ArticlePDF Available

Impact of fuel price on vehicle miles traveled (VMT): Do the poor respond in the same way as the rich?

Authors:

Abstract and Figures

The effects of fuel price on travel demand for different income groups reveal the choices and constraints they are faced with. The first purpose of this study is to understand these underlying choices and constraints by examining the variation of fuel price elasticity of vehicle miles travelled (VMT) across income groups. On the other hand, the rebound effect—increase in VMT as a result of improvement in fuel efficiency may offset the negative effect of fuel price on VMT. The second purpose of this study is to compare the relative magnitudes of the fuel price elasticity of VMT and the rebound effect. A system of structural equations with VMT and fuel efficiency (MPG, miles per gallon) as endogenous variables is estimated for households at different income levels from 2009 National Household Travel Survey. Higher income households show greater fuel price elasticity than lower income households. Fuel price elasticities are found to be −0.41 and −0.35 for the two highest income groups, while an elasticity of −0.24 for the lowest income group is identified. The rebound effect is found to be only significant for the lowest income households as 0.7. These findings suggest the potential ability of using fuel price as a tool to affect VMT. The study results also suggest possible negative consequences faced by lower income households given an increase in fuel price and call for more studies in this area.
Content may be subject to copyright.
Impact of fuel price on vehicle miles traveled (VMT):
do the poor respond in the same way as the rich?
Tingting Wang Cynthia Chen
Published online: 9 May 2013
ÓSpringer Science+Business Media New York 2013
Abstract The effects of fuel price on travel demand for different income groups reveal the
choices and constraints they are faced with. The first purpose of this study is to understand
these underlying choices and constraints by examining the variation of fuel price elasticity
of vehicle miles travelled (VMT) across income groups. On the other hand, the rebound
effect—increase in VMT as a result of improvement in fuel efficiency may offset the
negative effect of fuel price on VMT. The second purpose of this study is to compare the
relative magnitudes of the fuel price elasticity of VMT and the rebound effect. A system of
structural equations with VMT and fuel efficiency (MPG, miles per gallon) as endogenous
variables is estimated for households at different income levels from 2009 National
Household Travel Survey. Higher income households show greater fuel price elasticity than
lower income households. Fuel price elasticities are found to be -0.41 and -0.35 for the
two highest income groups, while an elasticity of -0.24 for the lowest income group is
identified. The rebound effect is found to be only significant for the lowest income
households as 0.7. These findings suggest the potential ability of using fuel price as a tool to
affect VMT. The study results also suggest possible negative consequences faced by lower
income households given an increase in fuel price and call for more studies in this area.
Keywords Vehicle miles travelled (VMT) Fuel efficiency
Fuel (gasoline) price Income Structural equations
Introduction
Amid the concerns for sustainable development, ranked on top are congestion, greenhouse gas
emissions, and climate change. The three are inherently related to each other—congestion, as
T. Wang (&)
University of Washington, Wilcox Hall 261, Seattle, WA 98195, USA
e-mail: wangtt@uw.edu
C. Chen
University of Washington, More Hall 133A, Seattle, WA 98195, USA
e-mail: qzchen@uw.edu
123
Transportation (2014) 41:91–105
DOI 10.1007/s11116-013-9478-1
the result of a much faster increase in travel demand than capacity directly contributes to
greenhouse gas emissions and consequently climate change. Reducing travel demand, in
particular, vehicle miles traveled (VMT), has been considered as a central strategy for sus-
tainable development.
There are many ways to reduce VMT and one promising strategy is to treat fuel price as
a policy tool—basic economics theory tells us that higher fuel prices should lead to lower
VMT. Yet, fuel is unlike other commercial products with many alternatives and the vast
majority of the Americans rely on cars and thus must use fuel for basic needs in life such as
going to work (American Community Survey 2007–2011). Thus, the observed change in
VMT as the result of a change in fuel price may reflect one’s choices to change travel
behavior; or, it may simply be a manifest of the constraints one is faced with. In response
to an increase in fuel price, higher income households may choose to reduce their VMT
even though they do not need to do so financially and lower income households may not be
able to reduce their VMT even though they want to, simply because they are already
travelling at a bare minimum to maintain a functional life in the society. Because of the
interplay between choices and constraints, it is possible that the effect of fuel price on
VMT can be nonlinear across income groups. The first purpose of this study is to explore
how the effect of fuel price on VMT may vary among households at different income
groups and understand the underlying choices made and constraints encountered.
Households may switch to more fuel-efficient vehicles and a rebound effect may occur.
A rebound effect is an increased level of VMT as the result of an improvement in fuel
efficiency (Berkhout et al. 2000; Binswanger 2001). Thus, the observed change in VMT
may be the result of both a negative fuel price effect and a positive rebound effect.
Ideally, people should behave uniformly in response to a change in fuel cost regardless
the source of the change—whether it’s a change in fuel price or in fuel efficiency.
Empirical results do not fully support this uniform response. Researchers who looked at
how a particular payment mechanism (e.g. paying with cash vs. credit card) might affect
spending behavior (Feinberg 1986; Hirschman 1979; Soman 2001) found that people spent
less when direct costs are involved (paying cash). The effect of an increase in fuel price on
travel cost can be directly felt by people at the gas pump, while the impact of an increase in
fuel efficiency is subtle and implicit, as one must calculate the miles gained or the fuel
saved. Therefore, we hypothesize that, in general, the positive rebound effect should be
less than the negative fuel price effect. A second purpose of this study is to compare the
relative magnitudes of the fuel price effect and the rebound effect. Understanding the
magnitude of the rebound effect itself has important policy implications—if the rebound
effect is sufficiently large, total fuel consumption as a product of fuel efficiency and VMT
may increase. Similarly, a pure reduction on the emission rate (brought by more fuel
efficient vehicles) may not reduce total emissions, if there is a significant increase in total
VMT.
The 2009 National Household Travel Survey (NHTS) data is used for this study. It is a
recent survey of a representative sample of the Americans and contains critical information
on VMT, fuel efficiency and fuel price at vehicle level. VMT is simultaneously modeled
with fuel efficiency in a structural equations system to separate the rebound effect (from
fuel efficiency to VMT) from the reverse relationship (from VMT to fuel efficiency).
The rest of the paper is organized into five sections. In ‘‘The effect of fuel price on VMT
and the rebound effect’ section, the related literature is reviewed. In ‘Empirical data’’
section, we describe the NHTS dataset. The model developed is presented in ‘‘Method-
ology’ section, followed by the model results in ‘Results’ section. Finally, conclusions
are summarized and policy implications are discussed in ‘Discussion’ section.
92 Transportation (2014) 41:91–105
123
The effect of fuel price on VMT and the rebound effect
Many studies (Dahl and Sterner 1991; Oum et al. 1992; Espey 1998; Goodwin et al. 2004)
studied the effect of fuel cost, which is fuel price ($/gallon) divided by fuel efficiency
(miles/gallon) on VMT. Using the data on state-level VMT and gasoline price for all 50
U.S. states and D.C. from 1970 to 1991, Haughton and Sarkar (1996) estimated that the
elasticity of VMT with respect to fuel cost fell between -0.16 and -0.07 for the short-run
and between -0.58 and -0.21 for the long-run. Based on the data from twelve countries
for the period from 1973 to 1992, Johansson and Schipper (1997) reported estimates of
average annual VMT elasticity with respect to fuel cost ranging from -0.47 to -0.06.
Employing a disaggregate dataset for U.S. households from 1984 to 1990, Goldberg (1998)
reported household quarterly VMT elasticity with respect to fuel cost as -0.2.
The related topic—the rebound effect—has also been studied (Greening et al. 2000;
Sorrell et al. 2009). Most of these studies calculated fuel cost elasticity of VMT, which
implicitly assume that all three—fuel cost elasticity, fuel price elasticity, and rebound
effect—are equal in magnitude. Greene (1992) regressed annual VMT of U.S. passenger
cars and light-duty trucks from 1966 to 1989 on fuel cost; he estimated the short-run and
long-run rebound effect to be the same as -0.13. Jones (1993) conducted a follow-up study
to re-examine the model used in Greene’s paper. He found that some implicit assumptions
of Greene’s model, in particular, the absence of lagged dependent variable, could not hold.
Therefore, he adjusted the long-run rebound effect to -0.3. In a recent study on rebound
effect, with data on VMT of U.S. from 1966 to 2001, Small and Dender (2007) estimated
the rebound effect as -0.045 and -0.22 for the short-run and long-run, respectively.
Common to these two strands of literature is that both represent the fuel price elasticity
or rebound effect by changes in VMT with respect to a change in fuel cost ($ per mile),
which is calculated by dividing fuel price ($ per gallon) by fuel efficiency (miles per
gallon). This practice implicitly assumes that the fuel price elasticity of VMT and the
rebound effect (defined as fuel efficiency elasticity of VMT) are of the same magnitude
(Sorrell et al. 2009). The underlying rationale is that rational motorists should respond to
changes in fuel cost uniformly regardless the source of the change—whether it’s a change
in fuel price or in fuel efficiency.
In reality, people may not behave uniformly in response to a change in fuel price versus
a change in fuel efficiency. A person’s experience of an increase in fuel price is both direct
and vivid—the change is immediately felt at the gas pump, while the effect of a change in
fuel efficiency is felt more subtly—one must calculate the extra mileage gained or the
saving on the fuel. Consequently, VMT likely responds more elastically to a change in fuel
price than to a change in fuel efficiency. That a salient and vivid experience influences
behavior more has been found in a number of cases from consumer behavior research. One
example is that people tend to spend more when using credit cards than cash (Hirschman
1979; Feinberg 1986). Soman (2001) explained that paying with cash was a more salient
experience to recall, because people had to write down the amount paid on checks or pay
by taking cash out of their purses, and this experience affected future spending aversely.
There are a limited number of studies that have attempted to separate the fuel price
elasticity of VMT and the rebound effect, and the findings on their relative magnitudes are
mixed. Using the U.S. Residential Transportation Energy Consumption Surveys between
1979 and 1994, Greene et al. (1999) estimated that the fuel price elasticity of annual VMT
per vehicle was of a similar magnitude compared to the rebound effect—both were around
0.29. Based on a panel dataset of German households between 1997 and 2005, Frondel
et al. (2008) also concluded there was no statistical difference between the fuel price
Transportation (2014) 41:91–105 93
123
elasticity of monthly VMT and the rebound effect, but their absolute value of fuel price
elasticity was found to be 0.58, which doubled Greene’s estimate of 0.29. The magnitude
of fuel price elasticity was found to be larger than that of the rebound effect in other
studies. Wheaton (1982) estimated the fuel price elasticity of VMT per vehicle as -0.5; his
estimate of the rebound effect, 0.063, was much smaller. Using a 9-year Consumer
Expenditure Survey data set of U.S. households, Puller and Greening (1999) modeled
annual non-business VMT of households as a function of real gas price and composite
MPG (defined as annual VMT divided by annual gasoline consumption); they found a fuel
price elasticity of -0.69 and a rebound effect of 0.41.
In theory, the fuel price elasticity of VMT should decrease as income increases,
assuming there are alternatives to fuel that people can switch to. Fuel, however, is not an
average commodity that has many alternatives. For many Americans, driving is a basic
need to maintain a normal life and thus fuel consumption is a captive choice for many in
many instances. Consequently, the observed elasticity of VMT with respect to income may
not be linear and monotonically decreasing, as the theory suggests. Indeed, both linear and
non-linear trends have been shown in empirical studies. Based on the UK National Travel
Surveys from 1988 to 1993, Blow and Crawford (1997) found household annual VMT
elasticity with respect to real cost of driving per mile (including fuel cost) decreased as
income increased. Similar results were reported by Santos and Catchesides (2005) using
the 1998 UK National Travel Survey. Using the 1997 Consumer Expenditure Survey, West
(2004) estimated elasticities of VMT with respect to operating cost (defined as cost per
mile, where cost includes those of fuel, vehicle maintenance and tire) for different income
deciles and found operating cost elasticity of VMT first decreased for the first eight deciles,
followed by an increase for the last two richest deciles. By entering an interaction term
between income and fuel cost in the model for VMT, Small and Dender (2007) found that
fuel cost elasticity of VMT diminished with higher income level. As mentioned earlier,
fuel cost is related to both fuel price and fuel efficiency and observing the trend in the
impacts of fuel cost on VMT with respect to income doesn’t distinguish the trends in fuel
price elasticity and rebound effect. In contrast, how the rebound effect varies with income
is a subject that has been much less studied. Using the mean income from 1966 to 2001,
Small and Dender (2007) calculated the rebound effect as 4.5 % for the short-run and
22.2 % for the long-run; when a higher mean income of a later period (from 1997 to 2001)
was used, the rebound effect was calculated to be 3.1 % for short-run and 15.3 % for long-
run. They then concluded that the rebound effect diminished with higher income level.
Empirical data
The 2009 National Household Travel Survey (NHTS) was used for this study. The original
survey dataset comprises 150,147 households with 309,163 vehicles. For this study, we
first removed those households owning zero-vehicle, because they didn’t yield valid
observations, and we then removed those owning vehicles marked as commercial vehicles
in the dataset, as the usage patterns of these vehicles do not represent those of personal
household vehicles. We also removed those households with missing values on VMT, fuel
efficiency and fuel price and household income, since these variables are of central interest
in our study. This results in a sample of 105,372 households as our study sample. To
capture the variations of fuel price elasticity and rebound effects across income groups, this
sample of 105,372 households is divided into five income quintiles (Table 1); the number
of households in each quintile is similar with each other.
94 Transportation (2014) 41:91–105
123
In the NHTS data, each vehicle in the vehicle file is associated with two kinds of annual
VMT: estimated VMT and self-reported VMT. The estimated VMT is derived by Oak
Ridge National Laboratory for four kinds of vehicles: automobile, van, sport utility vehicle
(SUV) and pick-up truck, while the self-reported VMT is surveyed for each recorded
vehicle. The estimated VMT was generated based on four pieces of information—a single
odometer reading, the self-reported VMT, the daily VMT (based on the designated travel
day) and the characteristics of the primary driver (Oak Ridge National Laboratory 2011).
The estimated VMT is likely a truer representation of the actual VMT than the self-
reported one, but the use of the estimated VMT in the model leads to the problem of
spurious regression when it will be regressed on a similar set of variables used to derive it.
Thus, we primarily use the self-reported VMT
1
(divided by the number of vehicles in a
household) as the dependent variable in this study.
Estimated EPA fuel efficiency was adjusted by Energy Information Administration
(EIA) according to the actual road condition and the driving pattern for each recorded
vehicle to obtain the in-use fuel efficiency for each vehicle in the NHTS vehicle file
(Energy Information Administration 2011). The fuel efficiency is then averaged among all
the vehicles in a household to derive the average MPG, which is used as the other
dependent variable in the study. The implications of using the arithmetic average MPG will
be discussed in more details in the last section.
The recorded fuel price for each vehicle was extracted from the EIA’s retail pump price
series for 2008 and 2009 taking account of its fuel type, fuel grade and the geographical area
where the household is located (Frondel et al. 2008). The fuel price is then averaged among
all the vehicles in a household to obtain the average fuel price. The age of each vehicle is
calculated based on its model year, which is available in the NHTS database. We use average
vehicle age as an explanatory variable in the study. In addition, a dummy variable
(HYBRID) is calculated; it is equal to 1, if the household owns one or more hybrid vehicles.
Four variables, namely RETIRED (a dummy variable indicating the presence of one or
more retired household members), household size, population density of the census tract
where household is located and employment density of the census tract where a household
is located, are extracted from the NHTS household file (each record represents a house-
hold) to describe the characteristics of each household and its surrounding built
environment.
Table 1 Number of households
by income groups Income Sample size
$0–24,999 17,193
$25,000–49,999 26,272
$50,000–74,999 17,471
$75,000–99,999 24,519
$100,000 and over 19,917
1
We were aware of the possible inaccuracy in self-reported figures. To ensure the quality of self-reported
VMT used, we only used the self-reported VMT of those households with a ratio between self-reported
VMT and the estimated VMT ranging from 0.25 to 4 or with the difference between self-reported VMT and
estimated VMT less than 10,000 (this criteria is used by Oak Ridge National Laboratory in identifying
outliers of self-reported VMT). Otherwise, VMT used was the estimated VMT. Households with the
estimated VMT only account for 5 % of the households in our sample.
Transportation (2014) 41:91–105 95
123
Table 2presents the basic descriptive statistics on vehicle and household characteristics
for each of the five income quintiles. VMT per vehicle increases as income increases. On
average, fuel efficiency (measured as MPG, miles per gallon) also increases as household
income rises. A parallel observation is that higher income households own newer vehicles,
which have higher fuel efficiency. Geographically, households in large metropolitan areas
have higher incomes than those in smaller urbanized areas or rural areas (as evident by the
higher population and employment densities associated with higher income households)
and fuel prices in large metropolitan areas are also higher. The percentage of households
owning hybrid cars also increases with income.
Household size increases with income for the first four quintiles. The household size for
the fifth quintile is slightly smaller than that for the fourth one. Reversing this trend is that
the number of the retired decreases as income increases.
Methodology
The theoretical framework of the model used in this study is household production theory.
Following Puller and Greening (1999), we posit that a household receives utility from
performing transportation services (e.g. driving) such that various needs and wants in life
can be fulfilled. Consequently, the demand for gasoline is a function of gasoline prices,
household income and other socio-demographic characteristics. Because gasoline con-
sumption is simply calculated as the total miles driven (VMT) divided by fuel efficiency
(MPG), the demand for gasoline can be decomposed into two inter-related equations: one
for VMT and the other for MPG. More specifically, the demand for VMT is a function of
gasoline prices, MPG, household income and other socio-demographic attributes, while the
demand for MPG is a function of gasoline prices, VMT, household income and other socio-
demographic characteristics.
Figure 1shows the interplay of fuel price, fuel efficiency and VMT. Fuel price is
expected to affect both VMT and fuel efficiency (represented by link 2 and link 1,
respectively), though through different forces—higher fuel price will reduce VMT but
stimulate the acquisition and the use of more fuel-efficient vehicles. The positive rebound
effect is represented by link 3. However, this rebound effect must be separated from the
reverse link from VMT to fuel efficiency, or link 4, as higher VMT encourages the use of
more fuel-efficient vehicles.
Figure 1illustrates the presence of two endogenous variables: fuel efficiency and VMT.
This requires the use of a structural equations system (SEM), which is designed to handle
multiple endogenous variables simultaneously (Golob 2003). Accounting for household-
level socio-demographics and other vehicle related variables (e.g. vehicle age), the rela-
tionships depicted in Fig. 1can be modeled in the following graphical-presented SEM:
In the SEM above, except for the dummy variables (variables ‘‘RETIRED’’ and
‘HYBRID’’), all variables are log-transformed such that the estimated parameters directly
represent elasticities. More specifically, a
2
is the fuel price elasticity of VMT and a
1
is the
elasticity of VMT with respect to fuel efficiency, or the rebound effect.
a
1
, representing link 3 in Fig. 1, should be positive and a
2
, representing link 2, is
expected to be negative. b
1
and b
2
correspond to link 4 and link 1 in Fig. 1, respectively
and hence, both are expected to be positive. The related variables can be classified into
three groups: socio-demographics, vehicle attributes and built environment.
In general, households with more members travel more, while retirement is associated
with less travel. Three other variables are also included in the equation of VMT per
96 Transportation (2014) 41:91–105
123
Table 2 Descriptive statistics for each of the five income groups
Variable name Notation Unit Mean
$0–
24,999
$25,000–
49,999
$50,000–
74,999
$75,000–
99,999
$100,000
and over
VMT per vehicle VMT Miles 8,161 9,438 10,590 10,850 11,650
Average MPG MPG Miles per gasoline equivalent gallon 20.56 20.72 21.00 21.08 21.25
Average fuel price PRICE Dollars per gasoline equivalent gallon 3.05 3.06 3.06 3.07 3.09
Average vehicle age VEHAGE Year 11.00 8.98 8.13 7.59 6.81
HYBRID
a
HYBRID Hybrid =1, other =0 0.02 0.03 0.04 0.05 0.08
Household size HHSIZE Person 1.86 2.12 2.44 2.83 2.81
RETIRED
a
RETIRED Retired =1, other =0 0.60 0.53 0.38 0.25 0.22
Population density POPDN Person per square mile 2,879 2,909 2,931 2,828 3,371
Employment density EMPDN Person per square mile 930 934 946 872 1,038
a
Dummy variables
Transportation (2014) 41:91–105 97
123
vehicle: average vehicle age, population density and employment density at census tract
level, because older vehicles and vehicles in denser areas are driven less.
Average MPG is related to characteristics of the household vehicle fleet. As an existing
vehicle becomes older and less fuel efficient, the likelihood of acquiring a fuel efficient
vehicle, for example, a hybrid car, is larger. In-use fuel efficiency also depends on the
geographic characteristics of where households live. Driving in a dense area, as indicated
by high population density and employment density in the MPG equation, likely experi-
ences a stop-and-go pattern, which can significantly reduce fuel efficiency.
Results
The two equations depicted in Fig. 2are simultaneously estimated with Maximum Likeli-
hood Estimation (MLE) for each income quintile separately in AMOS, a SPSS SEM
application software. The model is properly identified since the number of pieces of known
information or the total number of elements in the variance–covariance matrix (45) is greater
than that of unknown information or the number of parameters to be estimated (22). v
2
values
of all the models are significantly greater than the cut off value (13.091) at 0.05 level with 23
degrees of freedom. The root mean square error of approximation (RMSEA), another
measurement of overall goodness-of-fit, is estimated to be around 0.08 for all models; this
indicates an acceptable model fit (HOE 2008). The results are presented in Table 3.
Before we discuss the key findings on fuel price, fuel efficiency (MPG), and VMT, we
first describe the results on the variables measuring the socio-demographics of the
household and the built environment near home. In the equation for VMT, the estimates of
all household-level variables are significant and consistent with our expectation. More
household members will lead to more VMT; presence of one or more retired household
members is associated with a lower VMT. Households living in dense areas (characterized
by high population density and high employment density near home) tend to drive less; this
association is true except for the second and the fourth income quintile where the estimates
are insignificant. In addition, older vehicles are driven less.
Similarly, the estimates in the equation for fuel efficiency are also mostly significant.
More specifically, households living in densely developed areas are more likely to own fuel
efficient vehicles. This result is consistent with the previous finding that households in
dense areas are less likely to own pickup or vans (Bhat et al. 2009). Pickups or vans are
harder to maneuver and park due to the limited space available in a dense area. Compact
cars usually have higher fuel efficiency than bigger cars and this explains the impacts of
these two built environment variables on fuel efficiency. Fuel efficiency decreases sig-
nificantly with vehicle age. Hybrid cars are fuel efficient and thus the presence of a hybrid
vehicle in the fleet will significantly improve the average fuel efficiency.
Fig. 1 Interplay of fuel price, fuel efficiency and VMT
98 Transportation (2014) 41:91–105
123
Fig. 2 Structural equations system
Table 3 Estimation results for income quintiles
Dependent
variable
Independent
variable
$0–
24,999
$25,000–
49,999
$50,000–
74,999
$75,000–
99,999
$100,000
and over
VMT MPG 0.700* -0.011 -0.071 -0.203 0.085
PRICE -0.237* -0.125 -0.094 -0.406* -0.345*
HHSIZE 0.226* 0.143* 0.117* 0.035* 0.125*
RETIRED -0.136* -0.222* -0.222* -0.040* -0.205*
POPDN -0.004 -0.017* -0.010* -0.010* -0.022*
EMPDN -0.027* -0.019* -0.027* 0.004 -0.027*
VEHAGE -0.068* -0.128* -0.141* -0.173* -0.128*
CONST. 7.278* 9.691* 9.963* 10.602* 9.874*
MPG VMT 0.017 0.105* 0.115* 0.109* 0.083*
PRICE 0.437* 0.391* 0.490* 0.595* 0.447*
POPDN 0.003* 0.010* 0.010* 0.003* 0.009*
EMPDN 0.007* 0.008* 0.007* 0.001 0.003*
VEHAGE -0.086* -0.070* -0.065* -0.048* -0.037*
HYBRID 0.053* 0.106* 0.174* 0.158* 0.241*
CONST. 2.488* 1.646* 1.426* 1.416* 1.719*
Number of observations 17,193 26,272 17,471 24,519 19,917
Degrees of freedom 23 23 23 23 23
v
2
d 27,725 39,130 25,746 32,974 25,022
RMSEA 0.08 0.08 0.08 0.08 0.08
* Variables significant at 0.05 level
Transportation (2014) 41:91–105 99
123
We now discuss the identified relationships among fuel price, fuel efficiency, and VMT.
The coefficient associated with MPG in the VMT equation measures the rebound effect—it
is significant only for the lowest income quintile and positive as expected. The fuel price
elasticities are all negative as expected, though only those of the lowest, second highest and
highest income groups are significant. In the equation of fuel efficiency, higher fuel price
encourages the use of more fuel efficient cars. Except for the lowest income group, VMT
has a significant, positive effect on fuel efficiency for all other four income groups; this is
consistent with our expectation that households that drive more have an incentive to
acquire vehicles with higher MPG. The significance of this effect (from VMT to MPG)
also confirms the endogeneity of fuel efficiency and failure to capture it may result in over-
estimation of the rebound effect. The insignificant impact of VMT on MPG for the lowest
income households may be because they do not have the resources to acquire newer, more
fuel efficient vehicles. The price of vehicles of higher fuel economy is relatively high. As
an example, the hybrid models of 2013 Ford Fusion cost over three thousand dollars more
than their non-hybrid counterparts (e.g. $27,200 for Ford Fusion Hybrid SE; $23,830 for
Ford Fusion SE). In the future, as more hybrid cars enter the used car market, it’s expected
that they may become affordable to lower income households.
The estimated fuel price elasticity with respect to income displays an interesting,
nonlinear pattern. For the lowest income quintile, the estimated elasticity is significant and
negative: -0.24. For the second and third quintiles, they are insignificant. Then, a sig-
nificant elasticity of -0.4 is found for the fourth quintile, followed by a significant elas-
ticity of -0.35 found for the fifth quintile. Underlying these findings appears to be a
reversed U-shaped pattern (Fig. 3).
Exactly what are the behavioral mechanisms underlying this nonlinear pattern between
fuel price elasticity of VMT and income is unknown and requires a panel dataset that
follows households over a relatively long period. We offer some insights. Our thoughts
again center upon the constraints and the choices people faced when making travel related
decisions relating to VMT and vehicle acquisition and transaction decisions in response to
the changing fuel price. These constraints and choices relate to both time and money.
Lower income households may not have the resources to acquire newer, more fuel efficient
vehicles. Lower income groups may be time-strapped—the limited financial resources they
have may force them to live at locations that are spatially mismatched from where most of
the employment is concentrated (Kain 1992; McLafferty and Preston 1992). As shown in
Table 2, lower income groups tend to live in areas with lower density. In other words, even
though lower income households may want to reduce VMT in response to a higher fuel
price, they may not afford to do so, if they are already traveling at a minimum to maintain
basic functions in life. For higher income households, they are able to reduce VMT in
Fig. 3 Price elasticity of VMT across income groups
100 Transportation (2014) 41:91–105
123
response to a higher fuel price voluntarily if their existing travel patterns involve a fair
amount of discretionary travel. Exercising this choice will exhibit larger observed elas-
ticities. To look further into the issues of constraints and choices, we calculated some
descriptive statistics on maintenance
2
and discretionary
3
trips for each income group
(Table 4).
The statistics in Table 4appear to support our expectation that, in general, lower
income households conduct fewer maintenance and discretionary trips than higher income
households. This observation is reflected in the number, the length and the duration of
these trips. The differences in discretionary trips are larger than those in maintenance trips.
Given the fact that discretionary trips have more flexibility than maintenance trips, the
results suggest that higher income households may be more flexible in reducing VMT
faced with increasing gasoline price. The very long trip length, trip duration and large trip
rate of the fourth quintile may be explained by their large household sizes (Bawa and
Ghosh 1999) and the low densities near home, which both induce more travel (Ewing et al.
1996).
Having probed into the constraints of households with relatively low income, we also
notice that the fuel price elasticity for the lowest income groups, though smaller, is
comparable to that of the two highest income groups in terms of its absolute value. On one
hand, it is intuitive that these households are subject to more stringent financial constraints.
On the other hand, it suggests that the above reasoning we provided for the insensitiveness
of the 2nd and the 3rd income groups to price change is somehow less pronounced here.
Possibly, the price elasticity of the lowest income households is more a manifest of
financial constraints than that of a time-trapped travel. This certainly needs further
investigation in additional studies.
Table 4 Maintenance and discretionary trips statistics for income quintiles
$0–
24,999
$25,000–
49,999
$50,000–
74,999
$75,000–
99,999
$100,000
and over
Trip length (miles/household/day)
Maintenance 17.56 21.28 22.77 40.06 24.56
Discretionary 19.11 22.61 27.03 56.13 33.75
Trip duration (min/household/day)
Maintenance 51.66 58.85 59.90 102.80 60.66
Discretionary 56.75 59.96 68.25 134.70 87.43
Trip rate (trips/household/day)
Maintenance 2.72 3.56 3.94 7.78 4.50
Discretionary 2.52 2.56 4.03 7.91 3.74
2
In 2009 NHTS, trip purposes of maintenance trips are following: Day care, Medical/dental services,
Shopping/errands, Buy goods: groceries/clothing/hardware store, Buy services: video rentals/dry cleaner/
post office/car service/bank, Buy gas, Family personal business/obligations, Use professional services:
attorney/accountant, Attend funeral/wedding, Use personal services: grooming/haircut/nails, Pet care: walk
the dog/vet visits, Attend meeting: PTA/home owners association/local government, Transport someone,
Pick up someone, Take and wait, Drop someone off, Meals, Get/eat meal.
3
In 2009 NHTS, trip purposes of discretionary trips are following: Social/recreational, Go to gym/exercise/
play sports, Rest or relaxation/vacation, Visit friends/relatives, Go out/hang out: entertainment/theater/sports
event/go to bar, Visit public place: historical site/museum/park/library, Social event, Coffee/ice cream/
snacks.
Transportation (2014) 41:91–105 101
123
The findings on the rebound effects for different income groups appear to tell a different
story—only that for the lowest income group is significant. In other words, when there is
an improvement in fuel efficiency, there appears to be a significant rebound for households
whose income is $25,000 or less. This finding adds support to our hypothesis that the travel
demand of these households may be far from saturation. Getting access to more fuel
efficient cars provides the opportunity of fulfilling the latent travel demand without a
significant increase in travel budget. However, further studies are still needed to verify this
statement. Yet, households with income more than $25,000 are not likely to drive sig-
nificantly more when an increase in fuel efficiency occurs. This difference in findings
between the fuel price elasticity of VMT and the rebound effect appears to confirm our
expectation that the magnitudes of the two effects are likely different, since one (fuel price)
exerts an immediate, direct effect on cost while the impact of the other (fuel efficiency) is a
subtle one and requires calculation (Soman 2001). To further substantiate our speculation,
we specified another model with fuel price elasticity and rebound effect constrained to be
equal as done in many previous studies (e.g. Greene 1992). Our null hypothesis is that the
model with equal fuel price elasticity and rebound effect is equivalent to the model without
this constraint. Model comparison results indicate this hypothesis is rejected at 5 % level
for the first, fourth and fifth income quintile (pvalue =0.000). For the second and third
income quintile, this hypothesis can’t be rejected (with pvalue equals to 0.136 and 0.076,
respectively). However, for both income groups no significant price elasticity or rebound
effect is identified. Thus, at least for some subpopulations, the magnitudes of fuel price
elasticity and rebound effect are statistically different and a change in fuel price exerts
more influence on VMT.
Discussion
By analyzing the interplay among fuel price, fuel efficiency, and VMT and understanding
how the relationships vary with income, this study reveals the underlying complexities
involved—first, a seemingly reverse U-shaped pattern is identified between fuel price
elasticity of VMT and income, revealing the choices made and the constraints encountered
by people when making travel related decisions; second, fuel price elasticity of VMT and
the rebound effect are unlikely of the same magnitude, questioning the current practice of
equating the two.
The first finding suggests that an increase in fuel price likely ripples through the pop-
ulation unequally and probably unfairly. Households on the lowest end of the income
ladder are likely to be hit most. We showed that on average lowest income households
traveled less and made shorter trips and a large proportion of trips were made for main-
tenance purposes. Therefore, if the lowest income households were to reduce travel, they
would probably have to give up some maintenance needs, which could lend their lives less
prosperous. In response, they may have to reduce spending on non-transportation related
consumption, and some (e.g. food and education) may be critical for their long-term
prospect of moving up on the ladder of income and social status. These concerns call for
more research that repeats the current study in specific locales to further investigate the
possible non-linear relationship between fuel price elasticity and income. Equally impor-
tant is the call for researchers to look beyond transportation behavior and include non-
transportation consumption to understand the trade-offs people make within transportation
and between transportation and non-transportation consumptions.
102 Transportation (2014) 41:91–105
123
It should be noted that fuel price elasticity may be overestimated in our study due to the
use of arithmetic average MPG. Most households have more than one vehicle. In the long
term, they may substitute a more fuel efficient vehicle for a less efficient vehicle in
response to an increase in fuel price so that they could maintain their total VMT with the
same cost as before. With this substitution effect, the actual decrease in VMT in response
to an increase in fuel price would be less than what is observed in our dataset. Therefore,
the real fuel price elasticity is expected to be smaller than our estimates. However, as
indicated in one study by Mannering (Mannering 1983), the difference due to substitution
effect may be small for three reasons: (1) some vehicles were exclusively used by one
household member, which precluded the possibility of substitution; (2) the compatibility
between vehicle attributes and activity also limited substitution; (3) difference in the fuel
efficiency of household vehicles was negligible. In Mannering’s study, the median dif-
ference in fuel efficiency between household vehicles is 4.87 MPG. In our data, the median
difference in fuel efficiency of household vehicles is 4.4 MPG.
The second finding—the rebound effect is much smaller than the fuel price elasticity of
VMT for most income groups—is good news from a policy perspective. Looking into the
future, we are likely to witness further improvements in fuel efficiency and these
improvements will not likely induce a great amount of additional travel. The story for those
households whose income is under $25,000 is different: their fuel price elasticity of VMT
is found to be -0.237, corresponding to a rebound effect of 0.7. In other words, for them,
the rebound effect appears to be significantly larger than the negative price elasticity. This
difference suggests that this group of households’ travel is far from saturation and again
calls for more research that look into the constraints encountered by them and policies that
cater to their transportation and possibly non-transportation needs.
The study is limited in its use of a cross-sectional dataset to capture dynamic behavior,
in particular the rebound effect. In a true sense, the interest of this study—the interplay
among fuel price, fuel efficiency, and VMT—spans over time and involves both short-term
(reduction in VMT in response to an increase in fuel price) and long-term behaviors
(acquisition of fuel efficient vehicles in response to higher fuel price and more VMT).
Thus, in order to truly capture this dynamic behavior (to allow inter- and intra-person
comparisons), a panel dataset spanning over multiple years is needed. However, such a
nationally representative dataset is unavailable. The use of a cross-sectional dataset in this
case comes with several assumptions (Kitamura 1990): (1) VMT has an equilibrium given
a specific set of values of its contributing factors; (2) this equilibrium can be uniquely
determined by these values; (3) the convergence speed of VMT towards this equilibrium is
relatively faster than the change speed of its contributing factors. If these assumptions are
true, then, at any time point, the choices of VMT of a large enough population are in
equilibrium and the long-run elasticities can be approximated by the results derived from
cross-sectional data. These assumptions may not be true and thus, the elasticities obtained
in this study need to be verified if and when a panel dataset becomes available. We do note,
however, the rebound effect estimated for the whole sample is 20 %, which is in consistent
with the results in previous studies (Small and Dender 2007).
References
American Community Survey. http://www.census.gov/acs/www/data_documentation/2011_release/ (2007–2011).
Accessed 20 Mar 2013
Bawa, K., Ghosh, A.: A model of household grocery shopping behavior. Mark. Lett. 10, 149–160 (1999)
Transportation (2014) 41:91–105 103
123
Berkhout, P.H.G., Muskens, J.C., Velthuijsen, J.W.: Defining the rebound effect. Energy Policy 28, 425–432
(2000)
Bhat, C.R., Sen, S., Eluru, N.: The impact of demographics, built environment attributes, vehicle charac-
teristics, and gasoline prices on household vehicle holdings and use. Transp. Res. B 43, 1–18 (2009)
Binswanger, M.: Technological progress and sustainable development: what about the rebound effect? Ecol.
Econ. 36, 119–132 (2001)
Blow, L., Crawford, I.: The distributional effects of taxes on private motoring. Institute for Fiscal Studies,
London (1997)
Dahl, C., Sterner, T.: Analysing gasoline demand elasticities: a survey. Energy Econ. 13, 203–210 (1991)
Energy Information Administration: EIA estimation methodologies. http://nhts.ornl.gov/2009/pub/EIA.pdf
(2011). Accessed 30 June 2012
Espey, M.: Gasoline demand revisited: an international meta-analysis of elasticities. Energy Econ. 20,
273–295 (1998)
Ewing, R., Deanna, M., Li, S.-C.: Land use impacts on trip generation rates. Transp. Res. Rec. 1518, 1–6
(1996)
Feinberg, R.A.: Credit cards as spending facilitating stimuli: a conditioning interpretation. J. Consum. Res.
13, 348–356 (1986)
Frondel, M., Peters, J., Vance, C.: Identifying the rebound: evidence from a German household panel.
Energy J. 29, 145–163 (2008)
Goldberg, P.K.: The effects of the corporate average fuel efficiency standards in the US. J. Ind. Econ. 46,
1–33 (1998)
Golob, T.F.: Structural equation modeling for travel behavior research. Transp. Res. B 37, 1–25 (2003)
Goodwin, P., Dargay, J., Hanly, M.: Elasticities of road traffic and fuel consumption with respect to price
and income: a review. Transp. Rev. 24, 275–292 (2004)
Greene, D.L.: Vehicle use and fuel economy: how big is the ‘rebound’ effect? Energy J. 13, 117–144 (1992)
Greene, D.L., Kahn, J.R., Gibson, R.C.: Fuel economy rebound effect for U.S. household vehicles. Energy J.
20, 1–31 (1999)
Greening, L.A., Greene, D.L., Difiglio, C.: Energy efficiency and consumption—the rebound effect—a
survey. Energy Policy 28, 389–401 (2000)
Haughton, J., Sarkar, S.: Gasoline tax as a corrective tax: estimates for the United States, 1970–1991.
Energy J. 17, 103–126 (1996)
Hirschman, E.C.: Differences in consumer purchase behavior by credit card payment system. J. Consum.
Res. 6, 58–66 (1979)
Hoe, S.L.: Issues and procedures in adopting structural equation modeling technique. Quant. Method Inq. 3,
76–83 (2008)
Johansson, O., Schipper, L.: Measuring the long-run fuel demand of cars: separate estimations of vehicle
stock, mean fuel intensity, and mean annual driving distance. J. Transp. Econ. Policy 31, 277–292
(1997)
Jones, C.T.: Another look at U.S. Passenger vehicle use and the ‘rebound’ effect from improved fuel
efficiency. Energy J. 14, 99–110 (1993)
Kain, J.F.: The spatial mismatch hypothesis: three decades later. Hous. Policy Debate 3, 371–392 (1992)
Kitamura, R.: Panel analysis in transportation planning: an overview. Transp. Res. A 24A, 401–415 (1990)
Mannering, F.L.: An econometric analysis of vehicle use in multivehicle households. Transp. Res. A 17A,
183–189 (1983)
Mclafferty, S., Preston, V.: Spatial mismatch and labor market segmentation for African-American and
Latina women. Econ. Geogr. 68, 406–431 (1992)
Oak Ridge National Laboratory. Bestmile: developing a best estimate of annual vehicle mileage for 2009
NHTS vehicles. http://nhts.ornl.gov/2009/pub/BESTMILE.pdf (2011). Accessed 30 June 2012
Oum, T.H., Ii, W.G.W., Yong, J.: Concepts of price elasticities of transport demand and recent empirical
estimates: an interpretative survey. J. Transp. Econ. Policy 26, 139–154 (1992)
Puller, S.L., Greening, L.A.: Household adjustment to gasoline price change: an analysis using 9 years of
US survey data. Energy Econ. 21, 37–52 (1999)
Santos, G., Catchesides, T.: Distributional consequences of gasoline taxation in the United Kingdom.
Transp. Res. Rec. J. Transp. Res. Board 1924, 103–111 (2005)
Small, K.A., Dender, K.V.: Fuel efficiency and motor vehicle travel: the declining rebound effect. Energy J.
28, 25–51 (2007)
Soman, D.: Effects of payment mechanism on spending behavior: the role of rehearsal and immediacy.
J. Consum. Res. 27, 460–474 (2001)
Sorrell, S., Dimitropoulos, J., Sommerville, M.: Empirical estimates of the direct rebound effect: a review.
Energy Policy 37, 1356–1371 (2009)
104 Transportation (2014) 41:91–105
123
West, S.E.: Distributional effects of alternative vehicle pollution control policies. J. Public Econ. 88,
735–757 (2004)
Wheaton, W.C.: The long-run structure of transportation and gasoline demand. Bell J. Econ. 13, 439–454
(1982)
Author Biographies
Tingting Wang is a Ph.D. student at the University of Washington, Seattle. Her research interest lies in
travel behavior analysis. Her current research focuses on mining individuals’ behavioral patterns from large
data sets.
Cynthia Chen is an associate professor in Civil and Environmental Engineering at the University of
Washington, Seattle. Her past research effort spans in the area of travel behavior analysis and demand
forecasting, in particular, dynamic analysis of travel behavior, residential location choices, and innovative
use of survey techniques.
Transportation (2014) 41:91–105 105
123
... personal/household or vehicle-based) data. The usual variables analysed are fuel consumption (Wadud, Graham, & Noland, 2010), vehicle distance travelled (Small & Van Dender, 2007;Wang & Chen, 2014), car ownership (Johansson & Schipper, 1997) or traffic flow (Burke, Batsuuri, & Yudhistira, 2017). Many studies in Cluster 7 involved urban vs. rural issues and income elasticity analysis. ...
... congestion charges, fuel taxes) might be effective, but they are likely to be inequitable in many cities due to current dependence on the car (Moriarty, 2016). Fuel price effects vary at different income levels (Wang & Chen, 2014), as lower income households are likely to restrict mobility to necessary trips, whilst higher income counterparts can either afford discretionary travel, or invest in better fuel efficiency. While the land-use approach to reducing car dependence is often advocated by compact city theorists, housing-transport mismatch remains an issue for many cities, with the poor distributed to outer-suburban locations and prohibitive rents in locations with good non-car accessibility (Cao & Hickman, 2018;Wells, 2012). ...
Article
While the period of heightened oil prices in 2005–2014 has attracted considerable academic and policy attention, there is scant cross-disciplinary research considering energy price, transport and land use together. In car dependent societies many socio-economically disadvantaged areas often lack public and active transport accessibility. Emerging research focus has been placed on issues of transport equity including effects of fuel prices. We reviewed 45 years (1972–2017) of publications about fuel price impacts, transport, and urban context, drawn from Web of Science listed publications. Bibliographic citation analysis reveals eight major research clusters with a set of inter-city comparative studies at their centre. The historical evolution, geographical trends, research approaches and the key themes are revealed by context analysis. The paper highlights the need for further studies looking at energy, transport and land use interaction, and suggests a greater focus on transport equity.
... The influences of socioeconomic and individual factors on car ownership and use are well explored in many studies [48]. Income is found to be a determinant for car ownership and use both for work and non-work purposes, which also may be one of the most important factors [49,50]. Yet some researchers claim that the relationship between income and VMT is not significant [38]. ...
... Table 1 summarizes the existing studies. It shows that descriptive analysis and statistical models are most commonly used in the existing studies such as the negative binomial regression [1,32], the structural equation model [22,37,50,53,63,65], multilevel ordered probit model [23], the ordered logit model [36], the multinomial logit model [38], the ordinary least squares regression [38], and the logistic regression model [59,66]. The models can identify the influencing factors and measure the power of these factors simultaneously, which can help to understand the role played by built environment in influencing car ownership and use. ...
Article
Full-text available
Concerns about transportation energy consumption and emissions force urban planners and policy makers to pay more attention to the effects of car ownership and use on the environment in China. However, few studies have investigated the relationship between the built environment and car ownership and use in China, especially in mid-sized and small cities. This study uses Changchun, China as a case study and examines the potential impacts of the built environment and socio-demographics on car ownership and use for commuting simultaneously using Bayesian multilevel binary logistic models. Furthermore, the spatial autocorrelation of car ownership and use is recognized across traffic analysis zones (TAZs), which are specifically represented by the conditional autoregressive (CAR) model. The estimated results indicate that socio-demographic characteristics have significant effects on car ownership and use. Moreover, the built environment measured at the TAZ level still shows a significant association with other factors controlled. Specifically, it suggests that denser residential density, compact land use, better transit services and street connectivity can reduce car dependency more effectively. This study provides new insights into how the built environment influences the car ownership and use, which can be useful for urban planners and policy makers to develop strategies for reducing car dependency.
... A more precise estimation of the effects of a tax change requires knowing how different consumers change their gasoline consumption with a change in the price. For example, there is some evidence for the U.S. showing that higher income households have greater price elasticities than lower income households (Wang and Chen, 2014). Unfortunately, there is no empirical evidence for Chile in this regard and further investigations are required. ...
Article
Dos pistas: ni la clase media ni la clase baja. ¿Por qué, entonces, algunos parlamentarios insisten con tanto ahínco en rebajarlo?. Hasta marzo del año pasado, el impuesto a los combustibles –en rigor, el impuesto a las gasolinas– era de 6 UTM por cada metro cúbico. En ese mes, el gobierno lo rebajó transitoriamente por dos años, primero a 4,5 UTM y luego a 3,5 UTM. A partir del junio de este año volvió a 4 UTM, pero gracias a presiones se le fijó un tope de 4,5 UTM al menos hasta abril del 2010. Continuar leyendo...
... This effect is attenuated in the second and third quintile (--0.22% and -0.20%, respectively), but becomes stronger again in the fourth and fifth quintile (-0.26% and -0.29%, respectively). A similar pattern has also been found by Wang and Chen [11] with respect to vehicle miles travelled, based on cross-sectional data of the 2009 National Household Travel Survey (NHTS) from the USA. Due to lack of data we currently apply the elasticities reported by Wadud et al. [10], as well as our own estimations for a single household, but aim to provide own estimations for different household income quintiles in Austria once access to an extensive consumption survey in Austria is granted. ...
Conference Paper
Full-text available
We analyze the economic and environmental impacts of different CO2 tax (uniform or progressive) and rebate (reduction of VAT, social contributions or lump-sum payments) schemes with focus on private consumption (i.e. heating and mobility) as well as distributional impacts on different household income quintiles in Austria. We use the econometric input-output model DYNK to investigate these impacts. DYNK is able to consider macroeconomic feedbacks of CO2 taxes and accompanying rebate schemes. An energy module allows to link production and consumption activities with energy demand and associated GHG emissions and includes behavioral estimations with regard to energy demand for private household income quintiles that are fully integrated in the macroeconomic part of the model. First preliminary results indicate that a uniform CO2 tax on fossil fuel use for private consumption (including a tax rebate on VAT for other commodities) has a weak regressive impact on household incomes. The distributional impact of CO2 taxes differs between heating and mobility consumption.
... -0.35% demand for a 1% change in fuel price), this effect is attenuated in the second and third quintile (--0.22% and -0.20%, respectively), but becomes stronger again in the fourth and fifth quintile (-0.26% and -0.29%, respectively). A similar pattern has also been found by Wang and Chen (2014) with respect to vehicle miles travelled, based on cross-sectional data of the 2009 National Household Travel Survey (NHTS) from the USA. Due to lack of data we currently apply the elasticities reported by Wadud et al. (2012), as well as our own estimations for a single household, but aim to provide own estimations for different household income quintiles in Austria once access to an extensive consumption survey in Austria is granted. ...
Article
In this study we investigate changes in travel due to level 2 automation among owners of electric vehicles in California. Level 2 automation has the potential to reduce driver fatigue and make driving less stressful which could mean drivers choose to travel more. We use questionnaire survey data to investigate changes to long distance travel and annual vehicle miles travelled (VMT) due to automation. Results show those who report doing more long-distance travel because of automation are younger; have a lower household income; live in urban areas; own a longer-range electric vehicle; use automation in a variety of conditions; and have pro-technology attitudes and prefer outdoor lifestyles. We use propensity score matching to investigate whether automation leads to an increase in annual VMT. The results of this show 4059–4971 more miles per year among users of level 2 automation compared to owners of similar vehicles without automation.
Article
Achieving societal climate goals requires rapid reductions in greenhouse gas (GHG) emissions from transportation. Recent efforts by policymakers have focused on increasing consumer adoption of electric vehicles (EVs). Nevertheless, EV sales remain low. Worse, even if EV market share jumped dramatically, it would take decades to replace the existing vehicle fleet, during which time vehicle GHG emissions would continue, worsening climate change. Consequently, some argue for policies to accelerate the retirement of inefficient fossil-powered vehicles through ‘cash-for-clunkers’ (C4C) programs. We examine C4C policies through a behavioral model of vehicle fleet turnover and EV market development in the U.S. We find C4C policies can substantially reduce vehicle fleet emissions at reasonable cost per tonne of CO2. To meet emissions reductions goals, C4C policies should apply only when consumers replace their fossil-powered vehicles with EVs. C4C policies incentivizing EVs accelerate cost reductions through scale economies, charging infrastructure deployment, model variety, and consumer awareness, boosting EV adoption beyond the direct effect of vehicle replacement. The result is a substantial synergy amplifying the impact of C4C and lowering unit cost of emissions reductions. Cash-for-clunkers is further amplified when deployed together with complementary policies promoting renewable electricity production and a gas tax or carbon price.
Article
Drawing on the 2009 and 2017 waves of the National Household Transportation Survey, this paper is concerned with modeling the fuel price elasticity, allowing for differential estimates in its magnitude over time and across households. We find an elasticity close to zero for the year 2009, which increases to upwards of −0.3 by the year 2017. We explore the robustness of this result to different model specifications and estimation techniques, including instrumental variable estimation to account for the possible endogeneity of fuel prices, as well as quantile regression to account for heterogeneity according to driving intensity. While a similar pattern of increasing elasticity over time emerges across all these models, the quantile model suggests an inverse relationship between the magnitude of the elasticity and miles driven in 2017. As demonstrated with a back of the envelope calculation, one implication of this pattern is a more muted effectiveness of fuel taxation than implied by the estimates of a standard mean regression.
Article
Carbon dioxide emissions from New Zealand’s transport sector have experienced rapid growth since 1990. In this paper, we investigate the scope for a targeted price signal to curb emissions growth and help deliver on the country’s Paris Agreement pledges. Cost burdens on various income groups are investigated, and experts’ opinions are elicited to evaluate two types of carbon price policies (i.e. a carbon tax and an increase in the current ETS price). We use two methodological approaches in this paper. Estimates of the social cost of carbon and the price elasticity of fuel demand are used to understand the mitigation potential of a higher carbon price, while a multi-criteria analysis technique is used to understand experts’ preferences between an ETS and a carbon tax. The findings are that with a price elasticity of transport fuel demand of around −0.7, a carbon price between NZD 100 (USD 65) per tonne of carbon dioxide (tCO2) and NZD 235 (USD 153)/tCO2 could reduce transport emissions by between 33% and 44% in 2030, respectively, from the 2016 level. The (uncompensated) cost-burdens on lower income households due to a hypothetical price of NZD 100/tCO2 and NZD 235/tCO2 are estimated to be around NZD 531 (USD 345)/year and NZD 670 (USD 436)/year per household respectively. Experts marginally favoured a carbon tax over an ETS because the revenue collected through a carbon tax could be utilized in a more holistic way to offset the tax burden on lower income households and fund emissions reduction technologies and infrastructures. Key policy insights • A carbon price of NZD 235 (USD 153)/tCO2 implying a fuel price of NZD 3/litre (USD 1.95/litre) could reduce transport emissions by 44% in 2030 from the 2016 level. • A fuel price of NZD 3/litre is comparable to fuel prices in the early 1980s, when these were at a historic high. • A carbon price of NZD 235 (USD 153) would increase the annual domestic transport expenditure of lower income households by NZD 670 (USD 436), a 42% increase. • A carbon tax is marginally preferred by interviewed experts to an increase in prices under the current ETS.
Article
The rebound effect is the phenomenon underlying the disproportionality between energy efficiency improvements and observed energy savings. In road transport, the effect reveals the extent to which energy savings from improved fuel efficiency are forgone due to additional car travel. We present a meta-analysis of 74 primary studies containing 1120 estimates of the direct rebound effect in road transport to evaluate its magnitude and identify its determinants. We find that the short-run rebound effect is, on average, about 10–12%, whereas the long-run effect about 26–29%. However, variation of estimates is large and can mainly be explained by differences in the time horizon considered, the elasticity measure used, and the type of data and econometric approach employed in primary studies. We also find that the rebound effect is declining over time and that lower per capita incomes, higher gasoline prices and higher population density are associated with larger rebound effects.
Article
Full-text available
This paper assesses the regressive effects of gasoline taxation in the United Kingdom. When all households are considered, middle-income households suffer most of the burden. When only car-owning households are considered, gasoline taxation is strongly regressive. Low-income households that own a car are more severely affected than high-income households because they spend a larger proportion of their income on motoring. This conclusion is similar to that of previous studies on the topic.
Article
Full-text available
In the conventional four-step travel demand modeling process, the number of trips made by a household is modeled in terms of household size, income, and other sociodemographic variables; any effect of location, land use, or transportation service level is discounted. This is the same as discounting any effect of household accessibility to out-of-home activities as a factor in trip generation (accessibility depending on all three: location, land use, and transportation service level). In contrast to the practice of trip generation, theory tells us that trip rates must vary with accessibility, and some (not all) empirical studies have found that they do. In light of conflicting empirical studies, and the obvious need for more precise and policy-sensitive travel forecasts, this issue is revisited. The independent effects of land use and accessibility variables on household trip rates were tested for using data from Florida travel surveys. It was found that, after controlling for sociodemographic variables, residential density, mixed use, and accessibility do not have significant, independent effects on household trip rates. Conventional trip generation models, which generate person trips by vehicle (not by all modes) and do so without regard to residential location, may not be as bad as one would imagine a priori.
Article
Full-text available
This paper assesses the regressive effects of gasoline taxation in the United Kingdom. When all households are considered, middle-income households suffer most of the burden. When only car-owning households are considered, gasoline taxation is strongly regressive. Low-income households that own a car are more severely affected than high-income households because they spend a larger proportion of their income on motoring. This conclusion is similar to that of previous studies on the topic.
Article
Long-run fuel demand for cars is estimated by estimating separately total vehicle stock, mean fuel intensity, and mean annual driving distance, based on a new data set consisting of 12 OECD countries from 1973 to 1992. A large part of the estimated long-run fuel price elasticity arises from changes in mean fuel intensity, while the long-run income elasticity arises largely from changes in the number of cars. The effects of changed taxes on car ownership and use are significant, but smaller than from a change in fuel tax.
Article
Kenneth A. Small and Kurt Van Dender* Department of Economics University of California, Irvine Irvine, CA 92697-5100 ksmall@uci.edu, kvandend@uci.edu ... This version: April 10, 2006 (corrected July 17, 2006 and August 18, 2007) ... Shorter version published, Energy ...
Article
... This paper is a substantially revised version of the RWI : Discussion Paper No. 57, “ Identifying the Rebound – Theoretical Issues and Empirical Evidence from a German Household Panel ”. – All correspon- dence to Manuel Frondel, RWI Essen, Hohenzollernstr. ...
Article
The economic well-being of African-Americans and Latinos in the US depends critically on women's employment and earnings. Gender differences in labor market segmentation are central to the spatial mismatch debate, both in their effects on wages, occupation, and transportation access and their links to place-based variation in commuting and spatial access to employment. Using 1980 PUMS data for northern New Jersey, we examine the spatial mismatch between jobs and residences for employed black and Hispanic women and the links between labor market segmentation and spatial mismatch. Minority women have poorer spatial access to jobs than white women, but they typically have better spatial access to employment than minority men. The mismatch is greatest for African-American women, reflecting their heavy reliance on mass transit and poor spatial access to employment in all economic sectors. Latina workers have more localized labor markets, but also the lowest earning of any group. For Latinas, the primary problem is not spatial access to employment but rather a lack of access to well-paying jobs. These differences in mismatch reflect the combined effects of gender-and race-based segmentation and spatial access to employment and transportation. -from Authors