel Economy
Standards? (with Joshua Linn and Virginia McConnell). Economics of Energy & Environmental
Policy, 2018, 7(2).
New Markets for Credit Trading Under U.S. Automobile Greenhouse Gas and Fuel Economy
Standards (with Virginia McConnell). Review of Environmental Economics and Policy, 2017,
11(2): 207-226.
Fuel Prices, New Vehicle Fuel Economy, and Implications for Attribute-Based Standards (with
Joshua Linn and Virginia McConnell). Journal of the Association of Environmental and
Resource Economists, 2017, 4(3): 659-700.
Are Consumers Willing to Pay to Let Cars Drive for Them? Analyzing Response to Autonomous
Vehicles (with Ricardo Daziano and Mauricio Sarrias). Transportation Research Part C:
Emerging Technologies, 2017, 78: 150-164.
On the Importance of Baseline Setting in Carbon Offsets Markets (with Antonio Bento and Ravi
Kanbur). Climatic Change, 2016, 137(3): 625-637.
Designing Efficient Markets for Carbon Offsets with Distributional Constraints (with Antonio
Bento and Ravi Kanbur). Journal of Environmental Economics and Management, 2015, 70(2):
51-71.
The Welfare Effects of Allowance Banking in Emissions Trading Programs. Environmental and
Resource Economics, 2013, 55(2): 175-197.
Equivalencies in the Fishery (with Jon Conrad). Natural Resource Modeling, 2013, 26(2): 154-
163.
38
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Working Papers
Firms and Collective Reputation: a Study of the Volkswagen Emissions Scandal (with Rudiger
Bachmann, Gabriel Ehrlich, Ying Fan, and Dimitrije Ruzic)
Estimating Consumer Substitution between New and Used Passenger Vehicles, accepted,
Journal of the Association of Environmental and Resource Economists
Interpreting Tradable Credit Prices in Overlapping Vehicle Regulations (with Virginia
McConnell). 2020 RFF Working Paper 20-07, revisions requested.
Pass-Through and Welfare Effects of Regulations that Affect Product Attributes (with Joshnua
Linn and Katalin Springel). 2019 RFF Working Paper 19-07.
Have US Fuel Economy and Greenhouse Gas Emissions Standards Improved Social Welfare?
(with Joshua Linn and Katalin Springel). 2020 RFF Working Paper 20-06.
Other Publications
Federal Climate Policy 104: The Transportation Sector, 2021 RFF Explainer
The Potential Role and Impact of Electric Vehicles in US Decarbonization Strategies (with
Virginia McConnell). 2020 RFF Report 20-16.
Carbon Pricing 202: Pricing Carbon in the Transportation Sector (with Joshua Linn and
Kathryne Cleary). 2020 RFF Explainer.
What Does Ridesharing Replace? (with Jianwei Xing). 2020 RFF Working Paper 20-03.
Targeting Subsidies to Get More Electric Vehicles on the Road. RFF Resources Magazine
Issue 202, Fall 2019.
What Does an Electric Vehicle Replace? RFF Resources Magazine Issue 201, Summer 2019.
Comments to NHTSA and US EPA on the Safer Affordable Fuel-Efficient (SAFE) Vehicles Rule
for Model Years 2021-2026: Passenger Cars and Light Trucks (with Joshua Linn, Alan J.
Krupnick, and Virginia McConnell). Public Comments, October 2018.
Fleet Vehicles and Fuel Economy: How Do Fuel Prices Affect Vehicle Purchase Decisions for
Big Buyers? (with Virginia McConnell and Christy Zhou). RFF Resources Magazine Issue 197,
Spring 2018
The Benefits of Flexible Policy Design: US Energy Conservation Standards for Appliances (with
Josh Blonz and Karen Palmer). RFF Resources Magazine Issue 197, Spring 2018.
Comments on the US Department of Energy's Office of Energy Efficiency and Renewable
Energy's Request for Information on Energy Conservation Standards Program Design (with
Josh Blonz and Karen Palmer). RFF Report, February 2018.
39
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
The Effect of Standards for New Vehicle Fuel Economy and GHG Emissions on US Consumers
(with Joshua Linn and Christy Zhou) RFF Resources Magazine Issue 195, Fall 2017.
The Net Emissions Impact of HFC-23 Offset Projects from the Clean Development Mechanism
(with Clayton Munnings and Antonio Bento). 2016 RFF Discussion Paper 16-01.
Fuel Prices, Economic Activity, and the Rebound Effect for Heavy-Duty Trucks (with Joshua
Linn, Virginia McConnell and William Raich). 2016 RFF Discussion Paper 15-43 REV.
How Climate Change Affects Traffic Accidents (with Kevin Roth). RFF Resources Magazine
Issue 191, Winter 2016.
Comments on Midterm Evaluation Draft Technical Assessment Report for Model Year 2022-
2025 Light Duty Vehicle GHG Emissions and CAFE Standards (with Joshua Linn, Virginia
McConnell and Kenneth Gillingham). RFF Report, September 2016.
Do Low Oil Prices Undermine US Passenger Vehicle Fuel Economy Standards? (with Joshua
Linn and Virginia McConnell). RFF Policy Brief, July 2016, No. 16-08.
Comment on Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and
Heavy-Duty Engines and Vehicles; Phase 2. NHTSA-2014-0132-0109. The Potential for
Improvement in On-road Truck Fuel Economy: Evidence from the VIUS (with Jen He and
Virginia McConnell) https://www.regulations.gov/document?D=NHTSA-2014-0132-0109.
Research Funding
Oak Ridge National Laboratory, 2021-2022
Sloan Foundation, 2020-2021
Energy Foundation, 2020
Merck Family Fund, 2020
Georgetown Climate Center, 2019
United States Climate Alliance, 2019
Sloan Foundation, 2017-2019
Resources for the Future New Frontiers Competition, 2015-2016
Student Advising
Dissertation committee member for Kevin Ankney, Georgetown University
Dissertation defense external examiner for Cheng Xu, George Washington University, June
2019
40
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Teaching
Policy Analysis for Environmental and Natural Resource Management, University of Tennessee
Economic Perspectives on Natural Resource and Environmental Issues, University of
Tennessee
Environmental Economics, Ithaca College
Math Camp for incoming Masters and Ph.D. students, The Charles H. Dyson School of Applied
Economics and Management
Presentations
2020: ASSA Annual Meeting, San Diego, CA; Howard H. Baker Jr. Center for Public Policy,
University of Tennessee; Ford Motor Company; NBER Economics of Transportation in the 21st
Century, online; Exploring Innovative Transportation Policies, Resources for the Future, online
2019: Northeast Workshop on Energy Policy and Environmental Economics, Harvard University;
TE3 Conference on Transportation, Economics, Energy, and the Environment, University of
Michigan
2010-2018: Howard H. Baker Jr. Center for Public Policy Public Forum on Electric Mass Transit
as an Option for Urban Mobility, University of Tennessee; 6th World Congress of Environmental
and Resource Economists, Gothenburg, Sweden; 2018 TE3 Conference on Transportation,
Economics, Energy, and the Environment, University of Michigan, 2017 TE3 Conference on
Transportation, Economics, Energy, and the Environment, University of Michigan; AERE 6th
Annual Summer Conference, Pittsburgh, PA; Department of Transportation Workshop,
Washington, DC; AERE 5th Annual Summer Conference in Breckenridge, Colorado;
Transportation Research Board Annual Meeting, Washington, DC; The Economics of Low-
Carbon Markets, University of de Sao Paulo; MIT CEEPR Workshop, Cambridge, MA; Iowa
State University, Ames, IA; Wake Forest University, Winston-Salem, NC; California State - Long
Beach, Long Beach, CA; Resources for the Future, Washington, DC; AERE 2nd Annual
Summer Conference, Banff, Canada; 2nd Northeast Workshop on Energy Policy and
Environmental Economics, Cornell University, Ithaca, NY; EAERE Summer Conference,
Prague, Czech Republic; AERE 1st Annual Summer Conference, Asheville, NC; EAERE-FEEM-
VIU European Summer School, Venice, Italy; Eastern Economics Association Meetings, New
York, NY; United Nations Conference of Parties Climate Change (COP16) Conference,
Agricultural and Rural Development Day, Ideas Marketplace Presentation, Cancun, Mexico
Referee Experience
American Journal of Agricultural Economics, Climate Policy, Economic Inquiry, Energy
Economics, Energy Journal, Energy Policy, Environmental and Resource Economics,
International Economic Review, Journal of Choice Modeling, Journal of Environmental
41
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Economics and Management, Journal of the Association of Environmental and Resource
Economists, Journal of Natural Resources Policy Research, Journal of Public Economics,
Nature Climate Change, Nature Energy, Regional Science and Urban Economics, Research in
Transportation Economics, Resource and Energy Economics, Transportation Research Part A:
Policy and Practice, Transportation Research Part D: Transport and Environment
42
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
3. James Sallee
James M. Sallee
Department of Agricultural and Resource Economics
University of California, Berkeley Cell: 773-316-3480
207 Giannini Hall
sallee@berkeley.edu
Berkeley, CA 94720-3310
https://are.berkeley.edu/~sallee
CURRENT EMPLOYMENT AND AFFILIATIONS
Associate Professor, Department of Agricultural and Resource Economics
University of California, Berkeley
Previously Assistant Professor
2019—present
2015—2019
Research Associate, National Bureau of Economic Research
Public Economics and Energy and Environmental Economics Programs
Previously Faculty Research Fellow
2019—present
2010—2019
Research Associate, Energy Institute at Haas
Faculty Affiliate, E2e Program
Faculty Affiliate, Institute for Research on Labor and Employment
2016—present
2014—present
2016—present
PAST EMPLOYMENT
Assistant Professor, Harris School of Public Policy Studies 2008-2015
University of Chicago
Visiting Researcher, University of California Energy Institute 2010
EDUCATION
University of Michigan, Ph.D. in Economics (2008)
University of Michigan, M.A. in Economics (2005)
Macalester College, B.A. in Economics and Political Science, Summa Cum Laude, OBK (2001)
SELECTED PROFESSIONAL SERVICE
Member, National Academy of Sciences Committee to Review of Methods for Setting Building
and Equipment Performance Standard (2019-2020)
43
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
PUBLISHED ARTICLES (peer reviewed unless otherwise noted)
"Who Benefits When Firms Game Corrective Policies?" (with Mathias Reynaert) American
Economic Journal: Economic Policy. Forthcoming.
"The Use of Regression Statistics to Analyze Imperfect Pricing Policies" (with Mark R.
Jacobsen, Christopher R. Knittel and Arthur van Benthem) Journal of Political (May 2020),
pp. 1826-1876.
"Should Electric Vehicle Drivers Pay a Mileage Tax?" (with Lucas Davis) Energy Policy and
the Economy v. 1, Editors Matthew Kotchen, James Stock and Catherine Wolfram, NBER:
University of Chicago Press (2020), pp. 65-94.
"Flawed Analyses of U.S. Auto Fuel Economy Standards" (with Antonio Bento, Kenneth
Gillingham, Mark Jacobsen, Christopher Knittel, Benjamin Leard, Joshua Linn, Virginia
McConnell, David Rapson, Arthur van Benthem, and Kate Whitefoot) Science vol. 362,
issue 6419 (December 2018), pp. 1119-1121
"The Economics of Attribute-Based Regulation: Theory and Evidence from Fuel-Economy
Standards" (with Koichiro Ito) Review of Economics and Statistics 100, May 2018, pp.
319-336.
"Tax Incidence with Endogenous Quality and Costly Bargaining: Theory and Evidence from
Hybrid Vehicle Sales" (with Sumeet Gulati and Carol McAusland) Journal of Public
Economics 155, November 2017, pp. 93-107.
"Disparities in Complex Price Negotiations: The Role of Consumer Age and Gender" (with
Ambarish Chandra and Sumeet Gulati). Journal of Industrial Economics 64(2), June 2017,
pp. 235-74.
"Do Consumers Recognize the Value of Fuel Economy? Evidence from Used Car Prices and
Gasoline Price Fluctuations" (with Sarah West and Wei Fan) Journal of Public Economics
134, March 2016, pp. 61-73.
"Designing Policies to Make Cars Greener: A Review of the Literature" (with Soren T.
Anderson) Annual Review of Resource Economics 8, 2016, 157-80.
"The Intergenerational Transmission of Automobile Brand Preferences: Empirical Evidence
and Implications for Firm Strategy" (with Soren T. Anderson, Ryan Kellogg and Ashley
Langer) Journal of Industrial Economics, 63(4), December 2015, pp. 763-793.
"New Evidence on Taxes and the Timing of Birth" (with Sara LaLumia and Nicholas Turner)
American Economic Journal: Economic Policy, 7(2), May 2015, pp. 258-293.
"Rational Inattention and Energy Efficiency" Journal of Law and Economics, 57(3), August
2014, pp. 781-820.
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
"What Do Consumers Believe About Future Gasoline Prices? (with Soren T. Anderson and
Ryan Kellogg) Journal of Environmental Economics and Management, 66(3), November
2013, pp. 383-403.
"The Value of Honesty: Empirical Estimates from the Case of the Missing Children" (with
Sara LaLumia) International Tax and Public Finance, 20(2), April 2013, pp. 192-224.
"Car Notches: Strategic Automaker Responses to Fuel Economy Policy" (with Joel Slemrod)
Journal of Public Economics, 96(11-12), December 2012, pp. 981-999.
*Awarded the 2015 Atkinson Award (Best Paper in the Journal of Public Economics
2012-4)
"Financial Reporting, Tax, and Real Decisions: Toward a Unifying Framework" (with Douglas
A. Shackelford and Joel Slemrod), International Tax and Public Finance, 18(4), August
2011, pp. 461-494.
"Using Loopholes to Reveal the Marginal Cost of Regulation: The Case of Fuel-Economy
Standards" (with Soren T. Anderson) American Economic Review 101(4), June 2011, pp.
1375-1409.
"The Surprising Incidence of Tax Credits for the Toyota Prius" American Economic Journal:
Economic Policy, 3(2), May 2011, pp. 189-219.
"Forecasting Gasoline Prices Using Consumer Surveys" (with Soren T. Anderson, Ryan
Kellogg and Richard M. Curtin) American Economic Review Papers & Proceedings
101(3), May 2011, pp. 110-114. (Not Peer Reviewed)
"Fuel Economy Standards: Impacts, Efficiency, and Alternatives" (with Soren Anderson,
Carolyn Fischer and Ian Parry), Review of Environmental Economics and Policy, 5(1),
Winter 2011, pp. 89-108.
"The Taxation of Fuel Economy" Tax Policy and the Economy v. 25, Editor Jeffrey R. Brown,
NBER: University of Chicago Press, 2011, pp. 1-38. (Not Peer Reviewed)
"A Cautionary Tale About the Use of Administrative Data: Evidence from Age of Marriage
Laws" (with Rebecca M. Blank and Kerwin Kofi Charles), American Economic Journal:
Applied Microeconomics,*\ (2), April 2009, pp. 128 - 149.
"On the Optimal Allocation of Students and Resources in a System of Higher Education"
(with Alexandra M. Resch and Paul N. Courant) The B.E. Journal of Economic Analysis &
Policy (Advances Tier), 8(1), Article 11.
WORKING PAPERS
"Pigou Creates Losers: On the Implausibility of Pareto Improvements from Pigouvian
Taxation" (May 2019) NBER Working Paper No. 25831
45
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
AWARDS AND HONORS
Excellence in Advising Award: Outstanding Faculty Advisor (2018)
Club Six (2017, 2019 recognition for teaching scores above 6 out of possible 7, Haas MBA)
Hellman Family Faculty Fund Award (2017)
UC Regents' Junior Faculty Fellowship (2016)
Atkinson Award (2015, for best Paper in the Journal of Public Economics between 2012-
2014)
Best Teacher in a Core Course, The Harris School (2012, 2013)
John V. Krutilla Research Award from Resources for the Future (2009 - 2010)
National Tax Association Dissertation Award (2008)
National Science Foundation Graduate Research Fellowship (awarded 2003)
Population Studies Center Trainee Fellowship, University of Michigan (2003-2008)
GRANTS
Next 10 Foundation (Why are High Electricity Prices a Problem for Climate Change? 2019-
2021)
Giannini Foundation (Evaluating Optimal and Second-Best Nitrogen Regulations in California
Agriculture with Biogeochemical Simulations 2020-2021)
Institute for Transportation Research, UC Berkeley (Can Targeted Rebates Foster Equity in
Congestion Pricing Schemes? 2019-2020)
Giannini Foundation (An Optimal Tax Approach to Policy Problems in California Agriculture,
2019-2020)
Sloan Foundation (Heterogeneity, Equity and Energy Policy 2017-8)
France-Berkeley Fund Award (2016)
W.E. Upjohn Institute Early Career Research Grant (with Reed Walker) (2012)
TEACHING
University of California, Berkeley
Environmental and Resource Economics, ARE 261 (PhD)
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Empirical Energy and Environmental Economics, ARE 264 (PhD)
Environmental Economics, EEP 101/ECON 125 (Undergraduate)
Economic Analysis for Business Decisions (Core micro for MBAs), MBA201A
taught at Haas School of Business
University of Chicago (all for MPP students)
Policy Approaches to Mitigating Climate Change
Topics in U.S. Tax Policy
Empirical Methods in Policy Analysis II
Science, Technology and Policy
REFEREE
Editorial Board Member (2021-) Review of Environmental Economics and Performance
Editorial Council Member (2014-) Journal of the Association of Environment and Resource
Economists
American Economic Review, Journal of Political Economy, Quarterly Journal of Economics,
Econometrica, American Economic Review: Insights, Journal of Public Economics, American
Economic Journal: Economic Policy, American Economic Journal: Applied Economics,
Review of Economics and Statistics, RAND Journal of Economics, Journal of Environmental
Economics and Management, National Tax Journal, Journal of the Association of
Environment and Resource Economists, Journal of Labor Economics, International
Economic Review, European Economic Review, International Tax and Public Finance,
Journal of Law & Economics, Economic Journal, Energy Journal, Canadian Journal of
Economics, Nature, B.E. Journal of Economic Analysis & Policy, Economic Inquiry, Journal
of Human Resources, Economic Letters, Energy Economics, Environmental and Resource
Economics, Journal of Urban Economics, Journal of Policy Analysis and Management,
Transportation Research Part A, Journal of Population Economics, Environmental Policy and
Governance, Scottish Journal of Political Economy Grants: National Science Foundation,
European Science Foundation, Sloan Foundation, Smith Richardson Foundation, Time-
Sharing Experiments for the Social Sciences
SELECTED PRESENTATIONS
Invited 2018: Berkeley (Transportation Research Institute) 2017: Toulouse School of
Economics, UC Santa Cruz (Economics) 2016: Berkeley (Economics), UC Davis (ARE),
47
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Texas A&M (Economics), FGV Rio de Janeiro (Economics), Resources for the Future,
Arizona State 2015: LSE (Economics), Berkeley (Goldman), UCLA (Luskin), Colorado
School of Mines (Economics), Universidad de Chile (Business School), Pontificia
Universidad Catolica de Chile (Economics) 2014: Michigan (Ross), Berkeley (ARE),
University of Pennsylvania (Wharton), Berkeley (POWER Conference), Yale (FES),
Illinois (Economics), National Tax Association Spring Symposium, Federal Trade
Commission, EPA, University of Leuven (Economics), Universidad de Chile (Business
School), Pontificia Universidad Catolica de Chile (Economics) 2013: Georgetown
(Economics), Illinois (Economics), Wisconsin (Economics) 2012: Maryland (Economics),
Northwestern (Law), Universidad de Chile (Business School), Oxford (Business School);
2011: Columbia (Economics), Maryland (AREC), Syracuse (Maxwell), Illinois (Finance),
Ohio State (Economics), Illinois (Sustainability Center), NYU (Law conference),
University of Illinois at Chicago (Sustainability workshop), Treasury, EPA, Resources for
the Future (Conference); 2010: MIT (Economics), Yale (FES), Berkeley (ARE), Berkeley
(UCEI), NBER Tax Policy and the Economy, University of Chile; 2009: Cornell
(Economics), Minnesota (Applied Economics), North Carolina State University
(Economics), Berkeley (POWER Conference), University of Illinois at Chicago
(Economics), Macalester College (Economics); 2008: Resources for the Future,
University of Chicago (Harris), University of Pennsylvania (Wharton), University of British
Columbia (Economics), University of Kentucky (Martin/Economics), University of Indiana
(SPEA), University of California, Irvine (Economics), Treasury, Ford Motor Company
Conference 2019: NBER Energy and Environmental Economics (Spring), Association of
Environmental and Resource Economics, NBER conference on Economics of
Autonomous and Electric Vehicles 2018: ASSA (AEA), Heartland Environmental and
Resource Economics, National Tax Association 2017: National Tax Association, ASSA
(AEA) 2016: NBER EEE Summer Institute, Stanford Institute for Theoretical Economics,
National Tax Association, Heartland Environmental and Resource Economics 2015:
NBER EEE 2014: ASSA (AEA and AERE), NBER EEE, NBER Public Economics,
Oxford Tax Systems Conference, Michigan Tax Invitational 2012: NBER Public
Economics, National Tax Association, Michigan Tax Invitational 2011: National Tax
Association, ASSA, Association of Environmental and Resource Economics,
International Institute of Public Finance, University of California Energy Institute; 2010:
NBER Public Economics, Iowa State Bioenergy Camp; 2009: ASSA, National Tax
Association, Heartland Environmental and Resource Economics; 2008: APPAM,
National Tax Association; 2007: NBER Summer Institute (EEE), National Tax
Association, APPAM
48
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
Appendix B. Confil cts of Interest Declarations
1. Ashley Langer
\l/
ORGANIZATIONAL CONFLICT OF INTEREST CERTIFICATE
Customer: U.S. Environmental Protection Agency
Contractor: ICF Incorporated, LLC, 9300 Lee Highway, Fairfax, VA 22031
Prime Contract: EP-C-16-020, Work Assignment 4-04
Subject Re port: The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and
Scrappage
Subcontract/Peer Reviewer: Ashley Langer
In accordance with EPAAR 1552.209-70 through 1552.209-73, Subcontractor/Consultant certifies to the
best of its knowledge and belief, that (check one):
X No actual or potential conflict of interest exists.
Subcontractor/Consultant certifies that its personnel, who perform work on this contract, have been
informed of their obligations to report personal and organizational conflict of interest to Contractor and
Subcontractor/Consultant recognizes its continuing obligation to identify and report any actual or
potential organizational conflicts of interest arising during performance under referenced contract.
'ICF
An actual or potential conflict of interest exists. See attached full disclosure.
Su bco ntractor/Consu Ita nt
5/24/2021
Date
49
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
2. Benjamin Leard
\l/
~ICF
ORGANIZATIONAL CONFLICT OF INTEREST CERTIFICATE
Customer; U.S. Environmental Protection Agency
Contractor: ICF Incorporated, LLC, 9300 Lee Highway, Fairfax, VA 22031
Prime Contract: EP-C-16-020, Work Assignment 4-04
Subject Report:The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and
Scrappage
Subcontract/Peer Reviewer: Benjamin Leard
In accordance with EPAAR 1552.209-70 through 1552.209-73, Subcontractor/Consultant certifies to the
best of its knowledge and belief, that (check one):
X No actual or potential conflict of interest exists.
An actual or potential conflict of interest exists. See attached full disclosure.
Subcontractor/Consultant certifies that its personnel, who perform work on this contract, have been
informed of their obligations to report personal and organizational conflict of interest to Contractor and
Subcontractor/Consultant recognizes its continuing obligation to identify and report any actual or
potential organizational conflicts of interest arising during performance under referenced contract.
_ESenjamin Leard
Subcontractor/Consultant
5-15-21
Date
50
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
3. James Sallee
\l/
'ICF
ORGANIZATIONAL CONFLICT OF INTEREST CERTIFICATE
Customer: U.S. Environmental Protection Agency
Contractor: ICF Incorporated., LLC, 9300 Lee Highway, Fairfax, VA 22031
Prime Contract: EP-C-16-020, Work Assignment 4-04
Subject Report:The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and
Scrappage
Subcontract/Peer Reviewer: James M. Sallee
In accordance with EPAAR 1552.209-70 through 1552.209-73, Subcontractor/Consultant certifies to the
best of its knowledge and belief, that (check one):
X No actual or potential conflict of interest exists.
Subcontractor/Consultant certifies that its personnel, who perform work on this contract, have been
informed of their obligations to report personal and organizational conflict of interestto Contractor and
Subcontractor/Consultant recognizes its continuing obligation to identify and report any actual or
potential organizational conflicts of interest arising during performance under referenced contract.
An actual or potential conflict of interest exists. See attached full disclosure.
/
Subcontractor/Consultant
May 25,2021
Date
51
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
UNIVERSITY OF CALIFORNIA. BERKELEY
- mvm. -
I i * v F 5 • - SA N'
> \ t-arbar * • . ,• . islw;
lAMES M SiUXEE
Assocute Pf&FE&c*:
7l4UtavFJtMrs'HM.i.
TEL.(773)316-5450 -
May 30, 2021
Pacta Massoli and Sarah Lettes
ICF
IE: Peer Review
To Whom It May Concern:
I have conducted a peer review of "Hie Effects of New Vehicle Price Changes oil Nov and Used
Vehicle Markets and Scrappage" iRTl 0215574 004.028)
I do not have any conflicts of interest to report.
However, I wish to disclose thai tie I have collaborated professionaly with fee lead author of be report,
Mark Jacobsen of the University of California, Sail Diego. I had no involvement in Ms work cm Has
report, nor have 1 discussed it with mm or the other authors of tie report. I haw no financial interest
connected to tiie report ;n :my\vay.
Sincerely,
James Sallee
Associate Professor
52
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
SUMMARY NOTES:
PEER REVIEW FOR THE EFFECTS OF NEW VEHICLE PRICE CHANGES ON NEW AND
USED VEHICLE MARKETS AND SCRAPPAGE
MID-REVIEW MEETING, MAY 6, 2021 - 3:00 PM
Attendees:
Elizabeth Miller, EPA Ashley Langer, University of Arizona
Dana Jackson, EPA Benjamin Leard, University of Tennessee
Gloria Helfand, EPA James Sallee, University of California
Paola Massoli, ICF
Sarah Lettes, ICF
Welcome, Introductions, and Roles
Elizabeth Miller opened the meeting and all participants gave brief introductions of themselves,
their backgrounds, and their role in this Work Assignment.
Paola Massoli gave an overview of the peer review process as it applied to the report "The
Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage."
She also confirmed with the reviewers that everyone received their technical packages. Each
package included a charge letter, conflict of interest form, and the Report for review.
Overview of the Project and Report
Gloria Helfand gave a brief overview of the RTI report, and EPA's expectations of the peer
review process in terms of timeline and outcome. The meeting then turned to open discussion to
answer the initial peer reviewers' questions.
Open Discussion and Questions
James Sallee pointed out that this is the first peer review process where there is a kickoff
meeting amongst all reviewers and asked if peer reviewers would be prevented from talking to
each other during the review period. Paola Massoli responded that the EPA Handbook doesn't
prevent peer reviewers to interact, as long as the reviews are independently conducted.
Appendix C. Peer Reviewer Mid-Review Meeting
Notes
'ICF
53
-------
Peer Review of "The Effects of New Vehicle Price Changes on New and Used Vehicle Markets and Scrappage" - Final Report
EPA agreed that it would be up to the peer reviewers' good judgement to maintain an
independent behavior during the process.
James Sallee also asked clarifications regarding what constitutes a Conflict of Interest (COI)
and specified that he might have to disclose an existing professional relationship with one of the
authors of the RTI Report, whom he was not aware of when he accepted to perform the peer
review. James Sallee specified that he is not working with the RTI team, and there are is no
monetary COI. Paola Massoli responded that it is fairly common for professionals working in
one field to know each other and work together at some stage, and that the matter can be
solved by a simple disclosure statement where it is acknowledged that the individuals have
professionally interacted in the past.
Ashley Langer asked about the format of the answers to the report. Paola Massoli responded
that the peer reviewers should use the tabular format provided in the charge letter so that it is
easy to compile and summarize the answers to each question. Gloria Helfand added that
reviewers are encouraged to provide additional comments including edits to the draft Report, as
EPA wants to ensure that the report is thoroughly reviewed.
Benjamin Leard had no additional questions.
Schedule and Next Steps
Paola Massoli gave an overview of the project's schedule. She reminded the panel of the May
24, 2021 deadline for the reviews to be returned to ICF (to Paola Massoli and Sarah Lettes). If
additional review time is required, the reviewer should reach out to ICF so that an extension can
be coordinated and accommodated.
ICF will then compile all comments and share with EPA. Reviewers should also re-submit a
current CV or resume, and a cover letter that includes their name, name and address of their
organization. The completed COI form provided in the review package is also required.
During the review period, the reviewers will send any questions to ICF. ICF will forward the
questions to EPA as necessary. ICF will then share all questions and responses with the entire
review team.
54
------- |