Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: This research aims to quantify the effects of car sharing on car ownership, car use and CO2
Received 30 March 2016 emissions. The results are based on a survey amongst 363 car sharing respondents in the
Received in revised form 25 January 2017 Netherlands. We found over 30% less car ownership amongst car sharers and they drove
Accepted 8 February 2017
15% to 20% fewer car kilometres than prior to car sharing. The shared cars mostly replace a
Available online 3 March 2017
second or third car.
Due to reduced car ownership and car use, car sharers emit between 240 and 390 fewer
Keywords:
kilograms of CO2 per person, per year. This is between 13% and 18% of the CO2 emissions
Car sharing
related to car ownership and car use.
Environmental impact
© 2017 Elsevier B.V. All rights reserved.
1. Introduction
Car sharing recently has drawn a great deal of attention, partly because of its green image. For instance, the website
of Zipcar,1 one of the big car sharing companies, claims that ‘car sharing is about redefining transportation to make cities
better places for you and me, and it helps us to ensure a healthier future for the planet’. Metz (2013) states that car sharing
is environmentally beneficial as it reduces CO2 emissions.
Our research intends to add to the academic knowledge on the environmental impacts of car sharing by answering the
following question: What are the effects of car sharing on car ownership, car use and CO2 emissions?
Our research contributes to the academic literature in three ways. Firstly, contrary to most other research on this subject,
we considered the impacts of various forms of car sharing. Secondly, we took into account changes not only in car use, but
also in other transport modes, which would have been used if no car sharing programme would have been in place. Thirdly,
in addition to the CO2 emissions associated with car use, we also took into account those related to the construction and
demolition of cars. Our research focused on the Dutch situation.
There are various car sharing systems (for an overview, see Shaheen et al., 2012). For our study, we focused on formal,
organised car sharing by private consumers. We distinguished between business-to-consumer organisations (with vehicles
being owned by an organisation, such as Zipcar in Canada and the United States, and Car2go in several European and North
American cities) and ‘peer-to-peer’ car sharing organisations (private individuals offering their own cars for rent on online
platforms, such as Drivejoy in Spain and Snappcar in the Netherlands).
∗ Corresponding author.
E-mail addresses: hans.nijland@pbl.nl (H. Nijland), jordy.vanmeerkerk@pbl.nl (J. van Meerkerk).
1
http://www.zipcar.com/ziptopia/inside-zipcar/zipcar-green-how-car sharing-helps-the-environment-infographic.
http://dx.doi.org/10.1016/j.eist.2017.02.001
2210-4224/© 2017 Elsevier B.V. All rights reserved.
H. Nijland, J. van Meerkerk / Environmental Innovation and Societal Transitions 23 (2017) 84–91 85
Section 2 provides an overview of academic literature on the impacts of car sharing on car use, car ownership and CO2
emissions. Section 3 describes the research methods used. Results are shown in Section 4, and Section 5 presents conclusions
and discusses the research set-up.
2. Literature overview
A study by Shaheen and Cohen (2013) showed a steady increase in the number of countries participating in business-to-
consumer car sharing, as well as in the number of cars and car sharers involved. By 2014, approximately 5 million people
on 5 continents were sharing some 100,000 cars (Navigant Research, 2017). Peer-to-peer car sharing is growing even faster
(Shaheen et al., 2012). The environmental impact of peer-to-peer car sharing is not a widely researched subject, but one
might suppose that peer-to-per car sharing attracts a different type of users with different mobility patterns and different
cars than business-to-consumer car sharing does. Until now, research on the environmental impacts of car sharing has been
focused on business-to-consumer car sharing organisations. The environmental impact of car sharing has been examined
to various extents; some studies take full life-cycle impacts into account, including upstream (vehicle and fuel production)
and downstream (vehicle scrappage) effects, including or excluding the use of alternative modes of transport. Some studies
include the impact of reduced car ownership due to car sharing on the demand for parking (see e.g. Engel-Yan and Passmore,
2013 and Stasko et al., 2013). Only a few studies (Briceno et al., 2004; Hertwich, 2005) consider the rebound effects of
people saving on transportation expenditure by participating in a car sharing scheme and possibly spending that money on
other, non-transport forms of consumption. The additional emissions from those forms of consumption may have negative
environmental impacts.
Car sharing usually leads to a reduction in vehicle kilometres travelled (VKT) and reduced car ownership. Cervero et al.
(2007) did a survey amongst car sharers and a non-car-sharing control group in the San Francisco area, four years after the
City CarShare programme was inaugurated. In that survey, respondents were asked to fill out a questionnaire about their
shared car use over a 20-day period. A total of 619 responses were received. Car sharing was associated with reduced vehicle
ownership, decreased VKT and lower greenhouse gas (GHG) emissions. Martin and Shaheen (2011), in collaboration with
major car sharing organisations throughout North America, surveyed 9635 members of a car-sharing organisation about their
travel behaviour both during the year before they started car sharing and at the time of the survey. The results indicated
that the VKT dropped by 27% for those participating in a car sharing programme. They found that many participants in car
sharing schemes drove slightly more kilometres, but some drove considerably less. When also incorporating the kilometres
that would have been driven in the absence of a car sharing programme, they estimated a drop in VKT of 43%, whereas car
ownership dropped by 44%, either due to vehicles having been sold or vehicle purchases having been postponed. Based on
an internet survey amongst 1340 participants from all major car sharing organisations (except Zipcar) in the United States
and Canada, representing 5% of all members, Millard-Ball et al. (2005) found that each shared car replaced 14.9 privately
owned vehicles in North America, and, while individual VKT varied, net VKT decreased by 37%. An overview of impacts on
car ownership and car use in North American studies until 2011 is presented in Table 1 (from Shaheen et al., 2012).
Sioui et al. (2012) compared similar neighbouring households and found that those participating in car sharing pro-
grammes used their cars 3.7 times less often. Taking estimates on car use and car ownership from previous studies, Chen
and Kockelman (2016) performed an LCA of the potential impacts of car sharing. They found that participants in a car sharing
programme may reduce their average individual transportation energy use and GHG emissions by approximately 51%. These
energy and emission savings can be primarily attributed to modal shifts and foregone travel, followed by reduced demand
for parking infrastructure and lower fuel consumption.
Based on surveys amongst car sharers, Rydén and Morin (2005) found that 34% of the 301 respondents in Bremen
and 21% of the 272 respondents in Brussels disposed of their car, at least partly because of car sharing. They estimated
a reduction in car use in Brussels and Bremen of 28% and 45%, respectively. This represents an average decrease of 3000 km
per year, and the equivalent of 40%–50% decrease in CO2 , due to fewer kilometres driven and the use of smaller, more fuel-
efficient cars. Firnkorn and Muller (2011) interviewed 308 inhabitants of the German city of Ulm about their current mobility
behaviour. Those respondents were not participating in Ulm’s free-floating car sharing system (Car2go). The study modelled
the potential impact on CO2 emissions if the respondents were to start using Car2go, using three different scenarios, and
concluded that car ownership would probably decrease and annual CO2 emissions would decrease by 146–312 kg per person.
Doka and Ziegler (2001) compared the environmental impact, in Switzerland, of cars that were being shared to that of the
average Swiss passenger car. They conclude that the overall footprint of a shared car is 39% smaller than that of the average
privately owned car, but the study does not provide specific figures on avoided CO2 emissions.
86 H. Nijland, J. van Meerkerk / Environmental Innovation and Societal Transitions 23 (2017) 84–91
Table 1
Overview of the impact of car sharing in North America (Shaheen et al., 2012).
20000
15000
business-to-consumer
10000
Peer-to-peer
5000 Oneway
0
2008 2009 2010 2011 2012 2013 2014 2015 2016
Fig. 1. Number of shared cars in the Netherlands per type of car sharing organisation, between 2008 and 2016.
(source: http://kpvvdashboard-4.blogspot.nl/)
Fig. 1 shows that, in the Netherlands, peer-to-peer car sharing is growing fast.
In the Netherlands, car sharing first emerged in 1974, with the introduction of the Witkar (White Car) project in Ams-
terdam. This project ultimately failed in 1988. In the early 1990s, the former Dutch Ministry of Transport, Public Works and
Water Management started stimulating car sharing programmes, assuming these would contribute to the policy goals of
reducing both car use and the emission of pollutants (Ministerie van Verkeer en Waterstaat, 1988).
At that time, expectations with respect to car sharing were high. In 1993, for example, a feasibility study
estimated that, by 2010, car sharing could lead to a reduction of 3.5–4 billion car kilometres, with 40% of the
10 million Dutch drivers participating in a car sharing programme (AGV, 1993; Adviesdienst Verkeer en Vervoer,
1996). In 2016, these estimates appear to have been overly optimistic. At the moment, car sharing still is
only a niche market. The Dutch Government, however, is optimistic again about the future of car sharing and
its environmental impact. In June 2015, 29 Dutch organisations, including national and local authorities, lease
companies, insurance companies, and nature and environment organisations, entered into a Green Deal on car shar-
ing (https://www.rijksoverheid.nl/actueel/nieuws/2015/06/03/over-drie-jaar-honderdduizend-deelauto-s-in-nederland).
Its first objective is to support Dutch climate goals, the second to alleviate some of the urban mobility problems, and the
third for private parties involved to save money. The Green Deal’s short-term goal is to increase car sharing from around
15,000 cars in 2015 to 100,000 by 2018.
H. Nijland, J. van Meerkerk / Environmental Innovation and Societal Transitions 23 (2017) 84–91 87
Table 2
Distribution of the kilometres travelled by different modes of transport that were replaced by a shared car (in%).
Car 34
Train 41
Bus, tram, rapid transit 4
Bicycle 3
Car passenger 1
Other 2
Not travelled 15
3. Methods
We drafted a questionnaire and commissioned the Dutch Institute for Public Opinion and Market Research (TNS-NIPO) to
conduct a survey. From a sample of about 140,000 people representative of the Dutch population, TNS-NIPO first approached
everyone over the age of 18. They were asked if they had shared a car for at least one year, and, if so, whether they were willing
to participate in a survey. This resulted in 363 car sharing respondents. As the purpose of our study was to estimate changes
in mobility, participants were asked about their current car ownership and car use and that during the time before they
began car sharing (expressed as total numbers of cars and total number of kilometres travelled by car, per year, before and
after car sharing). For most respondents, this covered a period of one to three years. We also looked at what this ownership
and use would have been if they had not been car sharing. In spatial planning and economic studies, this is known as ‘the
counterfactual’.
For example, for participants who had owned a car before they started sharing and continued to do so, the effect of
car sharing was found to be zero. However, perhaps these respondents would have bought a second car if they had not
started car sharing. In such cases, there is an effect on car ownership, although it cannot be observed. Shaheen et al. (2012),
therefore, made a distinction between observed and unobserved effects, and both were included in our study. Moreover,
survey participants were not only asked about car use, but also about the use of other modes of transportation, as the shared
car likely fulfilled a mobility need that would otherwise have been fulfilled by bus, train or borrowed car. They were asked
about the length of their most recent two car-sharing journeys and how they would have travelled if they would not have
had a shared car at their disposal. The answers were used to calculate the use of the alternative modes of transport (see
Table 2) in the counterfactual. To answer the question about the impact of car sharing on mobility and the environment, we
compared the situations before and after car sharing, as well as a hypothetical counterfactual, for all respondents and per
type of car sharing organisation (peer-to-peer or business-to-consumer). The resulting difference, ideally, would represent
the impact. There are however also other factors that influence mobility behaviour. Major changes in people’s personal lives,
such as the start of cohabitation, a divorce, a new job, the birth of a child, are often reasons why people change their mobility
habits (e.g. see Oakil et al., 2014; Prillwitz et al., 2006; Verhoeven et al., 2005). Therefore, we addressed these issues and
analysed the results for all respondents as well as for those that did not encounter major life events during their period of
car sharing.
In our survey, 20% of respondents were a member of a peer-to-peer scheme, 50% of a business-to-consumer car sharing
organisation, and 30% participated in both. Analyses were performed for each of the three groups, as well as for all respondents
together (because the total number of respondents was relatively small).
Changes in mobility behaviour also cause changes in CO2 emissions. In order to determine the CO2 emissions, we took
into account the actual emissions involved in the use of various modes of transport. There are two commonly used methods
to calculate the CO2 emissions from various modes of transport:
- The tank-to-wheel method (TTW), which only calculates CO2 exhaust emissions. Advantage of this method is the large
availability of reliable data on various modes of transport. This method does not include emissions from electric trains, as
these emissions are emitted at the power plant and not by the train itself.
- The well-to-wheel method (WTW), which also includes the emissions related to fuel production (both for petrol and
electricity). Emissions due to train kilometres are also included.
The actual figures on emissions per passenger kilometre by car, train, bus and other modes of transport were taken from
Otten et al. (2015), which represents the state of the art in Dutch real world emission factors. Furthermore, we took into
account the emissions related to vehicle construction and demolition to calculate the effects of changing car ownership (in
addition to the emissions related to their use).
In our study, rebound effects were not taken into account.
4. Results
On the basis of this survey and of a larger TNS-NIPO panel, and considering age, gender, education and the degree of
urbanity of the residential area of participants, the number of ‘official’ car sharers in the Netherlands was estimated at
88 H. Nijland, J. van Meerkerk / Environmental Innovation and Societal Transitions 23 (2017) 84–91
Fig. 2. Impact of car sharing on the distribution of observed numbers of cars per household.
around 90,000 (i.e. excluding people sharing a car with friends, neighbours or family). This amounts to nearly 1% of all
people in the Netherlands who own a driving licence.
In our study, we distinguished two types of mobility effects; those on car use (and the use of other modes of transport)
and on car ownership.
Table 3
Changes in differences in car use and car ownership between types of car sharers.
Factor Group 1 Mean 1 Group 2 Mean 2 Mean difference t-value Sig. (2-tailed)
Change in car ownership Business-to −consumer −0.25 Peer to peer 0.00 −0.25a −2 0.014
(# cars in household)
Change in car use Business-to −consumer −2208 Peer to peer −1092 −1116 −0.741 0.459
(annual car km.)
Change in car ownership Business-to −consumer −0.25 Both −0,15 −0.10 −1.174 0.241
(# cars in household)
Change in car use Business-to −consumer −2208 Both −1462 −746 −0.532 0.595
(annual car km)
Change in car ownership Peer to peer 0.00 Both −0.15 0.15 1.274 0.205
(# cars in household)
Change in car use Peer to peer −1092 Both −1462 369 0.19 0.849
(annual car km)
T-tests were performed to analyse whether these differences would be statistically significant.
a
Means are different at the 0.05 significance level.
Table 4
Change in annual CO2 emissions due to changed car ownership and car use.
About 50% of respondents were participating in a business-to-consumer car sharing system, 20% in a peer-to-peer car
sharing system and 30% participated in both. The biggest changes in car ownership and car use were observed for the first
group, the business-to-consumer car sharing. The smallest changes were seen for the peer-to-peer car sharers. Table 3
shows that decrease in car ownership differed significantly between peer-to-peer and business-to-consumer car sharers.
Other differences were not significant (0.05). If we take into account that Dutch business-to-consumer car sharing led to the
largest decrease in car use and car ownership and that those cars generally were more fuel-efficient than the average car on
the Dutch roads, then it is reasonable to assume that the car-sharing-related CO2 reduction is larger for this group than for
peer-to-peer car sharing. Nevertheless, as peer-to-peer car sharing is growing faster, the total environmental impact of this
group may very well overtake the total impact of business-to-consumer car sharing.
Driving a car involves CO2 emissions. For car sharers, we found an average reduction in car kilometres of 1750 per year,
compared to the time before they started car sharing. This resulted in a 279 kg reduction in CO2 (with a 95% confidence
interval of 89–468 kg, taking a WTW emission factor of 220 g CO2 per vehicle km, which is, taking into account an occupancy
rate of 1.39 persons per car, 158 g CO2 per passenger km (see Otten et al., 2015)). However, a certain number of the car-
sharing kilometres, would previously have been travelled using more environmentally friendly modes of transport, or they
would not have been travelled at all. By choosing the car over other modes of transport, car sharers cause an additional
168 kg in CO2 emissions (with a 95% confidence interval of 125–212 kg).2 When we take both the number of car kilometres
driven and the change in mode of transport into account, the reduced car use of car sharers yields an annual CO2 reduction
of 110 kg.
However, car ownership also involves CO2 emissions, as vehicle manufacturing and demolition require resources and
energy, as well. Gbeghaje-Das, (2013) and Samaras and Meisterling (2008) found that 10% to 20% of the CO2 emitted during
the lifespan of a vehicle is related to its production and demolition, based on TTW emissions. If we assume that a car drives
250,000 km in total, with average TTW emissions of 181 g of CO2 /km (Otten et al., 2015), the CO2 emissions from driving a
car during its lifespan is 45,250 kg, i.e. 80% to 90% of total emissions. Emissions from manufacturing and demolition would
thus be 5028 to 11,312 kg. This equals 335–754 kg CO2 per year, assuming a 15-year lifespan. As car ownership decreased by
around 0.4 car per respondent, CO2 emissions decreased by 0.4*335 − 754, i.e. 134–301 kg CO2 per year. This figure should
be corrected for the increase in the ‘ownership’ of a shared car. We assumed 15 people would share one car, which means
that the respondents’ total emission reduction due to reduced car ownership was 14/15 * 134 − 301, i.e. 125–281 kg CO2 per
year.
90 H. Nijland, J. van Meerkerk / Environmental Innovation and Societal Transitions 23 (2017) 84–91
Table 4 summarises the impact of the change in car ownership and car use on CO2 emissions. It shows that car sharing
led to an average reduction of between 236 and 392 kg CO2 per respondent. This equals a reduction of around 13%–18% in
the emissions related to car ownership and car use.
This study included people who had experienced a life event over the surveyed period (197 respondents) and those who
had not (165 respondents). Analysing each group, we found that the impact on car ownership and car use was slightly larger
for those who had experienced a life event (car possession would decrease by 0.36 instead of 0.41 and car use would decrease
by 1600 instead of 1750 km/year, if we excluded those who experienced a life event). This could possibly be related to the
fact that some of such events, for example a divorce or job loss, also involve a loss of income. Seeing that financial motives
are the main reasons for car sharing, it is not unlikely that newly divorced or unemployed people would start car sharing.
- There is over 30% less car ownership amongst car sharers than before they began car sharing. The ‘traditional’ car sharers,
in particular, often indicated that they had since disposed of a privately owned car. The shared car mostly replaced a second
or third car.
- Car sharers drive around 15%–20% fewer car kilometres than before they started car sharing. This is mostly due to the fact
that, in general, once people have disposed of a car they tend to drive considerably less. The journeys they make by shared
car, would otherwise have been made predominately by train or by using a borrowed or rented car.
- Car sharers emit between 240 and 390 fewer kilograms of CO2 per person, per year. This is between 13% and 18% of their
CO2 emissions related to car ownership and car use. About one third to half of this reduction can be attributed to less car
use; the remainder to the lower degree of car ownership.
Our research yielded impacts that, at first sight, are in line with results from other studies (see also Section 2). Yet, if
we bear in mind that most other researchers did not include CO2 impacts due to changes in car ownership, and that those
changes account for roughly half of the CO2 impacts, it seems that our results are lower than those from most other studies.
We cannot explain this phenomenon.
There is some uncertainty involved, as our survey included fewer participants than those in comparable research abroad,
and it offered no certainty about causality–did respondents reduce the number of kilometres they were driving because
they started car sharing, or were other reasons involved? We asked respondents whether they had experienced major life
events (e.g. the start of cohabitation, a divorce, the birth of a child, a new job) since they began car sharing. After all, such
events often cause people to change their car ownership and/or mobility behaviour. For those respondents (about half of
the total), it was unclear whether the change in mobility could be attributed to the fact that they started car sharing or to
the particular life event. If we had excluded them from our analysis, the impact on car ownership and car use would have
been only slightly smaller.
In order to address the problem of causality, another set-up of the experiment would have been better (using a control
group or a longitudinal study). At the very least, our results are strongly indicative of car sharing being the reason for
differences in car ownership and car use.
Another point of caution is that we used the respondents’ own estimations about the number of car kilometres they had
previously been driving, even if that had already been some years ago. People’s memories could have been blurred or they
could have provided socially acceptable answers about their current car use. This last possibility could mean that the impact
of car sharing in this study has been overestimated.
Many respondents indicated that they would have bought an additional car if they had not started car sharing. Because of
a lack of reliable reference material, we decided against attributing any kilometres to the not-purchased, additional vehicle.
Yet, it is plausible that if they would have had such a vehicle at their disposal, they would have used it more often. This
would mean that the results presented here are more likely to represent an underestimation than an overestimation.
Our respondents answered the questionnaire from their personal perspective. However, decisions on car disposal or
acquisition are often made at household level. Impacts can also best be observed at household level. Therefore, it would
have been best to do the analysis at that level as well. Unfortunately, the available data did not suit such a method.
Our survey consisted of participants in various types of car sharing programmes. Given the limited size of our sample,
we did not analyse the impacts of each type of car sharer in full detail. However, this is a very interesting field of study, and
research in that area would certainly yield new insights.
Our research focused on private consumers. To date, environmental and mobility impacts related to corporate car sharers,
to our knowledge, have not been researched. This would certainly be a valuable topic for further research. As any insight on
this point is lacking, it is also not known whether municipal or national policies to promote this type of car sharing would
be worth considering.
2
The figures are based on emissions in actual practice, and have been corrected for capacity utilisation.
H. Nijland, J. van Meerkerk / Environmental Innovation and Societal Transitions 23 (2017) 84–91 91
References