Do you want my loyalty? Then understand what drives my trust – a conventional and Islamic banking perspective

Abstract

Purpose

The purpose of this research is to develop an enhanced understanding of the drivers of trust and loyalty in a conventional and Islamic banking setting.

Design/methodology/approach

The study’s sample included South African retail bank customers who had Islamic or conventional products and who were 18 years or older. A field services company collected data from respondents through the distribution of self-administered questionnaires and a total of 949 questionnaires were deemed suitable for data analysis. SmartPLS 3.2.7 and Hayes Process Macro for SPSS tested the study’s hypotheses.

Findings

Comparing conventional banking customers with Islamic banking customers, the path from trust to customer loyalty was statistically significantly different across customer type, while the paths between trust and customer orientation, information sharing, and service fairness were not statistically significantly different across customer type. A closer examination of the path coefficients reveals that the relationship between trust and loyalty is stronger for conventional banking customers than for Islamic banking customers.

Practical implications

The findings of the study guide both conventional and Islamic banks in South Africa on how banks should redesign their purpose as the providers of financial resources to their customer segments. It highlights the need for these banks to secure a more focused approach on how to deliver financial resources and consulting services to customers in a trusting, engaging and reliable manner.

Originality/value

The study provides insight into Islamic and retail bank customers’ perceptions of the drivers of trust and loyalty and how these constructs’ interrelationships differ between Islamic and conventional banking customers.

Citation

Roberts-Lombard, M. and Petzer, D.J. (2024), "Do you want my loyalty? Then understand what drives my trust – a conventional and Islamic banking perspective", International Journal of Islamic and Middle Eastern Finance and Management, Vol. 17 No. 3, pp. 532-551. https://doi.org/10.1108/IMEFM-10-2023-0412

1. Introduction

In a competitive banking environment, it becomes increasingly difficult for financial service providers, such as banks, to meet customers’ trust expectations. Especially considering that banks are increasingly unable to deliver on customers’ service engagement expectations in a trustworthy manner (Bizcommunity, 2022).

As such, scholars (Khan et al., 2023; Roberts-Lombard and Petzer, 2021) have indicated the need to explore trust as a relationship-building tool in numerous banking contexts. Therefore, this study aims to:

• Investigate the impact of selected trust antecedents in a conventional and Islamic banking context;

• Obtain an in-depth perspective on the mediating role of trust as a precursor to loyalty in a conventional and Islamic banking context;

• Develop an enhanced understanding of the moderation of bank charges and duration of support on the relationships between information sharing, service fairness, customer orientation, and trust; and

• Better understand the moderating role of customer type on the trust-loyalty relationship.

Considering the evolutionary needs of banking customers in multiple contexts, marketing scholars (e.g. Albaity and Rahman, 2021; Banahene et al., 2018) argue that continuous research on the factors that impact trust and ultimately the future loyalty intentions of customers is required. Nkwede et al. (2022) further argued the need to explore customer trust more in an African banking context in emerging markets. Against the background provided above, the following research questions were formulated:

RQ1.

Do selected trust antecedents (namely service fairness, customer orientation and information sharing), impact customers’ trust perceptions, do trust mediate the proposed relationships between its antecedents and loyalty, and does trust influence loyalty in conventional and Islamic banking in an emerging market environment?

RQ2.

Do bank charges and duration of support moderate the relationships between service fairness, customer orientation, and information sharing, and trust in the conventional and Islamic banking environment of an emerging market?

RQ3.

Does customer type moderate the trust-loyalty relationship of conventional and Islamic banking customers in an emergent market?

Considering the discussion above, limited research has been conducted in emerging markets that investigates the strengthening of customer loyalty through trust and selected drivers in both a conventional and Islamic banking context from a stimulus-organism-response (S-O-R) perspective. Previous research by Albaity and Rahman (2021) and van Esterik-Plasmeijer and van Raaij (2017) analysed trust as a mediating variable from a conventional or an Islamic banking perspective only. In addition, Izogo et al. (2017) conducted a study in Nigeria to explore the mediating role of trust as an organism, considering various stimuli and responses (loyalty) from a conventional banking perspective only. The limited focus of retail banking studies to explore relational banking from an S-O-R perspective in both conventional and Islamic banking secures six novel contributions made by this study:

1. First, this study frames the drivers of trust, and ultimately loyalty, from an S-O-R perspective, where selected stimulants (i.e., customer orientation, information sharing, and service fairness) are positioned as stimuli to trust as the organism, ultimately influencing loyalty as a response;

2. Second, the study identifies and investigates unique stimuli (i.e., customer orientation, information sharing, and service fairness) that could influence trust within an S-O-R context from a conventional and an Islamic banking perspective;

3. Third, the mediating effect of trust is researched from a multiple banking context perspective (both conventional and Islamic banking) to establish its influence as a mediator in a business-to-consumer (B2C) setting;

4. Fourth, the study analyses the moderating effect of bank charges and duration of support on the relationships between an organism (i.e., trust) and selected stimuli (i.e., customer orientation, information sharing, and service fairness);

5. Fifth, this research looks at the moderating effect of customer type on the relationship between an organism (i.e., trust) and a response (i.e., loyalty); and

6. Six, the study focuses on an emergent market in Africa, namely South Africa, where limited studies on conventional and Islamic banking segments have been conducted reflecting the trust stimuli and future loyalty intentions of banking customers in the dualistic banking industry, which have increased in a competitive nature over the past decade.

2. The stimulus-organism-response framework

S-O-R theory comprises three components – namely the stimulus, the organism, and the response – that are integrated and systematic in nature (Arora et al., 2020). The role of these components is to provide clarity on how different stimuli in an environment (stimulus) arise thoughtful (cognitive) or emotional (affective) outcomes in people (organism), which could result in specific behavioural responses (Kühn and Petzer, 2018). Therefore, this study applies the S-O-R framework to develop a deeper understanding of the impact of service fairness, customer orientation, and information sharing (stimulus) on trust (organism) and thus on loyalty (response) amongst South African banking customers in an emergent African market.

3. Theoretical model development

3.1 Validating the proposed direct effects

Shamsudheen and Rosly (2021) argue that when a bank has its customers’ best interests at heart, the customers’ trust level in the bank is positively impacted. Therefore, when bank employees engage with customers through professional interactive communication that is founded on friendly service, sound product and service knowledge as well as respect, the overall trust level that customers have in their bank is strengthened (Windarti et al., 2020). Consequently, the following hypothesis is proposed:

H1.

Customer orientation positively influences trust.

Through information sharing, service providers reflect their commitment to transparency in their business practices, thereby strengthening customer trust in the banking brand (Tiwari, 2022). Moreover, when customers receive information on products and services, they are better informed to make value-adding decisions, which could positively impact their trust levels in the bank (van Esterik-Plasmeijer and van Raaij, 2017). Therefore, it is hypothesised that:

H2.

Information sharing positively influences trust.

When a bank provides customers with products developed according to customer expectations and service experiences founded on the principles of fairness, reliability, and integrity, customers’ overall level of trust is strengthened (van Esterik-Plasmeijer and van Raaij, 2017). Therefore, when bank customers are of the opinion that they have been treated with respect by their bank and have received service speedily, they perceive such an experience as fair, which positively impacts their level of trust in their bank (Gokmenoglu and Amir, 2021; Ng et al., 2015). Hence, it is hypothesised that:

H3.

Service fairness (interactional and procedural) positively influences trust.

Trust is imperative when developing a long-term relational approach towards customers (Gokmenoglu and Amir, 2021; Sharif et al., 2023). Khan et al. (2023) agreed, asserting that trust in a bank is strengthened when the bank is concerned with its customers’ needs and engages with them in an honest and reliable manner. Through such an approach, future loyalty intentions are enhanced.Thus, it is hypothesised that:

H4.

Trust positively influences customer loyalty.

3.2 The mediating role of trust

When customers perceive their engagement experience with their bank as being positive, informative, and value-adding, they trust their bank more, thereby strengthening future loyalty intentions (Geçit and Taşkin, 2020). A study by Hussein et al. (2023) explored trust as a mediator between banking experience quality and loyalty in the banking sector of Indonesia. Moreover, a different study conducted in the South African retail banking industry investigated trust as a mediating variable in the relationships between brand relationship quality, customer advocacy, and brand loyalty (Quaye et al., 2022). Accordingly, it is hypothesised that:

H5.

The relationship between customer orientation and loyalty is mediated by trust.

When service providers (e.g. banks) keep their customers informed of information on new product and service developments, trust is strengthened, which could enhance customers’ future loyalty (Haron et al., 2020). Therefore, when a bank provides sound explanations whenever a customer is uncertain, the latter perceives the service engagement as being trustworthy, which stimulates the future loyalty intent (Alam et al., 2021). Considering this, it is hypothesised that:

H6.

The relationship between information sharing and loyalty is mediated by trust.

Hidayat and Idrus (2023) state that when customers perceive service engagement as being fair and supportive of their needs and expectations, trust in their bank is stimulated. As such, customers’ future intent to remain in a relationship with their bank is strengthened (Tran Xuan et al., 2023). Consequently, Farooq and Moon (2020) and Kwong et al. (2023) confirmed the mediating role of trust in a B2C environment in multiple settings. From the above, it is hypothesised that:

H7.

The relationship between service fairness (interactional and procedural) and loyalty is mediated by trust.

3.3 The moderating role of bank charges

When customers believe the pricing strategy of their bank in terms of bank charges/service fees/costs is reasonable for the services received, they also develop a deeper sense of trust in their bank (Jain and Lamba, 2020). Furthermore, Kaur and Arora (2023) contend that banking costs, perceived by customers as a price to pay for the service rendered, impact retail bank perceptions. As such, bank charges can impact customers’ future trust perceptions and intent to be committed to in a relationship with their bank. Considering this, it is hypothesised that:

H8a.

The relationship between customer orientation and trust is moderated by bank charges.

Scholars (Gladstone et al., 2022) argue that bank customers want to be regularly informed of new bank charges or changes to the pricing structure of their banking product to stay up to date on the cost implications to their banking package. Hence, it is important to note that the information banks share with their customers should reflect transparency, which can enhance customers’ overall trust level in the service provider (e.g. bank) (Losada-Otálora and Alkire, 2019). Thus, it is hypothesised that:

H8b.

Bank charges moderate the relationship between information sharing and trust.

Customers expect their bank to treat them with courtesy and address their queries in a professional and timely manner, which could stimulate their future trust in their bank (Gokmenoglu and Amir, 2021). In addition, Ahmad et al. (2024) have confirmed that bank-related charges impact customers’ perceptions of service experiences, which could guide their future willingness to be associated with their bank in multiple banking contexts. As such, it is hypothesised that:

H8c.

The relationship between service fairness and trust is moderated by bank charges.

3.4 The moderating role of duration of support

Research studies by Jadhav et al. (2023) and Menidjel and Bilgihan (2023) emphasise the critical importance of professionalism between stakeholders as a foundational element in driving future relational support. Furthermore, Dorai et al. (2021) concurred and established that the duration of a relationship functions as a moderator, which positively impacts customers’ willingness to associate with the business brand in the future. As such, it is hypothesised that:

H9a.

Duration of support moderates the relationship between customer orientation and trust.

The nature of the information that service providers communicate to customers and the duration of such information sharing influence customers’ preparedness to trust and connect with their provider (e.g. a bank) (Kosiba et al., 2020). Pan and Ha (2021) agreed that the length of the relational support provided through continuous engagement positively moderates customers’ willingness to be associated with a supplier in the future. Accordingly, the following hypothesis is proposed:

H9b.

The relationship between information sharing and trust is moderated by duration of support.

Itani et al. (2020) confirmed that in a service-orientated environment, the duration of time with which customers engage with providers, can influence the customer’s perceptions of service support received. As such, customers’ overall level of trust in their service provider is also influenced. Conclusively, scholars (e.g. Chi and Chen, 2019; Roberts-Lombard and Petzer, 2021) confirm customers’ willingness to strengthen future relational intent through durational support, especially if the latter is founded on a positive experience that stimulates future relational intention. Therefore, the following hypothesis is proposed:

H9c.

The relationship between service fairness and trust is moderated by duration of support.

3.5 The moderating role of customer type

Mahakunajirakul (2022) and Schirmer et al. (2018) opined that different customer types have varying perceptions of trust in service providers (e.g. bank), which impact their future loyalty intentions differently. Moreover, scholars (e.g. Banik and Rabbanee, 2023; Hirvonen et al., 2013) have validated the moderating role of customer type between multiple stimuli and a response outcome in a banking and small and medium-sized business environment. Consequently, customer type functions as a moderating variable on the positive relationships between service fairness, information sharing, customer orientation and trust. Hence, it is hypothesised that:

H10.

Customer type moderates the relationship between trust and loyalty.

South Africa’s traditional retail banking cohort encompasses customers who have bank accounts with conventional banks in South Africa. These customers access the banking system through mainstream banking brands in the country (The Banking Association South Africa, 2022). South African Islamic banking customers embrace sharia banking principles and access Islamic banking through conventional banks in South Africa offering Islamic banking windows or international Islamic banking brands operating in the country (The Banking Association South Africa, 2021). Since conventional and Islamic banking customers reflect two different banking customer types, they differ in their perceptions of trust, its antecedents, and outcome. As such, the following hypothesis is proposed:

H11.

The relationships depicted between the conceptual model’s constructs do not display equal regression weights for the different customer types.

Against the background provided above, the following Figure 1 is proposed.

4. Research methodology

4.1 Population and sample

The study’s population included South African retail bank customers who had Islamic or conventional banking products with South African banks, resided in the Gauteng province of South Africa, and were 18 years or older. A purposive non-probability sampling technique was applied to the study to ensure that the appropriate respondents (i.e. those holding a bank account with a South African bank) were included in the sample (Amin, 2021). A field services company collected data from eligible respondents using a purposive sampling procedure. Data were obtained from 949 respondents – 599 respondents with conventional banking products and 350 respondents with Islamic banking products. Considering the low incidence rate of Islamic banking customers, these respondents are much more difficult to find than conventional banking customers. Furthermore, it is more challenging to access this market segment to secure large numbers of responses. Yet, considering that the study incorporates 350 Islamic banking responses, it still reflects a significant number of Islamic banking customers, who can be deemed representative of Islamic banking customers in South Africa. The authors did not want to manipulate the data to create the impression that equal numbers of Islamic and conventional banking customers were secured. The difference in the group sizes reflected the reality with respect to the number of Islamic versus conventional banking customers in South Africa. The sample size of both groups was more than adequate for the statistical techniques used in this study.

4.2 Data collection and questionnaire design

Ten fieldworkers across the Gauteng province of South Africa gathered data from respondents. Fieldworkers had to identify individuals within their own network of friends, family, and colleagues who adhered to the stipulated criteria referred to above. The fieldworkers selected the respondents purposefully through their networks of friends, family, classmates, churches, mosques, neighbours, communities and colleagues who qualified to partake in the study. In addition, respondents were approached in public places, such as malls, parking areas, and social events. Respondents had to complete the self-administered paper-based questionnaires and return these to the fieldworkers. To increase the sample size, online surveys were used by sending a link to the survey to fieldworkers’ networks (e.g. friends, family, co-workers). Paper-based surveys were collected by fieldworkers handing out the surveys and collecting it within their networks or having respondents answer the surveys on the spot and collecting it at the same time. Fieldworkers checked that all questionnaires were correctly completed. Only fully completed questionnaires were used for coding the information on an Excel sheet. The collection of data was secured over an eight-week period.

The field services company had access to the respondents and has been involved with the researchers on numerous projects to collect data. The questionnaire included a consent statement followed by a screening question to ensure respondents qualified to partake in the study. The questionnaire comprised sections measuring respondents’ demographics, bank patronage habits, and perceptions of the study’s constructs. The constructs were evaluated using a seven-point unlabelled Likert-type scale, where 1 equalled “strongly disagree” and 7 equalled “strongly agree”. Table 1 provides insight into respondents’ overall demographic and bank patronage behaviour and for those with Islamic and conventional banking products. The items for the different constructs were adapted from various sources as follows, namely customer orientation (Cheng et al., 2008; Wray et al., 1994), information sharing (Chu and Wang, 2012; Ndubisi, 2007; Wong et al., 2007), service fairness (Giovanis et al., 2015), trust (Fornell and Larcker, 1981; Giovanis et al., 2015; McKnight et al., 1998) and loyalty (Mandhachitara and Poolthong, 2011) . Data collection was secured over a ten-week period in 2022. After the data was collected, it was coded and captured into SPSS.

4.3 Analysing the data

The evaluation of the measurement models (overall model, Islamic banking customers model, and conventional banking customers model) was undertaken using SmartPLS 3.2.7, which is considered suitable in instances where the data are not normally distributed. The three measurement models were evaluated for reliability, construct validity, and multicollinearity. In assessing the three structural models (H1 to H4), collinearity, structural model path coefficients, effect sizes (f²), coefficients of determination (R² value), and predictive relevance (Q²) were considered with the aid of SmartPLS 3.2.7. A mediation analysis with Smart PLS 3.2.7 with bias-corrected bootstrapping (n = 5000) was conducted to assess the intervening capabilities of trust as the “organism” in this S-O-R model (H5 to H7). A moderation analysis was undertaken using Hayes Process Macro for SPSS (Model 1) for H8 to H10 (Hayes, 2013) and ordinary least squares 95% confidence intervals were generated to determine whether interaction effects were significant or not. Bootstrapping resampling (n = 5000) was conducted and the bias-corrected confidence intervals were interpreted for significance of interaction effects. Using SmartPLS 3.2.7, a multi-group analysis assessed whether the conceptual model’s constructs displayed equal regression weights for the different types of banking customers (H11).

5. Results

5.1 Measurement model assessment

Table 2 shows the assessment of composite reliability (CR) and internal consistency as well as convergent validity of the three measurement models. As per Hair et al. (2014), the CR and Cronbach’s alpha values for each construct have a 0.7 cut-off. Furthermore, for convergent validity, the outer loadings for each item should be at least 0.6 and statistically significant (p-value < 0.05). The average variance extracted (AVE) for each of the study’s constructs ought to be higher than 0.5.

Table 3 provides insight into the assessment of discriminant validity. To assess discriminant validity, the researchers:

• applied the Fornell and Larcker (1981) criterion to determine whether AVE’s square root for each of the study’s construct was higher than its correlation with the other constructs in the study;

• considered cross-loadings by assessing whether the items measuring the study’s constructs exhibited higher loadings on the construct they measured than on other constructs and that the items did not have differences in loadings of less than 0.1 between the parent construct and the other model constructs; and

• evaluated the heterotrait-monotrait (HTMT) ratio to ensure the correlations between the construct pairs fell below 0.85 (Henseler et al., 2015; Sarstedt et al., 2021).

The variance inflation factor (VIF) values were also considered to assess the possible impact of multicollinearity, where a value of 5 and more signifies possible issues with collinearity (Hair et al., 2011). From the results presented in Table 3 there is adequate evidence of reliability (CR and internal consistency), construct validity (convergent and discriminant) and the absence of multicollinearity. Subsequently, the study’s main effects were tested.

5.2 Structural model assessment

For each model, the VIF values for the predictors, namely service fairness, information sharing, customer orientation, and trust, were smaller than 5 (Hair et al., 2011), meaning there were no collinearity issues. Table 4 provides the structural path coefficients. For the overall model, all paths were statistically significant, with the p-values < 0.05, t-values > 1.96, and each confidence interval not containing a zero. Service fairness (0.439; p-value ≤ 0.001) was a stronger predictor of trust than customer orientation (0.310; p-value ≤ 0.001) or information sharing (0.199; p-value ≤ 0.001). Trust was a strong predictor of loyalty (0.773; p-value ≤ 0.001). For the conventional banking customers model, all paths were statistically significant, with the p-values < 0.05, t-values > 1.96, and each confidence interval not containing a zero. Service fairness (0.430; p-value ≤ 0.001) was a stronger predictor of trust than customer orientation (0.310; p-value ≤ 0.001) and information sharing (0.218; p-value ≤ 0.001). Trust was a strong predictor of loyalty (0.797; p-value ≤ 0.001). For the Islamic banking customers model, all paths were statistically significant, with the p-values < 0.05, t-values > 1.96, and each confidence interval not containing a zero, except for the structural path between information sharing and trust (0.134; p-value = 0.047). Service fairness (0.457; p-value ≤ 0.001) was a stronger predictor of trust than customer orientation (0.339; p-value ≤ 0.001), and trust was a strong predictor of loyalty (0.719; p-value ≤ 0.001).

In the overall model, moderate variance was explained in respondents’ loyalty towards their bank (R² = 0.597) and substantial variance was explained in respondents’ trust towards their bank (R² = 0.769). Information sharing’s ability to predict trust was very small (f² = 0.051). The ability of customer orientation (f² = 0.123) and service fairness (f² = 0.300) to predict trust was medium. In contrast, trust’s ability to predict loyalty was large (f² = 1.483). Regarding the predictive relevance of the model or the model’s ability to correctly predict data not used during the estimation of the model, the Q² values were greater than zero, suggesting the model possessed predictive relevance for both trust (Q² = 0.643) and loyalty (Q² = 0.484).

In the conventional banking customers model, moderate variance was explained in all respondents’ loyalty towards their bank (R² = 0.636) and substantial variance was explained in respondents’ trust towards their bank (R² = 0.782). Information sharing’s ability to predict trust was very small (f² = 0.069), while the ability of customer orientation (f² = 0.144) and service fairness (f² = 0.286) to predict trust was medium. Contrastingly, trust’s ability to predict loyalty was large (f² = 1.746). Regarding the model’s predictive relevance or ability to correctly predict data not used during the estimation of the model, the Q² values were greater than zero, suggesting the model possessed predictive relevance for trust (Q² = 0.639) and loyalty (Q² = 0.530).

In the Islamic banking customers model, moderate variance was explained in all respondents’ loyalty (R² = 0.513) and trust (R² = 0.746) towards their bank. The ability of customer orientation (f² = 0.096) to predict trust was small, while with service fairness (f² = 0.330) it was medium. In contrast, trust’s ability to predict loyalty was large (f² = 1.051). Concerning the model’s predictive relevance or ability to correctly predict data that were not used during the estimation of the model, the Q² values were greater than 0, which suggests that the model possessed predictive relevance for trust (Q² = 0.638) and loyalty (Q² = 0.389).

H1, H2, and H4 were supported for the overall model, the conventional banking customers model, and the Islamic banking customers model. H3 was supported for the overall model and the conventional banking customers model, but not for the Islamic banking customers model.

5.3 Mediation analysis

Overall, trust was a partial mediator of loyalty’s relationships with customer orientation (0.240; 95% BCCI [0.171; 0.316]), information sharing (0.153; 95% BCCI [0.087; 0.218]), and service fairness (0.340; 95% BCCI 0.284; 0.396]). As for the conventional banking customers model, trust partially mediated loyalty’s relationships with customer orientation (0.247; 95% BCCI [0.160; 0.342]), information sharing (0.174; 95% BCCI [0.090; 0.251]), and service fairness (0.343; 95% BCCI [0.267; 0.415]). For the Islamic banking customers model, trust partially mediated loyalty’s relationships with customer orientation (0.243; 95% BCCI [0.137; 0.359]) and service fairness (0.326; 95% BCCI [0.245; 0.412]). However, trust did not mediate the relationship between information sharing (0.097; 95% BCCI [−0.003; 0.194]) and loyalty. Therefore, H5 and H7 were supported for all three models, but H6 could only be supported for the overall and conventional banking customers models.

5.4 Moderation analysis

Duration of support (H8a-c) and bank charges (H9a-c) were hypothesised as moderators between trust and customer orientation, service fairness, and information sharing for the Islamic and conventional banking customer models. Duration of support was divided into two groups: five years or less and more than five years. Bank charges were divided into: R100 or less; R101-R200; and R201 or more. In addition, customer type (Islamic and conventional banking customers) was hypothesised as a moderator between trust and loyalty (H10). Only statistically significant moderation effects were reported and the conditional effects were probed further.

5.4.1 Bank charges – conventional banking customers model.

The interaction effect of customer orientation and bank charges of R201 or more versus R100 or less (Int_2) was statistically significant at 0.145 (95% BCCI [0.032; 0.259]). As bank charges increased, the impact of customer orientation on trust increased in the conventional banking customers model.

The interaction effect between information sharing and bank charges of R201 or more versus R100 or less (Int_2) was statistically significant at 0.198 (95% BCCI [0.084; 0.312]). As bank charges increased, the effect of information sharing on trust increased in the conventional banking customers model.

The first interaction effect between service fairness and bank charges of R101-R200 versus R100 or less (Int_1) was statistically significant at 0.144 (95% BCCI [0.020; 0.268]). Furthermore, the second interaction effect of service fairness and bank charges of R201 or more versus R100 or less (Int_2) was statistically significant at 0.123 (95% BCCI [0.007; 0.238]). As bank charges increased from R100 or less to R101-R200, the effect of fairness on trust increased. However, this effect decreased as bank charges increased from R101-R200 to R201 or more.

Concerning bank charges moderating trust’s relationships with service fairness, information sharing, and customer orientation, no significant moderation effects could be found with the Islamic banking customers model. Therefore, H8a-c were supported for the conventional banking customers model, since bank charges moderated trust’s relationships with service fairness, information sharing, and customer orientation, but the hypotheses were not supported for the Islamic customers model.

5.4.2 Duration of support – conventional banking customers model.

The customer orientation by duration of support interaction was statistically significant at 0.156 (95% BCCI [0.019; 0.293]). When duration of support was low (−1SD or 1), customer orientation’s effect on trust was 0.789 (95% BCCI [0.739; 0.839]). When duration of support was medium (at the mean or 1.088), customer orientation’s effect on trust was 0.803 (95% BCCI [0.755; 0.850]). When duration of support was high (+1SD or 1.390), customer orientation’s effect on trust was 0.850 (95% BCCI [0.790; 0.909]). In short, customer orientation’s effect on trust increased as duration of support increased.

It is evident that the information sharing by duration of support interaction was statistically significant at 0.269 (95% BCCI [0.123; 0.414]. When duration of support was low (−1SD or 1), information sharing’s effect on trust was 0.721 (95% BCCI [0.672; 0.770]). When duration of support was medium (at the mean or 1.088), information sharing’s effect on trust was 0.745 (95% BCCI [0.698; 0.791]). When duration of support was high (+1SD or 1.390), information sharing’s effect on trust was 0.826 (95% BCCI [0.763; 0.889]). In short, information sharing’s effect on trust increased as duration of support increased.

Regarding the duration of support moderating the relationships between trust and service fairness, information sharing, and customer orientation, no significant moderation effects could be found in terms of the Islamic banking customers model. H9a and H9b were supported for the conventional banking customers model, since duration of support moderated the relationships between trust and customer orientation and information sharing, but the hypotheses were not be supported for the Islamic customers model. H9c was not supported, since duration of support did not moderate the relationship between service fairness and trust in any of the proposed relationships in either model.

5.4.3 Customer type – conventional banking customers and Islamic banking customers.

Trust by customer type interaction was statistically significant at −0.129 (95% BCCI [−0.233; 0.025]). The conditional effect of trust on loyalty was stronger for conventional banking customers at 0.764 (95% BCCI [0.699; 0.829]) than for Islamic banking customers at 0.635 (95% BCCI [0.554; 0.716])). From these findings, customer type moderated the relationship between trust and loyalty, thus supporting H10.

5.5 Multi-group analysis

When comparing conventional banking customers with Islamic banking customers, the path from trust to customer loyalty was statistically significantly different across customer type (p-value = 0.028), while the paths between trust and customer orientation (p-value = 0.766), information sharing (p-value = 0.312), and service fairness (p-value = 0.720) were not statistically significantly different across customer type. Closer examination of the path coefficients in Table 4 reveals that the relationship between trust and loyalty is stronger for conventional banking customers (β = 0.797) than for Islamic banking customers (β = 0.716). Therefore, the relationships depicted between the conceptual model’s constructs do not display equal regression weights for the different customer types, and H11 was supported.

In summary, concerning the direct effects of the study, customer orientation positively influenced trust, information sharing positively influenced trust, and trust positively influenced customer loyalty for all three models, namely the overall model, the conventional banking customers model, and the Islamic banking customers model. In addition, service fairness (interactional and procedural) was found to positively influence trust for the overall model and the conventional banking customers model, but not for the Islamic banking customers model.

Regarding mediation, the relationship between customer orientation and loyalty was mediated by trust and the relationship between service fairness (interactional and procedural) and loyalty was mediated by trust in all three models. The relationship between information sharing and loyalty was mediated by trust in the overall and conventional banking customers models.

In relation to moderation, bank charges moderated trust’s relationships with service fairness, information sharing, and customer orientation for the conventional banking customers model, but not for the Islamic customers model. Customer type moderated the relationship between trust and loyalty for both models. Finally, it was found that the relationships depicted between the conceptual model’s constructs did not display equal regression weights for the different customer types.

6. Discussion of results

Before this study was conducted, little attention was given to trust as an intervening variable from a conventional and Islamic banking perspective combined. Previous research has not extensively evaluated how multiple stimuli impact trust and customer loyalty from an S-O-R perspective, as this study argues (Islam et al., 2020; Jain et al., 2023). Therefore, this study’s results confirm the need for marketing literature to expand the exploration of trust antecedents and their impact on trust in a conventional and Islamic banking perspective, where both these banking types are imperative to the future economic survival of South Africa’s banking industry (Albaity and Rahman, 2021; Hussein et al., 2023; Khan et al., 2023). The results also secure a deeper understanding of the moderating role of bank charges and duration of support on the relationships between the stimuli of service fairness, information sharing, and customer orientation and trust as an organism in terms of conventional and Islamic banking. These findings add value to the relationship marketing field by confirming its relevance to the proposed relationships in an emerging market context, such as South Africa. Regarding customer type, the results show the conditional effect of trust on loyalty is stronger for conventional banking customers than for Islamic banking customers. As such, the study confirmed that customer type moderates the trust-loyalty relationship in both a conventional and Islamic banking context. This finding is significant, since it confirms a moderating influence in the relationship between trust and loyalty that has not been actively confirmed in a multi-bank and emerging market context in an emerging market.

7. Managerial implications

The study purports numerous recommendations for banks to consider. For example, banks need to understand that bank costs are an important factor that impacts conventional bank customers’ trust perceptions. However, this is not the case for Islamic banking customers, who illustrate a greater need for an engaging banking experience founded on trustworthy, friendly, and supportive banking practices, and reflects the values and principles of sharia banking. Therefore, Islamic banks need to focus less on managing their customer segments based on price, and more on the experiential benefit that customers receive when engaging with their bank. In terms of duration of support, conventional banks will only be able to retain their conventional customer base if they provide customers with professional service engagement founded on the principles of knowledge sharing, friendliness, care, and understanding. There should be a clear consultative approach towards customer management that could be secured in-house through consultative sessions with banking customers or online using virtual portals where artificial intelligence (AI) intervention, such as virtual consultants, can provide customers with standard answers to sophisticated financial queries or utilising chatbots to secure 24 / 7 interaction on general banking enquiries.

Concerning customer type, conventional and Islamic banking customers are impacted by trust to stimulate their future loyalty intentions. Therefore, South Africa’s banking sector needs to become more transparent through its banking practices by reporting to customers regularly – that is, through SMSes, social media engagement, or in-branch AI virtual communications – on bank policy changes, cost increases, service delivery transformation, and new strategies to be implemented to enhance the product and service experience. The ability to operate ethically, reliably, and transparently is increasingly becoming a differentiator in South Africa’s competitive business landscape, an emerging market.

8. Conclusion, limitations and areas for future research

The study is grounded within the S-O-R theory and makes a contribution to the marketing literature through a deeper understanding of consumer trust and loyalty perceptions when considering dual banking systems (namely conventional and Islamic banking). The study’s results show that the selected trust stimuli have a separate impact on the trust perceptions of conventional and Islamic banking customers. The key difference in the findings relates to the non-significant impact of information sharing on trust. In addition, the importance of trust in the development of long-term relationships with customers in a B2C context, was validated. It was confirmed that in an emergent market context, trust is critically important in driving future loyalty intention.

In terms of the study’s limitations, a limited number of antecedents to trust are included. Future studies could focus on trust precursors or use behavioural intention or purchase intention as an outcome. This study explored conventional and Islamic banking customers from a single setting perspective, since it was applied to one emerging market. Moreover, respondents were selected through the application of non-probability sampling to the study, thus limiting the generalisation of the results. Future studies could consider a comparative assessment of conventional and Islamic banking customers in an emergent and established market, between emergent markets, a different contextual setting for application or simply using a probability sampling technique in the study.

Figures

Figure 1.

Proposed conceptual model

Table 1.

Respondents’ demographics profile and bank patronage behaviour

Variables	Overall sample		Non-Islamic banking customers		Islamic banking customers
	N 949	100 (%)	N 599	63.12 (%)	N 350	36.88 (%)
Gender
Female	495	52.2	319	53.3	176	50.3
Male	454	47.8	280	46.7	174	49.7
Highest education level
Some schooling	165	17.4	100	16.7	65	18.6
High school completed	361	38.0	243	40.6	118	33.7
Post-school qualification	423	44.6	256	42.7	167	47.7
Employment status
Self-employed	217	22.9	105	17.5	112	32.0
Full-time employed	369	38.9	229	38.2	140	40.0
Part-time employed	90	9.5	62	10.4	28	8.0
Other (housewife, house husband, student, retired or unemployed)	273	28.7	203	33.9	70	20.0
Duration of support
Five years or less	844	88.9	548	91.5	296	84.6
More than five years	105	11.1	51	8.5	54	15.4
Bank charges (monthly)
R100 or less	472	49.7	302	50.4	170	48.6
R101-R200	224	23.6	142	23.7	82	23.4
R201 or more	253	26.7	155	25.9	98	28

Source: Researcher’s own construct

Table 2.

Measurement model results

Construct	Item	Overall model					Non-Islamic banking customers model					Islamic banking customers model
		Outer loading	p-value	Cronbach’s alpha	CR	AVE	Outer loading	p-value	Cronbach’s alpha	CR	AVE	Outer loading	p-value	Cronbach’s alpha	CR	AVE
Loyalty	Loyal1	0.920	0.000	0.890	0.932	0.820	0.919	0.000	0.906	0.941	0.841	0.923	0.000	0.856	0.913	0.777
	Loyal2	0.900	0.000				0.924	0.000				0.846	0.000
	Loyal3	0.895	0.000				0.909	0.000				0.874	0.000
Customer orientation	CO1	0.928	0.000	0.923	0.951	0.867	0.921	0.000	0.917	0.948	0.858	0.941	0.000	0.937	0.960	0.888
	CO2	0.945	0.000				0.941	0.000				0.954	0.000
	CO3	0.920	0.000				0.915	0.000				0.932	0.000
Information sharing	Info1	0.919	0.000	0.908	0.942	0.845	0.918	0.000	0.914	0.946	0.854	0.922	0.000	0.894	0.934	0.825
	Info2	0.893	0.000				0.906	0.000				0.865	0.000
	Info3	0.944	0.000				0.948	0.000				0.936	0.000
Service fairness	IF1	0.865	0.000	0.946	0.955	0.754	0.863	0.000	0.942	0.953	0.743	0.872	0.000	0.953	0.961	0.780
	IF2	0.899	0.000				0.891	0.000				0.913	0.000
	IF3	0.841	0.000				0.815	0.000				0.900	0.000
	IF4	0.894	0.000				0.894	0.000				0.897	0.000
	PF1	0.848	0.000				0.854	0.000				0.836	0.000
	PF2	0.871	0.000				0.871	0.000				0.871	0.000
	PF3	0.859	0.000				0.845	0.000				0.891	0.000
Trust	Trust1	0.913	0.000	0.938	0.955	0.842	0.899	0.000	0.929	0.950	0.825	0.944	0.000	0.954	0.966	0.878
	Trust2	0.931	0.000				0.920	0.000				0.954	0.000
	Trust3	0.934	0.000				0.927	0.000				0.949	0.000
	Trust4	0.893	0.000				0.888	0.000				0.900	0.000

Note:

All outer loadings are significant at p < 0.05, two-tailed

Source: Researcher’s own construct

Table 3.

Assessment of discriminant validity

Overall model
Fornell and Larcker criterion
Constructs	Customer loyalty	Customer orientation	Information sharing	Service fairness	Trust
Loyalty	0.905
Customer orientation	0.770	0.931
Information sharing	0.752	0.809	0.919
Service fairness	0.748	0.762	0.760	0.868
Trust	0.773	0.806	0.784	0.827	0.918
HTMT ratio
Loyalty
Customer orientation	0.847
Information sharing	0.833	0.880
Service fairness	0.813	0.815	0.818
Trust	0.844	0.865	0.847	0.877
Cross-loadings
Items measuring the constructs of the study exhibit higher loadings on the construct they are measuring than on other constructs and the items do not have differences in the loadings of less than 0.1 between the parent construct and the other model constructs
Multicollinearity
A VIF value of less than 5 is evident for each item measuring the study’s constructs
Non-Islamic banking customers model
Fornell and Larcker criterion
Loyalty	0.917
Customer orientation	0.781	0.926
Information sharing	0.767	0.781	0.924
Service fairness	0.763	0.764	0.772	0.862
Trust	0.797	0.809	0.792	0.835	0.909
HTMT ratio
Loyalty
Customer orientation	0.856
Information sharing	0.840	0.849
Service fairness	0.824	0.820	0.829
Trust	0.868	0.875	0.856	0.891
Cross-loadings
Items measuring the constructs of the study exhibit higher loadings on the construct they are measuring than on other constructs and the items do not have differences in the loadings of less than 0.1 between the parent construct and the other model constructs
Multicollinearity
A VIF value of less than 5 is evident for each item measuring the study’s constructs
Islamic banking customers model
Fornell and Larcker criterion
Loyalty	0.881
Customer orientation	0.747	0.942
Information sharing	0.714	0.869	0.908
Service fairness	0.716	0.760	0.734	0.883
Trust	0.716	0.803	0.764	0.813	0.937
HTMT ratio
Loyalty
Customer orientation	0.830
Information sharing	0.812	0.947
Service fairness	0.788	0.802	0.791
Trust	0.787	0.846	0.825	0.848
Cross-loadings
Items measuring the constructs of the study exhibit higher loadings on the construct they are measuring than on other constructs and the items do not have differences in the loadings of less than 0.1 between the parent construct and the model constructs
Multicollinearity
A VIF value of less than 5 is evident for each item measuring the study’s constructs

Source: Researcher’s own construct

Table 4.

Structural paths

Path	Coefficient	t-value	p-value	LLCI	ULCI	Result
Overall model
Customer orientation → Trust	0.310	6.669	0.000	0.219	0.404	Significant
Information sharing → Trust	0.199	4.777	0.000	0.116	0.281	Significant
Service fairness → Trust	0.439	11.858	0.000	0.367	0.511	Significant
Trust → Loyalty	0.773	44.753	0.000	0.736	0.805	Significant
Non-Islamic banking customers model
Customer orientation → Trust	0.310	5.404	0.000	0.197	0.424	Significant
Information sharing → Trust	0.218	4.296	0.000	0.117	0.315	Significant
Service fairness → Trust	0.430	9.174	0.000	0.338	0.524	Significant
Trust → Loyalty	0.797	39.274	0.000	0.754	0.834	Significant
Islamic banking customers model
Customer orientation → Trust	0.339	4.380	0.000	0.189	0.492	Significant
Information sharing → Trust	0.134	1.991	0.047	−0.004	0.268	Not Significant
Service fairness → Trust	0.457	7.602	0.000	0.339	0.573	Significant
Trust → Loyalty	0.716	22.914	0.000	0.649	0.772	Significant

Notes:

LLCI = lower-level confidence interval; ULCI = upper-level confidence interval; *two-tailed level of significance

Source: Researcher’s own construct