1 Introduction

In recent years, the phenomenon of sharing economy has emerged in many industries worldwide and businesses leveraging the sharing economy have flourished [1, 2]. Sharing Economy denotes the “collaborative consumption made by the activities of sharing, exchanging, and rental of resources without owning the goods” [3]. The collaborative consumption is the peer-to-peer-based activity of obtaining, giving, or sharing the access to goods and services, coordinated through community-based online services [1, 2]. Sharing is a phenomenon as old as humankind, while collaborative consumption and the “sharing economy” are phenomena born of the Internet age [1]. With the advancement of Information technology (IT), we have witnessed a flurry of emerging collaborative consumption, including sharing rooms (e.g. AirBnB), sharing cars and bikes (e.g., Relay Rides, Wheelz), and taxi services (e.g. Uber, Didi), etc.

Prior research has identified three key drivers of the sharing economy and collaborative consumption, which are changing consumer behavior, social networks and electronic markets, mobile devices and electronic services, which enabled the online interaction among consumers/users and collaborative consumption [4, 5]. Accordingly, existing research on collaborative consumption mainly focus on motivations of participation in the collaborative consumption with an emphasis on factors related to social network, electronic market, mobile devices and services, including trust, reputation of the platform, social capital/social ties [e.g. 6,7,8]. However, the first driver, i.e., changing consumer behavior has been largely ignored in prior literature, the existing literature on factors related to the first driver, remain insufficient.

Although the importance of collaborative consumption has been widely recognized, a comprehensive, yet theoretically solid framework of motivations to participate in collaborative consumption is still missing in the literature. Value is a central concept in consumer behavior and it directly explains why consumers choose to buy or avoid particular products or services [9,10,11]. Prior research also suggest that consumer value can be equally important for Internet commerce because it is critical to consumption behavior [12].

Therefore, to establish the theoretical linkage between collaborative consumption and consumer value, our study propose a research model to explain why consumers participate in collaborative consumption from a value co-creation perspective. Based prior literature on collaborative consumption and literature on consumer value and value co-creation, we identify five factors as key determinants of attitude towards collaborative consumption, including economic value, social value, entertainment value, convenience value, and trust.

2 Literature Review and Hypothesis Development

2.1 Collaborative Consumption

Collaborative consumption is not a niche trend anymore [5]. Instead, business leveraging collaborative consumptions has been flourished, for its large scale, large user volume, and profitable trend [6]. Table 1 synthesizes some recent research in collaborative consumption.

Table 1. Summary of selected literature
Full size table

While collaborative consumption has previously occurred mostly among close relationships such as family, kin, and friends, the Internet is deemed to engender the opportunity to engage more strangers in this activity [1]. Hence trust has been identified as a key factor in collaborative consumptions. In a study of Airbnb, Ert et al. suggested that the more trustworthy the host is perceived to be from her photo, the higher the price of the listing and the probability of its being chosen [14]. Tussyadiah also highlighted the importance of trust in peer-to-peer accommodation rental services [13].

Economic considerations, such as utility has also been identified in prior literature as a key determinant of consumer participation in various types of collaborative consumptions, such as car sharing [5] and accommodation rental [13]. Lamberton and Rose also showed the relevance of costs and benefits of sharing in promoting commercial sharing [7]. However, participating in collaborative consumption is not simply tied to a set of economic aspects of consumption, but also depends on the nature of the functional, social and individual utilities of the certain collaborative consumption. Factors such as sustainability and community have been identified [2, 13]. Nevertheless, research on the social aspect of collaborative consumption has largely been limited.

2.2 A Value Co-creation Framework for Collaborative Consumption

For consumption values which directly explain why consumers choose to buy or avoid particular products [9, 10], different types of values might play their unique roles in shaping consumers’ purchase choices. Participating in collaborative consumption is not simply tied to a set of economic aspects of consumption, but also depends on the social aspect in collaborative consumption.

On top of the traditional conceptualization of consumer value, the collaborative consumption in the current sharing economy environment often reflects the nature of value co-creation instead of value delivered largely by the firm themselves. Very often, firms do not create and deliver value to the passive customer, but rather through interaction and dialogue embeds value in the co-creation process between the firm and its active customer [15,16,17]. This moves the focus to a process of co-creating value through the exchange of resources with other consumers or partners to co-construct unique experiences [17, 18].

A review of the existing literature on collaborative consumption and consumer value shows that in comparison with social aspects, utility orientation dominates collaborative consumption research. Even for the limited social-related research, the positive aspect of social value, often reflected as maintaining interpersonal interconnectivity and social enhancement value [19] is dominated. However, in the offline context, people are particularly prone to avoid interacting with strangers due to “stranger danger” [1] and feelings of anxiety and uneasiness to meet and interact with them [20]. Similarly, not all consumers enjoy interacting with strangers in collaborative consumptions.

Hence, a comprehensive model which includes all relevant dimensions – psychological and functional needs – that constitute the value of collaborative consumption is still lacking. Considering all different aspects that constitute a customer’s perception of and willingness to participate in collaborative consumption, it is important to combine a set of value dimensions into one single framework, rather than treating each perceived value separately. Therefore, we propose a value framework in explaining consumer participation in collaborative consumptions, economic value, social value, entertainment value, and convenience value have been identified as key value dimensions. In line with prior literature, trust and perceived personal innovativeness have also been included in the proposed model. The research model and hypotheses are shown in Fig. 1.

Fig. 1.
figure 1

Research model and hypotheses

Full size image

Prior literature has identified economic gains as essential in gaining satisfaction and increasing the likelihood of people’s choosing collaborative consumption services [2]. Collaborative consumption is perceived as offering more value with less cost [6, 21]. Therefore, we propose that:

  • H1. Economic value is positively associated with consumer’ attitude towards participating in collaborative consumption.

Glind [58] highlighted that the main intrinsic motivation of people using collaborative consumption platforms were social, e.g., ‘meeting people’ or ‘helping out’. In terms of sharing, Prior studies indicate that social interactions and networks have a positive impact on sharing knowledge, i.e., more social interactions lead to a more frequent and intense knowledge exchange behavior [22, 23]. However, in terms of collaborative consumptions with strangers, research in the offline context suggested that people are particularly prone to avoid interacting with nearby strangers due to concerns such as fears of “stranger danger” [1], and feelings of anxiety and uneasiness to meet and interact with them [20]. Such research findings from offline context remind us the possibility that consumers online could behave similarly. Following prior empirical studies in collaborative consumption, we therefore propose that:

  • H2. Social value is positively associated with consumer’ attitude towards participating in collaborative consumption.

Uses and gratifications (U&G) paradigm explains what users can achieve when participating in SNSs from five dimensions, which can be seen as benefit factors attracting SNS users. The five dimensions are purposive value (i.e., informative and instrumental value), self-discovery, maintaining interpersonal connectivity, social enhancement, and entertainment value [19, 24]. Prior research in collaborative consumption also suggested that enjoyment of the activity is essential in gaining satisfaction and increasing the likelihood of people’s choosing collaborative consumption services [2, 5]. Accordingly, we propose that:

  • H3. Entertainment value is positively associated with consumer’ attitude towards participating in collaborative consumption.

The value of convenience has long been recognized in marketing and retailing literature [25,26,27]. For example, Anckar et al. [25] suggested that superior shopping convenience is an essential part of customer value. Szymanski and Hise [27] showed that convenience leads to E-satisfaction in online retailing. Yet few research in collaborative consumption has examined the role of convenience value. Service convenience facilitates the sale of goods and services in collaborative consumption. Because virtually all organizations create value for consumers through performances and because convenience is an important consideration for most consumers [28], we conjecture that convenience value could play an equally important in role collaborative consumption as other consumption situations. Therefore, we propose that:

  • H4. Convenience value is positively associated with consumer’ attitude towards participating in collaborative consumption.

The extant literature shows extensive support for the overall beneficial effect of trust to business transactions and consumptions [29, 30]. Empirical studies report that trust, by bringing about good faith in the intent, reliability, and fairness of partner behaviour [31, 32], reduces the potential for conflict [31]. A large amount of previous research has argued that trust, and relationship commitment, perform vital roles in promoting collaborative relationships [e.g. 33, 34].

Trust can be equally important in collaborative consumptions. Glind [58] highlighted that lack of trust as the important factors that deter the use of peer-to-peer services such as accommodation. Ert et al. showed that The more trustworthy the host is perceived to be from her photo, the higher the price of the listing and the probability of its being chosen in accommodation sharing [14]. Therefore, we propose that:

  • H5. Trust is positively associated with consumer’ attitude towards participating in collaborative consumption.

3 Research Methods

3.1 Data Collection

To conduct our research, we used a structured survey method. We recruit 278 students who has experience in participating in collaborative consumption from a major university in China for the survey research. A large-scale survey was carried out in December 2016. We send out invitations through the communication platform of Student’s Union and about 400 students registered for this survey. Finally, a total of 278 completed surveys were collected. Based on prior literature, it may be appropriate to utilize student subjects if the observation does not include phenomena [23, 35] such as social norms or political views that are structured over time [36, 37]. Finally, we have 278 complete and usable responses for data analysis. Table 2 reports the demographic description of those respondents.

Table 2. Demographic descriptions
Full size table

We examined the response bias issue with the procedure suggested by Armstrong and Overton [38]. We first compared the participants who registered but not completed the whole research process (either not login the APP, or not complete the final questionnaire) with those who complete the whole process on key demographic variables. The results of the Mann-Whitney tests show no significant differences, suggesting that response bias would not likely affect our findings [38].

3.2 Questionnaire Development

The questionnaire was developed through a two-stage process. We first extensively reviewed the literature to list the candidate constructs and measures that were used in prior research. A draft questionnaire was developed. Each item was measured in a seven-point Likert scale. In the second stage, four researchers reviewed the draft questionnaire, ranked each item according to their content validity, and suggested improvements in wording and the layout of the items. We also include two variables: personal innovativeness and familiarity as control variables in our research model. The instrument for the constructs in our research model and is presented in Table 3.

Table 3. Constructs and instruments
Full size table

4 Data Analysis and Discussions

We then conducted data analysis in accordance with a two-stage methodology [43]. The first step in the data analysis is to establish the convergent and discriminant validity of the constructs. We test the measurement model using Principal Components Analysis (PCA) in SPSS (SPSS, Chicago, USA) and Confirmatory Factor Analysis (CFA) in LISREL [44]. In the second step, following Cohen et al. [45]’s recommendations, a hierarchical multiple regression analysis was employed to test our hypotheses using SPSS.

In the first phase, we examine the data using PCA with Varimax rotation. Several items were removed one by one due to cross loadings and after careful examination between the wording of the items and the definitions of the construct. after removing those items, we identify 8 factors with eigen values greater than 1.0. All constructs explain 78.7% of the total variance (EFA results table is available on request).

We conduct the CFA analysis by creating a LISREL path diagram. Using LISREL for confirmatory factor analysis provides a more rigorous assessment of the fit between the collected data and the theoretical factor structure, and satisfies the minimum requirements of assessing the measurement properties of uni-dimensionality, convergent validity, and discriminant validity [46,47,48].

Uni-dimensionality is the degree to which items load only on their respective constructs without having “parallel correlational pattern(s)” [46, 49]. Uni-dimensionality cannot be assessed using factor analysis or Cronbach’s alpha, instead, covariance-based SEM (such as LISREL) provides the ability to compare alternative pre-specified measurement models and examine, through statistical significances and a wide set of the types of fit, which is better supported by the data [48]. A set of criterion has been proposed and adopted in the literature to assess uni-dimensionality, including model fit indices such as GFI, NFI, AGFI, and χ2 to show uni-dimensionality [48]. We apply the following indices and standards to assess model fit: goodness-of-fit index (GFI) and normed fit index (NFI) greater than 0.90, adjusted goodness-of-fit index (AGFI) greater than 0.80 [48], comparative fit index (CFI) greater than 0.90, and root mean square of approximation (RMSEA) lower than 0.08 for a good fit and lower than 0.05 for an excellent fit [50].

We conducted a confirmatory factor analysis (CFA) for the 8 constructs, namely, Attitude towards collaborative consumption (ATCC), Economic Value (ECVA), Social Value (SOVA), Entertainment value (ENVA), Convenience Value (COVA), Trust (TRUST), Personal Innovativeness (PEIN) and Familiarity (FAM). The means and standard deviation of each of the constructs are presented in Table 4.

Table 4. Descriptive statistics and correlations
Full size table

The CFA demonstrated good model fit, showing that over half the variance is captured by the latent construct [48, 49, 51].

Convergent validity is assessed using three criteria. First, standardized path loadings, which are indicators of the degree of association between the underlying latent factor and each item, should be greater than 0.7 and statistically significant [48]. Second, composite reliabilities, as well as the Cronbach’s alphas, should be larger than 0.7 [52]. Third, the average variance extracted (AVE) for each factor should exceed 0.50 [53]. As shown in Table 5, all path loadings are greater than 0.707 after removing CONVA1 and PERIN4, and all of them are significant. After evaluate the wording of these two items, we decided to keep these two items for further data analysis. The reliability measures are all above 0.7, and the AVEs are all above 0.5. Thus, convergent validity is established.

Table 5. Confirmatory factor analysis results
Full size table

To test the discriminant validity of each variable, the average variance extracted (AVE) and the inter construct correlation were compared. The results in Table 5 show that all AVEs for the latent variables were greater than the required minimum level of 0.5. Every construct had a larger square root of AVE than its correlations with other constructs. This result indicates that our measurement items have discriminant validity.

As with all self-reported data, there is the potential for the occurrence of common method variance (CMV), i.e., variance that is attributable to the measurement method rather than to the constructs the measures represent [54]. To address this issue, we used several procedural and statistical remedies.

First, we paid careful attention to the wording of the items, and developed our questionnaire carefully to reduce item ambiguity. These procedures would reduce the respondents’ evaluation apprehension and make them less likely to edit their responses to be more socially desirable, lenient, acquiescent, and consistent with how they think the researcher wants them to respond when crafting their responses [54, 55]. Second, we performed a Harman’s one-factor test via CFA by specifying a hypothesized method factor as an underlying driver of all of the indicators. The results revealed that the fit of the single-factor model was extremely unsatisfactory, indicating the common method variance is not a major source of the variations in the items [56]. Finally, following the literature, we used a marker variable to control for common method bias [57]. We used a statement in political ideology “do you believe that ‘from each according to his ability, to each according to his need’ is important” as the marker variable, as it was theoretically unrelated to many other variables [54, 57]. All significant correlations remained significant after the partial correlation adjustment. While the results of this analysis do not explicitly preclude the possibility of common method variance, they do suggest that common method variance is not of great concern in this study.

We conducted a 3-step hierarchical multiple regression analysis as recommended by Cohen [45]). According to Cohen et al. [45], hierarchical multiple regression analysis is best suited for identifying causal priority and removing confounding variables. This approach is appropriate when the independent variables need to be ordered in terms of the specific questions that are to be answered by the research study. In this study, after controlling for the possible confounding variables identified through the literature review, we sought to examine the antecedents of attitude towards collaborative consumptions. Through hierarchical multiple regression analysis, we could explore the change in the direct effects of value dimensions and trust when it meets other moderating variables. Therefore, following prior studies, we conducted the 3-step hierarchical multiple regression analysis recommended by Cohen, Cohen [45]). In Model 1 (including control variables only), Model 2 (including both control variables and direct effects), Model 3 (including control variable, direct and moderating effects), the R Square for attitude towards participating in collaborative cons, 0.541, and 0.563, respectively.

Hypothesis 1 that tests the influence of economic value on attitude towards participating in collaborative consumption is consistently significant in Model 2 (β = 0.12, p < 0.01) and Model 3 (β = 0.11, p < 0.01), supporting H1. Hypothesis 2 tests the influence of social value on attitude towards participating in collaborative consumption is consistently significant in Model 2 (β = −0.14, p < 0.01) and Model 3 (β = −0.14, p < 0.01), which is inconsistent with our H2 which hypothesizes a positive influence. The effects from entertainment value (H3) and trust (H5) are also significant. However, the hypothesize effect for convenience value is not significant (β = −0.062, p > 0.05), rejecting H4.

5 Discussions and Conclusions

In conclusion, this research proposes a value framework to key antecedents of consumer attitude towards participating in collaborative consumption.

The first contribution of this article is to establish a systematical framework to analyze the essential factors. Based on consumer value literature, we propose that four types of value, i.e. economic value, social value, entertainment value, together with trust, would affect consumer’s attitude towards collaborative consumption. The parsimonious model explains more than 50% variance of consumer’s attitude. Second, our results empirically confirm the effects from economic and entertainment value, which are consistent with prior literature. Finally, our results show that the effect from social value may not be always positive across different situations and collaborative consumption types. Although a large number of literature had found the positive effect from social value, the results from our study shows another possibility that social value could be negative in certain circumstances.