Development of a Motivation–Trust–Vulnerability (MTV) Framework for Cross-Border Online Shopping: A Cross-National Application to Chinese and German Consumers

According to the MA framework, the psychological driver that influences the degree to which an individual is inclined to perform a behavior is motivation (Rauch et al. 2015). Expectancy theory (Vroom 1964) argues that behavior results from conscious choices among alternatives whose purpose is to maximize benefits. In the context of cross-border online shopping, consumers are motivated by the anticipation of benefits which are related to cross-border online shopping., e.g., cheaper prices, greater product selection (Wagner et al. 2016). Previous studies empirically validate the motivational effect of shopping benefits on online purchasing behavior, and from the theoretical perspective, there is a consensus on the relationship (e.g., Forsythe et al. 2006). Therefore, we state that this is a well-known argument, and we assume that the relationship between consumers’ shopping motivation based on perceived benefits and shopping behavior will hold for cross-border online shopping. Thus, this direct effect serves merely as our baseline hypothesis, as we focus on the less understood interaction effects (Andersson et al. 2014). Our baseline hypothesis postulates the following:

Drawing on the MA framework, we first elaborate the relationship of ability and vulnerability in the context of cross-border online shopping and then argue why we include vulnerability in our framework. Ability refers to the physiological and cognitive capabilities that enable an individual to perform a behavior effectively (Rauch et al. 2015). To conceptualize ability, we follow previous research that regards ability as an individual’s knowledge and capability to acquire the relevant skills to carry out a particular task (e.g., Siemsen et al. 2008). Theoretically, this conceptualization of ability and its effects is rooted in Bandura’s (1977) self-efficacy theory, which states that individuals assess whether they have the required skills or knowledge desired to achieve their goals through their beliefs about their ability to successfully perform a particular behavior. Stewart and Pavlou (2002) suggest that a consumer’s ability to complete an online transaction represents an attempt by the consumer to acquire information in a structure in which the desired information is uncertain. For cross-border online purchases, consumers need to acquire even more knowledge and skills to compensate for the additional uncertainties of international online transactions (Safari and Thilenius 2013). For example, cross-border online shoppers need specific knowledge and skills to be able to calculate the total cost of a cross-border online purchase (which possibly includes the exchange rate, delivery costs, customs duty, and tariffs). If consumers do not possess this financial knowledge and cannot acquire these skills, they are financially vulnerable with regard to cross-border online shopping. Further areas of cross-border online shopping vulnerability involve legal (e.g., asserted rights), cultural (e.g., unknown language of customer service), physical (e.g., harmful products) and privacy (e.g., data skimming) factors (Guo et al. 2018).

To integrate the concepts of ability and vulnerability, we adopt the perspective of Shultz and Holbrook (2009), who state that lack of knowledge and the lack of capabilities to acquire the relevant skills and information to perform a specific task (i.e., lack of ability) create consumer vulnerability. In particular, Shultz and Holbrook (2009) suggest a two-dimensional conceptualization of vulnerability in which consumers are doubly vulnerable if they (1) do not know what is beneficial for them and (2) do not have the skills or other resources needed to acquire what would benefit them. Regarding this conceptualization, ability and vulnerability are the opposite ends on a continuum, i.e., higher levels of ability lead to lower levels of vulnerability and vice versa. This conceptualization is in line with Hill and Sharma (2020), who define consumer vulnerability as a state in which consumers are subject to harm because their access to and control over resources are restricted in ways that significantly inhibit their ability to function in the marketplace. With regard to this conceptualization lack of knowledge reflects a resource limitation, while lack of capabilities to acquire the relevant skills reflects restricted control (Hill and Sharma 2020). We include cross-border online shopping vulnerability in our framework because vulnerability is directly related to the uncertainties of cross-border online shopping, i.e., the losses related to obtaining the expected gains (Safari and Thilenius 2013). This new conceptualization of perceived vulnerability (reflected by lack of ability), which we empirically test in this research, refers to the uncertainties of cross-border online shopping and the related vulnerability of international online shoppers. It also has another advantage: Because it is often difficult for individuals to assess their own vulnerability, the evaluation of one’s own ability instead of vulnerability should avoid social desirability bias with regard to the interpretation of vulnerability as weakness or naivety (Jones and Middleton 2007).

Moreover, we propose that perceived vulnerability will have two contradictory effects on consumer behavior: a negative direct effect on cross-border online purchases and a positive moderating effect on the relationship between cross-border online shopping benefits and cross-border online purchases. A moderator variable is a variable that modifies or changes the strength or direction of the relationship between two other variables. We consider vulnerability as a moderator variable because vulnerability can influence the relationship between cross-border online shopping benefits and cross-border online shopping purchases. Consumers who perceive more uncertainties associated with online purchases will be deterred from online shopping (Forsythe et al. 2006). Therefore, we theorize that the more insecure consumers feel with regard to their skills and knowledge with regard to cross-border online shopping, the more vulnerable they will feel and the greater the likelihood that they will refrain from cross-border online purchasing. However, the existence of cross-border online shopping vulnerability might also cause consumers to focus on benefits that motivate cross-border online purchases and repress potential negative outcomes, thus leading to an overestimation of the benefits that they will obtain from cross-border online shopping (Jones and Middleton 2007). As our theoretical argument for the positive moderating effect of vulnerability on the relationship between cross-border online shopping benefits and cross-border online purchases, we propose that because vulnerability involves lack of knowledge and lack of skills, the perceived benefits that are easily accessible (e.g., cheaper prices) overlay the inherent vulnerability, increasing the motivating effect on cross-border online purchase behavior. This effect is due to an imbalance between gains and losses in the cognitive weighting mechanism (Kahneman and Lovallo 1993). In particular, potential losses are not perceived because they are not known (because of no previous experiences or information), or they are cognitively repressed as a strategy of dissonance reduction (Harmeling et al. 2015). Therefore, we conclude that vulnerable consumers, i.e., individuals who have less knowledge and skills with regard to cross-border online shopping, might be more motivated to cross-border online shopping because they directly experience certain benefits of cross-border online shopping without having the knowledge or the skills to assess the uncertainties. We can model the proposed relationship as a moderating (two-way interaction) effect in which the impact of perceived benefits on cross-border online purchases is strengthened when the level of vulnerability is higher. In summary, we propose the following:

We refer to trust theories, for example brought up by Morgan and Hunt (1994), to explain the effect of trust in our model. In online environments in which vendors’ true intentions are especially difficult to assess, trust is a crucial antecedent of behavioral activities (Bleier and Eisenbeiss 2015). As research on Internet shopping argues that shopping online inherently involves higher levels of uncertainty than does shopping at a physical store, we argue that cross-border online shopping involves higher levels of uncertainty than does domestic online shopping (Lim et al. 2004). Potential cross-border online shoppers often lack information on foreign online vendors (e.g., whether the store is legitimate, what payment service is available or whether the products offered are genuine or counterfeit). When information about foreign online vendors is lacking, trust serves as a key foundation on which online shoppers base their purchase decisions (Urban et al. 2009). Therefore, in the uncertain environment of cross-border e‑commerce, trust functions as a catalyst for transactions, as it reduces perceived risks (Pavlou 2003). Consequently, we theorize that consumers who generally trust foreign online vendors tend to be more tolerant of a higher level of uncertainty when transacting with foreign vendors, which should positively affect their cross-border online purchase behavior.

While we control for the above postulated direct effect of trust towards foreign online vendors, our focus is on the interaction effect of trust and perceived vulnerability on the relationship between cross-border online shopping benefits that motivate cross-border purchases, i.e., we postulate a three-way interaction in which trust moderates the moderating effect of perceived vulnerability (see Fig. 2). Our argumentation for this three-way interaction effect derives from the theoretical rationale of the relationship between trust and vulnerability. Trust is a way of dealing with vulnerability such that individuals who trust still feel vulnerable but the more they trust, the less they expect to actually be harmed (Tsui-Auch and Möllering 2010). Previous research discusses the relationship between perceived vulnerability and trust with regard to whether trust comes before or after vulnerability (Bigley and Pearce 1998). We follow the perspective of Mayer et al. (1995), who assume that trusting individuals start from a neutral position from which they decide to increase or decrease vulnerability. In this sense, trust is not risk taking per se; rather, it is the willingness to take risk (Mayer et al. 1995). Because trust increases one’s willingness to accept uncertainties and risks, we assume that trust towards foreign online vendors increases the moderating effect of perceived vulnerability, which—as stated in H2—moderates the relationship between cross-border online shopping benefits and cross-border online purchases (Rousseau et al. 1998). Accordingly, we postulate the following:

Our research model of the MTV framework for cross-border online shopping summarizes our theoretical reasoning, as illustrated in Fig. 2. In line with previous research, we combine three theories in our MTV framework to give a more complete account of the researched phenomenon (Sparrowe and Mayer 2011). By doing this, we also take into account that the underlying theoretical mechanism linking the relationship between moderator and dependent variable must differ from the theoretical mechanism that influences the main relationship (Andersson et al. 2014).

To test the generalizability of our framework, we select two culturally and economically distinct country markets: China and Germany. We select China and Germany because both countries show certain differences with regard to the cultural dimensions uncertainty avoidance and individualism–collectivism (see Table 1), which research identifies as being the most relevant to online shopping because of their link to the willingness to accept the potential risks of online shopping and to trust unknown online vendors (Lim et al. 2004). Additionally, we select China and Germany because both countries are among the five largest e‑commerce markets worldwide, indicating a large tendency to engage in and the relevance of online shopping (Eshopworld 2018). In particular, there are relevant differences in market size, economic conditions, market development and culture that can lead to discriminative market factors between Germany and China and that can affect cross-border online shopping. A comparison between these two country markets based on various economic and cultural criteria is presented in Table 1.

We generate data for our further analyses and hypothesis testing using two online questionnaires that are identical with regard to their content (both in national languages: German and Mandarin Chinese). For this purpose, we perform translation-back-translation (from German to Mandarin Chinese), conducted by four independent coders (Chinese native speakers living in Germany with fluent German language proficiency), to guarantee translation adequacy while considering the cultural context (Chidlow et al. 2014). We distribute the Chinese questionnaires via e‑mail and social networks, generating a convenience sample of Chinese online shoppers. To collect the German data, we use a local research agency panel of adult online shoppers. Before analyzing the data, we eliminate the data sets of all participants with a processing time that is less than half the median (less than 10 min) because this low processing time indicates a low level of engagement with regard to reading and answering all the questions. Furthermore, we conduct plausibility checks, e.g., eliminating the data of the participants whose answers contain variances of zero among all items. We obtain a data set consisting of N = 808 consumers, of whom 452 are from Germany (51.8% female, M_age = 44.02 years) and 356 are from China (54.5% female, M_age = 28.22 years). A total of 64.2% of the German and 53.4% of the Chinese respondents have already made a cross-border online purchase. To account for differences in demographics between the Chinese and German samples, we include gender, age, income and online shopping affinity (number of online purchases within the last twelve months) as control variables in our analysis.

We rely on established multi-item scales from previous studies that we identify and modify to fit the context of our study (see Table 2). To capture future cross-border online purchases, we adapt Pavlou’s (2003) online purchasing intention scale, which consists of three items. To account for actual cross-border online purchasing behavior, we use the answer to the question of whether the respondents have made cross-border online purchases in the past. This variable is a dichotomous variable reflecting past cross-border online purchase behavior (0 = not made cross-border online purchases; 1 = made cross-border online purchases). We evaluate satisfaction with cross-border online shopping, which is a variable that indicates an outcome of the expectations and actual performance of cross-border online purchases, only for the respondents who have already made cross-border online purchases in the past. To measure satisfaction, we adapt three items (e.g., “I think that I did the right thing when I purchased at foreign online vendors.”; α = 0.77) from Cronin et al. (2000). To measure perceived cross-border online shopping benefits, we adapt three items from Meuter et al. (2005) to the context of our study. Trust is measured using a three-item scale (Yoon 2009). Perceived vulnerability is measured as a reflective construct with a two-factor structure, with each factor consisting of three items, that encompasses perceived lack of knowledge and perceived lack of skills, as proposed by Shultz and Holbrook (2009). We adapt a measurement scale for perceived lack of knowledge and perceived lack of skills from Grewal et al. (2001) to address the lack of a compatible vulnerability scale in the literature that captures consumers’ perceived lack of knowledge and skills to successfully conduct cross-border online shopping. By measuring perceived vulnerability with items of ability dimensions, we decide to use a reverse-coded scale approach; studies usually apply such an approach to measure constructs that have negative connotations or that are difficult for respondents to assess, such as emotional stability (Cucina et al. 2019), job complexity (Morgeson and Humphrey 2006) or role ambiguity (Rizzo et al. 1970). We measure all constructs via seven-point Likert scales. Additionally, we pretest the measurement scales via an online survey with students from both countries to assess the reliability and comprehensibility of our adapted scales; the pretesting confirms the reliability and comprehensibility of our measurement instruments.

We investigate the dimensionality, reliability, and validity of our construct measures via exploratory and confirmatory factor analysis (CFA). The average variance extracted (AVE) values not less than 0.63 for all scales, the Cronbach’s alpha values of 0.83 and above and the composite reliability (CR) values of 0.84 and above are all satisfactory and reflect high levels of scale consistency (see Table 2). In addition, we assess all reflective scales for discriminant validity by applying Fornell and Larcker’s (1981) criterion, indicating that discriminant validity should not be a problem because no construct shares more variance with any other construct than with its own indicators (see Tables 3 and 4). To ensure that multicollinearity was not a problem within our study, we tested variance inflation factors (VIF), which showed all a measurement below 2, therefore under the recommended threshold of 10. Moreover, we checked the inter-construct correlations between all variables of our model as well as between of all control variables for both samples and found no indication for multicollinearity.

To conduct CFA of the measurement model on the Chinese and German datasets individually, we use the maximum likelihood (ML) estimation procedure with AMOS 25. The model yields an acceptable model fit with the sample data from China (χ²/df = 3.71; RMSEA = 0.09; CFI = 0.95; TLI = 0.93; SRMR = 0.05) and Germany (χ²/df = 2.58; RMSEA = 0.06; CFI = 0.99; TLI = 0.98; SRMR = 0.03). Each factor loading is statistically significant, and the standardized values are above the recommended threshold of 0.70 (Germany: 0.81 and above; China: 0.73 and above) for all items (Bagozzi and Yi 1988).

We perform tests of multigroup invariance to examine the equivalence of the proposed measurement model across the two country samples. Following the procedures suggested by Steenkamp and Baumgartner (1998), we first estimate a multigroup CFA model without any restrictions on the parameters across country groups. The overall model fit is sufficient, thus supporting configural invariance (global χ²/df = 2.52; RMSEA = 0.04; CFI = 0.97; TLI = 0.97; SRMR = 0.03). Next, we assess metric invariance by constraining the factor loadings in the two groups to be equal and comparing this model with another model in which the factor loadings are free to be estimated across groups. The results indicate that the two samples are not fully invariant because the constrained model has a significantly higher chi-square (∆χ² (20) = 197.5, p < 0.05). Therefore, we find that there is no full metric invariance (i.e., we cannot analyze a total model using the pooled data of both country groups). However, the test results indicate partial metric invariance; thus, we can estimate two independent models and compare the results (Awanis et al. 2017).

In our analyses, the evaluations of both the antecedents and the outcome measures in the model stem from the same person, which might produce common method bias (Chang et al. 2010). Following the suggestion of Podsakoff et al. (2003) to account for common method variance, we take several approaches and perform several tests. Ex ante, we assure the respondents of the anonymity and confidentiality of the study and indicate that there are no right or wrong answers and that they should answer as honestly as possible. Moreover, we counterbalance the order of questions relating to different scales and constructs and randomize the order of the items in our online survey. We also include a marker variable in our questionnaire that is conceptually independent of the latent variables in our study. Also, we examine the correlation of the marker variable with the other variables from the research model using the correlational technique, which is one of the options proposed by Simmering et al. (2015). Specifically, we choose a variable to measure the consumer’s charity participation likelihood (e.g., “How likely are you to donate money?”) on a 7-point Likert scale (1 = Not at all likely, 7 = Very likely) because it is theoretically unrelated to the constructs of our model. The marker variable is not significantly related to any of the variables in the model; therefore, the results of the marker variable testing provide further evidence that common method variance is not a serious problem in our study.

Our study contributes to international business and marketing theory by extending and respecifying the MA framework (Merton 1957) to advance our understanding of the effect through which vulnerability influences cross-border online shopping behavior. To understand this effect, we go beyond the simplistic argument “it depends” and consider the underlying interactions of the relationship between cross-border online shopping benefits and intention to understand the conditions under which this relationship applies (Andersson et al. 2014). Following Shultz and Holbrook’s (2009) conceptualization of vulnerability, we assume that the factor of ability is reciprocal with the concept of vulnerability in cross-border online shopping, meaning that a lower level of ability corresponds to a higher level of vulnerability. Therefore, this research also offers implications for the conceptualization of vulnerability. Here, we theorize and demonstrate the higher-order reflective structure of perceived vulnerability as a combination of perceived lack of knowledge and perceived lack of skills.

Our findings make an important contribution to online shopping research by identifying that vulnerability affects cross-border online shopping in two different ways: via a direct negative effect and via an indirect positive moderating effect. With the MTV framework, we therefore extend research on domestic online shopping that focuses solely on benefits and risks (e.g., Forsythe et al. 2006), but not on the lack of knowledge and lack of skills which create perceived vulnerability and affect cross-border online purchases. The introduction of a new conceptualization of perceived vulnerability into the international marketing literature and its integration in a holistic MTV framework offer a certain potential for helping to explain effectiveness in intercultural interactions.

In our two initial analyses, the MTV framework demonstrates its cross-cultural applicability and the potential to explain the effect of cross-border online shopping. Therefore, the MTV framework appears robust across countries with different cultural and economic backgrounds. Naturally, the usefulness of this new framework must establish its validity in a series of further studies. Because relationships depend on particular environmental conditions, the MTV framework might be appropriate to understand further motivation effects in situations that involve uncertain outcomes and high levels of vulnerability. Therefore, the MTV framework might also be applicable to further studies in international business where situations with increased uncertainty occur, for example, how the vulnerability (e.g., lacking language skills or limited knowledge about Chinese culture) of international managers operating in transition economies such as China and their trust towards the multinational corporation (MNC) affect their work motivations (Tsui-Auch and Möllering 2010).

The creation and empirical validation of the MTV framework is a step that is consistent with the development of a mid-range theory that links vulnerability perceptions to consumer behavior and international activities. Previous research on the topic of perceived vulnerability in international business studies is scarce and considers vulnerability mostly as an antecedent to managerial practices (Tsui-Auch and Möllering 2010). Derived from the MA framework, the combination of motivation and ability is primarily applied in the context of the main effect and as static factors influencing desired outcomes (e.g., Grewal et al. 2001; Bahadir et al. 2015). Although scholars agree that behavior across various contexts is a function of ability (or vulnerability) and motivation, there is no general agreement on the effect through which these factors operate (Siemsen et al. 2008). Indeed, it is possible that these factors combine in different ways across differing contexts. In line with Minbaeva et al. (2003), we show that effects not only result from the impact of individual variables, but also from the interaction of these variables. Moreover, mixed research results suggest that high levels of ability and motivation are not always valid precursors of behaviors (Moorman and Matulich 1993). For example, research finds that consumers who are moderately motivated and moderately able to perform some activities, perform most effectively (Bettman and Park 1980). These mixed findings indicate that with regard to the interaction of motivation and ability, a relationship of “the more, the better” might not exist in all cases. Our conceptualization of vulnerability as lack of ability might contribute to explaining these contradictory findings to some extent and help in examining the boundaries of the MA theoretical framework (Andersson et al. 2014).

A major implication of our findings for international business practice is that vulnerability may serve as an internationalization barrier to the cross-border activities of consumers that may counteract the cross-border e‑commerce efforts of companies to sell to international shoppers online. In particular, our findings suggest that managers must cope with vulnerability and take measures to decrease vulnerability. Marketing strategies should aim to neutralize feelings of vulnerability and to take advantage of consumer trust (Bart et al. 2005). The negative impact of vulnerability on cross-border online purchasing intentions indicates a chance for online vendors to increase consumers’ willingness to make cross-border online purchases by increasing knowledge and skills. Online vendors can increase knowledge by providing relevant information, for example, about the expected costs, delivery time, and return policy. Regarding skills, cross-border e‑commerce should be manageable by customers with the same set of skills that are necessary for domestic online shopping to reduce the uncertainties and barriers of cross-border online shopping. For example, online vendors should provide the same language, currency, and payment services as in the domestic market of the online shopper. Doing so is important because only customers who are satisfied with their cross-border online shopping experiences will continue to buy from foreign online vendors and thus represent loyal and long-term profitable customers.

Whenever, consumers are potentially vulnerable, the question arises, if marketing activities that target the vulnerable consumer group are unethical (Jones and Middleton 2007). Firms are urged to avoid intended or unintended unethical marketing strategies for cross-border e‑commerce, for example by concealing the actual place of location or displaying nontransparent delivery or return shipment costs. Moreover, retail managers are encouraged to develop a normative prescriptive framework for ethical conduct on the part of the cross-border e‑commerce business that considers vulnerability of foreign online shoppers. This measures would also help to increase trust in foreign online vendors, which as our findings suggest, is necessary to strengthen the positive relationship between cross-border online shopping benefits and the intentions to conduct cross-border online purchases (Bart et al. 2005). That trust is important for online transactions is well-known, but its interaction with perceived vulnerability underlines the relevance of this construct for online relationship building even more.

Additionally, cross-border online shopping vulnerability arises because there are almost no existing international rules or norms in place to protect cross-border online shoppers. Most existing trade agreements between countries were signed in the predigital era to cover traditional flows of goods. Because cross-border e‑commerce is related to direct shipments from foreign online vendors to customers in countries abroad, there is a lack of control and influence. This situation makes it even more important for governmental institutions and policy makers to help consumers develop the necessary knowledge and skills with regard to cross-border online shopping. Such knowledge and skills can be maintained by providing relevant information and educating consumers or by developing and offering tools that support online-shoppers, such as cross-border tax and tariffs calculators. Moreover, policy makers can take measures to help online shoppers to access the trustworthiness of foreign online vendors, to avoid a general frivolously trusting that can result in vulnerability.