Information about http://www.oilendgame.com/pdfs/TechAnnx/TechAnnx21.pdf
Winning the Oil End Game -- Technical Annex …
Tags: consumer surplus, conventional wisdom, efficiency improvement, energy outlook, flow model, government vehicle, inefficient vehicles, linear rate, market penetration, model overview, oilendgame, policy model, program increases, response sales, rocky mountian, state of the art technology, technical annex, vehicle policies, vehicle purchases, vehicle stock,Pages: 9
Language: english
Created: Tue Sep 21 15:56:15 2004
Display cached document
Page 1
Page 2
Page 3
Page 4
Page 5
Page 6
Page 7
Page 8
Page 9
Winning the Oil End Game -- Technical Annex www.oilendgame.org
Chapter 21
ROCKY MOUNTIAN INSTITUTE LIGHT-VEHICLE STOCK AND FLOW
POLICY MODEL OVERVIEW
The RMI light vehicle stock-and-flow model estimates the effects of the following light-
vehicle policies on efficient-vehicle market penetration.
Feebates
Previous studies have shown that greater than 90 percent of the efficiency improvement
from feebates results from manufacturers response to the feebate (product-mix), as
opposed to consumer response (sales-mix).i As a simplification and conservatism, we
assume light-vehicle supply equals demand and explicitly ignore the sales-mix response.
Our model takes feedback from previous years' sales to compute the next year's feebate
pivot point, which in turn affects consumer choice through a Benefit to Cost calculation
to the manufacturer. This calculation compares the fuel savings and rebates to the
consumer with the incremental cost of the Conventional Wisdom or State of the Art
technology to the manufacturer. Technological progress is modeled through the gradual
introduction of State of the Art technologies. The model then estimates the effect of
efficient-vehicle market penetration on vehicle stock dynamics through 2025 (including
fuel use, vehicle stock efficiency, and consumer surplus) based on EIA's Annual Energy
Outlook 2004 projected sales mix and vehicle miles traveled (VMT).
Low-income vehicle scrappage program with replacement
This policy applies only to Conventional Wisdom vehicles. We model one million vehicle
sales (above EIA's baseline projection) beginning in 2010, with a five year phase-in
period in which the number of vehicles introduced through this program increases at a
linear rate. For simplification we ignore the effect of removing inefficient vehicles from
the stock and focus instead on the much larger efficiency increase brought about by new
efficient-vehicle purchase.
Platinum carrot competition and government vehicle purchases
State of the Art with technology procurement assumes 3-year shortening of the time it
takes to capture 010% of the market as a result of secured market demand from three
policies (government procurement, golden carrot, and platinum carrot). Thus, we model
the guaranteed loans to directly increase the State of the Art retooling rate by pushing the
retooling end date from 2030 to 2024.
Secured debt financing to the manufacturer
We recognize the strained balance sheets of auto manufacturers,. Our model therefore
assumes secured debt financing to enable the manufacturers to finance the retooling
effort.
Assumptions
-5%/y discount rate to the consumer, fuel savings valued over 3 years.
-14-year vehicle life.
-Retooling: first State of the Art vehicle by 2010, 20 years to 90%; Conventional Wisdom
much earlier and steeper slope (because they incremental technologies currently in the
marketplace).
-No technology improvement for Conventional Wisdom or State of the Art vehicles while
EIA light vehicles improve per baseline (and include 1 million hybrids by 2025).
-Non-changing incremental costs for Conventional Wisdom and State of the Art.
-Per Davis (1995) and Greene (2004) 90%+ of improvement from feebates is due to
manufacturer (product-mix) improvement, while the remaining 10% is due to consumer
(sales-mix). For this reason we conservatively ignore the consumer response.
-Manufacturer response is modeled by Benefit to Cost ratio that compares consumer's
valuation of the technologies with consumers incremental technology cost (see Technical
Annex, Ch. 5, table 5-10 for more information).
-We assume EIA baseline for new light-vehicle sales volume, distribution of sales
between cars and light trucks, and VMT.
-Technological improvement will go entirely towards decreasing fuel consumption, not
increasing performance (except for the performance increase implicit in the EIA
projection). EIA projected vehicles and Conventional Wisdom/State of the Art
technologies assumed to be the same except for efficiency and price (consumers value of
performance increase = 0 in all cases).
-When introduced into the marketplace, State of the Art vehicles are purchased by buyers
who otherwise would have purchased Conventional Wisdom vehicles in the absence of
State of the Art technologies.
-Consumers will respond in the same way through the model to State of the Art vehicles
as they do to Conventional Wisdom vehicles.
-Indifferent manufacturers will adopt technologies 50% of the time.
Introduction
Consumers only consider the first three years of fuel savings in their automobile purchase
decision, and fuel efficiency ranks low on the list of considerations for new car
buyers.We examine a feebate system to solve this market failure by giving a rebate for a
portion of the previously undervalued fuel savings. Each feebate policy has a pivot point,
or fuel intensity (in gallons per mile) above which car buyers pay a fee and below which
car buyers receive a rebate. Both the fee and the rebate are based on the difference
between the new car's fuel intensity and the pivot point. Feebates work by increasing
market capture of existing efficient vehicles and, by changing the pivot point yearly,
work to incentivize manufacturers towards continued automobile efficiency. Unlike
efficiency standards, feebates are fully transparent to buyers and are continuous rather
than step functions, removing the incentives for gaming the policy. Unlike fuel taxes,
feebates provide a direct and undiscounted price signal at the time and place of buying
the car and unlike any other policy instrument, feebates reward continuous improvement.
We based the model on literature describing the adoption rate for individual fuel-efficient
technologies to model the adoption of our suite of Conventional Wisdom and State of the
Art technologies.
Methodology
We consider feebates on two distinct classes of vehicles (cars and light trucks) and over
two technology portfolios (Conventional Wisdom and State of the Art). Conventional
Wisdom and State of the Art are portfolios of technologies available to the consumer in
2025. At a respective 31 mpg and 74 mpg (EPA adjusted), they represent the upper end
of the new car efficiency distribution (see below).
MY 2004 through April combined car and light truck efficiencies
1400
light trucks
New sales thousands
1200
cars
1000
800
600
400
200
0
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
55
58
61
64
Efficiency (mpg)
Figure 21-1: MY 2004 cars include 657 models with a median fuel economy of 23 mpg,
a maximum of 63mpg, and a standard deviation of 5.45. MY 2004 light trucks include
488 models with a median fuel economy of 18 mpg, a maximum of 26 mpg, and a
standard deviation of 2.99.ii
In keeping with our assumption that consumer's utility function will not change, we
design a separate pivot point for cars and light trucks so as to not encourage shifting
between vehicle classes. We calculate the pivot point for the feebate in Equation 21-1 as
the difference between the average car and the Conventional Wisdom or State of the Art
car efficiency in gpm (gallons per mile), weighted by their respective populations for the
previous year.
Equation 21-1:
#Conventional Wisdom x (GPMcw PIVOT) = #AV x (PIVOT GPMav)
Rearranging Equation 21-1 implies:
PIVOT = [#Conventional Wisdom x GPMcw + #AV x GPMav] / ( #Conventional
Wisdom + #AV)
To keep the policy revenue neutral, and to create incentives to manufacturers for
installing new efficient technologies, we adjust the pivot point yearly based on previous
year's sales data and projected fleet efficiencies. To calculate the 2005 market share, we
use the seed values below for Conventional Wisdom and State of the Art light vehicle
sales in 2004:
Table 21-1: Estimated values for 2004 vehicle sales
car light
truck
Conventional 100,000 10,000
Wisdom
State of the 0 0
Art
The resulting rebate, or fee, is calculated at a variable rate per 0.01 gpm delta from the
pivot point (equation 21-2). As market penetration increases (mainly as a result of
increasing retooling rates for State of the Art vehicles), more cars will adopt fuel-efficient
technologies and will receive rebates, while fewer inefficient cars will be sold to pay the
complementary fees. Extremely efficient vehicles would receive a much larger rebate and
vehicles near the pivot point would receive a near zero rebate. The same would also be
true on the fee side, for a example a 2004 Toyota 4Runner would pay a fee of $429 in
2005 while the average fee would be $2, and the same Toyota 4Runner would pay $2,004
if it were sold in 2025, when the average fee would be $865. This moving pivot point is
an important instrument in feebates, beginning with small fees and large rebates when
volumes of efficient vehicles are small and increasing towards efficient technologies.
Equation 21-2:
$FEEBATE = RATE x [(GPMx - PIVOT) / 0.01]
Feebates can either be paid to the consumer or to the manufacturer. In theory, both
destinations would have the same effect to increase consumer choice for efficient
vehicles. We have learned from rebates to efficient appliances that the most value can be
captured from a program if incentives are applied throughout the value chain.
However, feebates given solely to the manufacturer would leave open the possibility of
rebates not being passed directly to the consumer (a lack of transparency). We
recommend feebates be given to the consumer at the point of purchase, to increase
transparency and prevent distillation of the rebate by the manufacturer. We imagine a
feebate program that is described directly on the sticker price, possibly in combination
with EPA's fuel efficiency labeling program.iii
We recognize that 90 percent of the increased efficiency that results from a feebate
program comes from a manufacturer's reaction to the increased market incentive for
efficient technologies. In reality, due to economic inefficiencies and indirect market
signals, we suggest feebates to effect the consumer decision directly and rely on
manufacturers to respond to the increased consumer demand for efficient technologies.
For these two reasons, we use the convention outlined in Davis et al. 1995 to calculate the
Benefit to Cost ratio for the manufacturer's adopting efficiency technologies (we
consider the entire Conventional Wisdom or State of the Art portfolio to be one efficiency
technology, as opposed to studying the effects of an individual technology as done in
Davis et al.).iv
Equation 21-3:
B/C = (PVFUELSAVE + Val$PERF + $FEEBATE)/TECHCOST
We define PVFUELSAVE as the discounted fuel savings (at a 5%/year real discount
rate) difference between a new Conventional Wisdom or State of the Art vehicle and the
average car or light truck projected by EIA for sale in that year. We only account for the
first three years of fuel savings using NRC's suggestion that consumers may only account
for this amount of savings when purchasing a new car.v Although it is important to note
this number is highly sensitive to perceived future fuel prices, rates of return on
investments in efficient technologies, and beliefs on the effect of fuel-efficient
technologies on used vehicle value. As the Conventional Wisdom or State of the Art
efficiencies remain constant while the EIA baseline projections increase, the
PVFUELSAVE decreases over time.
Val$PERF is the consumer's value of the performance change as a result of the fuel-
efficient technology. As we assume the Conventional Wisdom and State of the Art
vehicles to be equivalent to the average new vehicle projected by EIA except for the
increase in efficiency, we set Val$PERF to 0 for all calculations. $FEEBATE is the
change in feebate brought about by the introduction of Conventional Wisdom or State of
the Art technologies, as outlined in Equation 21-2. TECHCOST is the incremental cost of
the fuel saving technologies to the manufacture. We use the incremental technology costs
given in tables 5-4 and 5-10 of Technical Annex, Ch. 5:
Table 21-2: Incremental technology costs for Conventional Wisdom
and State of the Art vehicles
car light
truck
Conventional $806 $667
Wisdom
State of the $2167 $2805
Art
We further adopt the methodology outlined in Davis et. al 1995 to calculate market
penetration, M, of Conventional Wisdom and State of the Art vehicles that results from
the feebate policy.
Equation 21-4:
M = Mmax X Pmax X [1/(1+e^-2(B/C-1)]
Mmax is defined as the maximum market share of the technology. We assume Mmax to be
100 percent, as our Conventional Wisdom and State of the Art technology portfolios
include numerous technologies that can be substituted for each other to achieve a full
market share.
Pmax is defined as the retooling percentage that allows manufacturers to introduce new
technologies in their products. It includes both the engineering and manufacturing
improvements needed to bring new technologies to market. The retooling rate assumes
government financial support (i.e., removes financial constraint of the automotive
industry). Since Conventional Wisdom technologies are currently available to consumers,
we use a logistic function that is modeled after the introduction of incremental
technology adoption in the marketplace. Our Pmax function reaches 50 percent market
share by 2013 and achieves 99 percent by 2025.
Retooling Rates -- Conventional Wisdom
120% and State of the Art
100%
Percent Retooled
80%
Conventional Wisdom
60%
State of the Art - with
40% financing support
State of the Art - with
20% technology
procurement
0%
Figure 21-2: Retooling rates for Conventional Wisdom and2035 of2040 2045 2050
2010 2005
2015 2020 2025 2030
State the Art vehicles
with and without technology procurement.
For the State of the Art technologies, we opt against describing the Pmax function on
historical data due to the dramatically different manufacturing requirements compared to
an equivalent volume average vehicle plant. We base the Pmax function for State of the Art
technologies on our approximate judgment of the capital and operational constraints of
the automobile industry. The S-curve begins in 2010, reaches 50 percent by 2025, and
plants are fully retooled by 2035.
From equation 21-3 we see that if Mmax and Pmax are both 100 percent, and the Benefit to
Cost ratio is 1, we would see about half the manufacturers adopting the Conventional
Wisdom and State of the Art technologies, as they would be indifferent towards installing
the technology. The resulting market share percentage from Equation 21-4 is used to
calculate the number of Conventional Wisdom or State of the Art vehicles in the next
year.
We calculate the retail fuel savings of the program as the difference between the cost of
the program (incremental capital costs of the technologies) and the value to the consumer
(discounted value of the fuel savings over the 14-year vehicle life). While we only
consider the first three years of vehicle life in the Benefit to Cost calculation, when
calculating consumer surplus to society we include the entire vehicle life, as consumers
as a whole are indifferent towards who owns the vehicle but public goods accrue
regardless of ownership. Because vehicles have a remaining life after 2025, this policy
will save fuel beyond what is accounted for in this study.
Sanity-checking our light vehicle model:
Due to the significant differences in modeling environments between Greene et. al. (2004) and
RMI (2004), we conclude that under conditions of using various NRC input variables, our
simplified model computing MPG improvements from baseline that are 5%-25% within those
predicted by Greene et. al. implies that our model is robust.
As our model only considers the manufacturer (product-mix) response and explicitly ignores the
consumer (sales-mix) response, we expect to underestimate the feebate response by roughly 10%.
We then note that major differences between Greene et. al.'s static pivot point and their snap-shot
of one year approximately 10-15 years in the future gives opportunity only for approximate
agreement. Nevertheless, sanity-checking with our dynamic model is useful and gives the
expected (conservative) results vs. Greene et. al. In contrast with Greene et. al, we introduce
dynamics into our model in that we have a continuously improving baseline (EIA 2005-2025),
and a continuously increasing pivot-point to model the positive effect of feebates on
technological progress. While not used for comparison with NRC vehicles, we employ a gradual
introduction of SOA vehicles in modeling our Conventional Wisdom and State of the Art Vehicles
to represent (dynamics in) technological progress.
As a sanity-check, we have therefore introduced NRCs vehicles to our dynamic model to see how
closely their diffusion and resulting new-sales MPG agrees with Greene et. al.'s static model. We
approximated the point-estimates of NRC vehicles from each of the curves by introducing these
curves as single-point vehicles by inputting each curve's break-even vehicle under a given feebate
rate. This break-even point (single MPG at a given incremental cost) was identified by the
intersection of the marginal retail cost curve (i.e. cost faced by consumer) and the curve
representing marginal present value to consumer, i.e. that curve describing the present value of
the feebate and the three-year fuel savings (discounted at 6%, see Greene et. al., 2001).
Depending on the feebate rate and NRC cost curve, we find that new-sales MPGs from RMIs
model are between 5% above and 10% below the MPGs as predicted by Greene et. al. , as in the
table below:
Base Year (EIA AEO) Scenario 3 Scenario 6
N/A Feebate $500 Feebate $1000
N/A Average Cost Average Cost
MPG MPG MPG MPG MPG MPG
Greene RMI Greene RMI 2025 Greene RMI 2025
CARS 28.2 28.5 31.8 32.5 35.2 34.3
LIGHT TRUCKS 20.7 20.8 26 25.7 29.2 27.4
AVERAGE 24.3 24.1 28.9 27.9 32.3 29.6
Percent change in GPM from baseline Greene RMI Greene RMI
CARS 11% 12% 20% 17%
LIGHT TRUCKS 20% 19% 29% 24%
AVERAGE 16% 14% 25% 19%
note: fuel economies are EPA laboratory as presented in EIA Annual Energy Outlook (Greene et. Al. based on AEO 2002, RMI on AEO 2004).
When the fundamental differences are accounted for, we are comfortable with concluding that our
model applied in a dynamic setting gives results that broadly and to a sufficiently accurate degree
agrees with other and more sophisticated modeler's outputs.
i
W. Davis et. al., "Effects of Feebates on Vehicle Fuel Economy, Carbon Dioxide
Emissions, and Consumer Surplus," DOE/PO-0031, Office of Policy, U.S. Department of
Energy, Washington DC, 1995. Also shown in D. Green et. al., "Feebates, Rebates, and
Gas-Guzzler Taxes: A Study of Incentives for Increased Fuel Economy," En. Pol. In
Press, Oak Ridge, TN.
ii
EPA Fuel Economy Guide, http://www.fueleconomy.gov/
iii
EPA Fuel Economy Guide, http://www.fueleconomy.gov
iv
W. Davis et. al., "Effects of Feebates on Vehicle Fuel Economy, Carbon Dioxide
Emissions, and Consumer Surplus," DOE/PO-0031, Office of Policy, U.S .Department of
Energy, Washington DC, 1995.
v
NAS/NRC (National Academies of Science/National Research Council, "Effectiveness
and Impact of Corporate Average Fuel Economy (CAFE) Standards," National Academy
Press, Washington, DC, 2002. http://books.nap.edu/books/0309076013/html