Introducing a composite measure of trust in financial services

ABSTRACT Existing trust scales generally measure aspects of trustworthiness and not trust per se. Trust is a broader concept encompassing attributes of the trustee, trustor and the situation/context. The purpose of this study is to develop a composite measurement scale for trust in financial services that incorporates elements of these three facets of trust. The study draws on interdisciplinary theories and adopts a broadly quantitative approach to develop, test and validate a five-dimensional scale for measuring trust in financial services. The trust scale for financial services developed through this study has five dimensions (5Cs): character-competence, congruence, communication, commitment and context. The scale provides a holistic conceptualisation of trust and displays solid psychometric properties. A comprehensive interdisciplinary trust scale for financial services, with strong reliability and validity, holds important managerial implications, its ability to capture the attributes of the trustee, trustor and financial system attesting to its suitability as a diagnostic tool to measure trust more robustly. Our trust scale has significant practical implications, offering useful insights for policymakers, commercial organisations and other stakeholders. This is the first trust scale for financial services that captures the attributes of trustee, trustor and context, reconcilling tensions between the conceptualisation and operationalisation of trust.


Introduction
In the financial services sector, where products are intangible, difficult to evaluate and unpredictable in terms of outcomes, trust is pervasive (Ennew et al., 2011;Ennew & Sekhon, 2007;Sekhon et al., 2014) and levels of perceived risk relatively high (Allen et al., 2018). Such a situation is not surprising as research indicates that trust is inextricably linked with vulnerability (Gefen, 2002;Martin, 2018;Mayer et al., 1995) and the aforementioned high level of perceived risk (Hurley, 2012;McKnight et al., 2002;McKnight & Chervany, 2001;Rousseau et al., 1998). In this context, the core mediating effect of increased trust facilitates transactions, reducing the level of perceived risk and the requirement for deliberative effort (Elliott & Percy, 2007;Ennew & Sekhon, 2007). Trust also helps promote strong relationships and increases the willingness to purchase (Wang et al., 2015). In the financial services context, levels of trust have also been linked with the degree to which consumers are willing to delegate decision making on their behalf to organisations (Roy et al., 2020) and Huurne et al. (2017) identified trust as the 'key factor in overcoming uncertainty and mitigating risk ' (p. 485).
The concept of trust is rooted in multiple disciplines such as psychology, sociology, economics, social psychology and organisational behaviour (Mayer et al., 1995;McKnight et al., 1998;McKnight et al., 2002;McKnight & Chervany, 2001). This multidisciplinary perspective creates a tension between the conceptualisation and many operationalisations of trust (Rousseau et al., 1998). For example, although conceptually trust is a multidimensional concept, empirically, it is primarily treated as a unidimensional construct (McEvily & Tortoriello, 2011). Moreover, in Seppänen, et al's (2007) comprehensive review, several multidimensional trust scales fail to take into account the attributes of all parties and systems involved in the trust relationship. Many existing scales (e.g. Ennew et al., 2011) focus on the 'trustee' (or people and organisations to be trusted) and ignore the trustor's and contextual attributes, particularly in research focused on trust in business-to-consumer (b2c) contexts. Such discrepancies challenge the validity, precision and usefulness of existing trust scales when applied to a business-to-consumer setting.
Drawing on social psychology (Rempel et al., 1985), several trust scales have been designed to manifest theories of trustworthiness and interpersonal trust, focussing primarily on a trustee's characteristics such as honesty, integrity, benevolence and predictability (McKnight et al., 1998). In the context of financial services, these are integrity and consistency, concern and benevolence, expertise and competence, shared values and communications (Ennew et al., 2011;Ennew & Sekhon, 2007). Equally important, the other facets of trust, such as institutional trust (McKnight et al., 2002;McKnight & Chervany, 2001), have been overlooked by many trust scales. In the small number of studies where such aspects of trust have been covered (see Moin et al., 2015, for instance), they have been studied discretely, resulting in a limited understanding of relationships and interactions with other aspects of trust. In financial services, the role and importance of institutional trustthat is, trust in the financial system/structure from regulatory and company perspectivescannot be underestimated. Indeed, it has even been claimed that it is impossible to exaggerate the role of trust in financial services (Brychko et al., 2021). First, it is essential to safeguard consumers from the opacity and perceived risk associated with many of the products in the sector and the system within which they are marketed to consumers (Ennew et al., 2011;Ennew & Sekhon, 2007); and second, consumers' perception of the robustness of the financial system as a whole has a strong bearing on their decisions relating to trust (Sekhon et al., 2014).
The current study extends the debate on the measurement of trust in the context of financial services. It makes a contribution by offering a composite model of trust that combines elements of interpersonal trust (trust in the supplying organisation) with aspects of institutional trust (trust in the financial system as a whole), and the trustor's propensity/ commitment to take the risk by being vulnerable to the actions of trustees. After all, a combination of all three elements is required to provide the overall trust and reassurance needed for consumers to participate in a market characterised by a relatively high degree of risk and vulnerability (Gillespie, 2012;McKnight et al., 1998;Saunders et al., 2015). Our new measure consists of five dimensions, which we style the 5Cs: character-competence, congruence, communication, commitment and context. The scale offers a holistic conceptualisation of trust and incorporates trustor-and context-oriented attributes in addition to trustee-related attributes. The five-dimensional measure also captures the trustworthiness of financial institutions (attributes of trustees), the willingness of consumers to take the risk (attributes of trustors), and the robustness of structures in place within the financial system (contextual attributes). The study addresses the limitations of existing trust scales and provides interested academics and practitioners in the financial services sector and policymakers with a sophisticated composite tool to measure trust.

Conceptualising trust from an interdisciplinary perspective
Trust is best defined as the willingness of an individual, group, organisation or institution (i.e. a trustor) to accept risk or vulnerability arising from the actions of another party (i.e. a trustee) with the expectation that the trustee will not act contrary to the interests of the trustor (Martin, 2018;Mayer et al., 1995;Pirson et al., 2014). Although this is a widely accepted definition of trust, its conceptualisation is challenging due to the divergent bodies of literature that exist. McKnight et al. (1998) identified three primary deficiencies/ challenges concerning the understanding of trust: (a) there is a degree of construct confusion in the literature due to divergent conceptual definitions; (b) there is too little coverage or understanding of how trust is formed or about the basis of trust; and (c) the role of emotion in the formation of trust is mainly absent. Dietz and Den Hartog (2006) also provided a comprehensive review of trust measures, weighing them against the conceptualisation of trust and highlight challenges that arise from 'three constituent parts: trust as belief, as decision and as action' (p. 558).
Inquiry into the production and formation of trust (Anderson & Narus, 1990;Zheng et al., 2017) has shaped the conceptualisation of trust from a multidisciplinary perspective (Mayer et al., 1995;McKnight et al., 1998;McKnight et al., 2002;McKnight & Chervany, 2001;Rousseau et al., 1998;Tan & Sutherland, 2004;Zucker, 1986). However, there are some significant challenges in approaching trust through the multidisciplinary lens (Rousseau et al., 1998). Most management and social science researchers have conceptualised trust by drawing on psychology, sociology, social psychology, economics and organisational behaviour, striking a balance between the depth and breadth of interdisciplinary research. Each of these disciplines views trust somewhat differently, resulting in a richer understanding of this complex phenomenon through several schools of thought.
The first school of thought, aligned with the world-view of social psychologists (Rempel et al., 1985), views trust as a set of beliefs or expectations about trustees' motives or future intentions (Sitkin & Roth, 1993). This approach focuses on the attributes, or trustworthiness, of trustees. In the financial services sector, the trustees are institutions such as banks, building societies, insurance companies, investment companies, brokers and advisers, and credit card companies (Ennew & Sekhon, 2007). Trustors are the customers of those organisations.
Built around the worldview of psychologists (Rotter, 1971), the second school of thought views trust as 'an underlying psychological condition comprising the intention to accept vulnerability based upon positive expectations of the intentions or behaviour of another' (Zheng et al., 2017, p. 218). Thus, psychologists conceptualise trust as the willingness of trustors to depend upon trustees (Doney et al., 1998;Mayer et al., 1995) or as behavioural intention, that is, the willingness of trustors to deploy their expectations or beliefs (Luhmann, 2017;McAllister, 1995;Scott, 1995). This school conceptualises trust as a 'decision to accept the risks of dependence because of the expectation that others will act beneficially' (Alpenberga & Scarbroughb, 2018, p. 528). As previously explained, the trustor is the client of the financial services organisation in our context in question.
The third school of thought also uses the world view of the psychologist (Rotter, 1971) but looks at trust-related behaviour in taking actions that involve risk-taking (Moorman et al., 1992) and increase the vulnerability of one person to another (Deutsch, 1962). Finally, the fourth school of thought builds on sociologists' worldview (McKnight et al., 2002) and argues that contextual structures such as legal protections and safety arrangements play a role in creating an environment of trust. Thus, instead of focusing on trustee and trustor's attributes, emphasis is placed on the contextual or situational characteristics that will make an environment safe (McKnight et al., 1998). For example, trust in the context of developing countries (Saleh et al., 2014) is not the same as that in developed countries due to different safety arrangements and control mechanisms (Wang et al., 2015). In summary, trust is variously conceptualised according to the intentions of the trustee; the willingness of the trustor to depend on the trustee; the behavioural attributes of the trustor; and the broader context of the trust relationship. However, in our assessment, this rich conceptualisation manifested most comprehensively by McKnight et al. (2002) has not been subject to rigorous empirical testing across varying contexts. The authors acknowledge that their chosen context of giving advice in an online setting may have impacted the strength and significance of relationships they found or otherwise. The authors also recommended further tests on non-student subjects. In developing our new scale focusing on the context of financial services, we seek to address such concerns.

Rationale for a new trust scale: A composite approach
To account for the differing approaches to the conceptualisation of trust outlined in the previous section, it is clear that a nuanced, multi-disciplinary conceptualisation is required, where 'trust is depicted as occurring under conditions of risk which require the trusting party (the trustor) that are sufficient to prompt a willingness to become vulnerable to the trustee's future conduct.' (Saunders et al., 2015, p. 169). Drawing on Saunders et al. (2015) and other seminal papers, it can be deduced that in financial services, trust should involve at least the following process: . First, a rational choice made by financial services consumers (trustors) through an evaluation of the 'character and competence' of the financial institutions ( ' and [c] involves 'relational-based/ identity-based trust' (Dietz & Den Hartog, 2006;Rousseau et al., 1998;Saunders et al., 2015;Williams, 2001).
However, trust is most often operationalised as unidimensional: at least 46 dimensions of trust can be found in the literature, but in McEvily and Tortoriello's (2011) review, most studies, 161 out of 207 (78%) treat it as a unidimensional construct. Also, trust measurement predominantly looks at trustees' attributes from a social psychologists' perspective (Rempel et al., 1985). This marks an inconsistency between conceptualisation and operationalisation (McEvily & Tortoriello, 2011;Rousseau et al., 1998): the operationalisation of trust largely ignores the trustor and context-oriented attributes. Dietz and Den Hartog (2006) also found tension between the conceptualisation and operationalisation of trust and state: 'existing measures match the theory, but also show a number of "blindspots" or contradictions, particularly over the content of the trust belief, the selection of possible sources of evidence for trust, and inconsistencies in the identity of the referent.' (p.557). Where trust is viewed as a multi-dimensional construct, the other challenges of trust measurement that Dietz and Den Hartog (2006) highlighted are germane such as what is the relative importance of measures, are they measuring what they intend to measure and whether actions of trusting need to be included in the trust model. Table  1 gives an overview of the existing trust scales along with dimensions and the attributes these dimensions are intended to measure. Table 1 includes the major trust scales identified through reviewing the literature on measurement or operationalisation of trust as well as consulting the papers that have reviewed measures of trust, such as Dietz and Den Hartog (2006), McEvily and Tortoriello  Robinson (1996) Trust between employees and employers (trust seen as intraorganisational phenomenon) 1 7 items were used to measure trust (p. 583)

Yes
No No Brockner et al. (1997) Employees' trust in managers (trust seen as intra-organisational phenomenon) 1 3 items were to measure trust (p. 563)

Yes
No No Mayer and Davis (1999) Influence of effective performance appraisal system on trust for management (trust seen as intra-organisational phenomenon) 1 Trust: Shown as 1dimensional construct, measured by 4 items. Trustworthiness: Shown as 3-dimensional construct: Ability, benevolence, integrity.

Yes Yes No
Armfield et al.
Dental Trust Scale (DTS) measuring trust in the dental profession (trust between customers and specific professionals: marketing concern) 1 Trust in dentist Yes No No Ganesan (1994) Retail buyer and vendor relationship (trust seen as inter-organisational phenomenon)  Robort (performance, physical attributes), human (safety, experience), and external (task).

Yes Yes No
McKnight et al.
Trust in e-commerce (trust between consumers and vendors: marketing concern) 4 Disposition to trust, institutional trust, trusting belief, and trusting intention Yes Yes Yes Smith and Barclay (1997) Trust in selling partners' relationships (trust seen as inter-organisational phenomenon)

5
Honesty/ integrity, reliability/ dependability, responsibility, likeability and judgment Arguably, the majority of researchers are more interested in measuring 'trustworthiness' (Ennew & Sekhon, 2007;Ennew et al., 2011;Mayer and Davis, 1999) and not trust per se. This is a flaw of existing scales measuring trust in financial services. Arguably, trust as a concept is particularly crucial in the financial services arena. Here many products and services are characterised by complexity, a high degree of perceived risk and a significant level of fiduciary responsibility. Added to that, offerings are highly mental-intangible, and the benefits associated with products may not manifest themselves for several years. Finally, the financial services context is by no means immune from scandals, including the mis-selling of products and services. Furthermore, employing Dietz and Den Hartog (2006) identification of three broad stands of trust literature, it is noticed that more of the scale measures trust between the organisation (which fall under trust in the interorganisational setting); some measures trust within the organisation: employees trust upon their managers (which fall under trust in the intra-organisational setting); and only a few measures trust between organisations and their consumers/customers (marketing concern). The review of the suitability of the trust scales shown in Table 1 leads to the conclusion that a new composite scale to measure aspects of trust in financial services is beneficial. Whilst the scale derived by Ennew and Sekhon (2007) is highly relevant, it lacks incomprehensiveness and inclusiveness. It does not set out to measure the attributes of trustors (i.e. financial services consumers) or investigate the context in any detail, namely trust in the financial system or institutional trust. The McKnight et al. (2002) scale is comprehensive but does not account for the fact that financial services has unique attributes in terms of the type and nature of risks involved and the degree and relevance of institutional trust for the context. Thus, a new composite scale that builds upon the best and most relevant elements of these scales provides an extremely comprehensive and highly relevant composite measure of trust in the financial services sector.  (2007) Trust in financial services (trust between organisations and their customers: marketing concern) 5 Integrity and consistency, concern and benevolence, shared values, expertise and competence, and communications

Dimensions of trust
Most of the dimensions of trust scales mentioned in Table 1 are drawn from interpersonal trust theories rooted in social psychology and economics (McKnight & Chervany, 2001). These scales mainly deal with the trustee's attributes, collectively known as trusting belief (McKnight et al., 1998) or trustworthiness/ drivers of trust (Ennew et al., 2011;Ennew & Sekhon, 2007). The trustee's attributes foster a firm conviction within the trustor that the trustee has both character (i.e. will act morally and ethically) and competence (i.e. expertise to deliver what is expected) to protect the interests of the trustor. Among them, dimensions such as integrity, competency, credibility, benevolence, honesty etc. measure the cognitive-based trust, which involves accessing trustee's attributes by the trustors through rational decision making process (Isaeva et al., 2020) based on the credible information from others, also known as calculus-based trust (Rousseau et al., 1998). However, these are not sufficient to measure trust as the trustor's propensity to take a risk by placing trust upon the trustee is also a vital component. For instance, Friend et al. (2018) considered trustors' propensity to trust salespeople. Chang et al. (2016) also provide valuable insights into the relevance of capturing trustor's attributes while measuring mobile users' trust in smartphone social networking services. However, most existing trust measurements (see Table 1) overlook trustor's attributes except a few (e.g. McKnight et al., 2002).
Rooted in theories of institutional trust, structural assurance, and situational normality (McKnight et al., 1998;McKnight et al., 2002;McKnight & Chervany, 2001), contextual or situational attributes are crucial for in trust financial services (Moin et al., 2015), which is a highly regulated sector. Institutional arrangements such as guarantees, rights structures and safety nets provide a sense of structural assurance (McKnight et al., 1998;Shapiro, 1987;Zucker, 1986), and in the case of financial services, it refers to trust in the financial system as a whole (Lavezzoloa et al., 2018). At the macro-level, the robustness of the financial system depends on the policies and regulations to protect customers and the roles regulators (e.g. the Financial Conduct Authority and the Financial Ombudsman) play. At the micro-level, institutional trust in the financial system depends on whether financial institutions act honestly and ethically, to what extent they comply with the rules and regulations, and the quality of training for employees. The concept of contextual attributes relating to institutional trust is also supported by economic exchange theories, which provide a foundation for structuring economic/ financial transactions between two parties (Eisenhardt, 1989;Friend et al., 2018;Jensen & Meckling, 1976;Noorderhaven, 1992). An ideal trust scale for financial services should have the capability to measure the attributes of the trustee, the trustor, and the context, i.e. institutional arrangement.
In general, the existing trust scales, for the most part, measure the attributes of trustees (i.e. trustworthiness rather than trust) while ignoring contextual attributes and trustor attributes. Thus, the tension between differing conceptualisations and the operationalisation of trust remains unresolved. Recently, researchers such as Isaeva et al. (2020) have recommended that researchers aim for maximum clarity when discussing the theory underpinning trust and have implored scholars to account for the multilevel nature of trust in research, which we seek to do here. Figure 1 gives an overview of the conceptualisation and operationalisation of trust, substantiating the need for a composite measure of trust in financial services, which we develop in the remainder of this paper.

Item generation and content validity
The scale items were drawn from the interdisciplinary literature on trust (Mayer et al., 1995;McKnight et al., 1998;McKnight et al., 2002;McKnight & Chervany, 2001) and trust in the context of financial services (Ennew et al., 2011;Ennew & Sekhon, 2007;Moin et al., 2015;Sekhon et al., 2014). To measure the attributes of the trustee, 23 items were adapted from Ennew and Sekhon (2007)'s scale on trustworthiness dealing with interpersonal trust. A further 19 items were developed from the literature on dispositional trust and trusting intention to measure attributes of the trustor (e.g. the person taking a risk by trusting others). To measure the contextual attributes of trust, 8 items were derived from the literature on institutional trust and the roles of Financial Conduct Authority (www.fca.org.uk), and Moin et al. (2015), who studied structural assurance in the financial services sector. The initial pool consisted of a comprehensive list of 50 items.
Next, content adequacy (content validity) of the 50 scale items were assessed using four senior academics with extensive experience researching trust and five senior professionals from the UK financial services sector. Participants were asked to evaluate the representativeness of the trust items in capturing the attributes of the trustee, trustor and context as described above. A variant of Zaichkowsky's (1985) procedure was employed to narrow down the item pool. Trust items were kept if at least six of the nine judges rated them as 'somewhat representative' of the construct. As a result of this process, a final list of 40 items was retained.

Main study: sampling and data collection
For the main study, a purposive sampling procedure identified potential respondents. Two criteria were considered: participants should possess a basic knowledge of financial products and should have, as a minimum, a bank account (considered as a financial product). The first author had access to employees working for a large multinational company with branches in several major UK cities. Four hundred and twenty questionnaires were distributed via the company's internal mails, and 300 respondents completed the survey (71.43% response rate). The demographic profile of respondents is summarised in Table 2. Besides the convenience of access, this group of participants provides a good sample as the nature of their job, education, profession, etc. helps ensure that they are sufficiently informed. This scale development study is part of broader research seeking to develop an integrative and interdisciplinary brand-trust model and includes multi-item scales measuring other constructs such as brand experience (Brakus et al., 2009) and brand personality (Aaker, 1997). Respondents had to rate their level of agreement/disagreement with each statement using a 7-point Likert-type scale with anchors 1 = 'strongly disagree' and 7 = 'strongly agree'.

Exploratory factor analysis
To identify the dimensions of trust, principal component exploratory factor analysis with varimax rotation was conducted. Items with factor loading lower than 0.40, crossloading higher than 0.40 and/or item-to-total correlations lower than 0.50 were candidates for deletion (Hair et al., 2006). As a result, 12 items were dropped, resulting in a final five-factor model with the remaining 28 items, explaining 76.47% of the total variance. Factor loadings were high (≥ .537), and item-to-total correlations exceeded .50, indicating that the sample size did not affect the quality of the factor solutions (MacCallum et al., 1999). The first dimension was labelled as 'character-competence' and explained 23.211% of the variance (eigenvalue = 6.731); the second dimension was labelled as 'congruence' and accounted for 11.884% of the variance (eigenvalue = 3.446); the third dimension explained 8.731% of the variance (eigenvalue = 2.532) and was labelled as 'communication'; the fourth dimension explained 22.332% of the variance (eigenvalue = 6.476) and was labelled as 'commitment'; the fifth dimension explained 10.314% of the variance (eigenvalue = 2.991) and was labelled as 'context'. The trustworthiness of trustee is measured by character-competence, congruence and communication; the trustor's disposition to trust is measured by commitment, and trust in the overall financial system is measured by context. For simplicity and better memorability, we refer to the new measure as the 'Five Cs trust scale' (in short as 5CTS).

Confirmatory factor analysis and unidimensionality
The next stage is to establish the unidimensionality of the new scale. Unidimensionality is 'one of the most critical and basic assumptions of measurement theory' (Hattie, 1985, p. 139) and is considered as a 'logical and empirical necessity' (Bagozzi, 1980, p. 126). Unidimensionality refers to the existence of a single trait or construct underlying a set of items . Two conditions exist for measures to be considered unidimensional. First, an indicator should be significantly associated with the underlying latent variable and, second, the indicator must represent a single factor (Anderson & Gerbing, 1982;Phillips & Bagozzi, 1986). Following Pedhazur and Schmelkin (1991) and consistent with recent scale development studies (e.g. Joyner Armstrong et al., 2018), confirmatory factor analysis (CFA) was used to test for unidimensionality. A 28-item, five-dimensional (character & competence, congruence, communication, commitment and context) CFA model was estimated using LISREL 8.80. Standardised factor loadings, composite reliability and average variance extracted (AVE) are presented in Table 3. The overall fit of the CFA model was examined using commonly used parameters. Values for CFI (0.982), IFI (0.982) and NNFI (0.979) were above the ≥0.90 recommended cut-off value, and the root square error of approximation (RMSEA) was at 0.07 (below the accepted threshold of ≤0.08) (Hu & Bentler, 1998). The chi-square value (χ 2 = 896.46) did not exceed three times its degrees of freedom, df = 334 (Bollen, 1989). Overall, results indicate satisfactory measurement model fit.

Reliability assessment
Once the unidimensionality has been established, the next step is to assess the scale's reliability . The trust scale demonstrates high internal consistency with coefficient alpha for the five dimensions exceeding Nunnally's (1978) recommended value of 0.70 for new scales: character and competence (0.963), congruence (0.891), communication (0.896), commitment (0.952), and context (0.848). Besides, construct (composite) reliability (CR) was computed, consistent with recommended guidelines (e.g. Bagozzi & Yi, 1988;Baumgartner & Homburg, 1996;Medsker et al., 1994;Steenkamp & Van Trijp, 1991). Composite reliability was calculated using the squared sum of factor loadings for each construct and the sum of error variance terms (Fornell & Larcker, 1981;Werts et al., 1974). The scale sub-dimensions meet the minimum critical value for CR estimate 0.60 (Bagozzi & Yi, 1988), ranging from 0.857 to 0.956. Overall, results provide strong evidence of the scale's reliability.

Convergent validity
Having established unidimensionality and reliability, the next step is to check convergent and discriminant validity (Campbell & Fiske, 1959). Convergent validity is the extent scale items designed to measure a latent variable, statistically correlate (Hosany et al., 2015). Convergent validity was tested in three ways by: (i) checking the statistical significance of factor loadings ; (ii) assessing the magnitude of factor loadings (Hair et al., 2010;NeteMayer et al., 2003); and (iii) comparing average variance extracted (AVE) (Fornell & Larcker, 1981). First, all standardised confirmatory factor loadings are significant (p < 0.01), with t values greater than 2.57 (NeteMayer et al., 2003), ranging from 6.047 to 11.752. Second, the standardised factor loading for each item is substantial, ranging from 0.62 to 0.941 (Table 3). Finally, AVEs for the 5 dimensions exceed 0.50 (Fornell & Larcker, 1981). Together, these results provide evidence of convergent validity.

Discriminant validity
Discriminant validity is the extent to which scale items representing a latent variable isolate that construct from items representing other theoretical variables (Fornell & Larcker, 1981). Discriminant validity of the scale was first investigated by examining correlations between the 5 dimensions of the trust scale and another theoretical construct. In deciding on the other theoretical construct, we considered the intangibility and complexity associated with the financial services and brand experience plays an important role in promoting trust. Like trust, rooted in multiple disciplines, brand experience refers to consumers' subjective responses when exposed to brand-related stimuli, such as logo, name and advertisements (Brakus et al., 2009). Long-lasting brand experiences stored in consumer memory reduce the perception of risk. Brand experience consists of four facets: sensory, affective, intellectual and behavioural (Brakus et al., 2009). Trust and brand experience, in particular the affective component capturing emotions, although related, are theoretically distinct. Trust represents confidence in future outcomes (Hurley, 2012) and a willingness to take risks (Rousseau et al., 1998).
On the other hand, brand experience is evoked during consumers' interaction with brands (Brakus et al., 2009;Iglesias et al., 2011). According to the interdisciplinary perspective, trust can be seen through three lenses: cognitive, affective, and behavioural (Lewis & Weigert, 1985;Luo & Zhang, 2016). While cognitive-based trust helps in rational decision making and logical assessment, the emotional bond plays a role in creating the affectivebased trust (Isaeva et al., 2020). Our study was part of broader research that involves collecting data for brand experience and brand image for financial services brand, and we found the affective dimension of the brand experience as the closest construct to use for discriminant validity.
To test the discriminant validity of the trust scale, we used the affective dimension of the brand experience scale. The affective dimension (α = 0.847) was measured using 4 items adapted from Brakus et al. (2009). To establish discriminant validity of the trust scale, we follow Bagozzi et al. (1991) recommended procedure. Constructs were assessed in sets of two. For example, the 'congruence' dimension of trust was tested against the 'affective' dimension of brand experience. A series of one-and two-factor CFA models were conducted for every possible pair. For the one-factor models, the correlation between two constructs was set at 1.00, whereas the correlation parameter was freely calculated for the two-factor models . A chi-square difference test was performed between the one-factor and two-factor models. Discriminant validity is achieved if there is a significant difference in the chi-square statistic between the twoand one-factor models. From Table 4, all chi-square difference were significant (p < 0.001), and therefore establish discriminant validity of the trust scale. Discriminant validity was further assessed by comparing the squared correlation between a pair of constructs (shared variance) against the AVE for each of the two constructs (Fornell & Larcker, 1981). If for each pair of constructs, the shared variance is smaller than both the AVEs, indicating that the constructs exhibit discriminant validity. From Table 5, all AVEs are greater than the corresponding interconstruct squared correlation estimates (above the diagonal) and thus further support the discriminant validity of the trust scale (Fornell & Larcker, 1981).

Nomological validity
To establish the nomological validity of the scale, correlation analysis was performed following existing guidelines (e.g. Hair et al., 2010) and prior research (e.g. Seiders et al., 2007;Wong & Wan, 2013). Nomological validity is the extent to which a scale is related to another construct consistent with underlying theories or prior research (Bagozzi, 1980;Hair et al., 2010;Peter, 1981;Steenkamp & Van Trijp, 1991). In this study, to test for nomological validity, correlation analysis was conducted between the dimensions of trust and the 'sincerity' dimension from Aaker's (1997) brand personality scale.
Manifestation of character-competence, congruence and communication implies that financial services organisations (trustees) are sincere in carrying their business. From a logical point of view, this is also the case for trustors and the financial system in general. Thus, it is hypothesised that 'sincerity' is theoretically related to the trust scale. Sincerity (α = .932) was measured using 5 items adapted from Aaker (1997) brand personality scale. Correlation was performed between the trust scale sub-dimensions and theoretically related variable 'sincerity'. The correlation matrix (Table 6) establishes the

Discussion
In this study, we have sought to advance the operationalisation of trust in the context of financial services. Trust is particularly relevant to the financial services sector due to the complexity and intangibility attached to the products. Lack of trust can hurt the transactions of financial institutions and the performance of the global financial market. The role of trust has been portrayed well by Covey et al. (2012, pp. 13-14): The world's financial market's nearly collapsed last fall for one reason: lack of trust. Credit, the lifeblood of global economy, all but stopped flowing. Even big banks refused to lend to each other because they did not trust they would be repaid. We'd been taking trust for granted. Contacts back-up our deals, but who would sign them without trust in their counterparties. Trust is essential to building enduring connections with employees, suppliers, customers and the communities in which we do business. And it drives to risk-taking that leads to innovation and progress.
To date, the most noteworthy scale we have found for measuring trust in financial services is that developed by Ennew and Sekhon (2007) and further used by Ennew et al. (2011)   and Sekhon et al. (2014). However, this scale is limited in its comprehensiveness and omits aspects of trust attributable to trustors and context. The current student adds aspects of trust attributable to trustors and the situational context and follows a robust process to develop a new trust scale, which includes dimensions measuring the attributes of trustor and context. The study provides a comprehensive and inclusive approach to trust measurement in financial services. Our scale offers a composite and comprehensive measure of three essential aspects of trust. These are; the trustee's attributes or trustworthiness through character-competence, congruence, and communication; the trustor's attributes through commitment forms of trust identified by Dietz and Den Hartog (2006); and, institutional trust in financial services by measuring the structural assurance of the financial system (Mcknight et al., 1998;McKnight et al., 2002). Thus, we offer a comprehensive, integrative yet also tractable and practical scale for a composite measure of trust.
The trust scale developed through this research is parsimoniously represented in terms of a five-dimensional, 29-item measure. From a practical perspective, the 29-item, fivedimensional scale is sufficiently comprehensive to capture all the relevant aspects of trust and is relatively easy to administer. The 'five-C five-dimensional trust scale' (Table  7) has all the necessary psychometric properties, including unidimensionality, reliability and validity. Each dimension is theoretically consistent with the various studies on trust conducted previously.
The first dimension, 'character-competence', consists of two important attributes of the trustee: character and competence. Together, these two attributes provide the trustor with the necessary confidence to take the risk and trust their counterparts in the belief that their trust will not be abused or manipulated. Character provides confidence that the trustee has moral values to protect the interests of the trustor, deliver on promises and behave ethically; competence offers assurance that the trustee has the skills, knowledge and other abilities required to deliver on promises. The interdisciplinary trust literature emphasises the importance of honesty, integrity, goodness and morality as essential attributes of the trusteeall of which come under the umbrella of character. Character, in essence, is similar to goodness and morality (Giffin, 1967;Krackhardt & Stern, 1988;Ring & Van de Ven, 1994); and honesty (Blakeney, 1986;Gabarro 1978;Johnson-George & Swap, 1982;Rempel et al., 1985;Sato, 1988). Competence also has its roots in interdisciplinary literature. Mayer et al. (1995) called it 'ability', but Sirdeshmukh et al. (2002) and thereafter Coulter and Coulter (2003) settled on 'competence'. Competence has also been mentioned in the works of Anderson and Narus (1990), Baier (1986), Gabrro (1978, Kasperson et al. (1992), and Sitkin and Roth (1993). Barber (1983). Blakeney (1986) and Heimovics (1984) have also underscored the importance of competence but positioned it as 'expertness'. Competence' has even been found as a relevant dimension in a human-computer trust context (Gulati et al., 2019). Thus, character-competence comprises a critical dimension of our new trust scale, as trust depends on confidence in peoples' integrity and ability (Covey & Merrill, 2006).
The second dimension of the trust scale is congruence. According to a wide range of interdisciplinary trust literature, the trustor-trustee relationship is strongest when both parties share similar values (Morgan & Hunt, 1994). Sitkin and Roth (1993) reckoned an essential role for 'value congruence' in building trust. Congruence is similar to the 'shared values' mentioned by Ennew et al. (2011) and also close to benevolence (Dimoka, 2010). In the context of the financial services sector, it can be characterised as displaying values that are the same or similar between financial services institutions (trustees) and their customers (trustors). Congruence is particularly relevant to trustees, as it is the trustee who needs to demonstrate that he or she shares the same values as the trustor, which will encourage the trustor to place their trust in their counterpart. This notion indirectly links with concepts such as 'benevolence', 'caring' and 'concern' (Bonoma, 1976;Giffin, 1967;Heimovics, 1984;Holmes, 1991;Johnson-George & Swap, 1982;Kasperson et al., 1992;McGregor, 1967;Sato, 1988;Solomon, 1960). Congruence has strong roots in interdisciplinary literature, which further reinforces the multidisciplinary nature of this new trust scale. The third dimension is communication. Wetzel and Buckley (1988) have identified the passing of information between two parties as a critical element of human behaviour, and Anderson and Narus (1990) have argued that the flow of information is necessary for building relationships. Researchers have also identified communication as a critical antecedent for trust (Ennew et al., 2011). Clear communication creates transparency, one of the foundations of a trust relationship (Covey & Merrill, 2006). In financial services, communication is essential between institutions and consumers because most products are intangible and complex. Clear and transparent communication is not only required before but after sellingand indeed throughout the relationship lifecycle. Financial organisations are required to keep their customers updated about product features and risksall of which makes communication a substantial attribute of the trustee. In particular, financial advisers' communication style has been found to have a considerable influence on the formation of trust (Monti et al., 2014).
The fourth dimension of the trust scale is commitment. It is an attribute of the trustor rather than the trustee. Commitment is defined as trustors' willingness to take a risk and depend on others, that is, trustees (Doney et al., 1998;Luhmann, 2017;Mayer et al., 1995;McAllister, 1995;Scott, 1995). A commitment subscale has been developed to capture the intentional attributes of trustors (McKnight et al., 2002;McKnight & Chervany, 2001). Commitment is a crucial aspect of the trustor's decision to trust (Hurley, 2012) and reflects the trustor's trusting intent (Moody et al., 2017). Grounded in theories of interpersonal trust and the attributes of the trustor, commitment also manifests dispositional trust (the trustor's disposition to trust in particular cases). Whilst the first three dimensionscharactercompetence, congruence, and communicationreflect the world view of the first school of thought on trust, that of the social psychologists (Rempel et al., 1985), the fourth dimension, commitment, reflects the second and third schools of thought, drawing on the world view of the psychologists (Rotter, 1971).
The fifth dimension of the scale is contextan impersonal, situational attribute referring to trust in a system (Luhmann, 2017) or structure for example, the financial system. Trust scholars have acknowledged that the interactions between trustees and trustors take contextual factors into account (Zheng et al., 2017). McKnight et al. (1998) and McKnight and Chervany (2001) have popularised the concept of institutional trust, emphasising the need for impersonal structures to be in place to enable a person to anticipate a successful future outcome. McKnight and Choudhury (2006) have also stressed the importance of a robust structural assurance. In the setting of the financial services sector, when concerned with context the we focus on the robustness of the structure in place within an institutional system, also known as institutional trust. This has been found to be a highly important factor in influencing trust overall (Cheung & Lee, 2001;Doney & Cannon, 1997;Hosmer, 1995;Lee & Turban, 2001;Lewis & Weigert, 1985;Naquin & Paulson, 2003;Ratnasingam & Pavlou, 2003;Rousseau et al., 1998;Tan & Theon, 2001;Tan & Sutherland, 2004;Yoon, 2002). More recently, developments in the area of Fintech has empowered the financial institutions and government to use technologies and innovation in improving the overall safeguarding of the financial system (Nathan & Jacobs, 2020). A number of macro and micro factors provide further explanation for the context dimension. The macro factors are the role and power of regulators and the ombudsman in enforcing regulations to control the behaviour of financial organisations. Micro factors include the internal systems in place within financial services organisations to ensure they are treating customers fairly and dealing with complaints efficiently. Overall, this dimension reflects the fourth school of thought, built on the world view of the sociologists (McKnight et al., 2002).

Managerial and theoretical implications
A comprehensive interdisciplinary and composite trust scale for financial services, with strong reliability and validity, holds important managerial implications, its ability to capture the attributes of the trustee, trustor and financial system attesting to its suitability as a diagnostic tool to measure trust more robustly. Using consumer trust as an indicator, the scale also provides a means to benchmark various services offered to customers in the sector.
Our trust scale has significant practical implications, offering valuable insights for policymakers, commercial organisations and other stakeholders. It enables financial services organisations to measure and track trust in a more comprehensive and sophisticated, yet tractable, manner than was hitherto possible. Policymakers and firms will not only be able to monitor changes in perceptions of trust more accurately and regularly but will also gain an insight into the level of change in each of the five trust dimensionscharacter-competence, congruence, communication, commitment and contextusing the five-C trust scale.
Firms will thus benefit from better understanding of the reasons for changes in overall perceptions of trust and how a change in the attributes of the trustee, trustor, and situation can impact consumers' overall perception of trust. This will, in turn, help stakeholders develop a more effective strategy to restore or enhance consumer perceptions of trust, providing policymakers and firms with guidance as to where to focus their efforts.
Finally, the scale can also be used successfully in other contexts-services in generalthrough minor adaptation of the items in the context subscale. This will provide a more nuanced understanding of what constitutes consumers' overall perception of trust and the tools to measure it and be of interest to researchers and practitioners. Minor adaptation of the proposed scale will allow for its broader application across sectors where the perceived risk is high.
Our approach to trust literature focusing on reconciliation of conceptualisation and operationalisation of trust represents an addition to other research interested in the convergent theories of trust, given that due to disciplinary and epistemological differences the majority of studies adopt a divergent perspective. However, our novel scale, the 5CTS, represents a contribution which considers the critical characteristics of financial services' context, the complexity and features of financial products and the trusting relationships that take into account the measurement of both calculus-based/ knowledge-based and relational-based/ identification-based trust, as well as propensity to trust. Ultimately, we offer a scale which speaks to the convergent theories of trust in an integrative manner and which is a tractable yet comprehensive measure.

Limitations and directions for future research
The study has some limitations, which offers opportunities for future research. First, all respondents come from one country, British nationals, albeit with varying ethnicity. The UK is a developed country with a robust financial system, which will influence the notion of institutional trust. Therefore, we recommend further studies to validate the scale in another context (for example, developed, developing and under-developed countries with different financial systems).
Another issue relates to the demographic background of respondents in this study. Males (68%) are over-represented, and respondents are relatively young (median age range 18-35). While financial services organisations may be particularly interested in knowing the state of trust, or trust-related behaviour of millennials and generation Z, future research should consider stratified probability sampling to increase the generality of the findings. As for the future direction of research, the scale could also be used as part of a larger and broader study to examine the antecedents and trust outcomes in various contexts.

Disclosure statement
No potential conflict of interest was reported by the authors.