2
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
About KAPSARC
The King Abdullah Petroleum Studies and Research Center (KAPSARC) is a
non-prot global institution dedicated to independent research into energy economics,
policy, technology and the environment across all types of energy. KAPSARC’s
mandate is to advance the understanding of energy challenges and opportunities
facing the world today and tomorrow, through unbiased, independent, and high-caliber
research for the benet of society. KAPSARC is located in Riyadh, Saudi Arabia.
This publication is also available in Arabic.
Legal Notice
© Copyright 2021 King Abdullah Petroleum Studies and Research Center (“KAPSARC”).
This Document (and any information, data or materials contained therein) (the
“Document”) shall not be used without the proper attribution to KAPSARC. The
Document shall not be reproduced, in whole or in part, without the written permission
of KAPSARC. KAPSARC makes no warranty, representation or undertaking whether
expressed or implied, nor does it assume any legal liability, whether direct or indirect,
or responsibility for the accuracy, completeness, or usefulness of any information that
is contained in the Document. Nothing in the Document constitutes or shall be implied to
constitute advice, recommendation or option. The views and opinions expressed in this
publication are those of the authors and do not necessarily reect the ofcial views or
position of KAPSARC.
3
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
It is crucial to understand new vehicle buyers’ valuation of fuel economy to determine whether they are
myopic about future fuel costs. If consumers are very myopic, a fuel-price related policy instrument (such
as fuel tax or carbon tax) is unlikely to inuence their new vehicle purchasing decisions. We use a case
study of Indian two-wheeler buyers to estimate consumer sensitivity to future fuel costs when purchasing a
new vehicle. We build discrete choice models to estimate the discount rate that consumers apply to obtain
the present value of future operating cost at the time of vehicle purchase. High implicit discount rates imply
that consumers prefer current savings, consequently paying less attention to future costs. A low implicit
discount rate means that consumers are willing to pay more for an alternative vehicle that will produce
savings in the future. The models were built using nationally representative revealed-preference survey data
of more than 8,000 respondents who purchased new two-wheelers in 2018. Our results indicate that:
Indian two-wheeler buyers are not myopic about future fuel costs and ascribe a comparatively high
value to fuel economy.
The annual discount rate is less than 10% for Indian two-wheeler buyers with monthly household
incomes of more than 15,000 Indian rupees (INR) (~USD 215 in 2018; ~73% of the sample). This rate is
much lower than the estimated discount rate for four-wheelers in other regions. This result highlights the
need to formulate local policies based on respective market analyses, rather than borrowing policies
from other markets.
Our analysis also indicates that fuel economy valuation is the second most important factor inuencing
consumer purchase decisions, with vehicle styling being rst and comfort and brand being third and
fourth.
Key Points
4
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
India has the world’s third highest carbon dioxide
(CO2) emissions, after China and the United
States (Timperley 2019). The transportation
sector is the third largest contributor to carbon
dioxide emissions in India, accounting for roughly
11% of all carbon dioxide emissions in 2016
(Janssens-Maenhout et al. 2017). Road transport
accounts for around 94% of the total carbon dioxide
emissions of the transportation sector (Bhatt 2019).
The rapid increase in vehicle sales in India is
partly responsible for the increase in the country’s
carbon dioxide emissions. India recently displaced
Germany to become the world’s fourth largest
vehicle market (Gupta et al. 2018). Vehicle sales in
India are expected to increase even further, along
with rising personal incomes and rapid urbanization;
this increase has major implications for global
carbon dioxide emissions. Estimates from an
Indian government policy think tank suggest that
the number of on-road vehicles in the country and
passenger mobility-related carbon dioxide emissions
may triple by 2030 (NITI Aayog and Rocky Mountain
Institute 2017a).
Two-wheelers dominate the Indian passenger
vehicle market. In 2019, the sales share of two-
wheelers in the domestic passenger vehicle market
was 84%, compared with 13% for four-wheelers
and 3% for three-wheelers (SIAM 2019). Moreover,
annual sales of two-wheelers have almost doubled
during the last decade, from 11.8 million in 2010
to 21.2 million in 2019 (Statistical Research
Department 2020).1 Considering the large market
share of two-wheelers and their expected growth
rate, various policy levers, such as feebate2 policies,
are being considered to reduce the carbon dioxide
emissions associated with the Indian two-wheeler
sector (IEA 2020; NITI Aayog and Rocky Mountain
Institute 2017a, 2017b).
From a policy perspective, understanding the fuel
economy valuation of Indian two-wheeler buyers
is crucial to evaluating whether there is an ‘energy
efciency gap’ or an ‘energy paradox’, i.e., whether
Indian consumers are myopic about, and therefore
undervalue, future operating costs at the time of
purchase (Bento et al. 2012; Fuerst and Singh
2018; Gillingham and Palmer 2014; Gillingham et
al. 2019; Jaffe and Stavins 1994; Matsumoto and
Omata 2017; Orlov and Kallbekken 2019; Parry
et al. 2007; Yoo et al. 2020). If consumers are
found to undervalue future fuel economy at the
time of vehicle purchase, implementing policies
such as fuel economy standards3 that will help
consumers save money is reasonable (Allcott
and Wozny 2014; Chugh et al. 2011). Consumers’
fuel economy valuation has become even more
critical, given the Indian government’s recent policy
announcements aimed at securing energy security
by increasing the uptake of electric vehicles, which
have relatively higher upfront costs and lower
operating costs (Albrahim et al. 2019; IEA 2020;
Kumar and Alok 2020; Li and Wang 2019; Zhuge
et al. 2020). However, these types of analyses are
difcult to undertake in the Indian context due to
data availability challenges. We are only aware of
one study by Chugh et al. (2011) that estimates the
fuel economy valuation of Indian car buyers using
a hedonic price approach. Studies on preferences
for two-wheelers in a developing country, however,
are rare (Guerra 2019; Lin et al. 2013; Ye and Wang
2011).
Using nationally representative revealed-preference
survey data from more than 8,000 respondents who
purchased new two-wheelers in 2018, we estimate
consumers’ valuation of the fuel economy of two-
wheelers. We use discrete choice models and
estimate the discount rate that Indian consumers
use to obtain the present value of future operating
Executive Summary
5
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Executive Summary
costs at the time of purchase. Appropriate sampling
weights are used in the estimation to ensure that
the sample represents a population-level sales
proportion at the make-model level.
Our results indicate that the majority of Indian
two-wheeler buyers, ~73% of the sample, have a
discount rate below 10%. This indicates that they
are not myopic about future fuel costs. These
estimates could be used to forecast the demand for
new models in the two-wheeler market and to better
understand the effect of fuel prices on demand for
existing make-models. The results of the survey
analysis also indicate fuel economy to be among
the top two most important factors inuencing Indian
two-wheeler buyers’ purchase decisions.
6
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
This study used revealed-preference survey
data of 8,159 respondents from across India
who each purchased a new two-wheeler
vehicle for personal use between March and
October 2018 (at 2~6 months of ownership). The
survey was conducted between September and
December 2018. Respondents were contacted
through street intercepts or at other locations, such
as petrol stations and shopping malls. Detailed
interviews were conducted at a location and time
of their convenience. We also used market-level
sales data of two-wheelers to assess how well the
individual-level sample represents the population.
Both datasets were collected and provided by J.D.
Power, a global market research company. The
individual-level dataset contains information on the
attributes of the purchased two-wheelers, such
as the brand (make), model, segment (dened by
engine displacement), fuel economy and purchase
price. This dataset also contains the delivery date,
the number of months of ownership, the vehicle’s
mileage and the buyers' demographic characteristics
(e.g., income, gender and age).
As no information on the other models considered
by the respondents during purchase is available,
we created a choice set for each respondent.
Ben-Akiva and Boccara (1995) suggest a method to
latently generate a choice set, based on the two-
stage characterization choice process by Manski
(1977). However, this method is not scalable to a
large number of alternatives, as in this study (i.e., 70
make-models). Horowitz and Louviere (1995) and
Swait (2001) propose other methods of generating
a consideration set. However, these methods
also do not work in practice for a large number
of alternatives because they require enumerating
all possible choice subsets. In the absence of a
practical method of generating a choice set, we
assume that respondents considered all models
available on the market when they purchased their
two-wheelers. This means that all available models
in the market were included in the respondents’
choice set. Due to a substantial variation in the
reported purchase prices and fuel economy of
the models, we sourced data for the attributes of
each model from BikeWale, a platform that offers
comprehensive information about the two-wheeler
market in India.
Data Description
7
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Discrete Choice Model Formulation
Figure 1. An example of the nesting structure in the NL models.
For our model, we assume that consumers
choose an alternative that maximizes utility.
The indirect utility derived by individual i from
choosing alternative j is dened as follows:
!" =
!" +
e
!" = +
e
!"
1
where Vij is the systematic utility and
e
!"
is an
idiosyncratic error term. We assume that Vij is a
linear function of the alternative-specic attributes
Xij with marginal utility vector
. The distribution of
e
!"
leads to different models. Gumbel distribution of
independent error terms produces a conditional logit
(CL) model. After normalizing the scale of the error
term, the probability of choosing alternative j by
individual i is dened as follows:
!" =#!"
∑#!#
$∈&!
,
We note that the CL model is affected by an
independence-of-irrelevant-alternatives (IIA)
property (Train 2009). To address this challenge,
we also estimate a nested logit (NL) model
that considers a hierarchical tree structure and
addresses the IIA assumption by allowing for
correlation between the error terms of alternatives
(Wen and Koppelman 2001). Figure 1 shows one
possible nesting (tree) structure in the context of
the current case study, where an individual’s choice
set has 70 two-wheeler models. In this structure,
the rst level is dened according to makes (or
brands), and the error terms of models within each
nest (TVS, Hero, Honda, others) are assumed to
be correlated. In general, the correlation within
nest m is dened as
!= 1 − !
, where
!
is
the log-sum coefcient. To ensure that the model
is consistent with the random utility maximization
theory, the estimated log-sum coefcients should be
between 0 and 1. Both CL and NL can be estimated
using a maximum likelihood estimation. We estimate
these models using the mlogit package in R
software (Croissant 2012).
Choices
TVS Hero Honda Others
Model 70
Model 43Model 42Model 26Model 25Model 13Model 12Model 1
8
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Discrete Choice Model Formulation
Discount Rate
For decisions concerning future cash ow, rational
decision-makers are assumed to calculate the
net present value of benets and costs (Train
1985). Vehicle choice is one such example, as the
operating cost is incurred across the lifespan of the
vehicle but the vehicle price is paid at the time of
purchase. Vehicles can have higher upfront costs
but may yield lower operating costs due to improved
fuel economy, and vice versa. A rational user would
use a discount rate to compute the present value
of the operating cost, and would then consider a
trade-off between the purchase price and other
attributes. To represent this decision rule, we rewrite
the systematic part of the indirect utility equation in
the context of our case study:
= #$% +&'()$% +
where
!
and
!"#$
are marginal utilities of the
purchase price and present value of the operating
cost. Zij is a vector of other attributes (e.g., the top
speed and curb weight), with
as the marginal
utility vector.
Calculating PVOCij requires knowledge of the
discount rate, which is unknown to the researcher.
Helfand and Wolverton (2009) and Wang and
Daziano (2015) reviewed various methods for
estimating the discount rate in vehicle preferences.
We briey describe the endogenous discounting
method used in this study. If the vehicle life is long
enough (a long vehicle life assumption is tenable
for monthly cashow) and the fuel price is assumed
to be constant, PVOCij can be computed using the
capitalized worth approximation introduced by Train
(1985):
!" ≈ !"
,
where OCij is the uniform monthly operating cost
of alternative j for individual i, and r is the monthly
discount rate. If we dene
!"
as the parameter of
OCij, the systematic utility equation becomes:
= #$% +&'$% + ,
where
!" = #$!"
. For a rational consumer under
ideal market conditions,
!= "#$% = −
, where
is a marginal utility of the income. Therefore, the
maximum willingness to pay for marginal savings in
monthly operating costs can be expressed as:
= !
"#
= 1
"#
The monthly discount rate can be converted to
an annual discount rate (a) using the following
expression:
= (1 + )!" − 1.
Thus, the implicit annual discount rate can be
obtained as a by-product of the estimation of
discrete choice models, with covariates priceij and
OCij. A high implicit discount rate implies that two-
wheeler buyers care less about future savings or
costs. Therefore, they give a lower weight to the
monthly operating cost (i.e., fuel economy). This
phenomenon is known as the ‘energy paradox’
(Wang and Daziano 2015). Conversely, a low implicit
discount rate indicates that two-wheeler buyers are
willing to pay a higher upfront cost for an alternative
that provides future operating cost savings. In this
study, we assume that the marginal utility of the
purchase price decreases in household income
terms (Koppelman and Bhat 2006). Thus, the
discount rate also becomes:
!= "
!× #$
.
9
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Descriptive Statistics
In this section, we present a descriptive analysis
of the vehicle choice data. Table 1 summarizes
the average attribute values of the models in each
segment. The motorcycle (MC)-Economy and
MC-Executive segments have, on average, higher
fuel economies, lower prices, lower accelerations,
lower weights and lower top speeds compared with
other motorcycle segments.
Table 1 also shows the possibility of high
correlations between make-model-specic
attributes. To evaluate the extent of these
correlations, Table 2 presents pairwise correlations
between the attributes of all 70 make-models. The
results indicate that the magnitude of all correlations,
except for fuel economy, are above 0.75. The
presence of such high correlations between
alternative-specic attributes and no variation in
choice sets across respondents makes estimating
choice models using revealed-preference datasets
difcult (Haaf et al. 2016; Sheldon and Dua 2018).
Results and Discussion
Table 1. Mean values of model-specic attributes for segments.
Table 2. Correlation between model-specic attributes.
Segments Model count Fuel
economy
(km/l)4
Purchase
price
(INR)
Acceleration
(sec)
Weight
(kg)
Engine
power
(bhp)
Top speed
(km/hr)
MC-Economy 15 66 56,133 8.1 112 8.3 89
MC-Executive 758 65,571 7.0 121 10.3 95
MC-Premium 14 43 105,773 4.7 148 17.1 119
MC-Premium Plus 535 154,400 3.8 184 21.1 123
MC-Upper
Executive
947 88,411 5.3 142 13.1 112
SC-Economy 150 53,000 12.2 93 5.3 70
SC-Executive 14 48 56,000 10.1 106 7.6 83
SC-Upper
Executive
548 70,800 8.2 110 8.6 89
* MC: motorcycles, SC: scooters; Acceleration: time to accelerate from 0 to 60 kilometers per hour (km/hr); sec = seconds;
kg = kilograms; bhp = brake horsepower
Fuel
economy
Purchase
price
Acceleration Curb
weight
Engine
power
Top
speed
Fuel economy 1
Purchase price -0.73 1
Acceleration 0.42 -0.77 1
Curb weight -0.61 0.91 -0.82 1
Engine power -0.66 0.92 -0.82 0.92 1
Top speed -0.59 0.86 -0.88 0.85 0.92 1
10
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Results and Discussion
Survey Analysis
We also derive insights about the variation in
preferences of different demographic groups for
two-wheeler segments in Table 3. Of the sample,
5.3% of buyers were female but almost all of
them chose a two-wheeler from the scooter (SC)
segments. This may be because models in these
segments are convenient to drive and have better
storage facilities, automatic transmission and
step-through frames, which are more appropriate
for Indian women’s riding styles. Of the sample,
18.8% were consumers who already had at least
one vehicle in their household. As expected, the
proportion of vehicle owners is much higher in
the Premium Plus, Premium and Upper Executive
segments. While we see a similar trend for higher
income households, there are fewer buyers older
than 40 years in these luxury segments. These
ndings conrm our supposition that higher priced
two-wheelers are more likely to be owned by higher
income households, and that stylish two-wheelers
are more likely to be owned by younger consumers.
Table 3. Preferences of demographic groups for different segments.
Figure 2. Frequencies of main purchase reasons.
Female Car owner Monthly household
income > INR 50,000
Age > 40 years
MC-Economy 0.3% 12.4% 7.7% 14.2%
MC-Executive 0.0% 15.9% 8.6% 15.2%
MC-Premium 0.4% 24.3% 20.1% 4.2%
MC-Premium Plus 0.7% 39.2% 28.3% 8.2%
MC-Upper Executive 1.2% 18.6% 18.0% 7.8%
SC-Economy 39.8% 9.7% 10.7% 17.5%
SC-Executive 15.8% 18.1% 12.0% 15.6%
SC-Upper Executive 13.7% 20.8% 15.4% 15.4%
0% 5% 10% 15% 20% 25% 30%
Style/Looks
Mileage/Fuel economy
Comfort
Inconsistent colour
Engine performance
Recommended by friends/family
Popular/Seen frequently on road
Quality/Workmanship
Reliability/Durability
Low maintenance cost
Past experience with make/model
Service/Support
11
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Results and Discussion
Table 4. Segment-wise proportion of buyers who quote one of top three reasons behind the purchase.
Segments Style/Looks Mileage/Fuel economy Comfort Sample proportions
MC-Economy 11.0% 57.4% 22.3% 26.9%
MC-Executive 8.0% 14.6% 13.6% 11.9%
MC-Premium 28.3% 3.5% 12.2% 15.7%
MC-Premium Plus 8.9% 0.8% 3.5% 6.5%
MC-Upper Executive 10.5% 4.3% 6.2% 7.7 %
SC-Economy 0.8% 0.7% 3.6% 1.2%
SC-Executive 21.6% 13.7% 29.2% 21.4%
SC-Upper Executive 10.9% 5.0% 9.4% 8.7%
The survey also asked respondents about their
main reason for buying the specic two-wheeler.
The reasons given and their corresponding
proportions are presented in Figure 2. Figure 2
indicates that style, mileage, comfort, brand and
engine performance were the top ve reasons
given (in decreasing order). Bansal and Kockelman
(2017) used the opinions of experts to conclude that
Indian consumers prioritize fuel economy, price,
engine power and brand (in decreasing order of
importance) when purchasing two-wheelers. Three
of the four attributes highlighted by Bansal and
Kockelman (2017) are among the top ve factors
determined by the current study. However, the
experts interviewed in Bansal and Kockelman’s
(2017) study may have overlooked the importance
that Indian consumers attach to the style/look
of two-wheelers. This turned out to be the most
important factor in the current study.
We further investigate the segment-specic
preferences of respondents who stated one of the
top three reasons for purchase. The results are
presented in Table 4. As expected, the respondents
who quote style/looks as the most important
reason are more likely to buy two-wheelers in the
MC-Premium segment. Moreover, respondents
who cite fuel economy as the main reason behind
their purchase prefer the MC-Economy segment,
which has the highest fuel economy (Table 1).
Respondents considering comfort as a top reason
do not show a strong inclination toward any segment
but are slightly more inclined to buy two-wheelers
from the SC-Executive segment.
12
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Figure 3. Sample and sales proportions at make level.
Results and Discussion
Sampling Weights
To determine the representativeness of the sample,
we rst plot the make/brand-level sample and
sales proportions, presented in Figure 3. The plot
indicates that the sample mainly underrepresents
the Hero brand and slightly overrepresents all
other makes, except for Hero and Honda. As the
dataset used in this study was collected using a
choice-based sampling protocol, it is important to
include sampling weights in the estimation to ensure
consistent estimates (Manski and Lerman 1977).
To this end, we compute sampling weights for each
make-model to ensure that the choice proportions
in the sample are the same as the actual sales
proportions. The sampling weights vary from 0.14
for the most overrepresented model (Bajaj Avenger
220) to 3.09 for the most underrepresented model
(Hero HF Deluxe/Deluxe Eco/Deluxe i3S).
0% 5% 10% 15% 20% 25% 30% 35% 40%
Hero
Honda
Bajaj
TVS
Yamaha
Royal Enfield
Suzuki
Sample percent Sales percent
13
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Results and Discussion
Model Estimation Results
In this section, we discuss the results of the
conditional logit (CL) and nested logit (NL) models.
Having access to data on months of vehicle
ownership and mileage enabled us to compute
individual monthly operating costs. Decreasing
the marginal utility of the purchase price relative
to household income also guarantees enough
sampling variation in this variable. Given that two-
wheelers in the executive and economy segments
have substantially higher fuel economy than two-
wheelers in other segments, we anticipate that
buyers of these alternatives might ascribe greater
importance to fuel economy (Table 4). Therefore,
they may have lower discount rates. To address this,
we include the interaction of operating costs with the
economy-executive dummy in some specications.
The pairwise correlations between the covariates
are presented in Table 5.
We also attempted to estimate Indian consumers’
willingness to pay for other performance-based
attributes of two-wheelers, such as acceleration
and top speed. However, we encountered two
challenges. First, these attributes are highly
correlated with the purchase price (Table 2);
therefore, including them in the utility signicantly
affects the discount rate estimates. Second,
these attributes do not have variations across the
sample. Thus, the Hessian of the loglikelihood is not
invertible, leading to inference-related challenges.
These are common issues in revealed-preference
studies (Allcott and Wozny 2014; Haaf et al. 2016;
Mabit 2014; Sheldon and Dua 2020).
Table 5. Correlation between covariates.
Price/income OC OC x EC_EX
Price/income 1
OC 0.15 1
OC x EC_EX -0.22 0.23 1
* Price: purchase or ex-showroom price; income: monthly household income; OC: operating cost per month/1,000;
EC_EX: dummy, which is 1 if the alternative is executive or economy class.
14
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Results and Discussion
Tables 6 summarizes the results of weighted and
unweighted CL specications. Signs of marginal
utilities of purchase price and monthly operating
cost are as expected in all specications. The higher
loglikelihood value in the weighted CL indicates that
including sampling weights helps to better explain
Indian consumers’ preferences for two-wheelers.
Moreover, the marginal utility of the operating
cost, and thus, the discount rate estimates of the
weighted CL, are lower than those of the unweighted
CL. The sign on the interaction term is as expected
in both unweighted and weighted specications.
Including the interaction term does not affect the
marginal utility of the purchase price, because of the
low correlation between covariates.
Table 6. Results of the conditional logit model.
Specication 1 Specication 2
Unweighted Weighted Unweighted Weighted
Covariates Estimate z-value Estimate z-value Estimate z-value Estimate z-value
Price/income - 0.13 -9.2 - 0.13 -7.13 - 0 .13 -9.2 - 0 .13 -7. 2
OC -0.43 -5.3 -0.70 -7.18 -0.72 -7.5 -0.96 - 9.1
OC x EC_EX -0.25 -5.7 -0.31 -6.6
Loglikelihood -32,729 -28,519 -32,713 -28,497
McFadden
R-square 0.002 0.1304 0.0025 0.1311
# of observations 8,159 8,159 8,159 8,159
* Price: purchase or ex-showroom price; Income: monthly household income; OC: operating cost per month /1,000;
EC_EX: dummy; that is 1 if the alternative is executive or economy class.
** All specications have 69 alternative-specic constants, one for each make-model.
15
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Table 7. Results of the weighted NL model.
First level: Segment First level: Make
Same correlation
across nests
Different correlation in
each nest
Same correlation
across nests
Different correlation in
each nest
Covariates Estimate z-value Estimate z-value Estimate z-value Estimate z-value
Price/income - 0.16 -10.4 - 0.18 -9.3 - 0.12 -3.6 -0.31 -8.3
OC -1.18 -11.2 -2.06 -15.0 -3.61 -9.7 -2.30 -9.4
OC x EC_EX -0.38 -7. 9 -0.99 -11.9 -0.88 -6.7 -0.77 -7. 5
Log-sum coefcient Estimate Std. error Estimate Std. error Estimate Std. error Estimate Std. error
Segment 3.3 1.41
MC-EC-EX 4.1 1.39
MC-P-UE-PP 2.3 0.56
SC 130.0 1,184.91
Make 4.3 0.82
TVS 1.7 0.37
Honda 3.2 0.95
Hero 2.4 0.56
Other 3.1 0.47
Loglikelihood -28,486 -28,441 -28,779 -28,472
McFadden
R-square 0.1314 0.1328 0.1225 0.1319
# of observations 8,159 8,159 8,159 8,159
* Price: purchase or ex-showroom price; income: monthly household income; OC: operating cost per month /1000;
EC_EX: dummy; that is 1 if the alternative is executive or economy class; MC-EC-EX: nest having motorcycles in economy and
executive segments; MC-P-UE-PP: nests having motorcycles in prime, upper executive and prime plus segments; SC: nests
with all scooters.
** All specications have 69 alternative-specic constants, one for each make-model.
Results and Discussion
Alternative Specication
Table 7 shows the results of the weighted NL model.
We consider two nesting structures with brand/
makes and segments at the rst level. For each
hierarchical structure, we estimate two specications
with the same and with different error correlation
across all rst-level nests. The log-sum coefcients
across all specications are greater than one. This
indicates that the estimated model is not consistent
with the random utility maximization theory.
We also attempted various other specications at
the rst level. However, we could not nd a utility-
theory-consistent nesting structure. We explored
various cross-nested logit model specications
(Bierlaire 2006) but were unable to nd a utility-
consistent nesting structure. Therefore, we consider
the weighted CL (specication 2) as the nal
specication.
16
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Results and Discussion
Checking for Price
Endogeneity
Endogeneity arises if the unobserved factors
associated with vehicle choice are correlated with
observed attributes, such as price. If endogenous
regressors are present, the coefcient estimates of
the observed attributes are inconsistent. In vehicle
choice modeling, price endogeneity can be a
concern (Haaf et al. 2016) and statistical testing to
check for endogeneity is not possible in empirical
studies. However, given that alternative-specic
constants (ASCs) capture the part of the utility
governed by unobserved attributes, measuring
the correlation between purchase price and ASCs
indicates the extent of price endogeneity (Haaf
et al. 2016; Sheldon and Dua 2020). To check for
potential price endogeneity issues, we use the
same approach and nd a correlation between
price and the estimated ASCs of make-models. A
low correlation value (-0.06) for the weighted CL,
specication 2, shows that price endogeneity is
unlikely to be a concern in this analysis.
Discount Rate
Figure 4 shows the relationship between the annual
discount rate and monthly household income for
specication 2 of the weighted CL. In the executive
and economy segments, the annual discount rates
for households with monthly incomes of INR 10,000,
25,000 and 100,000 are 13.1%, 5.1%, and 1.2%,
respectively. These values are slightly higher —
17.7%, 6.1% and 1.7% — for all other segments.
Such low discount rates indicate that most Indian
two-wheeler buyers in our sample are not myopic
and ascribe a comparatively high value to fuel
economy.
Figure 4. The estimated discount rate (weighted CL, specication 2 in Table 6).
Household income (in thousands INR)
CL (Other segments) CL (Executive-economy)
20
Annual discount rate (%)
5
10
15
40 60 80 100
17
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Results and Discussion
Policy Implications
In this section, we discuss the importance of our
results within the context of the feebate policies
being considered by the Indian government to
expedite the adoption of fuel-efcient vehicles
(NITI Aayog and Rocky Mountain Institute 2017b).
The latest report by NITI Aayog mentions that auto
buyers apply a high discount rate. A feebate would
compel auto buyers to consider the vehicle’s entire
life-cycle’s fuel savings. However, we argue that,
while nding the pivot point5 and rate parameter6
when designing feebate policy, policymakers must
ensure that perturbation in the discount rates
due to a feebate policy should not be far from the
current discount rate perceived by auto buyers. This
consideration is particularly crucial in a democratic
country such as India, where policy implementation
can encounter political hurdles if many auto buyers
are worse off due to the policy (Ganguly 2019;
Phule 2019; Sharma 2019). Given that the discount
rate estimates of four-wheeler buyers do not
apply to two-wheeler buyers, and other countries
implementing feebate policies do not have such a
high proportion of two-wheelers, our discount rate
estimates would be valuable to Indian policymakers
when designing pivot points and rate parameters
of feebate policies for two-wheelers. For instance,
our results are not in agreement with the hypothesis
of the NITI Aayog report, which states that auto
buyers apply high discount rates. Their statement
might hold true for four-wheeler buyers but not
for two-wheeler buyers. Moreover, accounting for
the consumer heterogeneity in the discount rate
provides exibility for designing a progressive
feebate policy that involves higher rebates and lower
fees for lower income consumers (Sheldon and Dua
2019a). Similar progressive rebate designs have
also been reported in the literature and are currently
under pilot stage testing (Sheldon and Dua 2019b;
Sheldon and Dua 2019c) and/or fully implemented
(Colgan 2016).
18
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
This study uses discrete choice models to
analyze Indian consumers’ preferences for
fuel economy and other attributes when
purchasing two-wheelers. The results indicate
that most Indian two-wheeler buyers, ~73% of the
sample, have a discount rate below 10%. This
means that they are not myopic about the future
economy of their two-wheelers. These estimates are
useful in forecasting the demand for new models in
the two-wheeler market, and in understanding the
effect of fuel prices on demand for existing make-
models. The results of the descriptive analysis
are also relevant for vehicle manufacturers. For
instance, style/looks and fuel economy were found
to be the top two most important factors inuencing
Indian two-wheeler buyers’ purchasing decisions.
Conclusions
19
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
1 Younger buyers—30 years and younger—account for the majority of two-wheeler vehicle sales.
2 ‘Feebates’ is an incentive system that involves taxes on fuel-inefcient vehicles and rebates for fuel-efcient
vehicles.
3 Recent research suggests that, similar to a feebate, fuel economy standards impose a constraint on automakers
that creates an implicit subsidy for fuel-efcient vehicles and an implicit tax for fuel-inefcient vehicles (Davis and
Knittel 2019).
4 Measured at an average speed of approximately 31 km/h, as per the ofcial Automotive Research Association of
India test.
5 A pivot point (sometimes called a benchmark) denes for which vehicles suppliers pay fees and for which ones
they receive rebates (NITI Aayog and Rocky Mountain Institute 2017b).
6 A rate parameter determines the magnitude of the fee or rebate for each incremental difference from the pivot
point (NITI Aayog and Rocky Mountain Institute 2017b).
Endnotes
20
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
References
Albrahim, Mohammed, Ahmed Al Zahrani, Anvita
Arora, Rubal Dua, Bassam Fattouh and Adam
Sieminski. 2019. “An overview of key evolutions in
the light-duty vehicle sector and their impact on oil
demand.” Energy Transitions, 3(1–2): 81–103. https://
doi.org/10.1007/s41825-019-00017-7
Allcott, Hunt and Nathan Wozny. 2014. “Gasoline
prices, fuel economy, and the energy paradox.” The
Review of Economics and Statistics, 96(5): 779–
795. https://doi.org/10.1162/rest_a_00419
Bansal, Prateek and Kara M. Kockelman. 2017.
“Indian Vehicle Ownership: Insights from Literature
Review, Expert Interviews, and State-Level Model.”
Journal of the Transportation Research Forum,
56(2). https://doi.org/10.5399/osu/jtrf.56.2.4432
Ben-Akiva, Moshe and Bruno Boccara. 1995.
“Discrete choice models with latent choice sets.”
International journal of Research in Marketing, 12(1):
9–24. https://doi.org/10.1016/0167-8116(95)00002-j
Bento, Antonio M., Shanjun Li and Kevin Roth. 2012.
“Is there an energy paradox in fuel economy? A note
on the role of consumer heterogeneity and sorting
bias.” Economics Letters, 115(1): 44–48. https://doi.
org/10.1016/j.econlet.2011.09.034
Bhatt, Yagyavalk. 2019. “India’s transport CO2
emissions and value-added contribution.” King
Abdullah Petroleum Studies and Research Center.
URL: https://www.kapsarc.org/research/publications/
indias-transport-co2-emissions-and-value-added-
contribution/ (March, 2021)
Bierlaire, Michel. 2006. “A theoretical analysis of
the cross-nested logit model.” Annals of Operations
Research, 144(1): 287–300. https://doi.org/10.1007/
s10479-006-0015-x
Chugh, Randy, Maureen Cropper and Urvashi
Narain. 2011. “The cost of fuel economy in the
Indian passenger vehicle market.” Energy Policy,
39(11): 7174–7183. https://doi.org/10.1016/j.
enpol.2011.08.037
Colgan, Chuck. 2016. “Increased lower-income
incentives and reduced high-income caps start Nov.
1, 2016.” Center for Sustainable Energy. https://
energycenter.org/thought-leadership/news/california-
clean-vehicle-rebate-project-initiates-new-eligibility
Croissant, Yves. 2012. “Estimation of multinomial
logit models in R: The mlogit Packages.” R package
version 0.2-2. http://citeseerx.ist.psu.edu/viewdoc/
download?doi=10.1.1.303.7401&rep=rep1&type=pdf.
Davis, Lucas W. and Christopher R. Knittel.
2019. “Are fuel economy standards regressive?”
Journal of the Association of Environmental and
Resource Economists 6(S1): S37–S63. https://doi.
org/10.1086/701187
Fuerst, Franz and Ramandeep Singh. 2018. “How
present bias forestalls energy efciency upgrades:
A study of household appliance purchases in India.”
Journal of Cleaner Production, 186: 558–569.
https://doi.org/10.1016/j.jclepro.2018.03.100
Ganguly, Shreya. 2019. “Electric vehicles this week:
centre suspends ‘feebate’ policy ahead of elections
and more.” Inc42. https://inc42.com/buzz/electric-
vehicles-this-week-centre-suspends-feebate-policy-
ahead-of-elections-and-more/
Gillingham, Kenneth, Sébastien Houde and Arthur
Van Benthem. 2019. Consumer myopia in vehicle
purchases: evidence from a natural experiment (No.
w25845). National Bureau of Economic Research.
https://doi.org/10.3386/w25845
21
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
References
Gillingham, Kenneth and Karen Palmer. 2014.
“Bridging the energy efciency gap: Policy insights
from economic theory and empirical evidence.”
Review of Environmental Economics and Policy,
8(1): 18–38. https://doi.org/10.1093/reep/ret021
Guerra, Erick. 2019. “Electric vehicles, air pollution,
and the motorcycle city: A stated preference
survey of consumers’ willingness to adopt electric
motorcycles in Solo, Indonesia.” Transportation
Research Part D: Transport and Environment, 68:
52–64. https://doi.org/10.1016/j.trd.2017.07.027
Gupta, Shivanshu, Neeraj Huddar, Balaji Iyer and
Timo Möller. 2018. “The future of mobility in India’s
passenger-vehicle market.” McKinsey & Company.
https://www.mckinsey.com/industries/automotive-
and-assembly/our-insights/the-future-of-mobility-in-
indias-passenger-vehicle-market#
Haaf, C. Grace, W. Ross Morrow, Inês ML Azevedo,
Elea McDonnell Feit, and Jeremy J. Michalek.
2016. “Forecasting light-duty vehicle demand using
alternative-specic constants for endogeneity
correction versus calibration.” Transportation
Research Part B: Methodological, 84: 182–210.
https://doi.org/10.1016/j.trb.2015.11.012
Helfand, Gloria and Ann Wolverton. 2011.
“Evaluating the Consumer Response to Fuel
Economy: A Review of the Literature”, International
Review of Environmental and Resource Economics,
5(2): 103-146. http://dx.doi.org/10.1561/101.00000040
Horowitz, Joel L. and Jordan J. Louviere.
1995. “What is the role of consideration sets
in choice modeling?” International Journal of
Research in Marketing, 12(1): 39–54. https://doi.
org/10.1016/0167-8116(95)00004-l
International Energy Agency. 2020. India Energy
Policy Review 2020 - Event. https://niti.gov.in/sites/
default/les/2020-01/IEA-India%202020-In-depth-
EnergyPolicy_0.pdf
Jaffe, Adam B. and Robert N. Stavins. 1994.
“The energy paradox and the diffusion of
conservation technology.” Resource and
Energy Economics, 16(2): 91–122. https://doi.
org/10.1016/0928-7655(94)90001-9
Janssens-Maenhout, Greet, Monica Crippa, Diego
Guizzardi, Marilena Muntean, Edwin Schaaf, Jos
GJ Olivier, J. A. H. W. Peters and Klara M. Schure.
2017. Fossil CO2 and GHG emissions of all world
countries (Vol. 107877). Luxembourg: Publications
Ofce of the European Union. doi:10.2760/709792
Khan, Mohd, Akbar Ali and Venkata
Madhusudan Rao Datrika. 2018. “Two-Wheeler
Consumers’ Behaviour Towards Customer
Satisfaction.” (Working paper). https://doi.
org/10.29121/granthaalayah.v6.i2.2018.1536
URL: https://papers.ssrn.com/sol3/papers.
cfm?abstract_id=3112798.
Koppelman, Frank S. and Chandra Bhat. 2006. “A
self instructing course in mode choice modeling:
multinomial and nested logit models.” URL: https://
www.caee.utexas.edu/prof/bhat/COURSES/LM_
Draft_060131Final-060630.pdf
Kumar, Rajeev Ranjan and Kumar Alok. 2020.
“Adoption of electric vehicle: A literature review and
prospects for sustainability.” Journal of Cleaner
Production, 253: 119911. https://doi.org/10.1016/j.
jclepro.2019.119911
22
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
References
Li, Shuyu and Qiang Wang. 2019. “India’s
dependence on foreign oil will exceed 90% around
2025 – The forecasting results based on two
hybridized NMGM-ARIMA and NMGM-BP models.”
Journal of Cleaner Production, 232: 137–153. https://
doi.org/10.1016/j.jclepro.2019.05.314
Lin, Ru-Jen, Kim-Hua Tan and Yong Geng. 2013.
“Market demand, green product innovation, and rm
performance: evidence from Vietnam motorcycle
industry.” Journal of Cleaner Production, 40: 101–107.
https://doi.org/10.1016/j.jclepro.2012.01.001
Mabit, Stephan L. 2014. “Vehicle type choice under
the inuence of a tax reform and rising fuel prices.”
Transportation Research Part A: Policy and Practice,
64: 32–42. https://doi.org/10.1016/j.tra.2014.03.004
Manski, Charles F. 1977. “The structure of random
utility models.” Theory and Decision, 8(3): 229.https://
doi.org/10.1007/bf00133443
Manski, Charles F. and Steven R. Lerman. 1977. “The
estimation of choice probabilities from choice based
samples.” Econometrica: Journal of the Econometric
Society: 1977–1988. https://doi.org/10.2307/1914121
Matsumoto, Shigeru and Yukiko Omata. 2017.
“Consumer valuations of energy efciency
investments: The case of Vietnam's Air Conditioner
market.” Journal of Cleaner Production, 142: 4001–
4010. https://doi.org/10.1016/j.jclepro.2016.10.055
NITI Aayog and Rocky Mountain Institute. 2017a.
India leaps ahead: transformative mobility solutions
for all. URL: https://niti.gov.in/writereaddata/les/
document_publication/RMI_India_Report_web.pdf
NITI Aayog and Rocky Mountain Institute.
2017b. Valuing Society First: An Assessment
of the Potential for a Feebate Policy
in India. URL: https://rmi.org/insight/
india-leaps-ahead-important-policy-lever-feebates/
Orlov, Anton and Steffen Kallbekken. 2019. “The
impact of consumer attitudes towards energy
efciency on car choice: Survey results from Norway.”
Journal of Cleaner Production, 214: 816–822. https://
doi.org/10.1016/j.jclepro.2018.12.326
Parry, Ian WH, Margaret Walls and Winston
Harrington. 2007. “Automobile externalities and
policies.” Journal of economic literature, 45(2): 373–
399. https://doi.org/10.2139/ssrn.927794
Phule, Unmesh. 2019. “India’s feebate policy on hold
as EV plans to be delayed till general elections.”
ReviewTech. URL: https://www.reviewtech.in/indias-
feebate-policy-on-hold-as-ev-plans-to-be-delayed-till-
general-elections/
Sathish, M. and A. Pughazhendi. 2011. “A study on
consumer behaviour of automobile products with
special reference to two-wheeler in Tirunelveli City
Tamil Nadu, India.: TVs, 19(17.5): 1–5. https://doi.
org/10.15373/2249555x/dec2011/55
Sharma, Yogima. 2019. “With polls around the corner,
Govt stalls EV feebate plan.” ET Auto. URL: https://
auto.economictimes.indiatimes.com/news/policy/with-
polls-round-the-corner-govt-stalls-7500-cr-ev-feebate-
plan/68263976
Sheldon, Tamara L. and Rubal Dua. 2018. “Gasoline
savings from clean vehicle adoption.” Energy
Policy, 120: 418–424. doi: https://doi.org/10.1016/j.
enpol.2018.05.057
Sheldon, Tamara L. and Rubal Dua. 2019a. “Drivers
of new light-duty vehicle eet fuel economy in Saudi
Arabia.” KS—2019-DP54. Riyadh: KAPSARC. doi:
https://doi.org/10.30573/KS--2019-DP54
23
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
References
Sheldon, Tamara L. and Rubal Dua. 2019b.
“Measuring the cost-effectiveness of electric vehicle
subsidies.” Energy Economics, 84: 104545. doi:
https://doi.org/10.1016/j.eneco.2019.104545
Sheldon, Tamara L. and Rubal Dua. 2019c.
Assessing the effectiveness of California's “Replace
Your Ride”. Energy Policy, 132, pp. 318-323. doi:
https://doi.org/10.1016/j.enpol.2019.05.023
Sheldon, Tamara L. and Rubal Dua. 2020.
“Effectiveness of China's plug-in electric vehicle
subsidy.” Energy Economics, 88: 104773. doi:
https://doi.org/10.1016/j.eneco.2020.104773
Society of Indian Automobile Manufacturers (SIAM).
2019. Market share of vehicles in India, by segment
in FY 2019. URL: http://www.siam.in/.
Statistical Research Department. 2020.
“India – two-wheeler market share by
manufacturer FY 2018 | Statistica.” URL:https://
www.statista.com/statistics/610445/
two-wheeler-market-share-by-manufacturer-india/
Swait, Joffre. 2001. “Choice set generation within
the generalized extreme value family of discrete
choice models.” Transportation Research Part
B: Methodological, 35(7): 643–666. https://doi.
org/10.1016/s0191-2615(00)00029-1
Timperley, Jocelyn. 2019. “The carbon brief
prole: India.” URL: https://www.carbonbrief.org/
the-carbon-brief-prole-india.
Train, Kenneth. 1985. “Discount rates in consumers'
energy-related decisions: a review of the literature.”
Energy (Oxford), 10(12): 1243–1253. https://doi.
org/10.1016/0360-5442(85)90135-5
Train, Kenneth E. 2009. Discrete choice methods
with simulation. Cambridge University Press. http://
dx.doi.org/10.1017/CBO9780511805271
Wang, Chen and Ricardo A. Daziano. 2015. “On the
problem of measuring discount rates in intertemporal
transportation choices.” Transportation, 42(6): 1019–
1038. ht t p s: //d o i.o r g/ 10.10 07/ s11116 - 015 - 9 65 3 -7
Wen, Chieh-Hua and Frank S. Koppelman. 2001.
“The generalized nested logit model.” Transportation
Research Part B: Methodological, 35(7): 627–641.
https://doi.org/10.1016/S0191-2615(00)00045-X
Ye, Liang and Quan Wang. 2011. “Case study of
motorcycle use and policy analysis in Huizhou,
China.” Journal of Transportation Engineering,
137(11): 831–836. https://doi.org/10.1061/(asce)
te.1943-5436.0000263
Yoo, Sunbin, Arum Cho, Faris Salman and
Yoshikuni Yoshida. 2020. “Green paradox: Factors
affecting travel distances and fuel usages, evidence
from Japanese survey.” Journal of Cleaner
Production, 273, 122280. https://doi.org/10.1016/j.
jclepro.2020.122280
Zhuge, Chengxiang, Binru Wei, Chunfu Shao,
Chunjiao Dong, Meng Meng and Jie Zhang. 2020.
“The potential inuence of cost-related factors
on the adoption of electric vehicle: An integrated
micro-simulation approach.” Journal of Cleaner
Production, 250, 119479. https://doi.org/10.1016/j.
jclepro.2019.119479
24
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Notes
25
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
Notes
26
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
About the Authors
Prateek Bansal
Rubal Dua
Rico Krueger
Dan Graham
Prateek Bansal is a postdoctoral research fellow at Imperial College London,
working primarily on Bayesian machine learning methods and causal inference
models with applications for transport systems. He holds a Ph.D. from Cornell
University, an M.Sc. degree from The University of Texas at Austin, and a
B.Tech. from the Indian Institute of Technology (IIT), Delhi.
Rubal is a research fellow at KAPSARC working on vehicle
regulatory policy and shared mobility research from the consumer
perspective. He holds a Ph.D. from KAUST, Saudi Arabia, an M.Sc. from
the University of Pennsylvania, and a B.Tech. from the Indian Institute of
Technology, Roorkee.
Rico is a postdoctoral researcher in the Transport and Mobility Laboratory
at Ecole Polytechnique Fédérale de Lausanne. His research focuses on the
development of methods at the intersection of machine learning, econometrics
and statistics with applications for transport systems. He holds a Ph.D. in civil
and environmental engineering from The University of New South Wales,
Australia.
Dan is a professor of statistical modeling in the department of civil and
environmental engineering at Imperial College London (ICL) and a director
of the ICL Transport Strategy Centre. He holds doctoral degrees from the
Department of Mathematics at Imperial College London and from the London
School of Economics.
27
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
About the Project
Promoting the adoption of energy-efcient vehicles has become a key policy imperative for both
developed and developing countries. Understanding the impact of various factors on the adoption
rates of energy-efcient vehicles forms the backbone of KAPSARC’s research into light-duty vehicle
demand. These factors include (i) consumer-related factors, such as demographics, behavior and
psychographics; (ii) regulatory factors, such as policies, incentives, rebates and perks; and (iii)
geo-temporal factors, including weather, infrastructure and network effects. As part of the future
of transport and fuel demand initiative, our team is currently developing models at different levels:
micro-level models using large-scale data comprising new car buyers’ proles, and macro-level
models using aggregated adoption data to understand and project the various factors that affect the
adoption rate of energy-efcient vehicles.
28
Are Consumers Myopic About Future Fuel Costs? Insights from the Indian two-wheeler market
www.kapsarc.org
