See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/265133245

Understanding Car-Buying Behavior: Psychological Determinants of Energy

Efficiency and Practical Implications

Article  in  International Journal of Sustainable Transportation · January 2015

DOI: 10.1080/15568318.2012.732672

CITATIONS READS

38 6,957

3 authors, including:

Roland W. Scholz
Danube University Krems, Faculty of Economics and Globalization
532 PUBLICATIONS   17,162 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Collective Mind (2017-2020) View project

Integrated Systems Analysis and Modelling to Support Socio-ecological Transitions of Smallholder Systems to Sustainability View project

All content following this page was uploaded by Roland W. Scholz on 24 February 2019.

The user has requested enhancement of the downloaded file.

 Journal of Cleaner Production 15 (2007) 1076e1084
www.elsevier.com/locate/jclepro

Reducing energy consumption in road transport through

hybrid vehicles: investigation of rebound effects, and
possible effects of tax rebates
Peter de Haan*, Anja Peters1, Roland W. Scholz2
ETH Zurich, Department of Environmental Sciences, Institute for Environmental Decisions, Universitaetstrasse 22,
CHN J75.2, 8092 Zurich, Switzerland
Accepted 28 May 2006
Available online 26 September 2006

Abstract

Hybrid powertrains are considered to be a promising method to raise the overall fuel efficiency of passenger cars. They are predicted to enter
the automobile markets in increasing numbers in the near future. This paper investigates 2 aspects relevant to the promotion of hybrid vehicles as
part of an energy reduction strategy. First, are hybrid cars accompanied by rebound effects, which counteract their increased fuel efficiency?
Second, do tax rebates indeed lead to higher sales numbers? We present results from a survey conducted with all 367 buyers of hybrid sec-
ond-generation Toyota Prius in Switzerland in the first 9 months after market entry (response rate, 82.6%). As control group, questionnaires
were also sent to 250 Toyota Corolla (61% returned) and 250 Toyota Avensis (52%) buyers. Among other items, the surveys asked for the
car being replaced, if any. Main results are that the increase in car size for hybrid car buyers was lower than both, the market trend and the
control group. Increase in car size was lower for hybrid car buyers than for both control group and market trend. 6% of hybrid car purchases
did not replace a previously owned vehicle (control group 3%, P ¼ 0.05), the Swiss market average being 20%. Hence rebound effects could be
identified neither for vehicle size nor for vehicle ownership. As an energy policy measure, hybrid vehicles are eligible for tax rebates in parts of
Switzerland. We found evidence that these tax rebates indeed lead to significant higher sales. We present a rough cost estimate of such tax
rebates as CO2 abatement policy tool.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Hybrid cars; Direct rebound effect; Survey; Fuel consumption; Vehicle size; Car ownership; Vehicle transaction; Energy-efficiency; Early adopters; Car
ownership tax; Toyota Prius; CO2 abatement cost

1. Introduction important source for energy savings [2,3]. As there is no doubt

that the challenges will increase within the next few decades,
The continuing increase in energy consumption, the mid- and as the re-investment cycles of the passenger fleet, the pro-
depletion point of conventional oil and of the re-concentration duction capacities, and infrastructures have to be considered,
of crude oil production in the Near East will compel industri- early actions and long-term perspectives by governments and
alized as well as developing nations to make more efficient use technology producers become increasingly important. The
of energy [1]. Technological improvements still are the most transportation sector accounts for 21.8% of total primary en-
ergy consumption worldwide in 2000, and will account for
ca. 34% in 2050 (OECD countries: 28.1% and 40%, respec-
* Corresponding author. Tel.: þ41 44 632 4978; fax: þ41 44 632 10 29. tively) [1]. In the European Union (EU), road transport at pres-
E-mail addresses: [email protected] (P. de Haan), [email protected].
ch (A. Peters), [email protected] (R.W. Scholz). ent is the second-largest sector of energy consumption, but is
1
Tel.: þ41 44 632 6676; fax: þ41 44 632 10 29. one among the fastest growing sectors regarding CO2 emis-
2
Tel.: þ41 44 632 5891; fax: þ41 44 632 10 29. sions, with an increase of 20% between 1990 and 2000.

0959-6526/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jclepro.2006.05.025
 P. de Haan et al. / Journal of Cleaner Production 15 (2007) 1076e1084 1077

Various policy measures have been introduced in many rebound effects are areas of ongoing research [13,14]. Gener-
countries, and are considered for implementation in other ally, 3 different rebound effects might be induced [2]: in-
countries, to reduce CO2 emissions from individual road trans- creased demand for the same service as it has become
port. For this there are 3 main routes: (i) by reducing demand, cheaper (direct rebound effect), increased demand for other
i.e., the number of passenger kilometers; (ii) by reducing the services as money (i.e., purchasing power) has become avail-
energy intensity per passenger kilometer; and (iii) by reducing able (indirect rebound effect; also called secondary rebound
the CO2 intensity of energy. The latter can be achieved by the effect), and structural effects on larger parts of the economy
use of nuclear, alternative fossil (diesel and natural gas), or re- due to changed demand, production and distribution patterns
newable energy sources; it is not the topic of the present paper. (macro-scale rebound effect; also called economy-wide or ter-
Demand reduction, on the one hand, can be achieved with tiary rebound effect).
higher fuel taxes (energy tax or carbon tax), toll, congestion For example, if the energy efficiency of a car is increased
tax [4], road pricing [5], and also by subsidizing mass transit by technological innovations, 100 km can be driven with
(public transport). Since owning a vehicle is a good predictor less fuel and hence at a lower cost. This lower cost should
for vehicle use, reduction in the number of vehicles (through have the consequence that people demand more ‘‘car ser-
increases in sales tax or ownership tax) is often also regarded vices’’, i.e., drive more often, and/or drive longer, and/or drive
as an effective measure to reduce induced demand. with larger cars. Identification of occurrence, and, if present,
Reduction of energy intensity per passenger kilometer, on quantification of rebound effects are generally not straightfor-
the other hand, can be achieved by the following: ward. Most work has been done on the effects of the introduc-
tion of energy-saving technologies, e.g. space heating [15].
e increasing occupancy (promotion of car-pooling by privi- Rebound effects induced by costs savings were the first to
leges, e.g. high-occupancy vehicle lanes and high- be investigated and originated in economics, especially energy
occupancy/toll lanes [6,7]); economics. In close analogy, also the reduction of socio-
e reduction of energy-intensive traffic conditions like conges- psychological costs of ownership might be regarded as possi-
tion [8,4] and high-speed driving above 90e100 km h1; ble driver for rebound effects. For example, it may not be the
e reduction in car size, and/or engine downsizing; financial but the socio-psychological cost-of-ownership (due
e improvements to the internal combustion engine itself, or to neighborhood pressure, norms of a peer group, etc.) that
switch to other concepts like fuel cells [9]; prevents some people from buying sport-utility vehicles
e increasing the overall efficiency of the car: reduction of (SUVs). However, this could change as soon as SUVs with
energy consumption by auxiliaries, reduction of drag, im- hybrid powertrain enter the market. Of course, for other people
proved tire pressure, synthetic lube oil, etc.; and with different cultural and/or social backgrounds, owning an
e the use of hybrid powertrains [10] to recuperate braking SUV might instead be positive (e.g. as an expression of
energy. wealth), but this is not likely to be affected by the market entry
of hybrid SUVs.
Hybrid powertrains are energy-efficient since in addition to The present paper aims at identifying the occurrence, and,
recuperation, they also allow for an idle-off strategy, a higher if present, at quantifying 2 kinds of direct rebound effects that
share of engine modes with high thermodynamic efficiency, could possibly occur when buying hybrid cars: (i) people
and for engine downsizing (due to short-term additional power could tend to switch from small and/or already fuel-efficient
from the electric engine). Therefore, hybrid powertrains are cars to the new hybrid car, and (ii) average household vehicle
considered to be a promising option to decrease fuel consump- ownership (HVO) could increase if people either tended to
tion by passenger cars [11]. The few hybrid car models cur- purchase the hybrid car as an additional household vehicle
rently available have increasing sales numbers, and most without disposing of an already owned vehicle, or because
large car manufacturers have announced hybrid cars for the the availability of hybrid cars could make the decisive differ-
coming years. In 2004, 8 vehicle models utilizing hybrid- ence for a household to purchase a vehicle for the first time. In
electric powertrains were on the U.S. market and had sales of addition, we analyze the impact of car ownership tax rebates
88,000 units (0.52% market share) [12] forecast that by 2011, on the number of hybrid vehicles sold.
38 models will total to 535,000 sold units and 3% of U.S. sales. The paper is structured as follows. Section 2 presents the
Hybrid cars are an energy-efficient technology, however, research objective and research questions. The surveys are in-
from a macroeconomic point of view the introduction of troduced in Section 3. Section 4 investigates the presence of
more efficient products is often accompanied by rebound ef- the first rebound effect as mentioned above, possible above-
fects (also called take-back effects). If a product or service be- average increase in car size for hybrid car buyers. Section 5
comes more efficient (regarding energy use or the use of some analyses car transaction behavior, and investigates the pres-
other resources), it will also become cheaper: higher energy- ence of the second rebound effect, possible above-average in-
efficiency in production would mean lower sales price and creases in vehicle stock for hybrid car buyers. Section 6
hence increased demand. Higher energy-efficiency in the use provides an estimate of efficiency of car ownership tax rebates
phase would mean lower operation costs and hence increased for hybrid cars. Section 7 presents our concluding remarks
demand. This counteracts the positive effect of increased with regard to the rebound effects under investigation, and
efficiency. The definition, identification and quantification of consequences for future research.
1078 P. de Haan et al. / Journal of Cleaner Production 15 (2007) 1076e1084

2. Possible side effects of hybrid car purchases a hybrid vehicle out of quasi-rational cost reasoning where
in fact the economic analysis would suggest to do otherwise.
2.1. Possible increase in car size We adopt empty vehicle weight (curb weight) as surrogate
for ‘‘car size’’. For roughly the last 20 years, average empty
The first objective of the present study was to investigate vehicle weight of new car registrations has increased in
whether the following so-called direct rebound effect can be Europe (including Switzerland) and Northern America. A
detected, and, if so, can be quantified: an above-trend increase rebound effect can only be considered as present if an increase
in car size for hybrid vehicle buyers (i.e., the new hybrid ve- in empty vehicle weight (defined as the weight of the new car
hicle is larger than the average previously owned vehicle minus the average weight of previously owned vehicles which
that has been disposed off). If the use of a car costs less, peo- are replaced by a new car) is higher than this general trend.
ple will use more car services. This is called the direct rebound
effect (e.g. [2,9,16]), sometimes also called take-back effect. 2.2. Possible increase in vehicle ownership
An increased demand for ‘‘car services’’ can mean that
consumers either use the car more often (more trips, e.g. in- The second aim of the present paper is the detection, and, if
stead of public transport), or longer (driving longer trips, present, quantification of a second direct rebound effect: an
etc.), or that they use a larger car (here, larger might be either increase in average household vehicle ownership (HVO),
an increase in size, or an increase in relative power, or both). A i.e., that hybrid vehicles more often than conventional cars
highly fuel-efficient car will have lower operational costs, serve as an additional vehicle, thereby increasing the number
comparable mainly to cars which are much smaller than the of vehicles owned by a given household.
hybrid vehicle in question. So it could be expected that a cer- Again, as for possible increases in car size, there of course
tain percentage of hybrid vehicle buyers trade in a car belong- is no economic driver for any increase in car ownership due to
ing to a smaller segment (smaller car size). Though this might the mere presence of hybrid vehicles on the market, as hybrid
be intuitive, it should be noted that there is no economic driver vehicles by construction are at least as expensive as conven-
for such a behavior: the sales price for a hybrid vehicle is tional cars. So we should expect that if the above rebound ef-
higher than for a vehicle with conventional powertrain. fect is present, the reasons will be of socio-psychological
The surplus sales price of hybrid vehicles at present seems nature. Possibly the introduction of the Toyota Prius as the first
to be chosen by the manufacturer such that it compensates fully functional hybrid vehicle leads to surplus purchases in
partly if not completely for the future fuel reduction. In the those population groups for which owning a second (or third,
U.S., the Prius sells at EUR 3300 (using a USD/EUR exchange etc.) car is connected with socio-psychological costs, which
rate of 1.06) more than the Toyota Corolla (1.6 l VVT-i en- could be lower for fuel-efficient vehicles like hybrids.
gine, Linea Sol, 5-door, automatic transmission, hatchback), Even though it is simple to conclude for an entire popula-
which is not fully paid back over a vehicle lifetime of tion whether average HVO has increased or not, this same
250,000 km at current U.S. gasoline prices [17]. In Switzer- question is hard to answer for a sample of new car buyers
land, where fuel prices (EUR 0.9e1.0 per liter) are higher that answered a mail-back survey. In any population, young
than in Northern America but lower compared to surrounding households or individuals will at sometime purchase a vehicle
European countries, surplus price is EUR 4800, which does for the first time. And in correspondence, old households or in-
pay back after ca. 150,000 km, which is the average vehicle dividuals will at some point in life discard their last vehicle. So
lifetime in Switzerland. In Finland, where fuel prices are seen from the viewpoint of an individual, there is always
EUR 1.3 per liter gasoline, the price difference (before adding a first-time car purchase at first, followed by several car re-
the Finnish car acquisition tax of almost þ100%) is EUR placements, and possibly accompanied with the first-time pur-
8000, which does pay back after ca. 180,000 km. So the price chase of a second car, etc. One approach to test whether HVO
difference seems to have been set market-specific, compensat- increases for a given population of new car buyers is through
ing for expected reduction of (market-specific) fuel costs over simulation of the average car market. This is done in Ref. [16]
the vehicle’s (market-specific) lifetime. for the Toyota Prius. Another approach, adopted in the present
In the absence of an economic driver for a possible rebound paper, is to compare with a representative control group. We
effect, de Haan et al. [16] introduced the concept of a socio- will compare vehicle transaction behavior of Prius buyers
psychological driver that might still be present. Owning with that of Corolla and Avensis buyers.
a car is associated with socio-psychological ‘‘costs of car own-
ership’’, which would be reduced in the case of a hybrid vehi- 2.3. On the Toyota Corolla, Prius, and Avensis
cle, generally being accepted as an environmentally friendly
technology. The Toyota Prius is the first hybrid vehicle that reached
Another kind of socio-psychological reason could be that a sales volume of over 100,000 units, and that is successful
not all cost factors are perceived equally. It is generally ac- in all 3 major car markets, such as Japan, U.S., and Europe.
cepted that car drivers tend to overweight fuel costs, but to un- It was first introduced in Japan in 1997, with a redesign of
derweight running costs (lube oil, tire replacements, repairs, the batteries by the year 2000. In 2004, the second-generation,
etc.), car ownership taxes, and investment costs. This asym- reworked Prius entered the market. Only the year 2004 version
metrical cost perception might lead individuals to purchase of the Prius, called Prius 2 among experts, is the topic of the
 P. de Haan et al. / Journal of Cleaner Production 15 (2007) 1076e1084 1079

Table 1
Comparison of Toyota Corolla, Prius and Avensis (for Corolla and Avensis, numerous configurations can be ordered; we have selected the most sold engine
configuration of the body type being most comparable to the Prius, which is available as 4-door liftback sedan only)
Toyota Corolla Linea Sol Toyota Prius Toyota Avensis Linea Sol
Engine (fuel type) 1.6 VVT-i (gasoline) 1.5 VVT-i HSD (gasoline hybrid) 2.0 VVT-i (gasoline)
Body type, no. of doors Hatchback sedan, 4 þ 1 doors Liftback sedan, 4 þ 1 doors Liftback sedan, 4 þ 1 doors
Curb weight (kg) 1230 1375 1430
Length (m) 4.18 4.45 4.66
Rated power (kW) 81 57 þ 50a 108
Acceleration time 0e100 km h1 (s) 10.2b/11.8c 10.9d 9.1b/9.9c
CO2 per km 168b/190c 104d 191b/221c
Recommended sales price including taxese EUR 18,440b/EUR 19,330c EUR 24,140 EUR 24,000b/EUR 25,540c
a
Short-term additional power from electric engine.
b
Manual transmission, 5 gears.
c
Automatic transmission, 4 gears.
d
Continuously varying transmission (CVT).
e
In Switzerland in September 2004 (time of Prius survey), applied CHF/EUR exchange rate 1.57 (data source: Toyota 2005).

present study. The Prius is available with hybrid powertrain (compared to Western Europe) and by the share of diesel-
only, no version exists that would be equipped with a conven- powered passenger cars (Fig. 2, right-hand scale). The Swiss
tional internal combustion engine only. As control group, we diesel fleet is at a clearly lower level compared to the sur-
therefore, looked at buyers of the next-smaller Toyota car (Co- rounding countries, due to the fact that diesel fuel did not ben-
rolla) and next-larger (Avensis). Technical characteristics and efit from a reduced tax level as in most European countries.
sales prices of the Prius are listed in Table 1, together with fig- A sharp increase in the percentage of diesel passenger cars
ures on those configurations of Corolla and Avensis that can be since 2000 can be identified, however, and is forecast to reach
best compared to the Prius. In Switzerland in 2004, the Prius 2 and level off at some 50% by the year 2020. The share of all-
sold at about the same price as a Toyota Avensis (Table 1) and wheel driven cars (Fig. 2, left-hand scale) also is much higher
roughly EUR 4800 more as a comparable Toyota Corolla, (almost 20% compared to Western Europe with a steady in-
which is sometimes regarded as being the conventional coun- creasing share that reached 7% in 2003), which is probably af-
terpart to the hybrid Prius 2 from a technical perspective [17]. fected both by the wealth level and the presence of mountains.
It should be noted, however, that Prius and Corolla are distinct Hence, the Swiss car market cannot be regarded as being
cars and do not correspond in all respects. We argue that on the representative for Europe, and therefore, it is not a priori clear
automobile market, Prius is rather to be positioned between whether our results may be generalized.
Corolla and Avensis, regarding curb weight, regarding vehicle
length as surrogate for ‘‘size’’, and regarding acceleration 3. Surveys of new car buyers
time. It therefore, is hard to tell at what price a non-hybrid,
conventional Prius would sell. In most automobile magazines, Three separate surveys were performed. For each, the ad-
the Prius is classified as ‘‘mid-size’’ together with Avensis, dresses of new car buyers were kindly made available by
whereas the Corolla is mostly referred to as ‘‘lower mid- Toyota Switzerland. In September 2004, all Swiss buyers of
size’’ (other names are ‘‘compact class’’ or ‘‘B segment’’). the hybrid Toyota Prius 2 since its market entry in May

2.4. On the Swiss car market 2.1 110

left scale: avg. capacity [liter] Switzerland Western Europe average D/F/I/A
right scale: avg. power [kW] Switzerland Western Europe average D/F/I/A
Switzerland belongs to the richest countries in the world, 2.0 100
and among other sectors this affects the car market. Due to
the high Swiss gross domestic product per capita, passenger 1.9 90

cars bought in Switzerland are on average larger and have

higher engine capacity. As shown in Fig. 1 (right-hand scale), 1.8 80

the average new registered car in Switzerland has a higher

rated power than in Western Europe, the difference roughly 1.7 70

being a constant of 20 kW over the last 15 years. Also shown

1.6 60
are the market statistics for those countries neighboring Swit-
zerland (Germany, France, Italy, and Austria), whose popula-
1.5 50
tion-weighted average is very similar to the whole of 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Western Europe. The engine capacity (Fig. 1, left-hand scale)
Fig. 1. Evolution of engine capacity (in liters; left scale) and power (kW; right
reflects the same trend, with a difference of 300 ccm (slightly scale) of new passenger car registrations in Switzerland, Western Europe, and
decreasing) between Switzerland and the rest of Europe. This Switzerland’s neighboring countries D/F/I/A (weighted by car sales). Data
can for the smaller part be explained by the lower fuel taxes source from Ref. [21].
1080 P. de Haan et al. / Journal of Cleaner Production 15 (2007) 1076e1084

24
left scale: % 4x4 vehicles Switzerland Western Europe
60 whether an old vehicle had been disposed off (and if yes, what
average D/F/I/A
right scale: % diesel cars Switzerland Western Europe average D/F/I/A kind of car), and which other vehicles are owned by the house-
20 50 hold. Block B consists of 8 questions on car choice behavior
and Block C has 5 knowledge questions on the Toyota Prius.
16 40
Block D addresses the mobility behavior (7 questions on sea-
son ticket ownership for public transport, parking space avail-
12 30
ability, and commuting behavior). The remaining 7 questions
of block E serve socio-demographic statistics. The complete
8 20
questionnaire (in German or in French) may be obtained
4
from the corresponding author upon request. The present paper
10
deals with block A (Sections 2e5) and block C (Section 6)
0 0
only. We will report our findings on motives and preferences
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
of purchasers of hybrid cars (blocks B and D) elsewhere.
Fig. 2. Evolution of the market shares of 4-wheel driven (left scale) and diesel- In block A, people were not only asked to report brand and
powered (kW; right scale) new passenger car registrations in Switzerland, model name, but also engine capacity (in liters), gear type (au-
Western Europe, and Switzerland’s neighboring countries D/F/I/A (weighted tomatic or manual), fuel type (gasoline or diesel), and model
by car sales). Data source from Ref. [21]. year (vintage). These data allow for a precise identification
of any vehicle in the Swiss database on vehicle type registra-
2004 received a mail-out mail-back survey (for costs reasons, tions and provide us with all technical characteristics includ-
the Italian-speaking part of Switzerland was omitted from the ing size, weight, and fuel consumption of the vehicle. There
survey). Questionnaires were not numbered or otherwise iden- is also a redundancy such that missing data can often be filled
tifiable, and no reminder letters were sent out. Out of 367 in. For example, fuel type can in most cases be derived from
questionnaires, 303 (82.56%) were returned. engine capacity (and the other way round: imprecise engine
In May 2005, 2 other surveys of 250 Toyota Corolla and capacity can be corrected for using fuel type), automatic
250 Toyota Avensis buyers were performed. Their postal ad- gear often has not been sold for smaller engine sizes, etc.
dresses were randomly sampled from all buyers of the respec- The data quality of the returned questionnaires was unusu-
tive cars in the non-Italian speaking parts of Switzerland in 6 ally high and allowed for the precise identification of over
months before April 2005. 97% of all vehicles that were owned by the households prior
In the first 12e18 months after market entry in spring 2003, to the purchase of the Prius 2. For the remaining 2.9% of ve-
the waiting lists for the Prius were very long (often 9 months, hicles, we assigned the most sold car type for the brand, model
sometimes reaching 12 months). This likely discouraged many and vintage specified.
potential buyers. We therefore have chosen to conduct the sur-
veys on Corolla and Avensis buyers at a later time, in spring 4. Analysis of car size evolution
2005, because the waiting time between order and delivery
for the Prius in Switzerland was less pronounced in the period For the 3 car types, 153, 303, and 130 questionnaires (for Co-
from October 2004 to April 2005 (3e4 months on average), rolla, Prius, and Avensis, respectively) were used for the present
and was less likely to discourage potential buyers. analysis. Table 2 lists averages of engine capacity, rated power,
For both Toyota Corolla and Avensis, the market average of length and curb weight, alongside with fuel consumption (com-
builds, fuel types and gearbox types was presented in the sur- mon figure used in Europe) and fuel efficiency (common in
vey: both hatchback and station wagon models, both gasoline Northern America). We investigate whether car size increases
and diesel engines, and both manual and automatic gearboxes. more for hybrid car purchasers as compared to the control group
The monospace builds (marketed as Corolla Verso and Avensis of conventional Corolla and Avensis purchasers. As surrogate
Verso) were excluded, as they, in fact, are distinct car models. for ‘‘car size’’ we use both empty vehicle weight and empty ve-
Note that the Toyota Prius is on the market only as a gasoline hicle length. Table 3 lists the differences between those cars hav-
hybrid with hatchback and continuously variable transmission ing been replaced and the cars that have been replacing them. As
(CVT) (not to be confused with an automatic gearbox). can be seen, curb weight increased by 8.8 kg for Prius buyers.
Toyota Corolla and Avensis have been selected because they This is clearly below the before-mentioned market trend, and
are those non-hybrid vehicles on the market that are most sim- significantly lower than for the control group (54.4 kg,
ilar to the Prius. It can be argued that a large percentage of Co- P ¼ 0.01). Fig. 3 depicts the distribution of curb weight classes
rolla and Avensis buyers could actually have opted for the Prius, of those vehicles having been replaced. The chart shows that
as the following conditions are fulfilled: (i) need or willingness from this perspective, the Prius is closer to the Avensis.
to purchase a new car; (ii) availability of the necessary funds in The length of a vehicle in general shows a good correlation
the order of magnitude of EUR 20,000e27,000; (iii) Toyota as with car size segmentation. No significant increase in vehicle
acceptable brand; and (iv) looking for a mid-sized car. length could be observed (67 mm for Prius consumers, com-
The questionnaire consisted of 5 blocks with a total of 33 pared to 100 mm for the control group, P ¼ 0.256). Fig. 4 de-
closed questions and 2 additional open questions of general picts the distribution of vehicle length classes for the replaced
scope. Item block A (7 questions) deals with the car purchased, cars. Here, Prius figures are in between of Corolla and Avensis.
 P. de Haan et al. / Journal of Cleaner Production 15 (2007) 1076e1084 1081

Table 2
Average technical characteristics of the cars being replaced and the new cars having been bought
Corolla buyers Prius buyers Avensis buyers
Replaced car New Corolla Replaced car New Prius Replaced car New Avensis
Number 137 153 269 303 118 130
Engine capacity (ccm) 1672.3 1608.4 2021.7 1497.0 2010.0 2055.2
Rated power (kW) 64.7 86.8 100.1 57.0 96.5 106.1
Length (mm) 4216 4220 4383 4450 4451 4666
Curb weight (kg) 1194.0 1231.3 1366.2 1375.0 1353.3 1430.3
CO2 (g/km) 187.0 177.4 203.7 104.0 206.1 203.3
Fuel consumption (litre/100 km) 7.82 7.39 8.47 4.30 8.62 8.42
Fuel efficiency (U.S. mile per U.S. gallon) 30.3 32.0 27.9 55.0 27.4 28.1
Model year/vintagea 1995.5 2005.0 1996.7 2004.0 1997.4 2005.0
Annual mileage (km/a)b 13,981 15,704 19,350
Fuel type
Gasoline (%) 98 96 94 100 96 86
Diesel (%) 2 4 6 4 14
Gear type
Manual (%) 82 70 50 63 39
Automatic (%) 18 30 42 37 61
Gearless transmission (%) 0 0 8 100 0 0
Car shape
Station wagon (%) 36 17 20 19 57
Limousine/hatchback (%) 64 83 76 100 79 43
Cabrio (%) 0 0 4 2 0
a
Model year based on 257 (Prius), 133 (Corolla), and 116 (Avensis) vehicles.
b
Annual mileage based on 263 (Prius), 135 (Corolla), and 116 (Avensis) vehicles.

Hence, in summary, we conclude that no increase in car The same holds true for Avensis cars, the predecessors of which
size for hybrid car purchasers can be observed, on the contrary, were 7.6 years of age, but have been replaced by new Avensis
there are some indications that increase in car weight and cars with a rated CO2 emission level that is only 2.8 g km1
length are below-average for Toyota Prius buyers. lower, whereas the market average dropped by 22 g km1
On average, the vehicles being replaced by hybrid Prius over the time frame of 7.6 years. These differences arise
had a CO2 emission level of 203.7 g km1, compared to from the fact that most individuals increase car size with
104 g km1 for the Prius. On average for new registrations in each purchase, which, however, does not show up in the market
Switzerland, CO2 emissions dropped from 214.9 in the year average as young consumers fade in and old consumers fade
1996 to 191.9 g km1 in the year 2004 (test cycle: NEDC) out. So in order to assess possible above-average increases in
[18]. This corresponds to a decrease of 2.875 g km1 per year. car size for hybrid car purchases, one should make use of
It should be noted that for Corolla buyers, CO2 emissions control groups of conventional new car buyers. Comparing
dropped from 187 to 177.4 g km1, where the cars were being the increase in car size with market trend data, as done for ex-
replaced on average age of 9.5 years. This means that CO2 ample by the authors in Ref. [16], in fact is too strict: it is a suf-
emissions of the new Corolla cars are higher than expected. ficient but not a necessary proof of absence of a rebound effect.

Table 3
Mean differences in technical characteristics between replaced and new cars 35%
cars replaced by…
Difference between new Control group 30% Corolla
and replaced vehiclea Prius
Corolla Avensis Joined Prius Pb Avensis
25%
Number of cases 137 118 255 269
Engine capacity (ccm) 63.9 þ45.3 15.0 524.7 0.000** 20%
Rated power (kW) þ22.2 þ9.6 þ16.4 43.1 0.000**
Length (mm) þ4 þ215 þ100 þ67 0.256 15%
Curb weight (kg) þ37.3 þ77.0 þ54.4 þ8.8 0.010**
10%
CO2 (g/km) 9.6 2.8 6.3 99.7 0.000**
Fuel consumption (lit./100 km) 0.43 0.20 0.32 4.17 0.000** 5%
Fuel efficiency (U.S. mile per þ1.7 þ0.7 þ1.2 þ27.1 0.000**
U.S. gallon) 0%
Car age at time of replacement 9.5 7.6 8.6 7.3 0.467 <1000 1000– 1100– 1200– 1300– 1400– 1500– 1600– 1700– >1800
1100 1200 1300 1400 1500 1600 1700 1800
a
Only respondents considered that actually replaced a previously owned empty vehicle weight [kg]
vehicle.
b
2-Sided T test assuming not-equal variances (**significance 1% level, Fig. 3. Empty vehicle weight distribution of previously owned vehicles having
*significance 5% level). been replaced by Corolla, Prius, and Avensis buyers.
1082 P. de Haan et al. / Journal of Cleaner Production 15 (2007) 1076e1084

60% e First-time purchase: the household did not previously own

cars replaced by…
Corolla
a vehicle;
50% Prius e Stock increase purchase: the household already owned one
Avensis
40%
or more cars, of which none is disposed off, so the vehicle
stock increases by 1 vehicle.
30%
As can be seen, the percentage of so-called replacement
20% purchases is similar for all groups, with the Prius percentage
of 90.0% being between Corolla and Avensis percentages
10%
(89.5% and 90.8%, respectively; P ¼ 0.955 for a two-sided
Student’s t-test assuming non-equal variances). A difference
0%
<3.5 3.5– 3.9– 4.3– 4.7– >4.9 between hybrid and conventional cars can be found, however,
3.9 4.3 4.7 4.9
for the percentage of first-time purchases (only 3.3% for the
vehicle length [m]
Prius, but 7.2% for Corolla and 6.2% for Avensis; statistically
Fig. 4. Vehicle length (in millimeter) distribution of previously owned vehicles not significant with P ¼ 0.062). Correspondingly, the number
having been replaced by Corolla, Prius, and Avensis buyers. of cases where no vehicle has been disposed off is 3.3%
for Corolla and 3.1% for Avensis, but 6.7% for Prius
5. Analysis of household vehicle ownership evolution (P ¼ 0.050). This is not a proof of increasing HVO for the fol-
lowing reason: the Toyota Prius was very successful that the
In this section we investigate whether the vehicle transaction waiting time between order and actual delivery of the car
behavior of hybrid car purchasers does lead to an above- was up to 12 months for orders taken in the first 9 months after
average increase in average HVO, i.e., in the number of cars market entry. This of course made it impossible to purchase
per household on average. We again compare hybrid Prius a Prius for those potential buyers who were in urgent need
buyers with conventional Corolla and Avensis buyers. of a car. This fact, in our opinion, probably accounts in full
As discussed in the preceding section, the viewpoint of the for the difference in vehicle purchase behavior. This is also il-
individual is distinct from the market average. Whereas the in- lustrated by the Swiss market average of the percentage of new
dividual consumer will, on average, increase the number of ve- car purchases that do not immediately replace a previously
hicles owned up to the age of, say, 40e50 years, and will then owned vehicle, which is 20% [16], i.e., much higher than
decrease again, this individual pattern will not show up in the the corresponding figures for either the Corolla, Prius or Aven-
market average, as consumers fading in and consumers fading sis buyers (Table 4). There is a significant difference in aver-
out of the market will cancel each other out. age HVO after purchase (1.65 for hybrid buyers, 1.43 for the
For the vehicles having been replaced (technical characteris- control group, P ¼ 0.001). This can be related to higher in-
tics of which are listed in Table 2), Table 4 lists the vehicle come and education levels for hybrid buyers [22].
transaction behavior. We distinguish 3 types of vehicle
transactions: 6. Effect of car ownership tax reductions

e Replacement purchase: an old vehicle is disposed off; the In Switzerland, annual car ownership tax is regulated not at
vehicle stock remains constant; the federal level, but by the 26 cantons separately. Both tax
Table 4
Statistics on household vehicle ownership after the purchase of new Corolla, Prius, or Avensis car (#hh ¼ number of households)
b
Replacement # cars after purchase Control group P
a
Corolla buyers Avensis buyers Joined Prius buyers
# hh % # hh % # hh % # hh %
Disposal of previously owned vehicle 1 92 69 161 146
2 38 40 78 96
3 4 89.5 9 90.8 13 90.1 18 89.3 0.955
4 2 0 2 6
5 1 0 1 1
First-time purchase (no prev. veh.) 1 11 7.2 8 6.2 19 6.7 9 3.0 0.064
No vehicle disposed 2 5 4 9 15
3 0 0 0 4
3.3 3.1 3.2 7.7 0.050
4 0 0 0 3
5 0 0 0 1
Sum 153 100 130 100 283 100 299 100
a
For this table, 4 commercial fleet owners of Prius with over 30 vehicles each have been excluded.
b
2-sided T test assuming non-equal variances.
 P. de Haan et al. / Journal of Cleaner Production 15 (2007) 1076e1084 1083

level and tax base (either curb weight, engine capacity, rated of roughly EUR 107 per ton CO2. This figure is tentative only,
power, emission level, curb weight þ maximum payload, or as it might be argued that it is an underestimation but also that
a combination thereof) vary among the cantons. In September it is an overestimation. Since tax rebates also lead to additional
2003, 6 cantons had introduced tax reductions for environmen- indirect effects (increased awareness for fuel efficiency in the
tally friendly cars so that the Toyota Prius could benefit from public debate, etc.) that are not included in the above estima-
them: Geneva and Basel-Land have a tax rebate of 100% (lim- tion of cost efficiency, the actual figure could be lower. On the
ited to 3 years in Geneva). Lucerne offers an annual reduction other hand, it can be assumed that cantons which have a higher
of EUR 30. Zurich and Basel-Stadt have reductions of 50% share of urban population, i.e., with above-average education
and 10%, respectively. The canton of Ticino goes beyond levels, are also more likely to adopt hybrid-friendly tax re-
the other cantons by offering a cash reward for the purchase bates. Hence part of the higher sales numbers in cantons
of highly fuel-efficient (or electric) vehicles, and offers EUR with tax rebates might be attributable to the general environ-
1330 in the case of the Toyota Prius. mental values of the population and not be causally related
Block C of our questionnaire also asked, in the case of to the tax rebate itself, so the actual figure could be higher
Prius buyers only, in which canton the Prius buyer lives, and than estimated above.
whether he or she benefits from a car ownership tax rebate.
This allows us to give a rough, tentative estimate whether 7. Discussion, conclusions, and outlook
such tax rebates are an effective and efficient policy measure.
As our questionnaire was not sent to the Italian-speaking can- The goals of our research were to investigate the possible
ton of Ticino, this analysis accounts for 5 out of the 6 cantons presence of 2 rebound effects possibly associated with hybrid
listed above. cars: people could tend to upgrade from small or already fuel-
Out of 296 questionnaires with valid answers (7 question- efficient vehicles to the hybrid car, or households could tend to
naires had missing values on the canton), 120 (40.5%) lived increase the number of cars owned. We conducted a survey
in one of the 5 cantons with a tax rebate of some kind. The with all Swiss buyers of the hybrid Toyota Prius 2 since its
population share of these 5 cantons with respect to the whole market entry in 2004 (303 out of 367 questionnaires were
of Switzerland without Ticino is 35.2%. All Corolla and Aven- returned), and, as control group, surveys with 250 Toyota
sis (34.6%) from our control group were sold to these cantons. Corolla and 250 Toyota Avensis new car buyers, randomly
This amounts to a Prius purchase density of 0.0484 per inhab- sampled (153 Corolla and 130 Avensis questionnaires were
itant for cantons with tax rebate, and 0.0385 for cantons with- returned). Official registration data (including fuel consump-
out such a rebate. In other words, these tax rebates seem to be tion, size and empty vehicle weight) were matched to the
associated with a Prius sales level that is 25.7% higher. We replaced vehicles owned by the surveyed households prior to
tested this result for statistical significance using bootstrap the purchase of the new car.
resampling methods as proposed in Ref. [19]. The two-sided Our results show that neither the first nor the second re-
95% confidence interval for the ratio of Prius buyers living bound effect investigated could be detected. Increase in vehicle
in cantons with tax rebates, 40.5%, is [36.49%, 44.59%], length even was somewhat lower for hybrid buyers compared
which does not include the Swiss average of 35.2%. Com- to the control group (P ¼ 0.256, not significant). Increase in
puting the odds ratio (again based on a two-sided bootstrap vehicle curb weight was clearly lower for hybrid buyers
estimate using 100,000 samples) leads to a similar result: the (P ¼ 0.010). We conclude that hybrid car buyers on average
95% confidence interval for the odds ratio, 1.2582, is do not switch from small, already fuel-efficient cars to the
[1.0223, 1.5163] and hence does not include unity. new hybrid car.
Assuming an average duration of the tax rebate of 4 years Regarding vehicle ownership, the percentage of vehicle
(the canton of Geneva explicitly limits to 3 years), the average purchases where no previously owned vehicle had been dis-
tax loss for the government is about EUR 426 in total per Prius posed off, i.e., where the vehicle stock increased due to the
purchased (population-weighted average for the 5 cantons). To purchase of the new car, was twice as high for hybrid cars
compute the efficiency of these tax rebates, we have to realize in comparison to our control groups with conventional cars
that only 1 out of 5 Prius cars (corresponding to sales increase (P ¼ 0.050). We explain this with the extraordinary long wait-
of þ25%) would not have been purchased had the tax rebate ing time of up to 12 months for the Prius, which caused a pre-
not been present. A total tax loss of EUR 1658 is associated selection of potential buyers toward multi-car households.
with a net effect of 1 additional Prius in the market. Hence we conclude that the hybrid Prius was not associated
The market average in Switzerland of new gasoline car reg- with an increase in average HVO, confirming the findings of
istrations in 2004 corresponded to 196 g CO2 per km. Com- Ref. [16] while applying another method (survey of control
pared to this market average, the Prius with a CO2 emission group of conventional car buyers, vehicle ownership simula-
level of 104 g km1 will save 14.7 tons CO2 over 160,000 km tion model). Hybrid cars more often enter multiple-car house-
(Swiss average vehicle lifetime). holds (1.65 cars after purchase for hybrid buyers, control
Summarizing the above, rebates on car ownership taxes group 1.43, P ¼ 0.001), which can be explained by higher in-
seem to be effective, i.e., in our case an increased sales level come and education levels of hybrid buyers [22].
of Prius cars of 25% results. Their efficiency can be estimated We also analyzed the effect of car ownership tax rebates. It
at costs of EUR 1658 for 14.7 tons CO2, i.e., abatement costs could be shown that such tax rebates indeed do influence the
1084 P. de Haan et al. / Journal of Cleaner Production 15 (2007) 1076e1084

number of sales of hybrid vehicles. A rough estimate of the References

efficiency of such tax rebates indicates that abatement costs
are in the order of magnitude of roughly EUR 107 per ton [1] IEA, OECD. Energy to 2050: scenarios for a sustainable future. Paris:
CO2. This might seem rather high compared to carbon reduc- IEA Publications; 2003.
[2] Berkhout PHG, Muskens JC, Velthuijsen JW. Defining the rebound
tion measures in other economic sectors, but is competitive effect. Energy Policy 2000;28:425e32.
when compared to CO2 abatement policies focusing on road [3] Herring H. Energy efficiency e a critical review. Energy 2006;31:
transport. 10e20.
Overall, we conclude that hybrid cars indeed are suited to [4] Parry IWH. Comparing the efficiency of alternative policies for reducing
play a role, during the next 5 years, in energy policy schemes traffic congestion. J Public Econ 2002;85:333e62.
[5] Kim KS, Hwang K. An application of road pricing schemes to urban
aiming at reducing CO2 emissions from individual road trans- expressways in Seoul. Cities 2005;22:43e53.
port. Their very low fuel consumption does not motivate peo- [6] Dahlgren J. High-occupancy/toll lanes: where should they be imple-
ple to switch to larger cars or to buy additional cars. Hence mented? Transp Res A 2002;36:239e55.
hybrid vehicles are not only fuel-efficient on a per-vehicle ba- [7] Lam T. Evaluating value-pricing objects with both scheduling and route
sis, but constitute a technology effective in lowering overall choices. Reg Sci Urban Econ 2004;34:225e40.
[8] Harrington W, Krupnick AJ, Alberini A. Overcoming public aversion to
CO2 emissions. And the introduction of tax rebates for hybrid congestion pricing. Transp Res A 2001;35:87e105.
vehicles seems to be effective (significant higher sales num- [9] Nakata T. Energy modeling on cleaner vehicles for reducing CO2 emis-
bers in Swiss cantons having tax rebates), and also to be effi- sions in Japan. J Clean Prod 2003;11:389e96.
cient, with abatement costs per ton CO2 comparable to other [10] Nakata T. Analysis of the impact of hybrid vehicles on energy systems in
carbon reduction measures in the road transport field. Japan. Transp Res Part D 2000;5:373e83.
[11] Bitsche O, Gutmann G. Systems for hybrid cars. J Power Sources 2004;
In the near future, new cars equipped with hybrid technol- 127:8e15.
ogy will come to market in Europe. We intend to conduct anal- [12] JDP-LMC. Hybrid-electric vehicle outlook. Westlake Village CA,
ogous surveys for other hybrid cars, namely Honda Civic IMA U.S.A: J.D. Power-LMC Automotive Forecasting Services; 3 February
and Lexus RX400h. Especially the latter will be of interest 2005.
with regard to the postulated existence of rebound effects [13] Greening LA, Greene DL, Difiglio C. Energy efficiency and consumption e
the rebound effect e a survey. Energy Policy 2000;28:389e401.
due to reduced socio-psychological (rather than financial) [14] Grepperud S, Rasmussen I. A general equilibrium assessment of rebound
cost-of-ownership. effects. Energy Econ 2004;26:261e82.
Also, we plan to ask the Honda Civic and Lexus RX buyers [15] Haas R, Biermayr P. The rebound effect for space heating: empirical
(both hybrid and conventional powertrains) again, some 12 evidence from Austria. Energy Policy 2000;28:403e10.
months after purchase, how many miles they have driven the [16] de Haan P, Mueller M, Peters A. Does the hybrid Toyota Prius lead to
rebound effects? Analysis of size and number of cars previously owned
new car. This would allow for the investigation of the third by Swiss Prius buyers. Ecol Econ 2006;58:592e605.
possible rebound effect, increase of vehicle kilometers driven [17] Lave LB, MacLean HL. An environmentaleeconomic evaluation of
annually due to the fact that lower fuel consumption leads to hybrid electric vehicles: Toyota’s Prius vs. its conventional internal
cheaper operation cost. Ref. [20] hypothesizes that this re- combustion engine Corolla. Transp Res Part D 2002;7:155e62.
bound effect should in fact be present. [18] Auto-suisse. 9. Berichterstattung im Rahmen der Energieverordnung
über die Absenkung des spezifischen Treibstoffnormverbrauchs von
Personenwagen 2004. Bern, Switzerland: Auto-suisse, association impor-
Acknowledgements teurs suisses d’automobiles; 7 June 2005.
[19] Hanna SR. Confidence limits for air quality model evaluations, as esti-
Thanks go to Toyota Switzerland for financing the copying mated by bootstrap and jackknife resampling methods. Atmos Environ
1989;23:1385e98.
and mailing of the questionnaire to the Prius, Avensis and
[20] Binswanger M. Technological progress and sustainable development:
Corolla purchasers and for its translation from German to what about the rebound effect? Ecol Econ 2001;36:119e32.
French. Toyota Switzerland did not otherwise support this [21] ACEA. Association of European car manufacturers, <http://www.
research. acea.be>; 2005 [accessed 12.08.05].

View publication stats