Introduction

Shared economy, and carsharing services specifically, are on the rise globally (Kent and Dowling 2013; Shaheen and Cohen 2012; Ferrero et al. 2018; Zhou et al. 2020), although many barriers such as uncertainties about service characteristics still limit their use (Matyas et al. 2020). Carsharing contributes to more efficient car use by enabling the utilisation of a car without the necessity of owning one. Carsharing is believed to play an important role in the transition towards more environmentally friendly mobility thanks to its potential to reduce vehicle ownership (Becker et al. 2017; Namazu and Dowlatabadi 2018; Martin et al. 2010; Giesel and Nobis 2016; Le Vine and Polak 2019) and total kilometres travelled (Cervero et al. 2007; Nijland and Meerkerk 2017), and to replace older private vehicles with newer carsharing vehicles with higher fuel efficiency (Baptista et al. 2014). On the other hand, carsharing services substitute for some journeys otherwise undertaken by public transport, an effect most visible by free-floating carsharing (Silvestri et al. 2020).

The positive long-term impact of carsharing highly depends on the numbers of its users and their transport behaviour, among other things. The existing research has focused on carsharing users' typical characteristics to evaluate the market potential of carsharing services and, thereby, indirectly, its environmental impacts. The literature comparing effective or potential carsharing users with its non-users concludes that carsharing currently attracts mostly younger, highly educated people (e.g., Prieto et al. 2017; Becker et al. 2017) with moderate income (Martin et al. 2010), who tend to be more environmentally conscious than the rest of the society (Ramos et al. 2020).

Despite the number of surveys focusing on the profiles of carsharing users, no study, to our best knowledge, has focused on the two following critical issues: (i) attitudes of carsharing users towards sustainable urban mobility policies and their perception of car regulation measures (cf. Brůhová Foltýnová et al. 2020); and (ii) differences in the attitudinal profile, other personal and behavioural characteristics of carsharing users concerning different access to another car in the household. We therefore test the differences between carsharing adopters without another car at their disposal and users who have access to at least one car in their household. Do these two distinct groups differ in their socio-demographic and attitudinal profiles? Do they differ in their motives for joining carsharing and use of (shared) cars? Carsharing adopters with at least one available car in their household have a shared car complementing theirs. They might be more car-dependent, less environmentally conscious and less supportive of alternative modes to cars and restrictive policies tackling individual automobility than carsharing users without a car in their household.

Clarifying the profile of carsharing users in terms of their attitudes to sustainable urban mobility opens up a new field for debate over the potential of carsharing to enhance the political feasibility of restrictive policies against private car ownership and use, as they are the necessary precondition for achieving more sustainable urban futures. We also find this debate crucial because carsharing will never bring more sustainable mobility if a possible decrease in private car ownership and use by some carsharing users would lead to induced traffic (Noland 2002; Bucsky and Juhász 2022) and induced car ownership in other social groups. This might happen in the absence of accompanying restrictive measures with the aim of an overall decrease in car ownership and use. This is particularly the case of large cities where private car transportation in terms of the number of privately owned cars and traffic volumes is hitting its limits due to space constraints. If carsharing membership reduces car use on average (as suggested by Cervero et al. 2007; Nijland and Meerkerk 2017), the rest of society might take advantage of better opportunities for travelling by car thanks to the reduced traffic congestions and therefore travel time costs (a factor typically discussed in connection with the phenomenon of induced traffic). At the same time, as carsharing reduces members’ car ownership on average, it may simplify car ownership for other society members who might take advantage of more accessible parking thanks to decreasing costs of cruising for parking and parking tariffs if those are based on their market value (this phenomenon has not been discussed in the literature so far to our best knowledge). These side effects of new mobility opportunities need to be considered as they are the reason why introducing a policy intervention does not necessarily lead to a decrease in car use, congestion and related harmful environmental effects in total. Hörcher and Graham (2020) develop similar ideas as part of the transport policy debate over the potential introduction of subscriptions to mobility as service passes.

In this study, we analyse data collected from members of a carsharing company with a hybrid regime in between the station-based and free-floating systems and different vehicle types located in the largest cities of the Czech Republic. The Czech Republic is a country with a high vehicle ownership rate and positive attitudes towards cars (Brůhová Foltýnová et al. 2020), where carsharing service expansion has started only recently. Taking advantage of a statistical model using t-tests, we show differences between carsharing users claiming that they do not possess any other car in their household and those who have an additional car at their disposal (private or company car). We focus on their socio-demographic and economic profiles, their reasons for joining carsharing, opinions and attitudes towards mobility policy instruments, differences in use of shared cars and changes in their car use after entering carsharing services using both stated and revealed data. We apply ordered probit regression to test for statistical significance of factors which might affect the behaviour of carsharing adopters in terms of the change in kilometres driven by car as a result of joining carsharing.

The paper is organised as follows. Sect. "Literature review" provides a literature review. Sect. "Methodology" introduces our research methodology, gives information about carsharing in the Czech Republic and the reasons for our case study selection. Sect. "Results" provides results of our analysis, followed by Sect. "Discussion", discussing the results. Concluding Sect. "Conclusion" brings policy implications and future research needs.

Literature review

Many surveys have focused on carsharing users' travel behaviour and characteristics, mainly in Western European countries and North America, although some surveys from developing countries as emerging markets can also be found (e.g., Acheampong and Siiba 2020). The studies rely mostly on statistical analysis of data gathered by questionnaire surveys mainly using logistic regression or cluster analysis (e.g., Ramos et al. 2020; Alonso-González et al. 2020). Some studies complement quantitative data with qualitative in-depth interviews (e.g., Hartl et al. 2018) or rely solely on the content analysis of laddering data by creating a hierarchical value map (Schaefers 2013).

Regarding research into carsharing potential, some studies have analysed solely current carsharing users (e.g., Martin et al. 2010), some have aimed at intentions of current non-users to use carsharing services (Efthymiou et al. 2013; Prieto et al. 2017), and other studies have combined surveys of carsharing users and non-users (e.g., Burghard and Dütschke 2019; Ramos et al. 2020). Hjorteset and Böcker (2020) distinguish the following phases of enrolment in carsharing services: interest in carsharing, intentions to enrol, and the real decision to enrol.

The effects of carsharing on transport behaviour and the environment vary according to the type of carsharing services. A company offering two-seat cars in a free-floating system seems to further ease the lives of the members of car-owning households by providing them with more mobility alternatives. In contrast, a station-based carsharing company with different vehicle types seems to provide a better alternative to car ownership (Namazu and Dowlatabadi 2018), results also supported by Becker et al. (2017; 2018). Münzel et al. (2019) investigated differences in characteristics and motives of users of peer-to-peer and business-to-consumer modelsFootnote 1 in the Netherlands. Their findings suggest that adopters of both types of carsharing services are relatively similar in their characteristics but differ in the frequency at which they use carsharing.

Literature also shows a big dispersion in the replacement rates and effects of carsharing on private car ownership. Cervero et al. (2007) estimate that one shared car may replace nine private cars, Martin and Shaheen’s (2010) estimate is nine to thirteen private cars, Stasko et al. (2013) estimate the rate as high as one to fifteen. However, revealed preference studies among car sharing users that are often used in the papers estimating replacement rates may result in self-selection bias and lead to overly optimistic conclusions (Zhou et al. 2020). When the research is based on actual car registrations and empirical impact instead of answers from questionnaires, the results may differ. Kolleck (2021) investigated the impact of carsharing services in 35 German cities and concluded that the replacement rate for station-based carsharing is similar to other studies (one to nine) but there is no reduction in car ownership in free-floating systems. Zhou et al. (2020) used a sample of the general public and found no impact of carsharing on respondents’ vehicle ownership decisions.

Socio-demographic characteristics of typical carsharing users

A bulk of studies have explored socio-demographic characteristics such as age, education or household size and their relationship to the adoption of carsharing services. Efthymiou et al. (2013) found that younger people up to 25 years of age tend to use carsharing services more than older ones. These findings are also supported by Martin et al. (2010), who found that carsharing members were relatively young (the majority up to 30 years), well-educated (the majority holding a bachelor or master degree) with moderate incomes (most of the respondents up to 80 thousand USD per household of annual income, which means slightly above the median household income). Prieto et al. (2017), Becker et al. (2017) and Hjorteset and Böcker (2020) reported that most carsharing users were young and highly educated males, while Burghard and Dütschke (2019) found the highest use of carsharing among young car-free couples without children or young parents (these families tend to use carsharing as a supplement to their car). The role of education in carsharing use was also confirmed by Münzel et al. (2020). Hjorteset and Böcker (2020), Alonso-González et al. (2020) or Becker et al. (2017) conclude that carsharing users also own fewer private cars. To summarise, there is a general agreement that being a male, younger, more highly educated, and middle-income person increases the probability of carsharing adoption. The type of carsharing service (free-floating, station-based) is either not distinguished in most of the above studies or does not seem to play an important role in terms of the difference in socio-demographic characteristics among users of particular types of service. The conclusions of Becker et al. (2017) from free-floating carsharing are in line with other authors. Prieto et al. (2017) and Münzel et al. (2020) focused on comparison of business-to-consumer and peer-to-peer carsharing and did not find any significant differences in terms of socio-demographic characteristics between the two.

Attitudes and beliefs

General attitudes and personal beliefs also characterise a typical carsharing user while focusing on the association between environmental consciousness and interest in carsharing (e.g., Hjorteset and Böcker 2020). The environmental profile of carsharing users and non-users was further analysed by Ramos et al. (2020), who found that a larger share (81%) of carsharing users claim deep environmental concerns about car use in comparison with carsharing non-users (57%). Even within these shares, carsharing users tend to express deeper environmental concerns. The deeper environmental consciousness of carsharing users might be connected to the image of carsharing as being more environmentally friendly than car ownership and in particular with peer-to-peer systems being superior over business-to-consumer in this perception among carsharing users (Hartl et al. 2018). These results are in line with the findings of Burkhardt and Millard-Ball (2006). They are also supported by Clewlow (2016), who showed that carsharing users own more environmentally friendly cars on average (electric vehicles or other alternatives), if any, compared to the average population. However, it is unclear whether these consumers adopt these technologies based on their higher environmental sensitivity or experience with such cars in the carsharing system. On the other hand, Lamberton and Rose (2012) did not find that environmental concerns increase the likelihood of carsharing use, but that might be caused by the relatively low number of carsharing users in their sample.

Motivation for joining carsharing services

Motives for joining carsharing services are another broadly studied area of interest. Previous research usually acknowledges that the strongest reasons to join carsharing services are associated with cost savings (Lamberton and Rose 2012; Schaefers 2013) and the convenience that allows carsharing users to have a car whenever they need one, while at the same time avoiding maintenance and repair responsibilities (Schaefers 2013; Ramos et al. 2020). Contrarily, Hjorteset and Böcker (2020) did not find the financial criterion to be supportive of the intention to adopt carsharing.

Environmental consciousness as a direct motivator for enrolment in carsharing services has also been surveyed. Current results indicate that although carsharing users tend to be more environmentally conscious (see Hjorteset and Böcker 2020), their environmental attitudes are not the primary motivators of their enrolment in carsharing services (Hartl et al. 2018; Schaefers 2013). Moral satisfaction from making a greener choice is viewed as a "bonus" to other advantages, mostly cost savings. Emotional values are not as strong drivers of consumers' decisions as functional ones (Hartl et al. 2018). Another motivating factor discovered by Schaefers (2013) in the study of a free-floating system was the desire for self-expression by the visibility of the use of shared cars, labelled as a lifestyle motive.

Travel patterns

Some studies have also surveyed the travel patterns of carsharing users and their utilisation of shared cars. Recent statistics on European carsharing use from 22 operators of various types of carsharing services were summarised by Rodenbach et al. (2018), who found that carsharing is equally popular for short distances up to 10 km and medium/long distances over 50 km, with an average distance of 46 km per trip and the median being 25 km. In terms of average trip duration, there is a clear distinction between short-term and long-term carsharing. Over half of the trips were shorter than 60 min, but another 40% of trips were over three hours, resulting in a big difference between the average time of 412 min and the median of 49 min. Sprei et al. (2019) analysed free-floating carsharing data from 12 European and American cities. The average trip duration was slightly less than half an hour for the cities studied, while the average trip distance varied between 1 and 10 km in geometric distance between pick-up and drop-off locations instead of the actual distance travelled.

Methodology

Case study selection

Shared mobility is still a pioneer concept in the Czech Republic. Although it has been growing substantially in recent years, the volume of carsharing services is still low. There were nine shared cars in 2012, while five carsharing companies provided 1520 shared cars in 12 cities in 2022Footnote 2 (peer-to-peer carsharing is excluded but offers over 1500 additional vehicles). The current supply includes carsharing for individuals, companies and university students, offering conventional and electric cars and scooters. These activities have spread in the society with the extensive growth in car ownership–the number of cars in the Czech Republic has more than doubled since 1990, exceeding 6 million (the car ownership rate was 568 cars per 1000 inhabitants in 2020 compared to 233 car per 1,000 inhabitants in 1989; MoT CR 2021).

Our survey targets clients of the oldest Czech carsharing company “Autonapůl”, which was established in 2003. Nowadays, it operates 73 cars in 9 Czech cities; its car fleet is 2.5 years old on average. This company offers a combination of a free-floating and a station-based service, as with the exception of two-way carsharing, the clients can also use the option of returning the car elsewhere for an extra fee. The detailed conditions of the service vary across cities. The two-way carsharing system offered by Autonapůl is more flexible than a typical station-based system, as the company defines zones which act as “giant” stations for returning the car wherever convenient within such a zone (for instance, close to the client’s residence when returning the car). This flexibility is enabled by municipal policies usually offering free residential parking places for shared cars except reserved parking places. The zones usually cover two metro stations in the capital or an area similarly accessible by other public transport modes in other cities. For all trips, clients need to reserve a car in advance. Autonapůl has both company and individual clients. To answer our research questions, we focused on the latter group in our survey. The private clients use carsharing dominantly for non-business trips (the share of business trips in the total number of trips made by shared cars is 14% in Autonapůl).

Data collection and analysis

Data were collected using an electronic questionnaire distributed by Autonapůl directly to its members (in total, 770 e-mail addresses were contacted). We further used the company’s shared car use history database for the period between 2018 and 2019 (twelve months in total). The database provided individualised anonymised data on time and kilometres driven for all the car trips made. These revealed data were linked to individual responses of carsharing members from the questionnaire. The questionnaire was prepared using the SurveyMonkey software. The data collection took place from 18 September to 16 October 2019 with one e-mail reminder in the meantime, which increased the response rate by about one third. In cooperation with the carsharing company, we offered respondents who completed the questionnaire a chance to win gift vouchers for carsharing services as an encouragement to join the survey. The average time spent completing the questionnaire was slightly under 15 min.

We eventually obtained 366 completed questionnaires, which corresponds to a response rate of 47.5%. In this paper, we use only data from Czech-speaking respondents (N = 336) so that any potential non-standard behaviour of non-resident carsharing users is excluded (the questionnaire was developed separately in Czech and English languages so that we were able to collect data also from non-Czech speakers). Another 20 respondents did not reply to the question on the number of cars available in their household. These were also excluded from the analysis.

Our research focuses on two specific subgroups of carsharing users in relation to the availability of another car in the household: carsharing users without a car at their disposal and users who have access to at least one car in their household. The car at their disposal may be owned by the respondent or a household member or be a company car available for private use. To test for statistically significant differences between these two groups of carsharing users, we apply the standard statistical analysis of Welch’s two-sample t tests, where the tested variable is related to a yes/no binary variable expressing the availability of at least one additional car in the household beyond the shared car.

We tested the data for the non-response bias, as the respondents answering the questionnaire are often more active carsharing users than those who do not participate in the survey (Martin and Shaheen 2010). The Mann–Whitney test applied to our data confirmed a difference between these two groups (respondents and non-respondents). Even though the mean of kilometres driven by shared cars annually was almost identical for both groups (0.1% difference), the data differed significantly in the variance, which was almost twice as high in the group that did not participate in the survey. We expect that this bias is caused by two facts. Firstly, the carsharing company also offers vehicles to companies for business trips. These companies were excluded from the survey but their rides were included in the data about car trips without the possibility to distinguish them from rides of individual clients. Secondly, the clients only pay for the use of service without an obligation to pay a regular fee, which can result in a high number of clients who only used the service for a short period of time, use it very occasionally or had even terminated the membership before the survey took place. As a result, there is a higher number of users with very low annual kilometrage and several with a very high annual kilometrage in the group that did not participate in the survey. Since we did not compare the results to the general population but instead we focused on a comparison of the two subgroups of carsharing users (with and without a private car in the household), we do not expect any significant impact of this bias on our results.

The analysis of respondents’ attitudes to various transport modes and sustainable mobility measures covers the following topics: their opinions on and attitudes to car-restrictive measures in urban areas, support of alternatives to cars, individual freedom of modal choice (rejection of any regulations), perception of cars as a status symbol, and the potential of carsharing to reduce negative consequences of car use. The attitudes were measured using a set of statements developed for this study and tested for internal consistency using Cronbach's α (see Table 1). Respondents expressed their agreement with the statements on a 5-point Likert scale. The first two variables were calculated as a mean of multiple-scale items. Single statements measured variables capturing freedom of modal choice, the car as a status symbol and personal views on carsharing as reducing negative consequences of automobility. These possible motives are in line with most of the studies analysing the reasons for joining carsharing services (see Sect. "Literature review").

Table 1 Measurement details of variables characterising motives and attitudes
Full size table

In the regression model described in Sect. "Factors affecting change in kilometres travelled by car after joining carsharing", we explain what factors predetermine the change in the kilometres driven by car after joining carsharing. The variables listed below entered the model:

  • Change in the kilometres driven by car after joining carsharing (dependent variable): categorical variable with three categories (stated data): travelling approximately the same number of kilometres (used as a reference in the model), travelling fewer kilometres, and travelling more kilometres by car (either as a driver or as a passenger) after joining carsharing.

  • Kilometres driven by shared cars: revealed data on annual use of shared cars in total kilometres driven, provided by Autonapůl carsharing company.

  • Frequency of carsharing use: revealed data on annual use of shared cars in number of borrowings, provided by Autonapůl carsharing company.

  • Frequency of car use for private purposes now / before joining carsharing: stated data as an ordinal variable in frequency of car use for private purposes without taking into account trips made by shared cars, treated as an interval variable by using category mid-points (categories were at least 3 times a week ‒ once a week ‒ once a month ‒ twice a year ‒ once a year or less).

  • Car availability in the household: binary variable; 1 = at least one car is available in the household (either privately owned or as a company car).

  • Disposing of a car when joining carsharing: binary variable; 1 = the household got rid of a car by selling it or disposing of it in another way in connection to joining carsharing.

  • Avoiding buying a car when joining carsharing: binary variable; 1 = the household avoided buying an additional car thanks to joining carsharing.

  • Duration of carsharing use: stated data about the use of any carsharing company services in years, ordinal variable, treated as an interval variable by using category mid-points (categories were less than half a year ‒ 1 year ‒ 3 years ‒ 5 years ‒ more than 5 years; the mid-point for the last category was derived from the total number of years of the existence of Autonapůl, the oldest carsharing company in Czechia, and overall trends in the numbers of their clients).

  • Living outside a wider city centre: based on stated data in response to the question “How would you describe the place of your residence?” The category “Living outside a wider city centre” excludes categories “city centre” and “wider city centre”, and includes categories “city outskirts”, “village or small town” and “solitary house”.

  • Time to the nearest shared car: stated data about a typical time spent using the typical transport mode to the nearest shared car (generally, respondents use public transport or go on foot to the nearest shared car; they use a bicycle less frequently), an ordinal variable treated as an interval variable by using category mid-points (categories were less than 5 min ‒ 10 min ‒ 20 min ‒ 30 min ‒ more than 30 min; the mid-point for the last category was set to 45 min as not more than one hour can be expected as the pick-up time for a shared car).

  • Personal change after entering carsharing: binary variable; 1 = at least one type of personal change occurred as a reason for carsharing enrolment – moving, child birth, change of occupation or unspecified other change.

Results

Descriptive statistics

Our sample (316 respondents) is dominated by men (82.6%), with a majority of university-educated people (74.3%). This gender imbalance reflects the structure of the Autonapůl carsharing clients. Most of the respondents belong to the age category of 35–49 years (44.3%) or 18–34 years (43.4%). In comparison with the general population, ¾ of the sample have an average or below-average monthly income (approximately 1,200 EUR in the year of data collection) and they own fewer cars than the general population. The sample description and its comparison with the general population is summarised in Table 2, descriptive statistics of revealed data variables is presented in Table 3.

Table 2 Sample description
Full size table
Table 3 Descriptive statistics of revealed data variables
Full size table

Most respondents come from eight large Czech cities, while there is a dominance of respondents from Brno (44%) – the city where Autonapůl originally started – and Prague (30%). 66.8% of the respondents live in the city centre or wider centre; more than a quarter (27.8%) live in the suburbs. Over three-quarters of the respondents (77.8%) stated that they were members of Autonapůl carsharing only and had no previous experience with other carsharing companies.

Motivation to use carsharing and attitudes towards different transport modes

The respondents expressed the importance of their reasons for joining carsharing using a 5-point scale (5 = the main reason, 1 = no effect on the decision to join carsharing). Figure 1 demonstrates the median answer (bold line) and spread via 1st and 3rd quartile of each parameter (in this case, expressed importance of a reason) graphically. As Fig. 1 shows, the main reason for joining carsharing is “having access to a car without having to manage my own one” (mean = 4.17; we show means in parentheses below), followed by the reason “having a car available whenever I need one” (3.64). Other important motives include “being eco-friendlier” (3.47) and “reducing car use costs” (3.46). Contrarily, "having access to different car types” (2.52) is a factor less important for joining carsharing. The data indicate that carsharing members strongly prefer having a car available without the need to take care of their own car (which is also consistent with Schaefers 2013; and Ramos et al. 2020). Respondents also considered the economic and environmental consequences of car use. On the contrary, the flexibility in having access to different types of cars is not very important to them.

Fig. 1
figure 1

Reasons for joining carsharing

Full size image

In terms of their attitudes towards various transport modes and sustainable mobility measures, the respondents in general agreed that carsharing is an environmentally friendly mode of transport (4.29), they support car restrictions in urban areas (3.90) and measures supporting alternatives to cars (4.06). They also did not agree on the car being a status symbol (1.90). On the contrary, there was a disagreement about how strict the regulations of transport mode choices should be in terms of the freedom of transport mode choice (2.90). For more details, see Fig. 2.

Fig. 2
figure 2

Carsharing users’ attitudes towards sustainable urban mobility

Full size image

Differences in users' characteristics and attitudes in relation to car availability and avoidance of car purchase

This section focuses on differences between the following subgroups of carsharing users: (i) those who have a car available in their household and those who do not; (ii) those who avoided a car purchase thanks to enrolling in carsharing and those who did not.Footnote 3

The subgroups presented in Table 4 based on the availability of another car for private use and the avoidance of car purchase reveal only slight and statistically insignificant differences regarding their income, the number of household members and the number of children in households. Only those who avoided car purchase show surprisingly higher wealth than the rest. For the details, see Table 4.

Table 4 Comparison of carsharing users based on (a) availability of another car for private use; (b) avoidance of car purchase
Full size table

There are statistically significant differences between the analysed groups regarding carsharing use. The subgroup without any other car available uses carsharing substantially more often (on average 26 borrowings per year compared to 15 annual borrowings for those who have another car at their disposal), drives more kilometres using shared cars (on average 2082 km per year compared to 1374 km annually) and borrows shared cars for a longer time (on average 268 h per year compared to 119 h per year). Similarly, the subgroup of those who avoided the purchase of a car as a result of their enrolment in carsharing drives more kilometres per year (on average 2358 km per year compared to 1640 km annually) and uses the shared car more often in comparison to those who did not claim the avoidance of a car purchase (on average 35 borrowings per year compared to 18 borrowings).

The reasons for joining carsharing vary only slightly between the two groups. Those without a car have a statistically stronger preference for “having access to a car without having to manage my own car” (4.44 compared to 3.50), they put more emphasis on being eco-friendly by utilising carsharing (3.62 to 3.27), and they also care more about costs of car use (3.65 to 3.07). On the contrary, the respondents with a car available value access to different car types more (2.78 compared to 2.36). In relation to the avoidance of car purchase thanks to enrolling in carsharing, the statistically significant differences are also visible. Those who avoided a car purchase claim higher interest in having a car always available (3.79 to 3.48), and having access to different car types (3.05 to 2.32). Similarly to the car-free households, they perceive carsharing as an eco-friendly use of cars more likely (3.78 to 3.38) and one of their strongest motives is to reduce costs of car use (3.81 to 3.27).

Within subgroups in relation to car availability in the household, subtle but statistically significant differences can also be observed in attitudes. The subgroup without another car available are stronger supporters of car-restrictive measures in urban areas (4.04 compared to 3.63) and of prioritising alternatives to cars (4.19 compared to 3.79). These respondents also believe more in the positive environmental impacts of carsharing (4.40 compared to 4.04). On the contrary, they express less support to statements that the choice of transport mode is a freedom which should not be limited by any regulation (2.75 compared to 3.23). No statistically significant result was found within the subgroups in relation to (non)avoidance of car purchase.

Factors affecting change in kilometres travelled by car after joining carsharing

A majority of respondents claimed they changed their travel behaviour after joining carsharing regarding the number of kilometres driven (whether using a shared or a private car). This raises the question what are the factors influencing this change and in which direction. In our survey, respondents replied how their number of kilometres travelled by car changed (either as a driver or as a passenger) after joining carsharing: travelling fewer kilometres, approximately the same number of kilometres and more kilometres. We do not model these choices as ordinal. The reason is that factors behind a decrease in travelling by a car after joining carsharing may differ from factors behind an increase. We thus interpret the dependent variable as discrete and nominal, and we therefore decided to apply a multinomial logistic regression.Footnote 4 The reference category is “no change in kilometres”, the coefficients thus indicate whether the variable increases or decreases the probability of the two other responses (“increase in kilometres travelled” and “decrease in kilometres travelled”).

The results of the multinomial logistic regression are summarised in Table 5. None of the claimed reasons concerning carsharing adoption was found to have a robust statistically significant effect on the overall change in car use; these variables describing reasons for joining carsharing were therefore excluded from the final model.

Table 5 Multinomial logistic model results for change in kilometres driven by car after joining carsharing
Full size table

The model identifies the following statistically significant variables describing the change in kilometres travelled by car after joining carsharing: the frequency of trips by private car (both currently and also before joining carsharing), availability of a car in the household, whether the vehicle has been sold or otherwise disposed of after joining the carsharing, and typical travel time to the nearest shared car. The results indicate that a higher frequency of car use for private purposes now (after joining carsharing) increases the probability of reporting more kilometres after joining carsharing; at the same time, it decreases the probability of reporting fewer kilometres travelled by car for private purposes in total compared to no change. Respondents who often travelled by private car before joining carsharing have a higher probability of reducing the total number of kilometres driven after joining and a lower probability of increasing the overall kilometrage. Those who have a car at their disposal within their household have a lower probability of decreasing kilometres driven after joining carsharing. However, car availability is not a significant predictor of the probability of increasing the number of kilometres relative to no change. Getting rid of a car when joining carsharing (i.e., selling it or disposing of it in another way in connection to joining carsharing) is a statistically significant predictor of decreasing overall kilometres driven by car after joining carsharing. And finally, and in this case rather counterintuitively, but on a relatively low level of statistical significance, the respondents who have a longer trip duration to a shared car show a higher probability of increasing the kilometres travelled by car after joining carsharing.

Avoidance of buying a car after joining carsharing did not lead to a statistically significant change in the total kilometres driven by the respondents. This is consistent as those who did not reduce the number of cars in their ownership in connection to enrolment in a carsharing system cannot be expected to reduce their car use significantly. These respondents might even be expected to increase their car use if their avoidance of buying a car as a result of having the opportunity to share a car would be connected to a change in personal matters, such as a child birth.

We also controlled for the duration of carsharing use, as some studies suggest that the duration of carsharing enrolment might affect the overall car use in connection to the speed of different adaptation processes, typically the abolishment of private cars being a slow adaptation process (Firnkorn and Shaheen 2016). Also, according to Cervero et al. (2007), in the long run, carsharing might build more sustainable habits among carsharing users through their enrolment. On the other hand, the study of Jain et al. (2020) leads to the opposite conclusion of increasing car use in time. None of these findings were confirmed by our study as the variable duration of carsharing use was not statistically significant. Another control variable, living outside the wider city centre, also did not prove to be statistically significant in the model.

The non-significance of the variable personal change connected to entering carsharing may be explained by the fact that the personal change may be associated with an increase in car use, as it is typically a child birth in the household (5.6% of respondents), or a decrease in car use. Some of the respondents faced personal changes which might have led both to a decrease or increase in car use, such as a change in occupation (8.6% of respondents) leading to less/more commuting or moving the household (14.3% of respondents). The non-significance of both control variables connected to the use of shared cars seems surprising, either from the perspective of kilometres driven by a shared car or the frequency of shared car use, as we would expect these variables to have statistical significance on the change in total kilometres travelled by car in comparison to before joining carsharing.

Discussion

Our analysis of the profiles of carsharing users, in general, confirms the results of previous studies (Millard-Ball et al. 2005; Harmer and Cairns 2011; Loose 2010) that carsharing is more attractive to men than women, carsharing users are generally more educated, young or middle-aged, of moderate income, and own fewer cars on average. In terms of motives for joining carsharing, our findings also support the previous research in that the most significant motivators for joining carsharing are no need to manage one’s own car and having a car always available (Schaefers 2013; Ramos et al. 2020), as well as reducing car use costs (Lamberton and Rose 2012; Schaefers 2013). In our study, environmental friendliness also seems to belong among the primary motivators for joining carsharing. In terms of attitudes, we also confirm previous results that carsharing users perceive carsharing as an environmentally friendly transport mode as it can reduce many of the adverse effects of automobility.

This study highlights differences between user groups with and without an additional car in their household. Surprisingly, these groups do not seem to differ statistically significantly in their socio-demographic profiles. Households without an additional car are not statistically smaller or poorer than households using a shared car to complement privately owned cars. Apparently, with increasing mobility opportunities thanks to carsharing, families raising children can easily get along without a privately-owned car.

Statistically significant differences between user groups with and without an additional car appear in their motives for joining carsharing, their travel behaviour using shared cars and their attitudes. Some of these results are straightforward and expectable. Users without an additional car in their household use a shared car more often, for more kilometres and hours annually. They tend to be more motivated to become carsharing members due to gaining access to a car without having to take care of a privately owned one, reducing car use costs and becoming more environmentally friendly when using carsharing in comparison to a private car. Users with an additional car in their household demonstrated that having access to different car types was a slightly more important reason for joining. On the other hand, this motive was not decisive for either of the user groups.

Regarding carsharing users' attitudes towards sustainable urban mobility policies and their perception of car regulation measures, statistically significant differences between the defined groups were revealed. Users without an additional car in their household are more environmentally conscious and inclined towards sustainable urban mobility measures. These users also believe more firmly that carsharing can reduce many of the adverse effects of automobility. Users with an additional car in their household, contrary to the former group, are inclined to believe that the choice of means of transport should be entirely at each person's discretion and cannot be restricted by the state or any other institution.

An explanation of these differences can be derived from the availability of an additional car in the household. Carsharing members without an additional car in the household tend even more to support car alternatives and car-restrictive measures as both these approaches are clearly in their favour as well as in correspondence with their overall attitudes, whereas those having an additional car in their household might find some of these measures partly disadvantageous for themselves. On the other hand, both groups of carsharing users believe that carsharing is an environmentally friendly transport mode and both support car restrictions and alternatives to cars. Neither group perceives cars as a status symbol.

It remains an unanswered question whether carsharing helps its members form their attitudes towards urban mobility measures. The high acceptance of restrictive policies by all carsharing users may be a result of their relatively high level of environmental consciousness and awareness of the environmental consequences of private car use (e.g., Eriksson et al. 2006; Ünal et al. 2019) resulting from self-selection bias in joining carsharing (e.g., Hjorteset and Böcker 2020; Ramos et al. 2020). On the other hand, previous research indicates that trials can increase public acceptance of restrictive policies (Eliasson and Jonasson 2011), and carsharing might bring such an experience of private car mobility without owning a car.

In terms of car ownership impacts, we confirm the results of previous studies finding a decrease in car ownership among users of shared cars compared to the situation without available carsharing services. Car ownership of more than 40% of the respondents decreased compared to the situation without available carsharing services (either by selling their car, reported by 16.7% of respondents, or not buying a new one, claimed by 25.3%).

A different situation occurs with car use. The overall amount of kilometres travelled recorded various trends in relation to carsharing members’ characteristics. Namely, 33.5% of respondents claimed a decrease in kilometres travelled by car thanks to joining carsharing, while 37.8% claimed an increase in kilometres travelled. The availability of at least one additional car in the household is a strong predictor for not decreasing overall car use after joining carsharing. If carsharing adopters disposed of their car due to their carsharing membership, they tend to travel less by car after joining carsharing in comparison with the period prior to their enrolment in carsharing. Also those who travelled by private car relatively more often before joining carsharing have a higher probability of reducing the total number of kilometres driven after joining. On the contrary, those who utilised a private car rarely before joining carsharing tend to increase their car use thanks to the new opportunity for travelling by shared cars. Frequent users of cars for private purposes and those having at least one car in the household do not reduce the total kilometres driven after joining carsharing. These results are in line with the conclusions of Chapman et al. (2020), who found a significant reduction in car ownership among carsharing users as a necessary precondition for a positive influence of carsharing on reducing overall car use.

On the other hand, the results of our analysis are limited to the respondents’ claims about their overall trends in car use compared to the period before joining carsharing (if they travel with the car less, more kilometres or approximately the same number of kilometres annually) which does not allow us to consider individual differences in magnitudes of these trends. The majority of the variables entering our model was based on stated data, which further limits the plausibility of the analysis and opportunity for generalisation of our results. At the same time, our sample is limited to one carsharing company with a relatively small car fleet. As we described in Sect. "Data collection and analysis", our results may be negatively affected by the non-response bias of our respondents, as we did not capture mainly clients who use shared cars only rarely or have stopped using carsharing. On the other hand, in this case, we believe that this type of non-response bias did not influence our results based on the comparison of the sub-groups in our sample.

Our results on trends in car use correspond to Martin and Shaheen (2011), who found that if members dispose of their car, they largely reduce their car use. On the other hand, it does not mean that carsharing would not be an environmentally viable alternative for households owning a car too. Carsharing adoption responds to the increasing demand for car use of such households and might prevent an even larger increase in car use if an additional car was bought instead (Jain 2022). For these persons, carsharing will always be a more environmentally friendly way of car use than a new privately owned car. Indeed, some of the respondents (25.3%) claimed that carsharing led them to avoid buying a new car.

Conclusion

This study brings new insights into the profile of carsharing users in relation to the availability of an additional car in their households. It also touches the domain of carsharing users’ attitudes towards sustainable urban mobility measures, which we find crucial within the debate over the political feasibility of restrictive policies against private car use. We distinguish users without another car at their disposal and users who have access to at least one car in their household.

Although carsharing has a large potential to be an environmentally friendly way of car use, increasing the number of carsharing members cannot be considered a step towards more sustainable urban mobility per se. According to our results, carsharing helps support sustainable urban futures mainly if used by households without access to an additional car in their household. This factor, together with getting rid of a car when entering carsharing, is a strong predictor of decreasing kilometres travelled by car. At the same time, however, households with rare utilisation of cars prior to joining carsharing tend to increase their car use after enrolment. Households without access to an additional car seem to be less car-dependent on average than users utilising carsharing to extend their car use.

We must stress that the overall environmental impact of the introduction and rise of carsharing largely depends on other people’s induced car ownership and traffic as carsharing non-users might potentially take advantage of better access to parking as well as travelling if carsharing members reduce the number of cars owned by them and kilometres travelled by car. Therefore, a precondition for more sustainable urban futures thanks to carsharing is to accompany the rise of carsharing with restrictive measures against private car ownership and use, such as reductions to available parking capacity for private car transportation in urban environments accompanied by an increase in parking tariffs or various economic tools such as taxes imposed on car ownership, policies which policymakers often avoid for their low public acceptance.

Using these findings, we therefore aim to open the debate over the potential of carsharing to provide support to highly needed restrictive policies and increase their political feasibility. We find carsharing users to be highly positive towards pull as well as push urban mobility measures. Although those without an additional car in their household are statistically even more inclined towards supporting these measures, both carsharing member groups tend to agree with support of environmentally friendly transport modes as well as restrictions of private car use. We would like to encourage further research on the possible effect of the carsharing experience as stimulating people’s beliefs that a privately owned car is unnecessary for a high quality of urban life and as helping enhance the political feasibility of rather unpopular push measures that make car ownership and use less attractive. Further research is needed especially to find out whether carsharing experience leads to higher acceptance of car-restrictive measures for both groups of carsharing users – those who do not possess another car in their household and those who do.