Measuring and managing brand loyalty of banks` clients

The purpose of the study is to measure behavioral, attitudinal and other brand loyalty antecedents, and to develop an operating model for measuring and managing brand loyalty of commercial banks clients. A random sample of 500 members of the South African Commercial Institute, who are also commercial banks’ clients, received a 5-point Likert scale questionnaire to be completed online via Twitter and Facebook. About 196 people completed the questionnaire. The data possess construct validity and reliability (α ≥ 0.70). The results show that seven of the 12 original antecedents are banking related, namely five Attitudinal antecedents (r2 = 0.557) and two Other antecedents (r2 = 0.442). Behavioral antecedents were not important to bank clients. All the antecedents have factor loadings above 0.60, and there is a significant positive correlation between Attitude and the Other antecedents (r = 0.75; p ≤ 0.01). This means that the model is useful for managers in managing brand loyalty at their banks. It is also of value to researchers and academia looking to conduct further research on how to measure and manage brand loyalty. However, a caution is that the data originated from South African banks’ clients. Country-specific influences can cause different brand loyalty preferences among international banks’ clients.
AcknowledgmentI wish to acknowledge Mr. Sarel Salim for his contribution to administering the data collection for the original research on brand loyalty in banking (see also Salim and Bisschoff, 2014 in the reference list).


INTRODUCTION
South Africa has an exceptionally well-developed banking system and competition is rife in the banking sector, with a number of banks competing to remain relevant in the market. Most of these banks serve multiple market segments ranging from individual customers to high-end businesses, although a few South African banks have recently focused their marketing strategies and aims to penetrate specific market segments. Despite this strong competitive environment in the South African banking sector, virtual banks are now also entering the South African banking market, and this has increased competition even more.
South African banks compete using client service, quality products and customer loyalty, among other strategies, to retain their customers. However, although banks strongly focus on customer service and loyalty as competitive thrust, they only measure and manage customer service. None of the banks actively measures and manages brand loyalty to their bank. This strategic oversight is partly due to limited operating models or measurement tools, and the inability to accurately manage brand loyalty among their banks' clients. This study aims to set this oversight straight and provide bank managers with guidance and an operational tool to measure and manage brand loyalty of their banks' customers.

Historical overview of banking in South Africa
Banking in South Africa is controlled by the South African Reserve Bank. All banks are required to register with the Reserve Bank and acquire a bank license to be eligible to do business in the country. Banking regulations apply to local banks and foreign banks (such as Bank of Lisbon, JP Morgan Chase and others), as well as to all banks operating under the same legal regulations (South African Reserve Bank, 2020). Traditionally, banking in South Africa consisted of many smaller banks and building societies. Some of these building societies belonged to some of the four larger banks that dominated the market (Volkskas Bank, Nedbank, Standard Bank and First National Bank). In 1991, the Volkskas Bank group embarked on a merger and acquisition strategy, and incorporated the Bankorp Group (Bankfin, Senbank and Trust Bank), United Bank Allied Bank and some of the Sage group's interests into a new bank called Amalgamated Banks of South Africa (ABSA) to establish a new market force in South African banking.
Capitec Bank was launched in March 2001 by the PSG Group, and specifically targeted the lower-income market with simplified and low-cost banking. Their banks' products differ from the four traditional banks, and South Africans were introduced to concepts such as one account banking with their Global One account. Capitec explains their Global One as an account where a client gets a transaction account, four free savings plans, personalized credit options and insurance, all in one place. Clients can use their Capitec debit or credit cards to access their accounts, as well as manage their money matters easily from their mobile phone banking apps or on Internet banking anytime and anywhere. Despite rife competition, more banks entered the market; the latest entry is Discovery Bank, which is a business expansion project extension of South Africa's largest medical insurance group.
South Africa's banking sector is also on the forefront of technological advancement in world banking and is highly digitized, albeit this comes at a price. Banking services are relatively expensive compared to international banking fees. Internally, intense competition and digitalization have led Absa, FNB, Nedbank and Standard Bank to close down (or downsize) branches to control rising operating costs, as well as increased clients' use of online and mobile banking services. Other new entrants focus strongly on digital banking. These entrants include Discovery Bank, TymeBank and Bank Zero, as well as various mobile money apps and services provided by telecommunication and retail companies, where customers can pay a myriad of service suppliers. Clients can also draw cash from tellers at, for example, grocery stores while doing their weekly shopping. These service providers are forcing commercial banks to continue to modernize their technology platforms.

Measuring brand loyalty
Brand loyalty is part of a complex set of customer decision criteria. These criteria entail both internal and external influences. In more tangible buying behavior, consumers have the advantage of physically examining the products at hand before buying. Still, in the services industry, the characteristics of products fade into a more perceived realm. Banking (although banks call their offerings such as insurance, credit cards and current account "products") is a service.
Brand loyalty is defined as the tendency of customers to continuously prefer one brand over competitive branded products (TrackMaven, 2020). Netto (2020) explains that brand loyalty results in customers who are willing to wait in long lines to buy a specific product, customers who will travel long distances, and customers who are basically immune to promotions of competitive products of a similar nature. A company not only saves on acquiring new customers, but is able to service existing customers who return again and again because they have a loyal relationship with the specific brand; this is worth market share and increased profitability (Netto, 2020). Furthermore, the value of brand loyalty not only resides in the rebuy intentions of customers, but also because brand loyal customers are willing to pay higher prices and are less sensitive to price fluctuations (Nisa, 2019). This has just been confirmed by the Nielsen Report (2019) that states price is no longer the dominant consumer decision-making criterion in South Africa; brands and brand loyalty now strongly influence the decision-making process.
The original literature study by Moolla (2010) identified 54 popular brand loyalty antecedents that form established models for measuring and managing brand loyalty. In further scrutiny, Moolla (2010) reduced these antecedents to 26, based on their popularity of use and their level of success in the existing brand loyalty measurement models. The process of eliminating less important antecedents involved exploring a wide array of brand loyalty studies to identify the most important and widely used antecedents. A final list of 12 identified antecedents was retained after their relevance for modern measurement applications was confirmed by the literature: • Switching cost between different brands, products and services in some markets may be negligible, while other markets pose high costs to customers looking to change their brand or product (Bisschoff & Moolla, 2015 • Relationship proneness is defined as an individual customer's tendency to form a relationship with a brand or product (Kim, Morris, & Swait, 2008). Some customers are more prone towards relationships than others, and, therefore, relationship proneness is considered to be a personality trait of a consumer (Van der Westhuizen, 2018). Relationship proneness is also a conscious tendency as opposed to behavior linked to inertia or convenience (Schijns, 2003).
• Involvement refers to the continuous commitment of consumers' thoughts, feelings and behavior towards a brand or product. Involvement acts as a motivation to create an interest toward a product (Van der Westhuizen, 2018). Studies by Jacoby and Chestnut (1978), and others (Basson, 2014;Bisschoff & Moolla, 2015) report that a significant positive correlation exists between a product and brand loyalty. Similarly, Dick and Basu (1994), in their study, found a significant positive correlation between involvement and attitudinal loyalty. In this regard, Yasin and Shamim (2013) concur that higher levels of product or brand involvement should lead to higher brand loyalty levels).
• Perceived value refers to the overall value a consumer places on a product. This value is subjective since it is based on what the consumer's perceptions are or what he or she received in exchange for the price paid for the product (Punniyamoorthy & Raj, 2007 • Established repeat purchase behavior is fundamentally brand loyalty behavior. Repeat purchase, in essence, refers to how many times a customer re-purchase the same product or brand in any specific period (Fullerton, 2005). The advantage of behavioral brand loyalty, therefore, resides in the repetitive buying and consumption of the product or brand. A consumer develops a habit or systematically biased behavior due to his or her frequency of buying encounters (Ong, Lee, & Ramayah, 2018). Consumers seldom change their behavioral brand loyalty after the repeat buying behavior became a habit; commitment may even result in autonomous buying behavior of the specific product or brand (Kim, Morris, & Swait, 2008). The more established this habit becomes to buy a specific brand, the more difficult it becomes to change; this repeat purchase pattern then snowballs into higher levels of brand loyalty (Chaudhuri & Holbrook, 2002).
• Brand affect causes an emotional response when consumers use a product or brand (Chaudhuri & Holbrook, 2002). This affect can be positive or negative; customers are eager to experience a positive affect to avoid a negative affect. In practice, brand affect refers to the "feeling" that a brand produces when used by a customer; that is, for example, the sense of proudness when wearing running shoes of a top brand such as Nike. A significant positive relationship exists between a positive brand affect (or experience) and the consumer's willingness to buy. This positive affect also applies to the brand of the store image and leads to brand loyalty ( • Perceived brand performance is the customer's post-consumption perception on how well the brand (or product) performed. Have the brand delivered on its promise? (Bisschoff & Moolla, 2015). This experience is closely tied to the subjective, and the perceived performance is closely tied to the expected performance of a brand (Jacoby & Chestnut, 1978). Brand performance extends beyond the core brand or product attributes, as well as comprises intrinsic (effectiveness) and extrinsic (packaging) characteristics (Chaudhuri & Holbrook, 2002). Brand performance is also directly influenced by culture (Unurlu & Uca, 2017).
• Culture is a vital consumer buying behavior antecedent (Bisschoff & Moolla, 2015). Although young consumers entering the market at first remain loyal to known brands used at home, they may change their choice of brands once being exposed to other brands and influences. Trust could be instilled in a product or brand as a result of generations of use. Nostalgia is also a factor in maintaining brand loyalty of individuals loyal to classical brands (Kotler & Armstrong, 2019). If parents show brand loyal buying behavior, chances are that children will follow suit and one day be brand loyal customers. Thus, family influences play an important role in shaping buying behavior; this is true for products, services and brands. Culture also influences brand loyalty since some cultures are more brand loyal in nature. Different cultures also experience brand performance differently (Unurlu & Uca, 2017).
• Customer satisfaction is based on product performance and explains the feelings de-rived from actual performance compared to expected performance (Dick & Basu, 1994). A discrepancy between actual and perceived performance leads to satisfaction (positive gap), while dissatisfaction originates in a negative performance gap. Each product evaluation educates a customer and adjusts his/her expectations of a specific brand and product (Rather, 2018). Continued satisfaction leads to repeat purchases (or loyalty) towards a brand or product. Effective post-purchase complaint behavior also strengthens customer satisfaction; this leads to higher brand loyalty levels (Dilham, Sofiyah, & Muda, 2018).
The seminal study by Jacoby (1971), later confirmed by Jacoby and Chestnut (1978) and Aaker (1991;1996), showed that behavior (stochastic approach) and attitude (determinist approach) are the two main drivers of brand loyalty. Furthermore, these researchers found that attitude influences brand loyal behavior and identified significant positive and negative relationships (p ≤ 0.05; p ≤ 0.10) between these two drivers (Fischer, Völckner, & Sattler, 2010; Bandyopadhyay and Martell, 2007). In practice, this means that a positive attitude leads to increased brand loyalty. In contrast, negative attitudes towards a brand inversely affect brand loyalty behavior such as diminished repeat purchases of the brand (lower volumes of less frequent purchases). This is of great practical value for marketers and brand managers. Strong positive attitudes towards a brand allow a marketer to charge premium prices. Therefore, the brand becomes more profitable (Myanmar, 2018). Behavior, on the other hand, is strongly tied to market share. Consequently, positive behavior results in more frequent and higher purchase volumes (Myanmar, 2018). Ideally, positive brand attitudes could improve behavioral loyalty, and brand managers would be able to start asking premium prices in an increasing market share; this is how brand loyalty contributes to establishing a top brand. The brand loyalty antecedents are classified into attitudinal and behavioral antecedents (see Table 1). Two of the antecedents in the model are neither attitudinal nor behavioral; they are listed as "Other antecedents".

AIM
The aim of this study is to develop an operational management model to measure and manage brand loyalty among banks' clients.

RESEARCH METHODOLOGY
The theoretical model was developed from a myriad of seminal brand loyalty models developed as far back as 1940. Common brand loyalty antecedents and their respective relevant measuring criteria were identified and operationalized into a theoretical model to measure brand loyalty. The final theoretical model contains a questionnaire of 12 brand loyalty antecedents that are measured by 50 criteria. The antecedents and measuring criteria directly stem from the literature. This constitutes the questionnaire used to measure brand loyalty. The theoretical model (from Table 1) is shown in Figure 1.
The population consisted of all the members of the South African Commercial Institute. A random sample consisting of 500 members was drawn using a randomized computer algorithm. These members received the hyperlink to the questionnaires electronically via Twitter and Facebook (Salim & Bisschoff, 2014). They completed the questionnaire online via the Qualtrix (2019) questionnaire platform where data was automatically captured as soon as a respondent completed the questionnaire. 196 completed questionnaires were returned; this is a 39.2% response rate.
This questionnaire was empirically validated by Moolla and Bisschoff (2012b). Later, a study by Salim and Bisschoff (2014)  The questionnaire consists of 50 close-ended questions that had to be answered on a five-point Likert scale.

RESULTS
The suitability of the data for structural equation modelling requires verification. This verification involves sample adequacy, sphericity, reliability and, finally, multicollinearity of the data. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was used to measure sample adequacy. The KMO value for this study is 0.713 (decision rule: KMO ≥ 0.70). Sphericity was tested using Bartlett's test; the results show satisfactory values (p = 0.00) (decision rule: p ≤ 0.05) (Field, 2017). Regarding the reliability of the data, Cronbach alpha's coefficient indicates that the data are reliable (α = 0.71) (decision rule: α ≥ 0.70) (Cortina, 1993).
The data were also tested to rule out multicollinearity by using the variance inflation factor (VIF) and Slovin's tolerance thresholds in a linear regression model (Minitab, 2013). In this exploratory study, the VIF should ideally be below 3, preferably below 5 and definitely below 10, while the Slovin's tolerance threshold should ideally exceed 0.2, or preferably exceed 0.4 to prove that multicollinear-Note: * See Table 1 for the key to abbreviations. ity is not a problem (Statisticshowto, 2020). Table  2 shows the results of the multicollinearity tests. The results show that limited multicollinearity exists and that all the antecedents are within the ideal VIF and tolerance ranges (VIF ≤ 3; Slovin's tolerance ≥ 0.4). These results, in conjunction with the results from the sample adequacy, reliability and sphericity, show that the data are suitable for multivariate analysis and for structural equation modelling.
The structural equation model was developed using AMOS software, which is an extension of the IBM SPSS statistical software. Figure 2 shows the results of the scrutinized theoretical model. From the empirical model (Figure 2), it is clear that five of the 12 original antecedents that were evaluated were rejected. They are switching cost, brand performance, repeat purchase (all behavioral antecedents) and culture and brand trust (both are attitudinal antecedents). Interestingly, one behavioral antecedent (involvement) was worthy of retaining. However, on closer inspection, this antecedent loaded heavier onto the attitude category than on behavior. This means that banks' clients regard involvement as an attitudinal antecedent of brand loyalty and not a behavioral one. The two antecedents categorized as "other" antecedents (customer satisfaction and perceived value) were retained in the category "Other antecedents" in the model. Resultantly, the empirical model for brand loyalty of bank clients then consists of only two categories. They are the Attitudinal (r 2 = 0.557) and Other (r 2 = 0.442) categories. There is also a significant positive correlation between Attitude and the Other antecedents (r = 0.748; p ≤ 0.01).
A structural model should possess construct validity to be fit for use. This is achieved if both discriminant and convergent validly are proven in  Table 3 shows the convergent and discriminant validity calculations.
Convergent validity is achieved if the AVE value exceeds 0.5. The results show that the attitudinal antecedents (AVE = 0.560) achieved convergent validity, but that the other antecedents (AVE = 0.413) marginally do not achieve convergent validity. Regarding discriminant validity, the square root of AVE for each category (Attitude = 0.748; Other = 0.642) should exceed the correlation between the two categories (r 2 = 0.565). Both categories show that they do have discriminant validity. Thus, it can be concluded that, overall, strong indications of construct validity exist.  Table 4.
All but one of the model fit results are satisfactory and fall well above the required values as stipulated by the decision rules in Table 4. The RMSEA value, however, falls marginally outside the range (RMSEA ≤ 0.10) (Xia & Yang, 2019). This indicates a lower level of fit than the CFI.
In summary, the model fit is satisfactory. CFI, as a primary fit index, exceeds 0.95. This is supported by good fit indices by GFI and CMIN/df. TLI and RMSEA could have had better model fits. However, in defense of the model, it is noteworthy that this is an exploratory model, and it cannot be expected that the model fits superbly on all the indices. The model is of an exploratory nature and not yet a final, fully operationalized model.

CONCLUSION
With regard to data collection and the quality of the data used in this study, it can be concluded that the data collected was sufficient. Also, the questionnaire proved valid and reliable data. This means that the questionnaire and data collection methodology can be put in place and use by bank management to col- lect valid and reliable brand loyalty data among their customers. Therefore, the collection methodology can be duplicated for similar research.
Regarding the operational model to measure and manage brand loyalty, it is concluded that commercial banks' clients view attitudinal antecedents, customer satisfaction and perceived value as important brand loyalty traits, while the behavioral antecedents seem to be less important to manage brand loyalty. This means that bank management should focus their loyalty management interventions on attitudinal antecedents, customer service and their clients' perceived value to get the best returns on their managerial efforts. They should also include the antecedent Involvement (classified by theory as a behavioral brand loyalty antecedent) in their management strategies, as banks' clients clearly indicated that they view involvement as an attitudinal antecedent.
Regarding model validity, it is concluded that the model has good construct validity and possesses a good model fit. This means that the operating model can be confidently used by bank managers to measure and manage brand loyalty among their banks' clients.