Carsharing: the impact of system characteristics on its potential to replace private car trips and reduce car ownership

Abstract

This paper aims to explore the potential of carsharing in replacing private car trips and reducing car ownership and how this is affected by its attributes. To that affect, a stated choice experiment is conducted and the data are analyzed by latent class models in order to incorporate preference heterogeneity. The results show that around 40% of car drivers indicated that they are willing to replace some of their private car trips by carsharing, and 20% indicated that they may forego a planned purchase or shed a current car if carsharing becomes available near to them. The results further suggest that people vary significantly with respect to these two stated intentions, and that a higher intention of trip replacement does not necessarily correspond to higher intention of reducing car ownership. Our results also imply that changing the system attributes does not have a substantial impact on people’s intention, which suggests that the decision to use carsharing are mainly determined by other factors. Furthermore, deploying electric vehicles in carsharing fleet is preferred to fossil-fuel cars by some segments of the population, while it has no negative impact for other segments.

Full text

Vol.:(0123456789)
Transportation (2020) 47:935–970
https://doi.org/10.1007/s11116-018-9929-9
1 3
Carsharing: theimpact ofsystem characteristics onits
potential toreplace private car trips andreduce car
ownership
FanchaoLiao1 · EricMolin1· HarryTimmermans2· BertvanWee1
Published online: 11 October 2018
© The Author(s) 2018
Abstract
This paper aims to explore the potential of carsharing in replacing private car trips and
reducing car ownership and how this is affected by its attributes. To that affect, a stated
choice experiment is conducted and the data are analyzed by latent class models in order
to incorporate preference heterogeneity. The results show that around 40% of car drivers
indicated that they are willing to replace some of their private car trips by carsharing, and
20% indicated that they may forego a planned purchase or shed a current car if carsharing
becomes available near to them. The results further suggest that people vary significantly
with respect to these two stated intentions, and that a higher intention of trip replacement
does not necessarily correspond to higher intention of reducing car ownership. Our results
also imply that changing the system attributes does not have a substantial impact on peo-
ple’s intention, which suggests that the decision to use carsharing are mainly determined
by other factors. Furthermore, deploying electric vehicles in carsharing fleet is preferred
to fossil-fuel cars by some segments of the population, while it has no negative impact for
other segments.
Keywords One-way carsharing· Roundtrip carsharing· Car ownership· Electric vehicle
* Fanchao Liao
f.liao@tudelft.nl
Eric Molin
e.j.e.molin@tudelft.nl
Harry Timmermans
H.J.P.Timmermans@tue.nl
Bert van Wee
g.p.vanwee@tudelft.nl
1 Faculty ofTechnology, Policy andManagement, Delft University ofTechnology, P.O. Box5015,
2600GADelft, TheNetherlands
2 Urban Planning Group, Eindhoven University ofTechnology, P.O. Box513, 5600MBEindhoven,
TheNetherlands
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

936
Transportation (2020) 47:935–970
1 3
Introduction
Carsharing was introduced a long time ago (its earliest implementation was in the late
1940s) but only gained substantial attention and popularity during the past decade (Becker
etal. 2017). Thanks to the widespread use of smartphones, carsharing is becoming increas-
ingly convenient since real time information regarding the availability and location of
shared cars can be easily checked via mobile apps. In order to fulfill the diverse demand
of consumers, various carsharing schemes are offered in the market, which differ in terms
of pricing scheme and ways of organization (one-way vs. roundtrip carsharing). Since car-
sharing grants people access to cars without the responsibilities and hassles related to car
ownership such as regular maintenance and high parking costs, it serves as a viable sub-
stitute for conducting car trips and even buying a car for some people. Several empirical
studies found that carsharing users reduce their vehicle travel distance and even give up car
ownership (Martin etal. 2010; Millard-Ball etal. 2005; Shaheen and Cohen 2013).
The potential of carsharing in reducing car ownership gained considerable attention in
automobile industry and policy making. Since each shared car usually can serve more than
one person, a carsharing fleet is expected to replace more private cars than the number of
shared cars, consequently reducing the total number of cars. Therefore, car manufacturers
expect a “reasonable share” of their future profits will be coming from carsharing since
car ownership is likely to drop,1 and governments are focused on carsharing’s potential in
relieving the negative externalities brought by both the production and usage of cars, such
as pollution, CO2 emission, high parking pressure, etc.
In order for (potential) car owners to switch to carsharing and reduce car ownership,
the carsharing scheme has to be able to cover some trips which are currently (or expected)
conducted by the private car. Duncan (2011) investigated what kind of car trip patterns can
be cost-effectively accommodated by carsharing and derive the potential of joining car-
sharing by calculating the share of people with the compatible trip pattern. A similar study
by Schuster etal. (2005) simulated people’s choices between owning private car and car-
sharing by comparing their costs based on the car condition and trip pattern. However, cost
may not be the only consideration and people do not necessarily use carsharing to replace
private car trips even if it is slightly cheaper. Furthermore, those who can accommodate
more trips by carsharing are not necessarily more willing to reduce car ownership.
In order to decide whether there shall be policy incentives for carsharing, the govern-
ment needs information regarding the scale of impact of carsharing on car ownership.
Moreover, in order to understand how this impact can vary for different carsharing systems
and individuals, it is also necessary to know what factors affect people’s intentions of pri-
vate car trip replacement and car ownership reduction. Among all potential influential fac-
tors, carsharing system service attributes are especially of interest since they are within the
control of service providers.
Of all service attributes, the impact of deploying electric shared vehicle is particularly
worth investigating. Many governments have been promoting electric vehicles (EV) due
to the sustainability target and EVs have also entered carsharing service. If electric vehi-
cles are deployed in the carsharing fleet, the potential benefits of carsharing are further
enhanced. For example, many carsharing users still keep their private car (Martin etal.
1 “VW expects profits from car-sharing and ride-hailing”, https ://www.ft.com/conte nt/29097 c88-1bab-
11e7-a266-12672 48379 1a.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

937
Transportation (2020) 47:935–970
1 3
2010) and use carsharing services when their car is not available at the ideal time (e.g.
because their partner is using the car), a parking place is too hard to find, etc. In that case,
even if those people would not drive less due to carsharing, it can still reduce environmen-
tal impacts since most private cars are powered by fossil fuel. Moreover, deploying electric
vehicles in shared car fleets provides easier access to electric vehicles (EVs) for many peo-
ple who still have doubts towards adopting EV as a private vehicle (Zoepf and Keith 2016).
People may have less battery-related concerns (replacement costs, life expectancy, possible
decrease in range over time) for a shared car compared to a private car they have to pur-
chase, especially if they use shared cars for short urban trips; therefore, a carsharing fleet of
EVs may face less resistance from its potential users than the resistance EV has to confront
from its potential buyers. From the fleet owners’ perspective, EVs may also be a better
option because of their lower operational cost and positive environmental image compared
to internal combustion engine vehicles. There seems to be a possible synergy of carsharing
and electric mobility, therefore it is worthwhile to investigate how deploying electric vehi-
cle would affect potential carsharing users’ decision.
The aim of this paper is to investigate the effects of various carsharing system attributes
(including car fuel type) on people’s choice and propensity of joining and using carsharing.
We explore the potential of carsharing in both replacing car trips made by privately owned
cars and reducing car ownership. Furthermore, we identify different consumer groups
according to their heterogeneous preferences and describe each group based on individual-
related variables. Finally, we explore the relationship between people’s intention of using
carsharing to replace private car trips and the intention of reducing car ownership. For the
above purposes, we conducted a stated choice experiment and applied latent class models
to analyze preferences and categorize respondents. This paper contributes to the literature
by (1) exploring the impact of carsharing system attributes on the intention of replacing
private car trips and reducing car ownership under both roundtrip and one-way carsharing
schemes, especially the option of deploying electric vehicles in shared car fleet, (2) iden-
tifying different customer groups based on their preferences for carsharing and (3) exam-
ining the relationship between car owners’ intention of private trip replacement and car
ownership reduction.
The remainder of this paper is organized as follows: “Related work” section provides
a brief review on relevant literature; “Methodology” section introduces the methodology
including survey design, data collection and model estimation. “Results” section elaborates
the results we obtained from multiple analyses. The final section provides a discussion
regarding the policy implication of the results. Among others, we discuss the implications
of our results for the area of shared autonomous vehicles.
Related work
Most studies on carsharing potential user preferences focus on their decision to enroll as
carsharing member, which can be further categorized into three main types. The first type
utilizes revealed preference data in the region where carsharing is already available and
directly explore the influential factors on people’s membership (Becker etal. 2017; Ciari
et al. 2015; Juschten etal. 2017). This approach allows the investigation of the impact
of those service attributes which differ between carsharing stations or individuals: such
as access distance, number of vehicles in each station, etc. (Ciari et al. 2015; Juschten
etal. 2017). The second type studies the intention of joining carsharing systems without
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

938
Transportation (2020) 47:935–970
1 3
considering other transport options. The dependent variable is the intention to join car-
sharing, which is then analyzed by regression models to find individual-related variables
that significantly influence the intention to join (Efthymiou etal. 2013; Zhou and Kockel-
man 2011). These studies focus on the impact of individual characteristics on member-
ship decisions. Since they mostly focus on a single given carsharing system, their models
do not capture the marginal effects of carsharing system attributes. The third type mainly
uses stated choice experiments to study people’s choice between joining carsharing and
use other transport options. These experiments consist of several choice tasks that vary the
attributes of the carsharing system (and of other transport alternatives). This experimental
setting allows the preferences for carsharing system attributes to be captured (Kato etal.
2012; Le Vine etal. 2014b). A recent study of this type is Kim etal. (2017a, b) which
explores people’s choice between joining a carsharing system, buying a second car and
remaining the status quo. A context condition worth noticing in this study is that respond-
ents are assumed to own only one car in the household and have limited access to this
vehicle when needed (below 60%) in all choice tasks; however, this may not be the case for
many car owners. Despite its valuable contribution, this assumption of a specific context
may result in bias when evaluating the general potential of carsharing or even the mar-
ginal effects of attributes for the population at large. Besides, this study did not take into
account the impact of the fuel type of shared cars. In addition to these three types of stud-
ies, (Rotaris and Danielis 2017) applies a rather special approach which uses the general-
ized cost of carsharing to predict the probability of joining carsharing.
Previous research focusing on the impact of carsharing on car ownership mainly asked
current users of carsharing systems to report their (intentions of) ownership change after
joining carsharing (Cervero etal. 2007; Firnkorn and Müller 2011, 2015; Kim etal. 2015;
Shankar etal. 2015). Le Vine and Polak (2017) also estimated a regression model to see
what kind of carsharing users are more likely to reduce their car ownership. The effects
are usually expressed by how many private cars have been replaced by shared cars. The
estimated number of private cars replaced by each shared car is estimated to vary from 2.5
(Douma and Gaug 2009) to 13 (Martin et al. 2010). However, these studies share some
common limitations: first, some studies do not compare the car ownership changes of car-
sharing members with non-members; second, they focus on current carsharing users who
are considered to be the early adopters of the service and their behavior may not be repre-
sentative of the entire potential user group. Therefore, these numbers are likely to be over-
optimistic of the effects of carsharing (Tal 2009), which makes it difficult to extrapolate the
results to the total population and estimate the total potential of carsharing on car owner-
ship. As an exception, Klincevicius etal. (2014) used census data to explore the impact of
carsharing system on household car ownership.
Few studies investigated what extent carsharing can replace private car trips. An exam-
ple is Firnkorn and Müller (2011) which asked current car2go2 users what percentage of
current private car trips they plan to replace by car2go, which only provides a descriptive
analysis of the intentions of existing users. A much larger share of research investigated
people’s preferences for carsharing in a short-term mode choice for a given trip, but they
only looked at a specific trip context such as commuting (Kim etal. 2017b; de Luca and
Di Pace 2014), grocery shopping (Le Vine et al. 2014a) or park and carsharing service
2 A one-way free-floating carsharing service operated by Daimler.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

939
Transportation (2020) 47:935–970
1 3
(Cartenì etal. 2016); therefore, the results cannot be generalized to assess the total impact
of carsharing on replacing private car trips.
Consumer preferences and intentions regarding using carsharing to replace private car
trips and reducing car ownership are likely to be heterogeneous since carsharing is a niche
market (Bert etal. 2016) and there may only be a certain group of people who will seri-
ously consider carsharing as an option. Most above studies included various individual-
related variables in their models to capture their effects on carsharing decisions intentions,
but none have attempted to systematically classify people into groups with different prefer-
ence profiles. As mentioned in the introduction, our current study aims to address all the
above identified research gaps.
Finally, the intention of using carsharing to replace private car trips is usually studied
separately from the intention of reducing car ownership. As mentioned in the introduction,
some previous studies used “the compatibility of current car trip patterns with carshar-
ing” as a proxy for the possibility of switching away from owning car to joining carsharing
(Duncan 2011; Schuster etal. 2005). Another somewhat related study is Le Vine and Polak
(2017) which find that among current free-floating carsharing users, those who use the ser-
vice more often are also more likely to reduce their car ownership. However, to the best of
our knowledge, no study attempted to explore whether there is a relationship between the
intentions of trip replacement and car ownership reduction.
Methodology
Data collection andsample
Since we aim to investigate the impact of carsharing on car ownership, it makes sense to
narrow the research subjects down to potential consumers of cars. Therefore, our target
population is people who have a driver’s license and either own a car or intend to buy a
car within the following 3years. In addition, we only include respondents whose intended
purchase is a new car for private use. People who plan to acquire second-hand cars or com-
pany cars are excluded because these decisions may involve different considerations (e.g.
company car may not be financed by the user).
We used an existing Dutch national panel (Panelclix) to recruit respondents. These
panel members fill out questionnaires on a regular basis for a small reward. The members
who are invited to participate in our survey are selected at random from the Panelclix list.
Those who choose to participate, first answered a series of filter questions and only people
who fit our above requirements were asked to finish the entire survey. The data was col-
lected in June 2016 and the final sample consists of 1003 respondents.
Sample characteristics are listed in Table1. Comparing our sample to the Dutch car
owner data,3 we can see that our sample is fairly representative regarding employment sta-
tus and age, while being slightly over-represented by females (due to survey distribution
quota aiming to reach gender balance among respondents), and people with relatively low
income, which shall be taken into account when interpreting the results.
3 Only 10 people do not have a car right now.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

940
Transportation (2020) 47:935–970
1 3
Questionnaire design
Survey design
Since we are interested in exploring how individual-related variables affect carsharing
preferences and choices, we collected a wide range of information which may be related
to decision making of joining and using carsharing. Apart from the basic socio-demo-
graphic and socio-economic characteristics, we asked for information related to current
car ownership and travel behavior: respondents reported their current state of car own-
ership and the characteristics of the car they expect to purchase; they were also asked
about the frequency of their car trips for each different purpose (including commuting,
grocery shopping, other shopping and leisure) and frequency of using public transport
and bikes. If the frequency of car trips for a certain purpose is not zero, the respondent
is also asked to specify the distance, trip duration and parking time at destination of a
typical trip for that purpose.
Table 1 Sample characteristics
a We cannot find data for Dutch car owners regarding this variable. For household type, we used data for
the entire Dutch population except children. For education level, we used data for Dutch population above
15years old
Variable Level Percentage in
sample
Percentage in
Dutch car owners
Gender Male 51.7 62.7
Female 48.3 37.2
Age ≤ 35years 25.0 18.9
36–50years 24.0 30.2
51–65years 30.8 29.8
≥ 66years 19.2 21.1
Monthly net personal income < 1250 17.4 8.8
1251–2500 49.2 28.6
> 2500 33.4 62.5
Employment status Paid job 65.9 67.7
Students 3.6 1.6
Others 30.5 30.7
Household typeaSingle 16.8 22.9
Couple without children 40.9 35.5
Couple with children 31.1 37.5
Others 11.2 4.1
Education levelaWithout high education 56.6 71.1
With high education 43.4 28.9
Number of cars 0 1.0
1 68.4
2 27.6
Access to own car when needed (Almost) always 86.2
Most of the time 9.5
Not more than half 4.3
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

941
Transportation (2020) 47:935–970
1 3
In addition, we measured their familiarity and attitudes towards carsharing. We first
asked their previous experience with carsharing to see whether they have used, seen or
heard of carsharing. In total, 6% of the respondents are or have been carsharing members.
Considering that 1% of people over 18years old is estimated to use carsharing in the Neth-
erlands (Harms etal. 2016), carsharing users seem to be overrepresented in our sample, but
they still represent a very limited share of all respondents.
In order to measure respondents’ attitudes towards carsharing, we presented them with
4 statements about carsharing for which they respond on 5-point Likert scales that runs
from (1) totally disagree to (5) totally agree. The seminal work from Bergkvist and Rossiter
(2007) showed that if the construct consists of a concrete singular object (in our case being
carsharing) and a concrete attribute (attitude for a certain aspect), single items can have the
same predictive validity as multiple-item measurements; therefore we can still use it even if
the reliability is lower. Taking this into account, in order to capture the attitude of multiple
aspects with the least number of statements, the four statements are meant to cover aspects
of attitude different from each other.
Table2 presents the four statements and the distribution of their responses. In general,
carsharing does not have a negative image and people do recognize the environmental
friendliness of carsharing; however, on average people do not appreciate the convenience
brought by carsharing and still have a relatively strong attachment to car ownership. Two
statements are found to have high communalities; therefore, we generated a factor “hedonic
attitude” from these two statements. The other two statements measure the symbolic and
environmental attitude respectively. All factor and item scores are standardized for further
use.
Choice experiment design
The main part of the survey is a stated choice experiment which focuses on the decision
regarding the frequency of using carsharing and car ownership. As we mentioned in the
introduction, carsharing schemes can be categorized into two types, namely roundtrip and
one-way. The two most crucial differences between these two types are the following. First,
for roundtrip carsharing the shared car always has to be returned to its pick-up point while
this is not required for one-way carsharing; Second, roundtrip carsharing allows advanced
booking while one-way carsharing does not (booking time up to 30min). We decide to not
include both systems in the same choice task since we do not aim to study the competition
between roundtrip and one-way carsharing systems; besides, for those respondents who
are not that familiar with carsharing, learning about both schemes and trading off between
them is rather difficult and may lead to more misunderstanding and errors. Therefore, a
separate experiment was constructed for each scheme, and respondents were randomly
assigned to only one of the experiments. Before the start of the experiment, respondents
were introduced to the basic characteristics of the respective carsharing scheme.
In each choice task, respondents were asked to make a choice between two given alter-
natives which are a car and a carsharing scheme. The presentation of the car alternative
differs depending on the respondents’ condition: people who intend to purchase a car in
the near future (from now on referred to as prospective car buyers) were presented with a
car alternative of which the attributes describe the car they expect to purchase. This infor-
mation is collected from their answers to previous questions in the questionnaire. They
were asked whether they are willing to forego the car purchase and use the given carshar-
ing scheme instead. Other respondents (referred to as car holders) were only presented the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

942
Transportation (2020) 47:935–970
1 3
attributes of a carsharing scheme and answer whether they are willing to sign up for the
presented carsharing scheme and give up a car which they currently own. At the end of the
experiment, these car holders filled in the characteristics of their own car (or if they have
more than one car, the car which they are most likely to give up) and we assume that this is
the car with which they traded off in all choice tasks.
Table 3 lists the attributes that are varied in the experiment and their levels. In the
experiment for prospective car buyers, the attribute values of the expected car purchase are
based on the answers provided by respondents and fixed in all choice sets presented to the
respondent. The attributes for carsharing schemes are all varied by three levels except the
return location of one-way carsharing, their operationalization is further elaborated below:
• Fuel type of car This attribute is varied in the levels: (1) gasoline car, (2) electric car
with 100km of driving range and (3) electric car with 200km of range after full charge.
This allows investigating preference between gasoline vehicle and electric vehicle with
short and medium driving ranges.
• Purchasing cost: In case of roundtrip carsharing we set a deposit which is fully
refunded after the membership expires, while for one-way carsharing we specify a
one-time registration fee. This setting fits the current situation of existing carsharing
schemes in the Netherlands.
• Maintenance cost A monthly membership fee is also specified for both carsharing
schemes. The values for one-way carsharing are lower than that of roundtrip because
current one-way carsharing (such as car2go) do not charge any monthly fee while it is
common among roundtrip carsharing schemes.
• Operating cost The structures and levels of operating cost attributes of both carsharing
alternatives are based on the price levels of current carsharing schemes in the Nether-
lands.
• Access time to the shared car is also included as an attribute: since the position of
shared cars is not fixed at each time of use, the respondents are told that this is an aver-
age value.
• Car availability With respect to this attribute we use two different measures for the two
carsharing schemes based on their different booking mechanism. Since for roundtrip
carsharing it is possible to book a time slot in advance and check when cars are avail-
able, the measure we use is the difference between the initial ideal departure time and
the closest time slot available. For example, a “15 min difference from ideal time”
implies that on average a shared car is available only 15min earlier or later than the
Table 2 Statements used for attitude measurement and their responses
Cat-
egory
Statement Average score SD Factor loading
Sym-
bolic
Carsharing is for people who cannot afford cars 2.64 0.834
Environ-
mental
Carsharing is more environmentally friendly
than buying a car
3.46 0.850
Hedonic Carsharing causes more problems than owning
a car
3.31 0.816 0.617
I like the feeling of owning a car and carsharing
cannot match that
3.77 0.874 0.617
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

943
Transportation (2020) 47:935–970
1 3
Table 3 Attributes used in the choice experiment and their levels
Item Alternative Attribute Levels
Fuel type of car Buying (holding) a car Fuel type of expected (current)
car
Specified by respondent
Both carsharing Fuel type of shared cars Gasoline Electric 100km range Electric 200km range
Purchase cost Buying (holding) a car Price of expected (current) car (€) Specified by respondent
Roundtrip carsharing Deposit (€) 0 150 300
One-way carsharing Registration fee (€) 0 20 40
Maintenance cost Buying (holding) a car Cost of expected (current) car (€/
month)
Specified by respondent
Roundtrip carsharing Membership cost 0 10 20
One-way carsharing Membership cost 0 5 10
Operating cost Buying (holding) a car Fuel cost of expected (current)
car (€/km)
Specified by respondent
Roundtrip carsharing Distance cost (€/km) 0.20 0.25 0.30
Hourly cost (€) 2 4 6
One-way carsharing Minute cost (€) 0.20 0.25 0.30
Access time by walk-
ing
Buying (holding) a car To current parking location
(minutes)
Both carsharing To location of shared
car(minutes)
2 7 12
Availability of car Buying (holding) a car Availability of expected (current)
car
Expected: always available
Current: specified by respondent
Roundtrip carsharing Difference from ideal time
(minutes)
0 15 30
One-way carsharing Availability of shared car (%) 80 90 100
Return location of car One-way carsharing Return location of car Reserved parking spots for shared
cars
Reserved parking spots for shared cars + all
public parking spots
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

944
Transportation (2020) 47:935–970
1 3
initial ideal departure time. We only give the average value in order to control the com-
plexity of the experiment. Since one-way carsharing does not allow booking and one
can hardly do anything when no car is available (within reasonable walking distance),
its availability measure is straightforwardly defined as the probability of a shared car
appearing to be able to use when needed.
• Return location of car This attribute only applies to one-way carsharing. It has two
levels: (1) reserved parking spots for shared cars: this corresponds to one-way sta-
tions-based carsharing for which users have to park the car in the designated spots; (2)
reserved parking spots for shared cars + all public parking spots: this level represents
free-floating carsharing, which allows users to park the car anywhere allowed.
In addition to exploring to what extent carsharing can reduce car ownership, we were inter-
ested in exploring the potential of carsharing in reducing trips which would otherwise be
done by private fossil fuel cars. To that effect, respondents were asked to indicate for each
car sharing alternative to what extent they use it to replace their car trips (about which we
posed questions earlier in the survey). An answer was given for each of the four different
trip purposes using a 5-point scale ranging from 1 “never” to 5 “for all trips”. An example
of a choice task and questions is shown in Fig.1.
Both choice experiments were created using a D-efficient optimal design (Rose and
Bliemer 2009). The priors are mostly based on findings of previous research (Kim etal.
2017a) and assumed when not available. With this input, we used Ngene to construct the
two choice experiments and ended up with a 12-choice set design for each, which was
blocked into 2 blocks each with 6 tasks to which a respondent was randomly assigned.
Hence, every respondent faced 6 choice tasks. In the end, the one-way carsharing experi-
ment had 521 respondents in total while the roundtrip experiment received 482 responses.
Model conceptualization
Corresponding to the two questions in each choice task, we have two dependent vari-
ables. The first is an ordinal one measuring the extent to which the respondent is will-
ing to replace private car trips by carsharing. Although in each choice task we col-
lect responses for up to4 four common trips of different purpose (commuting, grocery
shopping, other shopping and leisure), we assume that all influential factors have the
same effect on these four responses and use a single model to describe these effects.
The second variable is dichotomous and denotes the choice whether to proceed with
a planned car purchase (or between keeping or shedding the current car). These two
dependent variables are indicators for the latent utility of each level of replacement
intensity or each choice.
Regarding the trip replacement intensity, we explore how its utility is determined
by the attributes of both the carsharing system and own car. We have already elab-
orated upon the carsharing system attributes in “Questionnaire design” section. The
own car attributes we concern are fuel costs and walking distance to the parking loca-
tion. Utility is also expected to be dependent on the trip characteristics as carsharing
may be more feasible and suitable for some trips than others. The characteristics we
4 If the respondent previously indicated that s/he never conducts or does not use a car to conduct a certain
type of trip, no response is collected for this trip.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

945
Transportation (2020) 47:935–970
1 3
investigated include trip frequency, duration, staying time at location and the purpose
of trip.
As for the choice of car ownership, the utility of choosing carsharing is also
assumed to be dependent on the attributes of carsharing system and own car. Although
more own car attributes are expected to be influential in this decision: apart from fuel
cost and distance to parking location, we also explore the effect of car price, monthly
maintenance cost and availability of own car.
The effects of these attributes and factors on utility are expected to be heterogene-
ous among people. Therefore, we assume that the entire population consists of several
classes: these effects are homogeneous within each class and vary between different
classes.
Finally, we are also interested in the role individual variables play in determining
class membership. In addition to the common socio-economic and socio-demographic
variables, we also investigated the influence of frequency of using bikes and public
transport and attitudes towards carsharing. Figure2 is an illustration of the conceptual
models for both trip replacement and car ownership.
Model specication
We applied latent class models to implement the above conceptualization. To be more spe-
cific, we estimated a latent class ordinal regression model for modelling trip replacement and a
latent class choice model for the car ownership model.
Let
yit
denote the response of respondent i in choice task t and m represent a specific cat-
egory of all possible responses. In the case of trip replacement intensity, m can range from
never (1) to all (5). In the case of car ownership choice, the respondent can choose either car
purchase or carsharing, therefore m can take two values. The final stated responses
yit
are indi-
cators of
𝜂m|zit
which indicates the latent systematic utility of each category (trip replacement
intensity) or alternative (car choice) of the response variable for subject i in choice task t.
In car ownership model, the value of this latent utility has the following form:
in which
𝛽m
and
𝛽mk
denote the alternative-specific constant and attribute effects respec-
tively.
zatt
itmk
represents the value of attribute k of alternative m in choice task t for subject
i. In latent class models, the entire sample population is assumed to belong to K different
latent classes which differ in their taste parameters. Therefore, the utility function of mem-
bers from class x is
which implies that a different set of
𝛽m
and
𝛽mk
will be estimated for each class x. The con-
ditional probability for the response follows the multinomial logit function:
(1)
𝜂
m
|
zit
=𝛽m+
K
∑
k=1
𝛽mkzatt
itmk +
𝜀
(2)
𝜂
m
|
x,zit
=𝛽xm +
K
∑
k=1
𝛽xmkzatt
itmk +
𝜀
(3)
P
yit =m

x,zit=
exp

𝜂mx,zit


M
m
�
=1
exp

𝜂m�

x,z
it 
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

946
Transportation (2020) 47:935–970
1 3
In the trip replacement model, the dependent variable is of ordinal level and the response
probability function is exactly the same as (3) while the “utility” function becomes
which applies the function of an adjacent-categories ordinal logit model (Agresti 2002).
zpred
itq
denotes the explanatory exogenous factor q which differs between choice tasks. These
factors are usually termed as “predictors” in latent class regression models.
𝛽xm0
and
𝛽xq
are class-specific intercepts of level m and effects of predictor q on utility which need to be
estimated.
For each subject i, the probability of belonging to a class x is predicted by its individual
characteristics
zcov
i
which are termed “covariates”. This probability function also takes the
form of a multinomial logit model:
Hence, for each class an intercept (
𝛾x0
) and a set of regression coefficients (
𝛾xr
) are
estimated. However, some individual-specific variables are dependent on other common
covariates (such as socio-economic characteristics) and thus cannot be considered as “truly
independent”; in contrast to the active covariates, these variables can be included as “inac-
tive” covariates. The name implies that these covariates do not affect the probability of
(4)
𝜂
m
|
x,zit
=𝛽xm0+
Q
∑
q=1
m𝛽xqzpre d
itq
(5)
P
x

zcov
i=
exp

𝛾x0+

R
r=1𝛾xrzcov
ir


S
x�=1exp

𝛾x�0+

R
r=1𝛾x�rzcov
ir

Fig. 1 An example of stated choice task (text translated from Dutch)
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

947
Transportation (2020) 47:935–970
1 3
class membership and are not included in the model estimation. Instead, we calculate the
distribution of inactive covariates for each class, which provides a richer profile of different
classes. In this study, urban density is included as an inactive covariate.
Finally, the probability of observing a certain sequence of responses can be written as
Model estimation
The latent class regression and the latent class choice model were each estimated sepa-
rately for one-way and roundtrip carsharing, hence, four models were estimated in total.
For the trip replacement model, we pooled and stacked the responses for each of the four
trips with different purpose in one dataset in order to estimate a single model for all trips as
we mentioned in conceptualization.
We used LatentGold (Vermunt and Magidson 2013) to estimate all four models. Effects
coding was used for all parameters of categorical variables. We used several criteria in
order to determine the optimal number of classes: first, statistical measures including ρ2
values and the Bayesian Information Criterion (BIC), which take both model quality and
parsimony into account; second, the interpretability of the estimated model, such as the
sign and size of coefficients; third, avoid solutions with classes which are not essentially
different from other classes; According to all the above criteria, for the latent class ordinal
regression model of trip replacement, we arrived at a 3-class structure; and we chose the
2-class solution for the latent class discrete choice model of car ownership.
(6)
P
(yi)=
S
∑
x=1
P(x
|
zcov
i)
T
∏
t=1
P(yit =m
|
x,zit
)
Covariates:
Individual
variables
Latent class
Ulity of
alternaves
Stated choices
Carsharing
system
aributes
Own car
aributes
Predictors
Covariates:
Individual
variables
Latent class
Ulity of
levels
Stated level of
use for trips
Carsharing
system
aributes
Trips
characteriscs
Own car
aributes
(a) (b)
Fig. 2 Model conceptualization: a trip replacement model; b car ownership model
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

948
Transportation (2020) 47:935–970
1 3
Results
This section elaborates the results from the estimated models. We first consecutively pre-
sent the results of the trip replacement model and the car ownership model; in the end, we
discuss the connection between these two choices.
Trip replacement model
Consumer groups andpreference heterogeneity
Based on their different preferences regarding the frequency of replacing private car trips
by carsharing, both in the model for one-way car sharing and the model for roundtrip car
sharing, the respondents can be categorized into three classes. Table 4 lists the result of
these two latent regression models. Both model fits are quite high and the pseudo R square
is significantly improved compared to the one-class ordinal regression model, which dem-
onstrates the power of the latent class model; the prevalence of non-trading behavior (see
below for a detailed description) can also be a reason for the high model fit.
We first briefly characterize each of the classes based on their indicated frequency of use
as presented in the top of Table5. Class 1 demonstrates an extremely low interest in using
car sharing both under one-way and roundtrip car sharing, which can be labeled as “own
car oriented”. When answering the questions about the share of their car trips they intend
to replace by carsharing, they choose the category ‘none’ for 95% of the time. In contrast,
Class 2 intends to replace a larger share of their trips by carsharing and can be described
as “CS-leaning”. Class 2 under roundtrip carsharing intends to replace more trips than the
same class under one-way car sharing. Finally, Class 3 intends to use carsharing the most
for replacing their car trips. They are likely to be frequent users for carsharing and are
termed “CS-enthusiasts”. Their responses lean more towards the extremes under roundtrip
carsharing; in other words, there are more responses for categories ‘none’, ‘most trips’ and
‘always’. This suggests that in case of roundtrip carsharing, the responses of Class 3 are
more divergent across different rating tasks, which implies that the choices are more sensi-
tive to changes in carsharing system attributes and/or different trip characteristics.
As for the size distribution between the three groups, Class 1 is bigger under roundtrip
carsharing (63.4%) than under one-way (54.7%), which implies that the latter seems to be
capable of attracting more subscribers. Class 3 under one-way carsharing (20.5%) also
takes a larger share than under roundtrip (13.9%).
Next, we describe how the trip replacement decisions of the three classes are differently
affected by carsharing system attributes, trip characteristics and their current car character-
istics. First, we focus on carsharing system attributes. The preference for vehicle type sig-
nificantly varies across the three groups. For both roundtrip and one-way carsharing, Class
1 prefers gasoline cars, while Class 3 does not have a significant preference over car types
used in carsharing systems. Class 2 prefers gasoline vehicles to EVs with only 100 km of
driving range under one-way carsharing. However, EVs with 200km range is even slightly
preferred to gasoline vehicles, suggesting that a driving range of 200km is sufficient to
meet consumer’s needs. Under roundtrip carsharing, Class 2 even prefers EVs with 100km
range to gasoline vehicles.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

949
Transportation (2020) 47:935–970
1 3
The taste parameters for the carsharing attributes costs, availability and access time also
differ across the three groups. For Class 1, all parameters have the expected sign and most
are statistically significant at a 95% confidence level both under one-way and roundtrip
carsharing. On the other hand, for both Class 2 and 3, only registration cost or access time
to shared car have a significant impact. It is worth noticing that the coefficient for flexible
return location of one-way carsharing is non-significant for all classes, which implies that
whether the one-way carsharing system is station-based or free-floating does not seem to
influence people’s trip replacement decisions. The prevalence of non-significance is prob-
ably due to the rather small size of these two predicted classes (especially class 3). Another
possible reason is that most system attributes genuinely do not have much impact on trip
replacement decisions of these two classes, at least not if the attribute values lie within the
range of levels varied in the choice experiment.
Two coefficients for Class 3 which are statistically significant have unexpected signs,
namely the membership cost of one-way carsharing and the distance cost of roundtrip car-
sharing; their size is however rather small in comparison to the constants and other attrib-
utes. A possible reason is that a small number of people associate low cost to low quality
(we did not specify the quality of the shared cars), despite the fact that we ask respondents
to assume the carsharing systems in the experiment are identical apart from the attributes
we describe. Since Class 3 is rather insensitive towards costs (all other cost attributes are
non-significant), these people may prefer a higher quality system. Hence, they may think
it is represented by high cost, which may explain the positive cost parameter. In general,
the parameter estimates of Class 3 in our model are not conclusive and shall not be overly
interpreted since the predicted class size is small. If we wish to obtain accurate parameter
values for this class, it is advisable to collect a larger sample or over-recruit people who
have strong intention to replace private car trips by carsharing.
Trip characteristics, including trip purpose, frequency, duration and staying time at
destination all influence the trip replacement decisions. Under one-way carsharing, their
impacts are vastly different between Class 2 and 3. Class 3 tends to use carsharing more
to replace trips which are more frequent, less than 1h and require a longer stay at the des-
tination, while this is the opposite for Class 2. Furthermore, Class 3 mostly tends replace
more grocery shopping trips, while Class 2 is willing to replace more shopping and leisure
trips. Under roundtrip carsharing, both Class 2 and 3 tend to replace more trips which last
between 16 and 30min and when the stay at the destination is less than an hour. Class 2
also replaces more frequent trips while there is no clear preference for Class 3.
It is worth mentioning that the parameter estimates cannot be directly contrasted with
the normal usage pattern of current carsharing systems. For example, a typical roundtrip
carsharing trip mostly has a parking time around 3h; while our model shows that Class 3
prefer parking time of less than 1h the most, which may seem contradictory. However, the
parameter estimates are class-specific and relative, while the revealed pattern is also related
to the distribution of trip characteristics among the population. Parking time between 2
and 4h is not significantly preferred by Class 3, but Class 3 members conduct significantly
more trips with 2–4h parking time (compared to trips with other parking duration), there-
fore this may still end up with a peak pattern of 3-h parking time even if there is no relative
preference between trips with different parking time. In addition, each trip’s utility score is
a combination of the coefficients of all its characteristics (duration, frequency, etc.). Most
of these trips also have a trip duration of less than 5min which has a large negative coef-
ficient, therefore these trips turn out to be less preferred.
The characteristics of the current car also have a significant impact on the intensity of
trip replacement. As expected, under both one-way and roundtrip carsharing, most groups
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

950
Transportation (2020) 47:935–970
1 3
Table 4 Parameters and z-values of the latent class ordinal regression model regarding choice of trip replacement
One-way Roundtrip
Class 1: own car
oriented
Class 2: CS-leaning Class 3: CS-enthusiast Class 1: own car
oriented
Class 2: CS-leaning Class 3: CS-enthusiast
Estimate z-value Estimate z-value Estimate z-value Estimate z-value Estimate z-value Estimate z-value
Intercept
None 6.1689 4.0617 1.1168 2.0873 − 2.0747 − 3.1032 0.4248 0.6012 0.0606 0.1200 − 0.5451 − 1.4350
A few trips 2.6668 3.4142 1.4202 5.2418 0.6273 1.9241 − 0.7403 − 2.0981 1.9085 7.3766 − 1.2773 − 5.9175
Half of trips 1.0804 5.7120 0.2165 4.0588 1.6205 27.4439 − 0.7636 − 5.3700 1.8126 26.8875 − 0.0036 − 0.0540
Most trips − 2.0645 − 2.6348 − 0.5262 − 1.9368 0.4459 1.3649 − 0.2936 − 0.7159 − 0.2477 − 0.9642 0.9968 5.1607
All trips − 7.8516 − 4.9611 − 2.2274 − 4.0845 − 0.6190 − 0.9579 1.3727 1.9305 − 3.5339 − 6.6522 0.8292 2.2073
Predictors
Gasoline car 0.1078 1.2823 0.0414 1.4205 0.0160 0.4549 0.2012 3.1487 − 0.1684 − 3.5631 0.0315 0.9007
EV with 100km
range
− 0.0110 − 0.1382 − 0.1198 − 3.7935 0.0037 0.1008 − 0.1674 − 2.4641 0.2234 4.7212 − 0.1127 − 3.0056
EV with 200km
range
− 0.0967 − 0.9827 0.0784 2.5975 − 0.0197 − 0.5656 − 0.0338 − 0.5495 − 0.0550 − 1.2391 0.0812 2.3566
Registration fee − 0.0137 − 3.6595 − 0.0044 − 3.1988 − 0.0015 − 0.9183
Deposit − 0.0003 − 1.0692 − 0.0002 − 0.8651 − 0.0003 − 1.5235
Membership
cost
− 0.1013 − 6.6516 − 0.0006 − 0.1147 0.0164 2.7399 − 0.0301 − 5.3611 − 0.0007 − 0.1810 − 0.0004 − 0.1303
Minute cost − 5.4655 − 3.2077 0.0420 0.0804 − 0.2501 − 0.4097
Distance cost/
km
− 1.8365 − 1.6389 − 0.7780 − 0.9769 1.3296 2.2146
Hour cost − 0.1161 − 4.1294 − 0.0195 − 1.0607 0.0096 0.7443
Availability 2.1669 3.0179 0.2839 1.1437 0.1459 0.4830
Difference from
ideal time
− 0.0167 − 4.9704 0.0013 0.4917 − 0.0012 − 0.6069
Access time − 0.0501 − 3.1288 − 0.0233 − 4.3191 0.0061 0.9592 − 0.0652 − 6.0483 0.0073 0.8620 − 0.0187 − 3.1484
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

951
Transportation (2020) 47:935–970
1 3
Table 4 (continued)
One-way Roundtrip
Class 1: own car
oriented
Class 2: CS-leaning Class 3: CS-enthusiast Class 1: own car
oriented
Class 2: CS-leaning Class 3: CS-enthusiast
Estimate z-value Estimate z-value Estimate z-value Estimate z-value Estimate z-value Estimate z-value
Flexible return
location
− 0.0011 − 0.0091 0.0434 1.0361 − 0.0051 − 0.1020
Trip frequency
Once a month
or less
0.0474 0.4053 0.0072 0.1474 − 0.2945 − 4.1457 − 0.0171 − 0.1552 − 0.4588 − 6.0288 − 0.1338 − 2.4368
2–3 times per
month
− 0.1804 − 1.3593 − 0.0788 − 1.6402 − 0.0859 − 1.4973 0.1965 2.2537 − 0.4052 − 5.9421 − 0.0262 − 0.5621
1–2 times per
week
− 0.0572 − 0.5540 − 0.0061 − 0.1371 0.1943 4.4169 0.3047 3.8054 − 0.1394 − 2.4563 − 0.0027 − 0.0671
3–4 times per
week
0.2878 2.5014 − 0.0591 − 1.0134 − 0.0353 − 0.5967 0.2500 2.6275 0.4233 5.4545 0.0860 1.6080
5 times per
week or more
− 0.0976 − 0.5625 0.1368 1.5168 0.2214 3.6273 − 0.7342 − 3.3210 0.5801 6.3311 0.0767 1.2101
Trip duration
5min or less 0.1019 0.6652 − 0.3671 − 5.4117 0.0418 0.6468 − 0.4250 − 3.2544 − 0.0140 − 0.1636 − 0.3944 − 5.8414
6–15min 0.0278 0.2613 − 0.2567 − 6.0551 0.1164 2.4803 − 0.1639 − 1.9971 0.0911 1.4743 0.0953 2.3107
16–30min − 0.0717 − 0.7030 0.1588 4.1041 0.0220 0.4378 0.0733 1.0191 0.1529 2.5248 0.0887 1.9909
0.5–1h 0.0687 0.6277 − 0.0288 − 0.5580 0.1151 1.9672 0.1902 2.2425 0.1033 1.2863 0.0456 0.8266
More than 1h − 0.1267 − 0.7970 0.4938 7.4362 − 0.2952 − 3.4819 0.3253 3.2754 − 0.3333 − 3.0532 0.1648 2.2497
Stay time at destination
Less than 1h − 0.0279 − 0.1918 0.4039 7.2866 − 0.4337 − 7.6581 − 0.0683 − 0.6167 0.2912 4.4694 0.1154 2.1640
1–2h − 0.0366 − 0.3514 − 0.0576 − 1.4699 − 0.0888 − 1.9581 − 0.0542 − 0.7067 − 0.0109 − 0.1904 − 0.0915 − 2.2997
2.1–4h − 0.1311 − 1.2444 − 0.1276 − 3.2853 0.0968 2.1078 0.0030 0.0403 − 0.1588 − 2.8565 0.0439 1.0501
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

952
Transportation (2020) 47:935–970
1 3
Table 4 (continued)
One-way Roundtrip
Class 1: own car
oriented
Class 2: CS-leaning Class 3: CS-enthusiast Class 1: own car
oriented
Class 2: CS-leaning Class 3: CS-enthusiast
Estimate z-value Estimate z-value Estimate z-value Estimate z-value Estimate z-value Estimate z-value
More than 4h 0.1955 1.4315 − 0.2187 − 3.7882 0.4258 7.5235 0.1194 1.3138 − 0.1215 − 1.6638 − 0.0678 − 1.2967
Purpose
Commuting − 0.1617 − 1.0942 − 0.1883 − 2.7894 − 0.1154 − 2.0038 − 0.2351 − 2.0664 − 0.2998 − 3.9339 − 0.2401 − 4.2650
Grocery shop-
ping
− 0.2925 − 2.0712 − 0.3547 − 6.7061 0.2315 4.6349 0.1895 1.8146 − 0.2037 − 3.0274 0.0725 1.4435
Shopping 0.0669 0.5710 0.1874 4.2895 − 0.0568 − 1.1622 0.0231 0.2837 0.1591 2.5287 0.0519 1.1482
Leisure 0.3873 3.9913 0.3556 8.7694 − 0.0593 − 1.2499 0.0226 0.3121 0.3444 5.9160 0.1158 2.6312
Fuel cost per km 0.7860 1.3119 0.3276 1.0417 1.3371 6.5150 − 2.3798 − 3.4516 3.9435 10.4177 − 3.6562 − 12.1785
Parking distance 0.1272 14.6152 0.0696 10.1080 − 0.0359 − 8.3331 0.0360 2.8963 − 0.0015 − 0.1956 0.0304 6.2414
Pseudo
R-squared
0.6866 0.6835
Pseudo
R-squared
without latent
class
0.0708 0.0458
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

953
Transportation (2020) 47:935–970
1 3
Table 5 The within-class distributions of choices and covariates of the trip replacement model
One-way Wald p value Roundtrip Wald p value
Own car CS-leaning CS-enthusiast Own car CS-leaning CS-enthusiast
Frequency of use
None 95% 30% 1% 95% 6% 19%
A few trips 3% 41% 15% 4% 38% 6%
Half of trips 1% 15% 49% 1% 45% 15%
Most trips 1% 10% 22% 0% 10% 34%
All trips 0% 4% 13% 0% 1% 26%
Mean 1.07 2.18 3.30 1.07 2.61 3.41
Covariates
Gender 1.0 0.600–0.0 0.990–
Male 52% 56% 56% 49% 54% 50%
Female 48% 44% 44% 51% 46% 50%
Age 10.4 0.006** 20.0 < 0.001**
Mean 50.02 50.41 42.92 51.38 45.77 49.86
Education 7.3 0.120–21.1 < 0.001**
Low 23% 21% 23% 19% 26% 17%
Middle 36% 36% 32% 38% 35% 18%
High 40% 43% 46% 42% 39% 65%
Income 7.0 0.130–9.1 0.058*
Low 19% 20% 12% 16% 17% 17%
Middle 51% 47% 49% 53% 42% 41%
High 30% 33% 39% 30% 41% 42%
Household 14.9 0.021** 18.5 0.005**
Single 19% 17% 14% 19% 6% 19%
Couple without kids 44% 38% 28% 45% 41% 36%
Single or couple with kids 32% 34% 51% 29% 48% 44%
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

954
Transportation (2020) 47:935–970
1 3
Table 5 (continued)
One-way Wald p value Roundtrip Wald p value
Own car CS-leaning CS-enthusiast Own car CS-leaning CS-enthusiast
Others 5% 11% 6% 7% 6% 1%
Employment status 13.2 0.040** 13.0 0.043**
Employed 62% 67% 83% 63% 67% 68%
Student 3% 5% 2% 4% 4% 1%
Retired 21% 22% 9% 22% 20% 19%
Others 14% 5% 7% 11% 9% 11%
New purchase planned 8.6 0.014** 4.3 0.120–
Yes 74% 80% 91% 74% 83% 84%
No 26% 20% 9% 26% 17% 16%
Frequency of using public transport 23.9 0.047** 25.6 0.029*
(Almost) everyday 1% 5% 3% 1% 1% 4%
1–6days per week 11% 20% 26% 9% 20% 19%
Less than once per week 88% 75% 71% 90% 79% 76%
Frequency of using bikes 20.1 0.130–14.9 0.380–
(Almost) everyday 20% 26% 11% 21% 31% 25%
1–6days per week 38% 40% 51% 35% 35% 41%
Less than once per week 42% 34% 38% 44% 34% 33%
Symbolic attitude 5.2 0.073* 14.2 0.001**
Mean − 0.04 − 0.04 0.12 − 0.08 0.20 0.11
Environmental attitude 3.1 0.210–5.8 0.055*
Mean − 0.02 0.09 – 0.23 0.03 − 0.12 0.35
Hedonic attitude 32.4 < 0.001** 21.9 < 0.001**
Mean 0.25 − 0.20 –0.42 0.14 − 0.42 0.06
Urban density (inactive)
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

955
Transportation (2020) 47:935–970
1 3
Table 5 (continued)
One-way Wald p value Roundtrip Wald p value
Own car CS-leaning CS-enthusiast Own car CS-leaning CS-enthusiast
Rural 33% 40% 37% 33% 33% 25%
Small city 50% 44% 46% 52% 50% 54%
Big city 17% 17% 17% 15% 17% 21%
**Significant at p < 0.05, *Significant at p < 0.1, –Not significant
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

956
Transportation (2020) 47:935–970
1 3
tend to use carsharing more to replace their car trips if they currently (or are expected to)
have higher fuel cost or a longer parking distance.
Personal characteristics
The class membership model reveals the impact of personal characteristics on class mem-
bership. Table5 displays the individual variables included in the model and their Wald
statistics and p value. The within-class percentage distribution of each individual covariate
is also presented.
Class 3 has the largest share of people who are younger, highly educated, earning high
income, employed, have kids and use public transport more often under both one-way and
roundtrip carsharing (a couple of effects are not statistically significant though). By con-
trast, the composition of Class 1 is mostly opposite to Class 3 in terms of these individual
characteristics. In other words, the covariate distribution of Class 1 and 3 lie on different
ends of the spectrum. For example, with respect to employment status, Class 3 have the
highest percentage of employed people while Class 1 have the lowest. Consequently, the
covariate distribution of the Class 2 mostly lies between Class 1 and 3. The only exceptions
are age and educational level: under roundtrip carsharing Class 2 is the youngest and least
educated; on the other hand, under one-way carsharing it is the oldest. There is no signifi-
cant difference in the distribution of gender and urban density across the three groups.
Since Class 2 and 3 indicate their intention to use carsharing and are likely to enroll
for carsharing membership, we can contrast their characteristics to the previous findings
in carsharing members. We confirm the typical image of CS users: younger than average,
well-educated, have higher income, employed and more likely to have children (Becker
etal. 2017; Le Vine and Polak 2017). Becker etal. (2017) also found that people who are
employed tend to use one-way carsharing more frequently: although there is no discernible
different between Class 2 and 3 regarding the employment status for roundtrip carshar-
ing, we do find that under one-way carsharing Class 3 has a much higher percentage of
employed people than Class 2. Finally, while most studies find carsharing members are
predominantly male (Becker etal. 2017; Juschten etal. 2017 and its citations), we do not
find any significant impact of gender on the intention of trip replacement.
We now focus on the impact of attitude on class membership. All three attitudes have a
significant influence in case of roundtrip carsharing, while only symbolic and hedonic atti-
tude are relevant under one-way carsharing. Surprisingly, the average attitude is not always
congruent with the preferences of every group. In the model of one-way carsharing, while
Class 3 has the highest preference for carsharing, they attach a more negative symbolic
value to carsharing compared to the other two groups. This counter-intuitive result sug-
gests that this negative connotation is not strong enough to deter Class 3 away from using
carsharing. Under roundtrip carsharing, Class 2 recognizes the environmental-friendliness
of carsharing the least while they intend to replace more trips than Class 1, which suggests
that the replacement is not motivated by environmental considerations.
Discussion
The impact of carsharing system attributes on the intended frequency of private car trip
replacement is rather limited according to our model. For Class 1, although all coefficients
are significant, group members choose to never use carsharing to replace their private
car trips in 95% of their responses to trip replacement questions. Therefore, the effect of
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

957
Transportation (2020) 47:935–970
1 3
promoting carsharing usage is expected to be rather limited for this class if the perfor-
mance of carsharing systems is not drastically increased (beyond the range we tested). For
the other two classes, only shared car type, registration cost and average access time to
shared car are significant predictors. A previous study based on an existing carsharing sys-
tem also finds that the distance to carsharing stations is a significant determinant of car-
sharing membership (Juschten etal. 2017). In general, most attributes regarding costs and
availability of car do not have a significant impact on trip replacement decisions.
We mentioned above that the preference for car type differs for Class 2 in two schemes:
for one-way carsharing EV with 200km range is their favorite, while for roundtrip carshar-
ing EV with only 100km range is already preferred over gasoline vehicle. This may be
explained by the characteristics of the trips for which they prefer to use carsharing. Under
roundtrip carsharing, Class 2 uses it more for trips of middle length (16–30min), for which
100km range is less likely to be a problem; on the other hand, they use it mostly for longer
(more than 1h) trips under one-way carsharing and 200km seems to be sufficient to meet
their requirements.
Different classes also vary in terms of their attitudes towards carsharing and how their
own car characteristics affects their willingness to use carsharing in replacement of pri-
vate car. These coefficients may reveal the respondents’ motivation of using carsharing.
For example, under one-way carsharing, Class 2 members who currently have higher fuel
costs intend to use carsharing to replace more trips, which is probably motivated by saving
operation cost of car trips.
Apart from the socio-demographic variables and attitudes, we also examined the trip
patterns of each class in order to explore whether those who show higher intention have
a trip pattern more “compatible” with carsharing. The trip characteristic distributions of
all classes are almost identical. For roundtrip carsharing, Class 3 only stands out with the
highest share of trips with parking time between 2 and 4h (29.4% vs. average of 26.4%),
which matches the typical trip pattern of each carsharing system. Class 3 of one-way car-
sharing has the highest share of frequent trips (at least 3 times per week, 33.5% vs. average
of 24.5%), this demonstrates that the flexibility of one-way carsharing makes them more
suitable for accommodating frequent trips such as commuting. In general, it seems that
Class 3 does not have any distinct trip pattern which can explain their high intention of trip
replacement.
Car ownership model
This section looks at people’s choice regarding whether they will use carsharing to replace
their expected car purchase or current car. Table6 presents the estimated choice model and
Table7 presents the distributions of covariates within each class. We found that a two-class
model structure best describes the behavior. We first estimate a full model, and in the final
model we constrain those parameters which are not significantly different across classes
to be equal. The final model fit is high and the improvement from basic multinomial logit
model is also significant. However, since most attributes are non-significant, the model
fit is mostly contributed by the constants. This is mainly caused by non-trading behavior
which will be discussed later in detail.
For both one-way and roundtrip carsharing, Class 1 and 2 are labeled as “Ownership-
Oriented” and “CS-Oriented” according to their choice patterns. The choice responses are
rather extreme for both classes: when answering whether to obtain or give up ownership
of a current (or intended) car if carsharing becomes available, Class 1 choose to keep the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

958
Transportation (2020) 47:935–970
1 3
car or go through the planned car purchase in over 97% of responses, while Class 2 opt for
carsharing and forego the planned car purchase or replace one of their current cars in the
vast majority (over 70% for one-way carsharing and 85% for roundtrip) of choice tasks.
This implies that non-trading behavior is prevalent in the sample. Some research suggests
that these observations shall be discarded (Hess et al. 2010) which can improve model
fit (Wardman and Ibáñez 2012); however, it can be an expression of genuine preferences
(Börjesson etal. 2012): given the attribute range in choice experiment design, when none
of the other alternatives are more attractive than the alternative which the respondent sticks
to, non-trading behavior is observed.
Comparing the two models, Class 1 takes a dominant share in both models (78.3% and
82.5%); while Class 2 of one-way carsharing (21.7%) is slightly larger in size than that of
roundtrip carsharing (17.5%). The model results therefore suggest that the potential of both
types of carsharing in reducing car ownership is on par with each other.
Now we briefly discuss the taste parameter for service attributes. The fuel type of the
shared car does not have any significant impact on the final choice for both classes under
both one-way and roundtrip carsharing. Except for the registration fee of one-way carshar-
ing, none of the taste parameters for carsharing system attributes are significant for Class
2, while access time is significant for Class 1 both for roundtrip and one-way carsharing. In
addition, monthly membership cost and car availability also have significant impact under
roundtrip carsharing. Similar to the trip replacement model, these non-significant param-
eters may be a true reflection of people’s preferences: when considering whether to use
carsharing and forego a planned car purchase (or shed an owned car), carsharing system
attributes genuinely do not play an important role as long as they are not extremely high or
low. It may also be explained by two other reasons: first, Class 1 hardly trade-off between
attributes across choice tasks; second, the size of Class 2 is limited.
All variables with respect to the current (or expected) car (car price, fuel cost, main-
tenance cost and access time to one’s own car) are significant for at least one class in the
model for one-way or roundtrip carsharing. This implies that these factors influence the
decision regarding whether to use carsharing and reduce car ownership. For Class 2 in both
models, people who (are expected to) have a more expensive car are less likely to forego
their ownership: this suggests that an expensive car may be more than a tool for transport
and bears a symbolic value, which was revealed by previous studies (Steg 2005).
Several socio-economic variables account for preference heterogeneity (see Table 7).
Both under roundtrip and one-way carsharing, employed people are more likely to be a
member of Class 2. Under one-way carsharing, Class 2 is also younger. Gender, education,
household composition, income are non-significant predictors for class membership in both
models. There is no significant difference between the two classes regarding urban density
distribution either. Although higher education and higher income can lead to a higher pos-
sibility to join and use carsharing, they are also found to have positive impact on the prob-
ability to maintain car ownership (Le Vine and Polak 2017), which may explain why their
effects in our model become non-significant. In contrast, (Le Vine and Polak 2017) also
found that people with children are more likely to join carsharing and reduce car owner-
ship: although our Class 2 under one-way carsharing still has a much larger share of people
with children, we do not find this strengthened impact. Under both one-way and roundtrip
carsharing, people who are expected to buy a new car have a higher probability of belong-
ing to Class 2 than those who do not. This was expected since the potential buyers are
asked whether they will forego the planned purchase while the others answer whether they
will shed a current car. It is certainly easier to give up a purchase which has not been mate-
rialized than giving up a car one already owns.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

959
Transportation (2020) 47:935–970
1 3
People’s travel patterns also have substantial influence on class membership. Class 2 has
more frequent public transport users under one-way carsharing and contains more intensive
bike users under roundtrip carsharing. This suggests that a multi-modal person is more
likely to forego car ownership when carsharing becomes available.
Of the three attitudinal items, only the hedonic attitude has a significant effect on class
membership, which shows that the decision of giving up car ownership is influenced by the
individual’s attachment to car ownership, while one’s perception regarding the environ-
mental friendliness and symbolic image of carsharing do not have much impact.
Table 6 Parameters and z-values of the latent class discrete choice model regarding choice of car ownership
One-way Roundtrip
Ownership-oriented CS-oriented Ownership-oriented CS-oriented
Estimate z-value Estimate z-value Estimate z-value Estimate z-value
Alternative specific constant
Buy car 0 0 0 0
Carsharing − 5.0970 − 4.0945 1.3890 1.2209 − 1.2586 − 1.4300 2.1376 4.6842
Attributes
Gasoline car 0.1709 1.4110 0.1709 1.4110 0.0615 0.3990 0.0615 0.399
EV with 100km
range
− 0.1334 − 1.0610 − 0.1334 − 1.0610 − 0.0467 − 0.2952 − 0.0467 − 0.2952
EV with 200km
range
− 0.0375 − 0.3088 − 0.0375 − 0.3088 − 0.0148 − 0.0999 − 0.0148 − 0.0999
Registration fee − 0.0130 − 2.3194 − 0.0130 − 2.3194
Deposit − 0.0004 − 0.6359 − 0.0004 − 0.6359
Membership cost − 0.0010 − 0.0516 − 0.001 − 0.0516 − 0.0443 − 2.5986 0.0219 1.2260
Minute cost 2.6725 1.2711 2.6725 1.2711
Distance cost/km − 0.7843 − 0.3159 − 0.7843 − 0.3159
Hour cost − 0.0842 − 1.3943 − 0.0842 − 1.3943
Availability 0.7878 0.7706 0.7878 0.7706
Difference from
ideal time
− 0.0295 − 2.7693 0.0086 0.7470
Access time − 0.1703 − 2.6912 − 0.0208 − 0.8580 − 0.1115 − 3.0738 0.0144 0.4104
Flexible return
location
0.0177 0.1077 0.0177 0.1077
Current/expected car characteristics
Car price 0.0162 1.1668 − 0.0471 − 4.2881 − 0.0032 − 0.3117 − 0.0191 − 3.4387
Fuel cost 0.3208 0.1386 − 0.6997 − 0.9110 0.0105 0.3134 − 0.1114 − 2.6893
Parking distance 0.2334 8.4361 − 0.0515 − 3.7330 2.0985 1.4541 5.4214 2.8559
Maintenance cost − 0.0024 − 3.2154 − 0.0003 − 0.4654 − 0.0024 − 1.5916 − 0.0008 − 0.7355
Log-likelihood − 730 − 674
Null log-likelihood − 2166 − 2004
McFadden rho-
square
0.663 0.664
Log-likelihood of
MNL
− 1407 − 1336
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

960
Transportation (2020) 47:935–970
1 3
Table 7 The within-class distributions of choices and covariates of the car ownership model
One-way Wald p value Roundtrip Wald p value
Ownership-
oriented
CS-oriented Ownership-
oriented
CS-oriented
Choice
Buy the car 98% 28% 97% 14%
Give up purchase 2% 72% 3% 86%
Covariates
Gender 0.6 0.430–0.3 0.610–
Male 52% 60% 50% 51%
Female 48% 40% 50% 49%
Age 7.0 0.008** 1.5 0.220–
Mean 49.91 44.25 50.40 47.63
Education 3.8 0.150–3.6 0.170–
Low 23% 20% 21% 20%
Middle 36% 32% 33% 42%
High 40% 48% 46% 38%
Income 2.0 0.360–4.1 0.130–
Low 18% 18% 17% 17%
Middle 50% 47% 50% 44%
High 32% 35% 33% 40%
Household 5.8 0.120–1.6 0.650–
Single 19% 13% 17% 15%
Couple without kids 41% 31% 41% 48%
Single or couple with kids 34% 47% 35% 32%
Others 6% 9% 6% 6%
Employment status 6.2 0.1* 7.3 0.064*
Employed 66% 73% 62% 73%
Student 3% 6% 4% 1%
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

961
Transportation (2020) 47:935–970
1 3
Table 7 (continued)
One-way Wald p value Roundtrip Wald p value
Ownership-
oriented
CS-oriented Ownership-
oriented
CS-oriented
Retired 20% 14% 22% 16%
Others 11% 7% 11% 10%
New purchase planned 14.3 < 0.001** 4.8 0.03**
Yes 75% 92% 76% 85%
No 25% 8% 24% 15%
Frequency of commuting trip by car 11.7 0.069* 9.7 0.140–
5 times per week or more 30% 33% 29% 27%
1–4 times per week 27% 34% 30% 34%
Less than once per week 9% 8% 6% 7%
None 34% 25% 35% 31%
Frequency of using public transport 12.4 0.086* 6.2 0.520–
(Almost) everyday 2% 3% 2% 2%
1–6days per week 14% 26% 12% 16%
Less than once per week 84% 72% 87% 82%
Frequency of using bikes 8.7 0.270–20.0 0.006**
(Almost) everyday 21% 15% 21% 38%
1–6days per week 39% 47% 40% 16%
Less than once per week 40% 38% 38% 46%
Symbolic attitude 0.4 0.510–1.1 0.220–
Mean 0.01 − 0.10 0.00 0.04
Environmental attitude 1.3 0.250–0.2 0.640–
Mean 0.02 − 0.22 0.05 0.01
Hedonic attitude 13.2 < 0.001** 8.5 0.004**
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

962
Transportation (2020) 47:935–970
1 3
Table 7 (continued)
One-way Wald p value Roundtrip Wald p value
Ownership-
oriented
CS-oriented Ownership-
oriented
CS-oriented
Mean 0.10 − 0.34 0.06 − 0.30
Urban density (Inactive)
Rural area 36% 35% 32% 31%
Small city 48% 44% 51% 55%
Big city 16% 21% 17% 14%
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

963
Transportation (2020) 47:935–970
1 3
Relation betweenthetrip replacement andcar ownership decisions
A main motivation for car ownership is to conduct trips by car. If carsharing becomes
available and can also fulfill this functional use of car ownership, it is likely to reduce the
need for car ownership. Therefore, a plausible conjecture is that the intention of reducing
car ownership is related to people’s willingness to replace their private car trips by carshar-
ing. In electric vehicle adoption research, several studies assume that the acceptance of
EV is strongly related to the inconvenience caused by EV which stems from the mismatch
of the limited driving range of EV and the travel pattern (such as long trips) of individu-
als (Tamor etal. 2015; Tamor etal. 2013). Similarly, as we mentioned in the introduction,
there have been previous research effort to measure the potential of carsharing by calculat-
ing how many people’s current travel patterns are economically compatible with carsharing
(Duncan 2011; Schuster etal. 2005): the inherent assumption is that the choice of giving
up car ownership depends on the extent to which carsharing can cover (replace) one’s cur-
rent trips. In this section, we aim to investigate whether people who intend to use carshar-
ing to replace more of their private car trips are more willing to give away (one of) their
car.
We assign each respondent to a class in both classifications (trip replacement and car
ownership) according to the posterior probabilities of belong to a particular class based
on their responses and individual characteristics, and explore whether there is any relation
between these two class memberships. Tables 8 and 9 display the cross tables of people’s
membership under the two classifications for one-way and roundtrip carsharing. We can
derive some interesting insights from the two tables:
(1) Some people who choose carsharing to replace very few of their car trips are still
willing to give up their car ownership, albeit the share is rather small (10% of the total
sample). A possible explanation for this may be that only a small share of the current
car trips absolutely needs to be done by driving; with the support of carsharing in
fulfilling this essential need, this group can shed a car and turn to other travel modes
such as biking or public transport to conduct the trips which were previously conducted
by private car. This group therefore may correspond to those carsharing users who
reduced their mileage of car trips after joining carsharing scheme and give up their car
(Millard-Ball etal. 2005).
(2) In the case of one-way carsharing, the preference regarding trip replacement is in line
with their preference for giving up car ownership: CS-enthusiasts has the largest per-
centage which are CS-oriented than other two groups. Le Vine and Polak (2017) also
found that those who use free-floating carsharing more are also more likely to reduce
their car ownership.
(3) In the model of roundtrip carsharing, the percentage of CS-enthusiasts (Class 3 in the
trip replacement decision) who are also CS-oriented (Class 2 in the car ownership
decision) is lower than CS-leaning (Class 2 in the trip replacement decision). This
may seem surprising since CS-enthusiasts on average are willing to use carsharing to
replace more of their private car trips than CS-leaning. However, in Table5 we can see
that the CS-leaning class of roundtrip carsharing has the lowest hedonic score which
indicates their low attachment to their own car compared to the other two groups. This
shows that the decision of reducing car ownership does not solely depend on the prac-
tical consideration such as how many current trips can the carsharing scheme serve.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

964
Transportation (2020) 47:935–970
1 3
Other factors may override the importance of the practicalities, such as emotions and
attachment towards car ownership.
It is expected that one-way free-floating carsharing is the most popular scheme due to its
flexibility comparing to one-way station-based and roundtrip carsharing. Our results show
that indeed a larger group of car drivers is interested in using one-way carsharing for some
of their car trips compared to roundtrip; however, one-way carsharing does not show sig-
nificantly higher potential in reducing car ownership. The relation between usage intensity
and car ownership reduction is also different for the two types of carsharing systems. Fur-
thermore, the perks of free-floating carsharing—free parking at all public parking spots—
do not show any effect on the probability of joining or using carsharing. We expect car-
sharing (especially one-way) is most likely to be operated within big cities with a certain
level of urban density to be profitable. People living in highly urbanized areas may also
have a higher interest in carsharing due to the limited parking and expensive parking fee in
these areas. However, we discerned no significant difference in the overall preferences for
carsharing between people living in areas with different levels of urban density (the urban
density level variable is non-significant in the membership function).
Conclusions anddiscussion
Conclusion
This study aims to investigate car drivers’ intention of replacing private car trips by car-
sharing and reducing car ownership when a carsharing scheme is available, which shed
light on the potential of carsharing among all car drivers. Latent class models are estimated
to identify groups with different preference profiles. We found that for both intentions of
trip replacement and car ownership reduction, people vary significantly with respect to
their preferences for carsharing in general, its system attributes and how the characteristics
of their own car affect their preferences. In total around 40% of the entire sample (CS-
leaning and CS-enthusiasts) indicate that they may be willing to use carsharing to replace
at least some of their private car trips. About 20% (CS-oriented) are likely to give up a
planned car purchase or shed a current car when a suitable carsharing system becomes
available. These numbers can be regarded as an upper limit of the potential for carsharing
in replacing private car trips and reducing car ownership since models calibrated by stated
preference data tend to overrate the preference for new products (see “Limitations and rec-
ommendations for research” section). This variance of preference may be attributed to the
difference in socio-economic condition, travel pattern and carsharing-related attitudes, con-
firming similar findings in previous studies.
We also examined the impact of carsharing system attributes on these two intentions.
As for the fuel type of the shared cars, it does not make any difference in the decision of
giving up car ownership. Regarding the trip replacement decision, EVs are even preferred
to gasoline vehicles by some classes; however, EVs with limited range is less preferred
when carsharing is mostly used for long trips, but a driving range of 200km is enough to
compensate in this case. Based on these findings we may conclude that consumers in gen-
eral do not show resistance and even demonstrate preference for electric vehicles. Regard-
ing other system attributes such as costs and availability, if the current performance level
is already acceptable (within the range of our experiment), a further improvement of the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

965
Transportation (2020) 47:935–970
1 3
system performance in these aspects do not seem to have a significant impact on facilitat-
ing carsharing to replace car ownership or private car trips.
As for the relation between the decision of trip replacement and forgoing car ownership,
people who use carsharing to replace more trips are not necessarily more willing to give up
their car, which indicates that these two effects of carsharing must be studied separately.
By looking at the two decisions together, we arrive at a more detailed classification of the
population and a richer picture of people’s preference profiles regarding carsharing. We
reveal groups such as heavy carsharing users who still want the guarantee of their own car
and people who are willing to give up their car even when carsharing cannot replace most
of their car trips (which they may conduct by other travel modes instead).
Policy implications
We can derive several policy implications from the findings. First, our study reveals that
the potential for car sharing is quite large: as explained above 40% of the entire sample
indicate that they may be willing to use carsharing to replace at least some of their private
car trips, and 20% are likely to give up a planned car purchase or shed a current car when
a suitable carsharing system becomes available. This implies that policies stimulating car
sharing can have substantial societal relevant advantages, related to owning and using cars,
as explained in the introduction. Policies to stimulate car sharing can, for example, be the
provision of designated parking facilities (pull) but also the introduction of more paid park-
ing in residential areas (push). Secondly, deploying electric vehicles has no negative or
even slightly positive impact on increasing carsharing use, which confirms the potential of
carsharing in reducing car trip emissions. This not only is relevant because shared vehicles
can be EVs reducing the environmental pressure of car use, but it is also relevant because
an increase in EV sales in the fleets of shared vehicles can stimulate EV sales in the entire
Table 8 The distribution within the two classifications for one-way carsharing
Car model class Trip model class Own-car
oriented
(%)
CS-leaning (%) CS-
enthusi-
asts (%)
Class size (%)
Ownership-oriented % within trip model class 91.6 70.5 54.2 78.2
CS-oriented % within trip model class 8.5 29.5 45.8 21.8
Class size % of total 54.7 24.8 20.5 100.0
Table 9 The distribution within the two classifications for roundtrip carsharing
Car model class Trip model class Own-car
oriented
(%)
CS-leaning (%) CS-
enthusi-
asts (%)
Car class size (%)
Onwership-oriented % within trip model
class
89.5 67.0 76.1 82.5
CS-oriented % within trip model
class
10.5 33.0 23.9 17.5
Trip class size % of total 63.4 22.7 13.9 100.0
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

966
Transportation (2020) 47:935–970
1 3
car fleet, because due to scale effects (more sales) prices of EVs will go down, and the rela-
tively high purchase costs are a barrier for many people to buy an EV (Liao et al. 2017).
Thirdly, the potential of CS to reduce car ownership reduces the environmental impact of
car ownership—note that producing cars also results in environmental pressure. And park-
ing pressure can to some extent be reduced via increased levels of sharing. So, policies that
stimulate CS might have environmental benefits via reduced ownership levels. Fourth, for
one-way carsharing systems, in contrast to station-based setting, free-floating setting usu-
ally require much more government cooperation since it demands access to public parking
spots. However, our results suggest that consumers do not really appreciate the extra flex-
ibility brought by free-floating. This suggests that a station-based one-way carsharing sys-
tem (such as Autolib in Paris) is a better option which is easier to implement and does not
reduce utility for its users. Fifth, reducing user costs or increasing the availability of shared
cars seem to have little or no impact on mid-term decisions such as the extent of replacing
private car trips and reducing car ownership. Therefore, these strategies are probably not
useful if the goal is to facilitate more trip replacement and car ownership reduction. Sixth,
our results reveal that potential consumers’ preferences regarding carsharing are highly
heterogeneous. Certain groups have more favorable attitudes and preferences towards car-
sharing and may be more susceptible to carsharing promotion policies/strategies, thus it is
recommended that they are given higher priority in such promotion. Furthermore, since
the groups which intend to use carsharing to replace more private car trips do not neces-
sarily overlap with the group which is more willing to reduce car ownership, campaigns
and advertisements promoting carsharing should choose target groups depending on their
specific goal. Seventh, because our study shows that young people are more than average
inclined to become users of CS systems, such systems may lead to postponed car owner-
ship, or even to an overall reduction of the desirability of owning a car, as debated in the
literature on ‘peak car’ (Goodwin and van Dender 2013).
Limitations andrecommendations forresearch
This study has several limitations. First, since carsharing is still a niche market, despite
the fact that we collected a sample of average size, the number of respondents who are
potentially interested in carsharing is rather limited; we also observed the prevalence of
non-trading behavior among the general population. This may lead to statistical insignifi-
cance of some attributes, predictors and covariates. If we wish to have better estimates
of the preference coefficients of the potentially interested group in order to fine tune the
carsharing scheme services, we need a sample which is more targeted towards the poten-
tially interested customers. However, this was not the main aim of this study, which was to
examine the potential of carsharing among the general population of car drivers. Second,
stated preference method is known to result in inflated willingness-to-pay for some socially
desirable behaviors (Axsen etal. 2015), and the online survey we used for data collection
is known to result in even more positive responses than other types of surveys such as face-
to-face interviews (Efthymiou and Antoniou 2016). Therefore, our results may be over-
optimistic in evaluating the potential of carsharing. Thus, while we find that the carsharing
potential is rather limited in the general car driver population, it may even be more limited
than we find here. Third, in this explorative study we simplified some aspects of the choice
problems: for example, we did not consider the uncertainty of remaining range when some-
one takes an electric shared car with limited range. Neither did we consider more flexible
pricing structure (such as different price for driving and parking). Finally, a large part of
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

967
Transportation (2020) 47:935–970
1 3
our respondents resides in rural areas. Although they seem to have no significant difference
in terms of the intention of trip replacement and car ownership reduction and a fair share
of them seem to be quite positive towards carsharing, it shall be kept in mind that some of
the service attribute levels we used in the experiment are not economically feasible to be
realized in those areas.
More future research is needed in order to better investigate people’s preferences and
the possible benefits of all types of carsharing. Comparing the usage pattern of roundtrip
and one-way carsharing is an interesting direction which is of high practical relevance. For
example, if one-way carsharing scheme is especially often used for shopping trips, then
the carsharing scheme can set up more stations (parking docks) around shopping centers.
The potential of peer-to-peer (P2P) carsharing in rural area is also worth investigating: our
results show that people living in rural area seem to be as interested in carsharing as peo-
ple from urban areas; however, as we mentioned in the limitations, the carsharing systems
in our experiment may not be feasible or profitable in rural area and P2P carsharing may
be the only option. Therefore, it is important to examine people’s preference for P2P car-
sharing. Furthermore, if we wish to arrive at a more realistic forecast of the potential of
carsharing, we may combine revealed preference data with stated choice data in the model
estimation. Finally, the introduction of shared autonomous vehicles will also further com-
plicate or even completely change the entire picture. Many researcher, planners and policy
makers now envision a prospect in which car ownership is vastly reduced because peo-
ple on a large scale will make use of shared autonomous cars. However, our results pose
doubt on this prospect: most people prefer to remain owning a car and only intend to make
limited use of carsharing to replace their trips, and this preference is not very sensitive to
improvements of carsharing systems. It is more likely that as long as cars are affordable
and parking regulations with respect to car parking do not dramatically change, people will
continue to own and use private cars even when shared autonomous cars become available
on a large scale. Therefore, more behavioral research is needed to investigate the feasibility
and possibility of the rosy future scenario promised by the introduction of shared autono-
mous vehicles.
Acknowledgement We would like to thank the three anonymous reviewers for their helpful comments.
Authors’ contribution FL conceived of the study, participated in experiment design, performed statistical
analysis, drafted manuscript. EM helped shaping research design, participated in experiment design, exten-
sively edited the manuscript. HT commented on experiment design, edited the manuscript. BW helped shap-
ing research design,commented on experiment design, edited the manuscript.
Compliance with ethical standards
Conict of interest All authors declare that they have no conflict of interest.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 Interna-
tional License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution,
and reproduction in any medium, provided you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license, and indicate if changes were made.
References
Agresti, A.: Categorical Data Analysis. Wiley, New York (2002)
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

968
Transportation (2020) 47:935–970
1 3
Axsen, J., Bailey, J., Castro, M.A.: Preference and lifestyle heterogeneity among potential plug-in electric
vehicle buyers. Energy Econ. 50, 190–201 (2015). https ://doi.org/10.1016/j.eneco .2015.05.003
Becker, H., Ciari, F., Axhausen, K.W.: Comparing car-sharing schemes in Switzerland: user groups
and usage patterns. Transp. Res. Part A Policy Pract. 97, 17–29 (2017). https ://doi.org/10.1016/j.
tra.2017.01.004
Bergkvist, L., Rossiter, J.R.: The predictive validity of multiple-item versus single-item measures of the
same constructs. J. Mark. Res. 44, 175–184 (2007). https ://doi.org/10.1509/jmkr.44.2.175
Bert, J., Collie, B., Gerrits, M., Xu, G.: What’s ahead for car sharing? The new mobility and its impact on
vehicle sales. Boston Consulting Group. 23 February 2016. Available online: https ://www.bcg.com/
publi catio ns/2016/autom otive -whats -ahead -car-shari ng-new-mobil ity-its-impac t-vehic le-sales .aspx
(2016). Accessed 9 Oct 2018
Börjesson, M., Fosgerau, M., Algers, S.: Catching the tail: empirical identification of the distribution of the
value of travel time. Transp. Res. Part A Policy Pract. 46, 378–391 (2012). https ://doi.org/10.1016/j.
tra.2011.10.006
Cartenì, A., Cascetta, E., de Luca, S.: A random utility model for park and carsharing services and the
pure preference for electric vehicles. Transp. Policy 48, 49–59 (2016). https ://doi.org/10.1016/j.tranp
ol.2016.02.012
Cervero, R., Golub, A., Nee, B.: City carshare: longer-term travel demand and car ownership impacts.
Transp. Res. Rec. J. Transp. Res. Board 1992, 70–80 (2007). https ://doi.org/10.3141/1992-09
Ciari, F., Balac, M., Balmer, M.: Modelling the effect of different pricing schemes on free-floating carshar-
ing travel demand: a test case for Zurich, Switzerland. Transp. (Amst) 42, 413–433 (2015). https ://doi.
org/10.1007/s1111 6-015-9608-z
De Luca, S., Di Pace, R.: Modelling the propensity in adhering to a carsharing system: a behavioral
approach. Transp. Res. Procedia 3, 866–875 (2014). https ://doi.org/10.1016/j.trpro .2014.10.065
Douma, F., Gaug, R.: Carsharing in the twin cities: measuring impacts on travel behavior and automobile
ownership. In: 88th Annual Meeting of the Transportation Research Board, Washington D.C. (2009)
Duncan, M.: The cost saving potential of carsharing in a US context. Transp. (Amst) 38, 363–382 (2011).
https ://doi.org/10.1007/s1111 6-010-9304-y
Efthymiou, D., Antoniou, C.: Modeling the propensity to join carsharing using hybrid choice models and
mixed survey data. Transp. Policy (2016). https ://doi.org/10.1016/j.tranp ol.2016.07.001
Efthymiou, D., Antoniou, C., Waddell, P.: Factors affecting the adoption of vehicle sharing systems by
young drivers. Transp. Policy 29, 64–73 (2013). https ://doi.org/10.1016/j.tranp ol.2013.04.009
Firnkorn, J., Müller, M.: What will be the environmental effects of new free-floating car-sharing sys-
tems? The case of car2go in Ulm. Ecol. Econ. 70, 1519–1528 (2011). https ://doi.org/10.1016/j.
ecole con.2011.03.014
Firnkorn, J., Müller, M.: Free-floating electric carsharing-fleets in smart cities: the dawning of a
post-private car era in urban environments? Environ. Sci. Policy 45, 30–40 (2015). https ://doi.
org/10.1016/j.envsc i.2014.09.005
Goodwin, P., van Dender, K.: “Peak car”—themes and issues. Transp. Rev. 33, 243–254 (2013). https ://
doi.org/10.1080/01441 647.2013.80413 3
Harms, L., Jorritsma, P., Berveling, J., Meerkerk, J. Van, Nijland, H.: Carsharing in the Netherlands :
user characteristics and mobility effects. In: 14th World Conference on Transport Research
(WCTR), Shanghai (2016)
Hess, S., Rose, J.M., Polak, J.: Non-trading, lexicographic and inconsistent behaviour in stated
choice data. Transp. Res. Part D Transp. Environ. 15, 405–417 (2010). https ://doi.org/10.1016/j.
trd.2010.04.008
Juschten, M., Ohnmacht, T., Thao, V.T., Gerike, R., Hössinger, R.: Carsharing in Switzerland: identify-
ing new markets by predicting membership based on data on supply and demand. Transp. (Amst)
(2017). https ://doi.org/10.1007/s1111 6-017-9818-7
Kato, H., Inagi, A., Igo, T.: Awareness and potential choices of carsharing: comparative analysis of
data from four Japanese cities. In: Proceedings of the 91st Annual Meeting of the Transportation
Research Board, Washington D.C. (2012)
Kim, D., Ko, J., Park, Y.: Factors affecting electric vehicle sharing program participants’ attitudes about
car ownership and program participation. Transp. Res. Part D Transp. Environ. 36, 96–106 (2015).
https ://doi.org/10.1016/j.trd.2015.02.009
Kim, J., Rasouli, S., Timmermans, H.: Satisfaction and uncertainty in car-sharing decisions: an integra-
tion of hybrid choice and random regret-based models. Transp. Res. Part A Policy Pract. 95, 13–33
(2017a). https ://doi.org/10.1016/j.tra.2016.11.005
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

969
Transportation (2020) 47:935–970
1 3
Kim, J., Rasouli, S., Timmermans, H.J.P.: The effects of activity-travel context and individual attitudes
on car-sharing decisions under travel time uncertainty: a hybrid choice modeling approach. Transp.
Res. Part D Transp. Environ. 56, 189–202 (2017b). https ://doi.org/10.1016/j.trd.2017.07.022
Klincevicius, M.G.Y., Morency, C., Trépanier, M.: Assessing impact of carsharing on household car
ownership in Montreal, Quebec, Canada. Transp. Res. Rec. J. Transp. Res. Board 2416, 48–55
(2014). https ://doi.org/10.3141/2416-06
Le Vine, S., Polak, J.: The impact of free-floating carsharing on car ownership: early-stage findings from
London. Transp. Policy (2017). https ://doi.org/10.1016/j.tranp ol.2017.02.004
Le Vine, S., Adamou, O., Polak, J.: Predicting new forms of activity/mobility patterns enabled by
shared-mobility services through a needs-based stated-response method: case study of grocery
shopping. Transp. Policy. 32, 60–68 (2014a). https ://doi.org/10.1016/j.tranp ol.2013.12.008
Le Vine, S., Lee-Gosselin, M., Sivakumar, A., Polak, J.: A new approach to predict the market and
impacts of round-trip and point-to-point carsharing systems: case study of London. Transp. Res.
Part D Transp. Environ. 32, 218–229 (2014b). https ://doi.org/10.1016/j.trd.2014.07.005
Liao, F., Molin, E., van Wee, B.: Consumer preferences for electric vehicles: a literature review. Transp.
Rev. 37, 252–275 (2017). https ://doi.org/10.1080/01441 647.2016.12307 94
Martin, E., Shaheen, S., Lidicker, J.: The impact of carsharing on household vehicle ownership. Transp.
Res. Rec. J. Transp. Res. Board 2143, 150–158 (2010)
Millard-Ball, A., Murray, G., Schure, J. Ter, Fox, C., Burkhardt, J.: Car-Sharing: Where and How it Suc-
ceeds, TCRP Report, Washington D.C. (2005)
Rose, J.M., Bliemer, M.C.J.: Constructing efficient stated choice experimental designs. Transp. Rev. 29,
587–617 (2009). https ://doi.org/10.1080/01441 64090 28276 23
Rotaris, L., Danielis, R.: The role for carsharing in medium to small-sized towns and in less-densely popu-
lated rural areas. Transp. Res. Part A Policy Pract. (2017). https ://doi.org/10.1016/j.tra.2017.07.006
Schuster, T., Byrne, J., Corbett, J., Schreuder, Y.: Assessing the potential extent of carsharing: a new
method and its implications. Transp. Res. Rec. J. Transp. Res. Board 1927, 174–181 (2005). https ://
doi.org/10.3141/1927-20
Shaheen, S.A., Cohen, A.P.: Carsharing and personal vehicle services: worldwide market develop-
ments and emerging trends. Int. J. Sustain. Transp. 7, 5–34 (2013). https ://doi.org/10.1080/15568
318.2012.66010 3
Shankar, G., Clewlow, R.R., Mokhtarian, P.L., Widaman, K.F.: The effect of carsharing on vehicle hold-
ings and travel behavior: a propensity score and causal mediation analysis of the San Francisco Bay
Area. Res. Transp. Econ. (2015). https ://doi.org/10.1016/j.retre c.2015.10.010
Steg, L.: Car use: lust and must. Instrumental, symbolic and affective motives for car use. Transp. Res.
Part A Policy Pract. 39, 147–162 (2005). https ://doi.org/10.1016/j.tra.2004.07.001
Tal, G.: Evaluating the effect of car-sharing: exploring the gap between what we know vs. what we need to
know and its effect on optimism bias. In: Transportation Research Board 88th Annual Meeting (2009)
Tamor, M.A., Gearhart, C., Soto, C.: A statistical approach to estimating acceptance of electric vehicles and
electrification of personal transportation. Transp. Res. Part C Emerg. Technol. 26, 125–134 (2013).
https ://doi.org/10.1016/j.trc.2012.07.007
Tamor, M.A., Moraal, P.E., Reprogle, B., Milačić, M.: Rapid estimation of electric vehicle acceptance using
a general description of driving patterns. Transp. Res. Part C Emerg. Technol. 51, 136–148 (2015).
https ://doi.org/10.1016/j.trc.2014.10.010
Vermunt, J.K., Magidson, J.: Technical guide for Latent GOLD 5.0: Basic, advanced, and syntax. Statistical
Innovations Inc, Belmont, MA (2013)
Wardman, M., Ibáñez, J.N.: The congestion multiplier: variations in motorists’ valuations of travel time
with traffic conditions. Transp. Res. Part A Policy Pract. 46, 213–225 (2012). https ://doi.org/10.1016/j.
tra.2011.06.011
Zhou, B., Kockelman, K.M.: Opportunities for and impacts of carsharing: a survey of the Austin, Texas
market. Int. J. Sustain. Transp. 5, 135–152 (2011). https ://doi.org/10.1080/15568 31100 37171 81
Zoepf, S.M., Keith, D.R.: User decision-making and technology choices in the U.S. carsharing market.
Transp. Policy (2016). https ://doi.org/10.1016/j.tranp ol.2016.01.010
Fanchao Liao is a Ph.D. candidate in the Transport and Logistics section at Delft University of Technology.
Her main research interests lie in modeling decision-making behavior.
Eric Molin is an associate professor of Travel Behavior Research at Delft University of Technology. He
received a Ph.D. degree at Eindhoven University of Technology and a Masters of Arts degree in Sociol-
ogy at Radboud University Nijmegen, with a specialization in social research methods. He specialized in
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

970
Transportation (2020) 47:935–970
1 3
developing advanced stated choice experiments. Topics of research involve travelers’ preferences regarding
multi-modal transport systems and innovative transport technologies, in particular travel information ser-
vices, alternative fueled vehicles and automated driving.
Harry Timmermans is Head of the Urban Planning Group of the Eindhoven University of Technology, the
Netherlands. He has research interests in modeling decision-making processes and decision support systems
in a variety of application domains, including transportation. He is editor of the Journal of Retailing and
Consumer Services, and serves on the board of several other journals in transportation, geography, urban
planning, marketing, artificial intelligence and other disciplines. He is Co-chair of the International Asso-
ciation of Travel Behavior Research (IATBR), and member of several scientific committees of the Trans-
portation Research Board. He has also served as member of conference committees in transportation and
artificial intelligence.
Bert van Wee is professor in Transport Policy at Delft University of Technology, the Netherlands, faculty
Technology, Policy and Management. In addition, he is scientific director of TRAIL research school. His
main interests are in long-term developments in transport, in particular in the areas of accessibility, land-use
transport interaction, (evaluation of) large infrastructure projects, the environment, safety, policy analyses
and ethics.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center
GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers
and authorised users (“Users”), for small-scale personal, non-commercial use provided that all
copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of
use (“Terms”). For these purposes, Springer Nature considers academic use (by researchers and
students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and
conditions, a relevant site licence or a personal subscription. These Terms will prevail over any
conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to
the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of
the Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may
also use these personal data internally within ResearchGate and Springer Nature and as agreed share
it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not otherwise
disclose your personal data outside the ResearchGate or the Springer Nature group of companies
unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial
use, it is important to note that Users may not:
use such content for the purpose of providing other users with access on a regular or large scale
basis or as a means to circumvent access control;
use such content where to do so would be considered a criminal or statutory offence in any
jurisdiction, or gives rise to civil liability, or is otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association
unless explicitly agreed to by Springer Nature in writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a
systematic database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a
product or service that creates revenue, royalties, rent or income from our content or its inclusion as
part of a paid for service or for other commercial gain. Springer Nature journal content cannot be
used for inter-library loans and librarians may not upload Springer Nature journal content on a large
scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not
obligated to publish any information or content on this website and may remove it or features or
functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke
this licence to you at any time and remove access to any copies of the Springer Nature journal content
which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or
guarantees to Users, either express or implied with respect to the Springer nature journal content and
all parties disclaim and waive any implied warranties or warranties imposed by law, including
merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published
by Springer Nature that may be licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a
regular basis or in any other manner not expressly permitted by these Terms, please contact Springer
Nature at
onlineservice@springernature.com