Trust has long been regarded as one of the most critical variables for developing and maintaining well-functioning customer–seller relationships (Celuch, Bantham, & Kasouf, 2011; Moorman, Deshpande, & Zaltman, 1993; Sharma & Patterson, 2000; Thomas, 2009). Research has shown that trust may lead to higher customer relationship commitment and loyalty, among other effects (e.g., Eisingerich & Bell, 2007; Morgan & Hunt, 1994; Selnes & Sallis, 2003). While recognizing the importance of relationship trust, past research (e.g., Driscoll, 1978; Grayson, Johnson, & Chen, 2008) suggests that trust relates not only to customer trust in individual companies. Trust also relates to the broader business context in which customer–seller relationships may develop. Consistent with this suggestion, this paper distinguishes between two kinds of trust: narrow-scope trust and broad-scope trust (BST). The often-cited definition proposed by Sirdeshmukh, Singh, and Sabol (2002), which defines narrow-scope trust as “the expectation held by the customer that the service provider is dependable and can be relied on to deliver on its promises” (p. 17), is adapted. Based on this definition, BST is defined as the expectation held by the customer that companies within a certain business type are generally dependable and can be relied on to deliver on their promises. Notably, this definition is consistent with previous research suggesting that BST (or “generalized trust”) is a generalized expectancy that the promise of a group can be relied upon (Rotter, 1980; Siegrist, Gutscher, & Earle, 2005).
The purpose of the present study is to model and investigate the direct and moderating effects of BST on customer–seller relationships. While many studies have examined the role of narrow-scope trust in customer–seller relationships, only one previous study (i.e., Grayson, Johnson, & Chen, 2008) has investigated the role of BST in customer–seller relationships. Based on two rival sociological perspectives (functionalist and institutional theory, respectively), Grayson, Johnson, and Chen (2008) found that narrow-scope trust mediates the influence of BST on customer satisfaction. While past research has considered the direct and indirect influence of BST on relationship satisfaction, no research has examined whether BST may moderate the relationships between relationship variables. This is an important shortcoming since if moderating effects occur it means that customers reward relationship experiences and perceptions—such as satisfaction and narrow-scope trust—differently depending upon the perceived level of BST. This research seeks to address this shortcoming in the literature on BST and customer–seller relationships. Notably, it is also shown that BST positively influences customer satisfaction and narrow-scope trust. This study is based on two surveys. Survey 1 comprises 1155 bank consumers, whereas survey 2 comprises 817 insurance consumers. The remainder of the paper is organized as follows. First, the theoretical framework and hypotheses are introduced followed by a review of the methods used to test the hypotheses. Next, the results are presented. Finally, the implications of the findings are discussed and suggestions for future research are provided.
THEORETICAL FRAMEWORK AND RESEARCH HYPOTHESES
Sellers and customers not just exchange services and money but also often create ongoing, and even trusting, relationships of mutual benefit as suggested in the marketing relationship approach (Johnson & Selnes, 2004; Mohr, Fisher, & Nevin 1996; Mohr & Nevin, 1990; Morgan & Hunt, 1994; Palmatier, 2008; Sheth & Parvatiyar, 1995; Vargo & Lusch, 2004; Ward & Dagger, 2007). Although a large number of conceptualizations of “relationship marketing” have been proposed, marketing researchers seem to agree that (a) relationship marketing focuses on the individual customer–seller relationship; (b) both parties in a relationship must benefit for the relationship to continue; (c) the relationship is often longitudinal in nature; (d) the focus of relationship marketing is to retain customers (Bejou, 1997; Cox & Walker, 1997; Grönroos, 1994; Hunt, Arnett, & Madhavaram, 2006; Peterson, 1995). Especially social exchange theory (Thibault & Kelley, 1959) and relational contracting (Macneil, 1974, 1980) have been employed to model and understand customer–seller relationships. Social exchange theory holds that interactions between people often are of mutual interest to both parties and that they are likely to continue interacting as long as they both believe that it is beneficial (Thibault & Kelley, 1959). Relationships are assumed to grow, deteriorate, and dissolve as a consequence of such interactions. In a similar vein, relational contracting holds that exchange behavior is often characterized by whole person relations, extensive communications and significant elements of noneconomic personal satisfaction (Macneil, 1974). The application of these theories has resulted in a strong focus on variables such as trust, satisfaction, and loyalty within the relationship marketing approach (e.g., Hunt, Arnett, & Madhavaram, 2006; Morgan & Hunt, 1994; Mohr & Nevin, 1990; Nijssen, Singh, Sirdeshmukh, & Holzmoeller, 2003). Drawing on previous marketing relationship research a baseline model is initially developed (in the following referred to as the “STL baseline model”) comprising the marketing relationship constructs satisfaction (S), narrow-scope trust (T), and loyalty (L; Figure 1).
The STL (satisfaction, narrow-scope trust, and loyalty) baseline model is consistent with prior research (Johnson & Selnes, 2004) suggesting that satisfaction and loyalty constitute main competitive advantages that may be gained from developing relationships with customers. On a similar note, satisfaction, narrow-scope trust, and loyalty can be seen as dimensions indicating “relationship quality,” that is, the strength of the relationship between customer and seller (Huang, 2008). The main purpose of the present research is not to investigate and discuss linkages between the three relationship constructs. Instead, a baseline model is utilized, which comprises what is generally believed to be among the most important relationship marketing constructs as well as specifies the most broadly recognized interconstructs relationships (e.g., Anderson & Srinivasan, 2003; Eisingerich & Bell, 2007; Homburg & Giering, 2001). Similar to previous research concerning context effects in customer–seller relationships (Nijssen et al., 2003), the STL baseline model is then used as a basis for modeling the effects of BST by investigating the direct and moderating effects of BST on the relationships included in the STL baseline model.
By maintaining that “consumers enter into relational exchanges with firms when they believe that the benefits derived from such relational exchanges exceed the costs” (Hunt, Arnett, & Madhavaram, 2006, p. 76), marketing relationship theory basically takes a value-approach to marketing. Gaining value will improve customer satisfaction and stimulate repurchasing (or loyalty). Since the value is more connected with ongoing exchanges than with a specific transaction, relationship marketing is most reasonable applied when there is an ongoing desire for the product or service in question. Thus, although relationship marketing is not appropriate for all consumer markets, it is argued that the relationship marketing approach is particularly applicable to the financial services sector, as financial services can be characterized as highly intangible, complex, high-risk, and often long-term service-based offerings, wherein relationship participation is central to service delivery (e.g., Devlin, 1998; Guenzi & Georges, 2010; O'Loughlin, Szmigin, & Turnbull, 2004). Moreover, consistent with the relationship marketing approach, recent empirical results suggest that customers are often loyal to their financial service provider (Finextra, 2009), confirming the presence of ongoing relations. In the following, the STL baseline model constructs are further conceptualized and the expected interconstruct relationships are discussed. Also, the type of BST (i.e., formal vs. informal BST) considered in the present study is clarified.
Satisfaction has attracted attention for many years (e.g., Fornell, Johnson, Anderson, Cha, & Everitt, 1996; Homburg, Koschate, & Hoyer, 2006). Research suggests that satisfaction has impact on Return on Investment (ROI) (Anderson, Fornell, & Lehmann, 1994), shareholder value (Ittner & Larcker, 1996), higher marketshare and profit (Homburg & Rudolph, 2001), and overall firm performance (Anderson & Sullivan, 1993). Although previous research results are mixed concerning the relationship between satisfaction and narrow-scope trust (e.g., Grayson, Johnson, and Chen  found that narrow-scope trust positively influenced customer satisfaction), in the baseline model satisfaction is expected to positively influence narrow-scope trust. This expectation reflects the proposition that trust is an aggregate evaluation at some higher level than satisfaction (Ravald & Grönroos, 1996; Selnes, 1998). As suggested by Selnes (1998) and Sabel (1993), narrow-scope trust is derived not only from experiences or episodes within the relationship but also from a type of cultural context of how business partners are expected to behave. Moreover, the expectation concerning the relationship between satisfaction and narrow-scope trust is based on past research suggesting that satisfaction antecedes trust (Aurier & N'Goala, 2010; Bearden & Teal, 1983; Omar, Wel, Musa, & Nazri, 2010; Ouyang, 2010; Zboja & Voorhees, 2006), that satisfaction develops in the initial stages of marketing relationships and trust develops in the intermediate stages (Geyskens, Steenkamp, & Kumar, 1999; Leisen & Hyman, 2004), and that satisfaction positively influences narrow-scope trust because it increases consumers’ confidence that they will be treated fairly and that the seller cares about their interests (Ganesan, 1994). In the STL model, satisfaction is also expected to positively influence loyalty. Past research suggests that loyalty is difficult to achieve without customers having some degree of satisfaction (Omar et al., 2010) and that satisfied customers are more motivated to continue the relationship with the supplier (Halimi, Chavosh, & Choshali, 2011; Selnes, 1998; Sharma & Patterson, 2000). Satisfaction may be conceptualized as a facet (attribute-specific) or as an overall (aggregate) characteristic. Also, the characteristic can be viewed as transaction-specific (encounter satisfaction) or as cumulative (satisfaction over time). Similar to past relationship and service-related research (Dimitriades, 2006; Levesque & McDougall, 1996), satisfaction is in the present study conceptualized as an overall, cumulative customer evaluation toward a financial service provider.
Narrow-scope trust is being regarded as one of the most critical variables for developing and maintaining well-functioning relationships (Moorman, Deshpande, & Zaltman, 1993; Morgan & Hunt, 1994; Sharma & Patterson, 2000) and is likely to be especially important in financial customer–seller relationships because financial companies have an implicit responsibility for the management of their customers’ funds and the nature of financial advice supplied (Harrison, 2003). Moreover, financial services are high in credence properties since even in the usage situation they can often not be evaluated by the customer because of their long-term nature (Darby & Karni, 1973) and because customers may lack the competencies to confidently evaluate the financial consequences of the services, thus elevating the potential importance of trust in financial customer–seller relationships. While a large body of research exists within the concept of narrow-scope trust, with different points of view being advocated, this study adapts the often-cited definition proposed by Sirdeshmukh, Singh, and Sabol (2002) and conceptualizes narrow-scope trust as “the expectation held by the consumer that the service provider is dependable and can be relied on to deliver on its promises” (p. 17). Past research has recognized narrow-scope trust as an important determinant of relationship loyalty (Eisingerich & Bell, 2007; Morgan & Hunt, 1994; Ouyang, 2010). When a service provider builds consumer trust, the perceived risk associated with the specific service provider is likely reduced since the consumer can more confidently predict the future behavior of the service provider (Sirdeshmukh, Singh, & Sabol, 2002). In a similar vein, Gwinner, Gremler, and Bitner (1998) suggest that consumers may receive psychological benefits (e.g., reduced anxiety), among other benefits, as a result of having developed a trustful relationship with a particular provider. Such benefits may motivate customers to continue the relationship with the supplier.
Customer loyalty has been identified as a strong determinant of profitability (Verhoef, 2003) and competitiveness (Kotler & Singh, 1981) and has become a top priority in service industries (N'Goala, 2007). Two main approaches to loyalty have evolved in the literature: behavioral and attitudinal approaches. While the behavioral approach defines loyal customers as their intent to stay with an organization, or whether they have repurchased its offerings, the attitudinal approach recognizes not only the behavioral dimension, but also the attitudinal dimension of loyalty (Bodet & Bernache-Assollant, 2011; Brunner, Stöcklin, & Opwis, 2006). Similar to recent research on customer–seller relationships (Bell, Auh, & Smalley, 2005), this study takes a behavioral intentions perspective of loyalty rather than a repeat purchase perspective to avoid confusing spurious loyals—those who have a low relative attitude toward the relationship but are constrained to repeat purchase (Dick & Basu, 1994)—with genuinely loyal customers. Thus, loyalty is conceptualized as a customer's expected propensity to stay with a particular service provider.
While narrow-scope trust has been extensive investigated within the relationship marketing literature, BST is clearly under-researched. BST can be regarded as “formal” or “informal.” Formal BST is the belief that proper impersonal structures are in place to enable one to anticipate a successful future endeavor (McKnight, Cummings, & Chervany, 1998). Formal BST is also referred to as “system trust” thereby underlining that it relates to the customer's views regarding the formalized regulation of a particular activity system (Grayson, Johnson, & Chen, 2008). Informal trust (also referred to as “generalized trust”; Humphrey & Schmidt, 1996) concerns whether the entities in a system can be trusted, regardless of sector or context. Informal trust is an expectation that system entities will generally abide by commonly held social norms, roles, and ethical dictates. People who have informal trust expect system entities to function as they “should” (Muhlberger, 2003). In this paper, informal trust is considered. This is because informal trust is more directly related to the behavior of companies than formal trust, which also concerns trust in legal rules and public authorities. As stated above (informal), BST is conceptualized as the expectation held by the consumer that companies within a certain business type are generally dependable and can be relied on to deliver on their promises. Both direct and moderating effects of BST on the specified STL model relationships are included in the proposed research model (Figure 1). In total, the research model comprises six research hypotheses, which are divided into two groups depending upon whether they are related to the direct or moderating effects of BST. Background evidence for the proposed hypotheses is provided in the following.
It is predicted that satisfaction is positively influenced by BST. When BST is low, it means that not every service provider can be trusted to deliver satisfying services and therefore the consumer faces the problem of avoiding pitfalls in the marketplace (Tan & Vogel, 2008). However, this problem might not be easily solved. Several research results and financial reports point to the fact that many consumers possess highly limited knowledge about financial products (e.g., Estelami, 2005; N'Goala, 2007; OECD, 2006; Perrin, 2008). Thus, the consumer risk that her/his interests are currently not being properly served. On a similar note, Sjöberg (2001) found that lack of trust in the general honesty of people was positively associated with risks perceived. In such incidents, past research suggests that in order to maintain self-confidence and to avoid cognitive dissonance the consumer will assign external blame (Blount, 1995; Gotlieb, 2009; Shaver, 1985; Todd & Gigerenzer, 2003), which may reduce relationship satisfaction. In contrast, high BST means lower general uncertainty about the service outcome, which reduces potential cognitive dissonance and the need to assign external blame. This expectation is consistent with past research. In a study of the role of trust in medical relationships Hall, Camacho, Dugan, and Balkrishnan (2002) found that general physician trust relates positively to satisfaction with individuals’ personal physician. Thus, the following hypothesis is presented.
H 1a. BST has a positive influence on relationship satisfaction.
Two competing views on the relationship between BST and narrow-scope trust can be identified (Grayson, Johnson, & Chen, 2008; Luhmann, 1979; Rousseau, Sitkin, Burt, & Camerer, 1998). Both views are based on sociologically oriented theories. (1) Functionalism seeks to explain the relationship between different parts of a social system and how these parts relate to the system as a whole (Davis & Moore, 1966; Parsons, 1951, 1967). The functionalist perspective holds that in all social systems there are a number of functional prerequisites—such as allocation and performance—that must be met if the system is to function effectively and to survive. All roles must be filled and according to the functionalist perspective, they will be filled by those best able to perform them. In order to accomplish this, all complex societies need some mechanism that reduces uncertainty and ensures effective role allocation and performance. In that respect, BST serves as an uncertainty reducing mechanism, which facilitates successful negotiations among economic parties. The roles taken on by the parties and the institutions of a social system are regarded as interdependent. A change in one part of the social system therefore means that other parties of the system may need to modify their behavior. In incidents where broad scope is insufficient (i.e., at a low level), economic parties may therefore compensate for this by developing narrow-scope trust. In other words, narrow-scope trust is formed where it is needed (Luhmann, 1979) suggesting the existence of a negative relationship between BST and narrow-scope trust. This view on trust is also consistent with the perspective found in neoclassical economics in which people only trust others if needed and if it is in their own self-interest to do so (Fetchenhauer & Dunning, 2009). (2) The institutional perspective argues that the processes and structures that are established within a society, or a community, act as authoritative guidelines for social behavior. Social behavior needs to be legitimized by the rules and norms that exist in the broader social environment (Scott, 2004). In a similar vein, Meyer and Rowan (1977) argue that institutional rules function as myths, which organizations incorporate in order to gain legitimacy, stability, and enhanced survival prospects. Also, DiMaggio and Powel (1983) suggest that organizations that operate outside of accepted norms in the organizational field face isomorphic pressures. According to DiMaggio and Powel organizational legitimacy is therefore closely linked with survival. Thus, if trust is common within a business type, it encourages the development of trust in customer–seller relationships suggesting the existence of a positive relationship between BST and narrow-scope trust. In their recent empirical study Grayson, Johnson, and Chen (2008) found that BST positively affected narrow-scope trust, thus providing support to the institutional perspective. Consistent with this finding, Pennington, Wilcox, and Grover (2003) found that system trust positively influenced perceived trust in a vendor. These findings are adapted in this study.
H 1b. BST has a positive influence on narrow-scope trust.
Many financial services are experiential in nature, characterized by technical complexity, information asymmetry, and may even contain credence properties (Darby & Karni, 1973; Guenzi & Georges, 2010; Romàn & Ruiz, 2005). Thus, the information available for decision making may be too vague, or too imprecise, to calculate the probabilities of different outcomes of switching to another service provider and because of this choice complexity a customer may perceive considerable risk in switching to an alternate service provider. In this regard, past research indicates that BST may be applied as a choice heuristics (Siegrist, Gutscher, & Earle 2005; Sjöberg 2001), which can be regarded as “inferential rules of thumb” (Allison, Worth, & King, 1990). This is because consumers may rely on BST to reduce the complexity they are faced with when choosing among various services (Siegrist & Cvetkovich, 2000). Thus, high BST is likely to facilitate consumers’ consideration of alternate service providers without having extensive knowledge about individual service providers (Hall et al., 2002). On the other hand, when BST is low, it means that consumers cannot just rely on all services having satisfying outcome characteristics. That is, when BST is low consumers are faced with a higher choice complexity, which in turn increases costs of switching (Sharma & Patterson, 2000), when considering switching to an alternative service provider than when BST is high (Siegrist, Gutscher, & Earle, 2005). Hence, while high BST may facilitate consumer considerations of other service providers, low BST may prevent consumers from considering switching to another service provider. In line with these considerations, the following hypothesis is proposed.
H 1c. BST has a negative influence on relationship loyalty.
The specification of the moderating effects of BST on the relationships in the STL model draws on attribution theory. Specifically, it is suggested that BST will negatively moderate the relationships in the STL model. Attribution theory describes consumers’ evaluation of causality in a postbehavior context on the basis of different situational contexts (Fiske & Taylor, 1991; Tomlinson & Mayer, 2009; Weiner, 1985, 1986). Weiner (1986) suggests that an individual's perception of an outcome leads to a general emotional reaction of pleasure, or displeasure, which causes the individual to identify the outcome's cause. Kelley (1967) has conceptualized this as the “process by which an individual interprets events as being caused by a particular part of an environment” (p. 193). Weiner (1985) states that the causes of all outcomes can be decomposed into a set of points on three orthogonal continua, or causal dimensions. These continua are (a) locus—the prior outcome's causal agent relative to the decision maker; (b) stability, stable to unstable—the likelihood that a prior outcome's causal agent will persist in the future; and (c) controllability, controllable to uncontrollable—the decision maker's degree of influence over the causal agent. Attribution research can be useful in exploring how consumers explain experiences within customer–seller relationships. Locus of causality relates to the location attributed to the cause of an outcome. It could be an internal position (the cause is located in the consumer her-/himself or in one of her/his decisions), external (located in the company that offers the service), or situational (located in environmental effects) (Oliver, 1993; Ryu, Park, & Feick, 2006). In that respect, consumers will distinguish between causes that are internal, external, and situational. Locus of causality is in particular relevant in the present study because it explicitly distinguishes between situational causes (i.e., BST) and causes (i.e., satisfaction, narrow-scope trust) that are more directly related to the individual seller that offers the service. Attribution theory suggests that consumers will try to understand success or failure in terms of locus of causality indicating that BST may be taken into account by consumers when attributing the cause of their relationship experiences (Cox & Walker, 1997).
Attribution theory predicts that consumers are more likely to evaluate a supplier positively when they make higher external attributions and lower situational (or internal) attributions toward a positive experience (Weiner, 1986). This is because the supplier is viewed as more responsible for the positive experience when external attributions are made, whereas the supplier is perceived to be less responsible for the positive experience when situational (or internal) attributions are made. Several insightful studies have investigated trust using an attribution theory approach. As an overall conclusion, these studies indicate that trust in a relationship is enhanced to the extent that the other's trustworthiness can be ascribed to factors that are internal to the trustee, rather than situationally driven (Kruglanski, 1970; Malhotra & Murnighan, 2002; Strickland, 1958; Tomlinson & Mayer, 2009). For example, in a study of the effects of contracts on interpersonal trust Malhotra and Murnighan (2002) found that the use of binding contracts to promote or mandate cooperation will lead interacting parties to attribute others’ cooperation to the constraints imposed by the contract rather than to the individuals themselves, thus reducing the likelihood of relationship trust developing. Their results also suggest that contracts not only impeded the development of trust but also diminished existing trust. The use of binding contracts seems to have kept interacting parties from seeing each other's cooperative behaviors as indicative of trustworthiness. In a similar vein, empirical findings concerning the behaviors of team members suggest that if an individual is deemed not to be responsible for her/his unfavorable behavior (i.e., the unfavorable behavior can be attributed to situational effects) then prosocial behavioral responses from peers are more likely (Weiner, 1985).
Drawing on such insights, it is predicted that in an environment where BST is low, consumers should be expected to be more likely to attribute negative experiences to situational causes and less likely to attribute negative experiences to external causes (i.e., poor performance by the company that offers the service) compared with environments where BST is high. In a similar vein, when BST is low, consumers should be expected to be less likely to attribute positive experiences to situational causes and more likely to attribute positive experiences to external causes (i.e., good performance by the company that offers the service) compared with environments where BST is high. Specifically, in trying to assess the causes for their level of satisfaction, customers may evaluate their experiences in the light of the perceived trustworthiness of available alternative choices (i.e., other banks or insurance companies). In incidents where BST is low, attribution theory suggests that consumers would be more inclined to attribute positive experiences to their current bank or insurance company, which in turn may enhance narrow-scope trust and loyalty; and vice versa when BST is high. Thus, it is expected that BST would negatively moderate the relationships between satisfaction and narrow-scope trust and between satisfaction and loyalty, respectively. It is also argued that BST should be expected to negatively moderate the relationship between narrow-scope trust and loyalty. This is because when BST is low, the consumer should be expected to be more likely to attribute trust to the customer–seller relationship than to a situational cause; and vice versa when BST is high. Thus, the consumer would probably be more likely to convert narrow-scope trust into relationship loyalty under conditions of low BST than under conditions of high BST. In sum, the following hypotheses are proposed.
H 2a. The influence of satisfaction on narrow-scope trust is negatively moderated by BST, such that satisfaction has a greater positive effect on narrow-scope trust when BST is low compared to high.
H 2b. The influence of satisfaction on loyalty is negatively moderated by BST, such that satisfaction has a greater positive effect on loyalty when BST is low compared to high.
H 2c. The influence of narrow-scope trust on loyalty is negatively moderated by BST, such that narrow-scope trust has a greater positive effect on loyalty when BST is low compared to high.
Two financial service industries were selected for this study: banks and insurance companies. The use of multiple service industries provides a robust test of model relationships by allowing greater variability in study constructs (Sirdeshmukh, Singh, & Sabol, 2002). For each industry, a two-step procedure was utilized to sample respondents from Capacent Epinion's online panel of approximately 30,000 (Danish) consumers. In the first step, stratified random samples of 2382 (bank sample) and 2049 (insurance sample), respectively, consisting of respondents aged 18+ were drawn from the online panel, reflecting the distribution of gender, age, and educational level in the population (aged 18+) as a whole. In the second step, respondents were contacted by e-mail, and asked to respond to the screening question: “Have you recently been in contact with your current (main) bank/insurance company?” (Yes/No/Not currently engaged with a bank/an insurance company) to ensure that only ongoing relationships were included in the samples.
In the final samples (bank sample, n = 1155; insurance sample, n = 817), 52.3% (bank sample) and 51.8% (insurance sample), respectively, were women and average age was 46.5 (bank sample) and 45.1 (insurance sample) years with a range between 18–85 (bank sample) and 18–86 (insurance sample) years, respectively. It was investigated whether the profiles of the study samples deviated from the country population aged 18–85 and 18–86, respectively, on gender and educational level. χ2-Tests of differences between samples and population frequencies on each of these criteria produced p-values >0.12 for both samples, indicating that the samples reflected the demographic profile of the studied country population.
All measurement items were based on prior research, modified to fit the financial service context of this study where relevant. The final items for each construct are summarized in Appendix. All items were identical across industries. Narrow-scope trust was measured by three items adapted from the trust in the organization scale developed by Tax, Brown, and Chandrashekaran (1998). A 3-item scale adapted from De Wulf, Odekerken-Schröder, and Iacobucci (2001) measured satisfaction. The two loyalty intentions items developed by Sirohi, McLaughlin, and Wittink (1998) along with one additional item measured loyalty, whereas three items based on Tax, Brown, and Chandrashekaran (1998) measured BST. In the questionnaires, the items of the four scales were presented in random order.
Validation of Measurements
Confirmatory factor analysis (CFA) was conducted on the four latent factors, with each indicator specified to load on its hypothesized latent factor. Raw data were used as input for the maximum likelihood estimation procedure (Gerbing & Anderson, 1988) using the pooled sample of respondents. Table 1 summarizes the CFA results.
Model fit (pooled sample): χ2 = 289.22 (df = 48, p < 0.01), CFI = 0.93, NFI = 0.92, RMSEA = 0.076.
*One item for each construct was set to 1.
The measurement model yields a chi-square of 289.22 (df = 48, p < 0.01). However, the Hoelter (0.05; Hoelter, 1983) estimate (=202) suggests that the lack of absolute fit can be explained by sample size. Thus, since the chi-square test is highly sensitive to sample size (MacCallum & Austin, 2000) other fit measures are given greater prominence in evaluating model fit (e.g., Ye, Marinova, & Singh, 2007). The root mean square error of approximation (RMSEA = 0.076), the comparative fit index (CFI = 0.93), and the normed fit index (NFI = 0.92) show an acceptable degree of fit of the measurement model (Bagozzi & Yi, 1988). Composite reliability, which represents the shared variance among observed items measuring an underlying construct (Workman, Homburg, & Jensen, 2003) was examined. All reliabilities exceeded 0.80, indicating good reliability of measured constructs (Bagozzi & Yi, 1988). Finally, extracted variance was greater than 0.5 for all latent constructs, which satisfies the threshold value recommended by Fornell and Larcker (1981).
In order to investigate discriminant validity the method proposed by Fornell and Larcker (1981) was applied. According to this method, the extracted variance for each individual construct should be greater than the squared correlation (i.e., shared variance) between constructs. An examination of Tables 1 and 2 shows that the extracted variance for each of the constructs exceeded the squared correlation.
Table 2. Correlations and Descriptive Statistics
Correlations adjusted for common method bias are reported above the diagonal; zero-order correlations are reported below the diagonal. CMV, common method variance.
Averaged scale means are reported; all items were measured on 7-point Likert scales.
Pooled sample correlations and descriptive statistics are reported.
Moreover, to further test discriminant validity, the baseline measurement model was compared to alternative models where covariances between pairs of constructs were constrained to unity (Anderson & Gerbing, 1988). In every case, the restricted model had a significant (p < 0.05) poorer fit than the unrestricted model suggesting sufficient discriminant validity.
Common Method Variance
Initially, a CFA approach to Harmon's one-factor test was used as a diagnostic technique for assessing the extent to which common method bias may pose a serious threat to the analysis and interpretation of the data (Kandemir, Yaprak, & Cavusgil, 2006; Ramani & Kumar, 2008). The single latent factor accounting for all the manifest variables yielded a chi-square value of 2208.24 (df = 54, p < 0.01). A chi-square difference test between the chi-square values of the two models suggested that the fit of the one-factor model was significantly worse than the fit of the four-factor model (Δχ2 = 1919.02, Δdf = 6, p < 0.01) indicating that the measurement model was robust to common method variance (CMV).
Next, the marker variable test suggested by Lindell and Whitney (2001) was conducted. In accordance herewith, a CMV-marker variable that is theoretically unrelated to at least one of the utilized research scales was used. Specifically, a 3-item scale measuring customers’ “propensity to use the Web when searching for financial information’ (α = 0.85; see Appendix) was chosen. This construct can be considered theoretically unrelated to BST (rxy = 0.06) as it does not relate to the magnitude of information search but merely to the use of a particular search instrument. Next, the correlations among study constructs with the CMV marker partialled out of each correlation were calculated (Table 2). An inspection of Table 2 shows that all significant zero-order correlations (reported below the diagonal) remained significant when adjusted for CMV (reported above the diagonal) suggesting that the results cannot be accounted for by CMV (Lindell & Whitney, 2001). In summary, based on the results of the conducted tests, CMV does not appear a problem in this study.
Models and Hypotheses Testing
Table 3 displays the results from estimating the hypothesized model of Figure 1.
Table 3. Estimated Coefficients for the Influence of Broad-Scope Trust on Customer Satisfaction, Narrow-Scope Trust, and Loyalty
*Significant on the 1% level; **significant on the 5% level.
The moderating effects were formed using the residual-centered, two-step procedure recommended by Little, Bovaird, and Widaman (2006). First, for each of the two interactions, that is, interactions involving BST and either satisfaction or narrow-scope trust, each of the nine possible product terms was regressed onto the first-order effect indicators of the two constructs. Second, for each of these regressions, the residuals were saved and used as indicators of the interaction construct. This method, which is facilitated by the relatively large sample sizes of this study, is regarded superior to more common path models, because it incorporates measurement error. Accounting for measurement error is beneficial because measurement error in exogenous and endogenous variables can attenuate regression coefficients and induce biased standard errors, respectively (Kaplan, 2009). The hypothesized model was fitted simultaneously to the bank and insurance industries samples using multiple-group latent variable structural equation modeling (SEM) analysis. Initially, a fully restricted model was estimated holding all paths invariant across the two data sets. Next, using a chi-square difference test, it was investigated whether paths with significant test statistics varied across subsamples. All of the released paths failed to enhance model fit suggesting that the investigated path coefficients did not differ significantly across the two subsamples. The chi-square statistic was 4262.55 (df = 825, p < 0.01) indicating that the model fails to fit in an absolute sense. However, since the χ2-test is very powerful when n is large, even a good fitting model (i.e., a model with just small discrepancies between observed and predicted covariances) could be rejected. The more robust fit indexes (CFI = 0.90, NFI = 0.91, RMSEA = 0.066) indicated an acceptable model fit. In addition, the NNFI, which is thought to be sensitive to both explanation and parsimony, equals 0.90, suggesting that the model shows an appropriate balance between these competing goals. To test the improvement in fit due to BST, the STL baseline model (omitting any relationships involving BST, but retaining all other paths) was estimated simultaneously for the two financial industries. Compared with the proposed model the results suggest that baseline model had inferior fit statistics: χ2 = 614.12 (df = 69, p < 0.01), CFI = 0.88, NFI = 0.88, RMSEA = 0.103, NNFI = 0.88. The results indicate that all relationships in the STL model, except for the satisfaction-loyalty path (bank industry: β = 0.04, p = 0.53; insurance industry: β = 0.04, p = 0.82), are significant and positive as expected. Satisfaction positively influenced narrow-scope trust (bank industry: β = 0.72, p < 0.01; insurance industry: β = 0.77, p < 0.01) and narrow-scope trust positively influenced loyalty (bank industry: β = 0.68, p = 0.02; insurance industry: β = 0.76, p = 0.02).
More important, the results suggest that BST positively influenced satisfaction (bank industry: β = 0.48, p < 0.01; insurance industry: β = 0.45, p < 0.01) and narrow-scope trust (bank industry: β = 0.18, p < 0.01; insurance industry: β = 0.21, p < 0.01) but that the influence on loyalty was nonsignificant (bank industry: β = −0.08, p = 0.31; insurance industry: β = −0.06, p = 0.45), although the coefficient was in the expected direction for both industries. Thus, H1a and H1b were both supported in the study, whereas H1c was rejected. Several significant moderating effects were obtained. Providing support for H2a, BST negatively moderated the relationship between satisfaction and narrow-scope trust (bank industry: β = −0.09, p = 0.05; insurance industry: β = −0.10, p = 0.02). In contrast to the expectation that a negative moderating effect would occur, BST positively moderated the relationship between satisfaction and loyalty (bank industry: β = 0.11, p < 0.01; insurance industry: β = 0.11, p < 0.01), rejecting H2b. BST negatively moderated the relationship between narrow-scope trust and loyalty (bank industry: β = −0.14, p = 0.01; insurance industry: β = −0.15, p < 0.01). Thus, H2c was supported.
Mediation Analysis and Test for Competing Models
The specified STL baseline model suggests that narrow-scope trust partially mediates the relationship between satisfaction and loyalty. However, the results revealed no direct link between satisfaction and loyalty. To further investigate the role of narrow-scope trust as a potential mediator of the satisfaction-loyalty relationship, four conditions under which full or partial mediation can be satisfactory documented were investigated (Baron & Kenny, 1986; Hair, Black, Babin, & Anderson, 2009). (a) The independent variable must be related to the mediator; (b) the independent variable must be related to the dependent variable; (c) the mediator must be related to the dependent variable; (d) full mediation is indicated if the effect of the independent variable is no longer significant when the mediating variable is added, whereas partial mediation is suggested if the effect of the independent variable is reduced but remains significant. The first condition was satisfied since satisfaction was significantly related to narrow-scope trust (bank industry, β = 0.66, p < 0.01; insurance industry, β = 0.68, p < 0.01). The second condition was met since satisfaction was significantly related to loyalty (bank industry, β = 0.57, p < 0.01; insurance industry, β = 0.56, p < 0.01). The third condition was also satisfied since narrow-scope trust was significantly related to loyalty (bank industry, β = 0.69, p < 0.01; insurance industry, β = 0.68, p < 0.01). The fourth condition is satisfied if the relationship between satisfaction and loyalty is either reduced (partial mediation) or become nonsignificant (full mediation) when the mediator (i.e., narrow-scope trust) is added to the model. The results revealed that the significant relationship between satisfaction and loyalty became nonsignificant when the mediator was included. Notably, this holds true across the two financial industries investigated and regardless of whether BST was included in the model (bank industry, β = 0.04, p = 0.53; insurance industry, β = 0.04, p = 0.82), or not (bank industry, β = 0.16, p = 0.43; insurance industry, β = 0.12, p = 0.47). Thus, the conditions specified suggest that narrow-scope trust fully mediates the relationship between satisfaction and loyalty. To inspect whether the model needs to be respecified as a consequence of this result, a competing model leaving out the relationship between satisfaction and loyalty was tested. Initially, a chi-square difference test suggested that the removal of the satisfaction-loyalty relationship from the model did not improve model fit (Δχ2 = 0.21, Δdf = 1, p = 0.65). Further, the results for all the competing model hypotheses tests were identical to the results reported above suggesting that the results are robust with respect to whether the satisfaction-loyalty relationship is included in the model or not. A second competing model, which reversed the hypothesized relationship between satisfaction and narrow-scope trust (i.e., in the second competing model narrow-scope trust (a) influenced loyalty through satisfaction and (b) directly influenced loyalty), was also specified. This model yielded the following fit to the data: χ2 = 4443.23 (df = 825, p < 0.01), CFI = 0.90, NFI = 0.90, RMSEA = 0.072 indicating that the hypothesized model was a better fit to the data than the second competing model.
This study investigated the direct and moderating influence of BST on customer–seller relationships. In that respect, a distinction was made between two forms of trust: BST and narrow-scope trust. This study provides evidence that the explanation and understanding of relationships between customer satisfaction, narrow-scope trust, and loyalty is significantly enhanced by inclusion of the potential direct and moderating effects of BST. The estimation of the STL relationships in the framework indicated that the relationships between constructs were significant as expected, except for the nonsignificant direct relationship between satisfaction and loyalty. Although the nonsignificant direct relationship between satisfaction and loyalty was unexpected, similar results concerning the satisfaction-loyalty relationship has been evidenced in previous research, which often fails to show a strong association of satisfaction and loyalty (Jones & Sasser, 1995; Neal, 1999; Nijssen et al., 2003; Oliver, 1999). In the present study, the lack of a direct relationship between satisfaction and loyalty may be related to many consumers perceived complexity of financial markets (Estelami, 2005; OECD, 2006). This is because perceived complexity emphasizes the importance of trust in a service provider (Guenzi & Georges, 2010) suggesting that although satisfaction may be an important ingredient for the emergence of loyalty, it does not necessarily translate into customer loyalty. This is consistent with the finding that narrow-scope trust fully mediates the relationship between satisfaction and loyalty indicating that this form of trust has assumed tremendous significance in financial customer–seller relationships as it also has been noted by previous research (e.g., Romàn & Ruiz, 2005). Also, past research has indicated that loyalty is more likely driven by satisfaction in industries where transactional relationships are common, whereas loyalty is more likely driven by trust in industries where enduring relationships (e.g., customer–seller financial relationships; Guenzi & Georges, 2010; Finextra, 2009) are common (Garbarino & Johnson, 1999; Johnson & Selnes, 2004).
The results suggest that BST positively influences satisfaction and narrow-scope trust, whereas no direct influence of BST on loyalty was detected although the coefficients were in the expected direction. This finding might be related to the observation that financial customer–seller relationships are often relatively close and long term. Early in a relationship, narrow-scope trust is most likely based primarily on general trust in financial institutions, but as the relationship continues, a divergence is more likely between BST and narrow-scope trust (Hall et al., 2002). This is because as individual relationships develop, the personalization that occurs may boost the level of narrow-scope trust (Gutek, Bhappu, Liao-Troth, & Cherry, 1999). Notably, this suggestion is consistent with the present study results. A comparison of BST with narrow-scope trust shows that narrow-scope trust is approximately one-fifth higher on average than BST (5.65 vs. 4.74, p < 0.01; Table 2). Similar results have been observed in other fields, such as medicine, where people typically have stronger trust in their individual physician than in physicians in general (Hall et al., 2002). Specifically, it is proposed that as the divergence between narrow-scope trust and BST increases consumers may be more reluctant to switch financial service provider, which in turn may diminish the negative influence of BST on relationship loyalty. However, these are speculations and it is therefore suggested that future research collects longitudinal data to examine the possible decreasing influence of BST on loyalty as customer–seller relationships develop.
The results indicate that BST has a positive moderating influence on the relationship between satisfaction and loyalty. Thus, by taking into account the contextual effect of BST, the present study contributes to the explanation of the satisfaction-loyalty relationship in customer–seller relations, although the moderating effect was in the opposite direction than predicted. Two different possible explanations for this finding are suggested. First, when BST increases consumers may be inclined to have a more optimistic view of continuing their financial relationships. This is because they need to resolve the cognitive dissonance that would otherwise exist if they believed, in a situation with high satisfaction, that their financial service provider is not better than average to be continuously engaged with. Second, the moderating influence of BST on the satisfaction-loyalty relationship may be affected by other variables. For instance, applying moderated regression analysis, Bloemer and Kasper (1995) found that involvement positively moderated the relationship between satisfaction and loyalty. Also, in an empirical study of car customers Homburg and Giering (2001) found that variety seeking, age, and income were important moderators of the satisfaction-loyalty link, whereas less evidence for moderating effects was detected for gender and involvement. Based on the findings in the present study, the result that BST positively moderates the satisfaction-loyalty link may encourage the collective of financial managers to reduce customer-switching rates through investments in BST.
As expected, the findings also indicate that BST has a negative moderating influence on the relationship between satisfaction and narrow-scope trust and a negative moderating influence on the relationship between narrow-scope trust and loyalty. These findings have important implications for financial managers. While managers should be concerned that a low level of BST may have a direct negative influence on satisfaction and narrow-scope trust, they should also be concerned when BST level is high or on the increase since BST negatively moderates the relationships between customer satisfaction and narrow-scope trust and between narrow-scope trust and loyalty. Since BST is an environmental effect, these negative effects hold true even for service providers who have not actively participated in improving BST. On the other hand, with higher levels of BST managers may benefit from an improved relationship between satisfaction and loyalty. Thus, the results stress that BST may significantly influence customer–seller relationships beyond the direct and indirect influence, which previous research has focused on. Moreover, an equal importantly, the results point to the complexity of the moderating influence of BST on customer–seller relationships. However, because BST positively influences narrow-scope trust and customer satisfaction managers have hardly any choice other than to encourage the development of both types of trust, even though developing BST may negatively moderate some of the relationships between customer–seller relationship constructs. In order to deal with (especially) the negative moderating effects of BST, it is suggested that future research looks further into possible background explanations for their presence. This study drew on attribution theory in order to develop a theoretical understanding of the moderating effects of BST on the STL baseline model relationships. However, it is not suggested that this study provides a definitive background understanding of the complexity of the proposed relationships. Indeed, the coefficient concerning one of the specified hypotheses was in the opposite direction than expected. Future research should therefore regard the propositions put forward in this study as starting points for a further understanding of the role of BST in customer–seller relationships, which is clearly an under-researched topic. In relation hereto, future research could also collect longitudinal data to assess the long-term influences of BST on narrow-scope trust, customer satisfaction, and loyalty. Such an investigation could validate the notions that BST positively influences satisfaction and narrow-scope trust and provide further evidence for the important role of BST as a moderator of the STL model relationships. Longitudinal studies would also help understand whether the nature of the effects obtained in this study is indeed long term.
There are several limitations of this study that should be acknowledged. Customers were approached via online surveys; they may behave differently when engaging in specific relationship settings. Thus, although a survey is generally accepted as a means of data collection, there is little control over the contextual setting and over the response behavior of customers (Kozup, Creyer, & Burton, 2003). This study concentrated on analyzing the consumer population of one society and in two industries. Although the investigated financial industry types are present in most societies and even though their service offerings are most likely guided by similar financial and economic principles, this could mean that the results may suffer from a lack of generalizability when other countries and/or industries are considered. Future research is also called upon to take into account cultural characteristics such as, for example, the degree of customer uncertainty avoidance, among others. According to Hofstede (2001), uncertainty avoidance reflects a society's tolerance for uncertainty and ambiguity. Since trust may decrease uncertainty, financial customers within uncertainty avoiding societies may put higher emphasis on trust (broad-scope and narrow-scope trust) when compared to less uncertainty avoiding societies. Also, this study used perceptive measures, which could be threatened by biased responses. Future research could examine this issue by manipulating BST in an experimental setting. Such an experimental study would also replicate the present cross-sectional survey results in a more controlled laboratory setting, and thus provide stronger evidence for the direction of causality in the proposed research model. This study included only one consequence of consumer satisfaction and narrow-scope trust: relationship loyalty. Other possible consequences of satisfaction (e.g., word-of-mouth communication) and narrow-scope trust (e.g., commitment) could be included in the baseline model to further explore the possible moderating effect of BST on customer–seller relationships.
The data for this study were collected in collaboration with the Danish Money and Pension Panel.
Items used to measure the constructs in the study
*Item reverse coded.
Identical measurement items were applied across the two investigated financial industries.
X1. In general, I believe that financial companies cannot be relied upon to keep their promises*
X2. In general, I believe that financial companies are trustworthy
X3. Overall, I believe financial companies are honest
X4. I believe that my [financial service provider] cannot be relied upon to keep its promises*
X5. I believe that my [financial service provider] is trustworthy
X6. Overall, I believe my [financial service provider] is honest
X7. I am satisfied with the relationship I have with my [financial service provider]
X8. As a regular customer, I have a high quality relationship with my [financial service provider]
X9. I am happy with the effort my [financial service provider] is making towards regular customers like me
X10. I plan to terminate the relationship with my [financial service provider]*
X11. I'm considering changing [financial service provider] within the next twelve months*
X12. I consider myself as a loyal customer to my [financial service provider]
Propensity to use the Web when searching for financial information (CMV marker)
CMV1. When searching for financial information in general
CMV2. When searching for financial information relating to prespecified financial services
CMV3. When searching for information that compares financial services