M-insurance in a generic sense is any microinsurance product whose marketing involves the MNOs as the conduit in delivering some components of an insurance value chain to prospective clients. The M-insurance market has been scaling up globally with a myriad of new products being rolled out annually. It is estimated that as of June 2016, there were over a 100 M-insurance services, and 52.7 million policies issued (denoting a whopping 263%, annualised growth) across 30 countries globally – powered by the disruptive technologies and innovations of MNOs (GSMA, 2014, 2016).

Arguably, its intended purpose of creating a niche market to amplify financial inclusion ostensibly for demographics within low-income brackets is being met. It is evidenced that there has been a global pattern of financial liberation and economic empowerment in regions of socio-economic commonalities (such as Africa, Asia and Latin America) where mobile-enabled insurance, credit and savings products are visibly and ubiquitously present. The composition of insurance services offered by the M-insurance industry range from life, health and accident to agriculture (GSMA, 2016).

The unique presence of microfinance schemes in the aforementioned geographical settings is against the backdrop of unmet conditions attached to the product offerings of general insurance service providers by a chunk of the populace. The M-insurance model, however, is appealing and attractive to the underserved and the low-income clusters of the society. It is designed to accommodate the unique socio-economic characteristics of these vulnerable demographics. Relative to orthodox insurance product offerings, the M-insurance products have a peculiarly low distribution cost, and thus make it feasible to spread financial services to a chunk of the populace in order to manage their household shocks and financial risk (GSMA, 2016). India and Bangladesh are two countries outside the African enclave with huge footprints of M-insurance services (GSMA, 2014).

According to (National Communication Authority, 2015) report, Ghana has a mobile network penetration rate of approximately 115%; and the country is adjudged as one of the most promising and vibrant mobile markets in African. There are six licensee MNOs in Ghana – Mobile Telephone Network (MTN) (45.6% market share), Vodafone (23% market share), Tigo (13.85%), Airtel (12.4%), Glo (4.8%) and Expresso (0.4%) (Mobile Insurance and Risk Management in Ghana, 2015).

The microinsurance industry in Ghana assumed a propitious dimension in the wake of Tigo’s and MTN’s launch of M-insurance products in 2010 and 2011, respectively. M-insurance is an innovative insurance delivery system through which the MNOs market insurance products to the wider general public. It plays overarching roles within the Ghanaian and similar developing economies’ microinsurance ecosystem as it is estimated that 60% of lives were insured (in Ghana) via M-insurance in 2014 alone. The M-insurance models are twofold, namely, the strategic model⁴ and the transactional model.⁵ In a strategic model, the insured clients recognise the insurance products to be owned by the MNOs because of the product’s implementation, branding, advertisement and marketing mechanism (Mobile Insurance and Risk Management in Ghana, 2015; Tellez & Zetterli, 2012).

Strategic M-insurance products are grouped into two: loyalty (otherwise known as Freemium) and paid products. Under the loyalty products option, MNOs pay the premium and offer the insurance to subscribers as a monthly freebie, aiming at incentivising subscribers to increase the average revenue per user (ARPU), and secondly, to eschew a possible switch to a competitor. Subscribers are, however, eligible for the monthly insurance coverage only if a top-up to a certain threshold is made.

Under the paid products regime, subscribers personally pay for the insurance through mobile money or direct payment (airtime). Automatic payment is made if there are sufficient funds in the subscribers’ M-wallet or airtime (Mobile Insurance and Risk Management in Ghana, 2015).

Studies indicate that there are an estimated 2.7 million M-insurance policyholders, and three service providers (MNOs) – Tigo, MTN and Airtel: are presently operating in Ghana. Also, there are three insurers in the Ghanaian M-insurance market – Prudential, Enterprise Life and Unique Trust (UT) Life – and three Technical Service Providers (TSP) namely, Barmore Insurance Marketing Association (BIMA), Mobile Financial Services (MFS) Africa and MicroEnsure (Mobile Insurance and Risk Management in Ghana, 2015; Tellez & Zetterli, 2012; Tellez, 2013). Relative to other enclaves, sub-Saharan Africa is extremely active in M-insurance provision (see Table 1) because of the financially inclusive nature of the model. The current M-insurance products in Ghana are predominantly, hospitalisation, funeral, life, personal accident and property covers (see Table 2), with monthly premiums payment from mobile wallets or airtime deduction (Mobile Insurance and Risk Management in Ghana, 2015; Ndurukia et al., 2017).

The apparent socio-economic commonalities within the enclaves with huge footprints of M-insurance, thus, make the findings our study transferable.

In order to have a competitive edge in a highly charged competitive environment such as the M-insurance market, industry players must constantly pursue top notch strategic plans aiming at securing CL. In other words, service providers must innovatively differentiate themselves from rivals; and also create a long-lasting client-company dyad (Morgan & Hunt, 1994 [The Apostle Model of Customer loyalty]).

Customer Relationship Management (CRM) and Consumer Analysis are targeted at CL and, as such, it has become the fulcrum of attention among the academics and professionals within the realms of consumer service management. Theories abound, therefore, with evidence of the inextricable satisfaction–loyalty duo (Cronin & Taylor, 1992; Lewin, Rajamma, & Paswan, 2015). Fundamentally, CRM and Consumer Analysis encapsulate strategies adopted by managers to provide top-notch client-focused services to existing and prospective clientele, ostensibly to ignite clients’ retention in general (Fornell, 1992).

Clients’ loyalty is defined as ‘a deeply held commitment to rebuy or re-patronise a preferred product/service consistently in the future despite situational influences and marketing efforts having the potential to cause switching behaviour’ (Oliver, 1999). That is, CL connotes hanging on during trial moments; not scanning for competitors regardless of a dire situation; making an effort to communicate with the service providers in order to build on successes chalked up and also surmount possible threats. It also involves offering excellent sentimental rating scores and Word-of-Mouth (WOM) publicity. Studies, however, centre on the behavioural–attitudinal–composite trio of loyalty (Parasuraman, Zeithaml, & Berry, 1994a; Richard & Zhang, 2012; Wang, Ngamsiriudom, & Hsieh, 2015; Zhang, Fu, Cai, & Lu, 2014).

In the Apostle Model of customer loyalty, four kinds of client are distinguished – loyalists, mercenaries, defectors and hostages. It advances that a highly satisfied client will always be loyal and also be willing to serve as an apostle to a product or service offering. There is plethora of studies with empirical support that have examined the satisfaction–loyalty–performance trio construct owing to its overarching role in consumer psychology (Helgesen, Ivar, & Nesset, 2010; Smith & Wright, 2004). Gremler and Brown (1996) established that CL is important for the survival of businesses; hence its antecedents ought to be explored.

Prior researchers (Adnan, Suleman, & Bilal, 2015; Lai-Ying, Teck-Soon, Voon-Hsien, & Keng-Boon, 2015; Thakur, 2016) opine that satisfaction is an antecedent of customer loyalty. In other words, highly satisfied clients are more likely to repurchase and become apostles of a brand. It is, however, imperative to acknowledge that satisfaction is not the singular antecedent of CL. Other antecedents such as trust, value proposition, etc. have been observed to correlate positively with CL disposition (Chen & Hu, 2012; Hartono & Raharjo, 2015; Yieh, Chiao, & Chiu, 2007).

This study, therefore, seeks to examine the overarching roles of institutional-based trust, Service Quality (SERQUAL), Clients’ satisfaction and corporate identity and their interactions in determining CL within the M-insurance ecosystem (see Figure 1).

Institutional-based trust is a form of trust that connotes the perception or the disposition of an individual relative to an institution’s holistic operational structure. Institutional-based trust theorists advance that strong institutions through cognitive, regulative and normative frameworks are able to set the tone that ultimately triggers a domino effect for permanent trust-relations among the actors of the market. Institutional-based trust is fundamentally the relationship between two entities – an individual and an institution – and it is expressed through value judgement (McKnight & Chervany, 2002; Pavlou & Gefen, 2004). The disposition relationship between these two parties may be negative, positive or even neutral. Therefore, in the context of this study, institutional-based trust is defined as the dynamic trust relationship between clients and service providers of the M-insurance industry. McKnight and Chervany (2002); Gefen, Karahanna, and Straub (2003); and Ratnasingham, Gefen, & Pavlou, (2005) decomposed institutional-based trust into structural assurance (SA) and situational normality (SN) and opined that it is the antecedent of continuous service utilisation and CL.

Structural assurance is defined as the disposition that an enabling condition (servicescape and/or experiencescape) or a congenial atmosphere that assures clients maximum security for the actualisation of their economic goals are guaranteed by the service provider. Situational normality, on the contrary, is the belief of the presence of propitious and promising structures for the success of a business, and also evidence of the assurance that systems will work normally as perceived or expected by clients. Situational normality deepens and bolsters the trust disposition of service providers (Pavlou & Gefen, 2004). Thus, institutional-based trust is the outcome of SA and SN. In other words, SA and SN (as explanatory variables) complementarily translate into institutional-based trust (as the outcome or response variable).

Yang, Pang, Liu, Yen, & Tarn (2005) posit that clients’ privacy and security dispositions affect their trust relative to mobile payment systems. Thus, by extension, privacy and security contribute immensely to the institutional-based trust in the M-insurance industry. Again, clients of M-insurance companies who perceive the presence of SA will also believe that the legal provisions relative to institutions within the M-insurance ecosystem (i.e. MNOs, insurers, TSPs and banks) work in their best interest. SA builds clients’ trust and confidence, and as such makes them believe that there will always be SN whenever something goes haywire (Al-Ghazali, Rasli, Yusoff, & Mutahar, 2015; Coetzee & Eloff, 2005; Wang et al., 2015). Evidentially, inadequate privacy and security results in dissatisfaction and its antecedent customer churn (PricewaterhouseCoopers [PWC], 2014).

Zucker (1986) advances that institutional-based trust triggers a lasting relationship and continuous interactions between the market forces. Against this backdrop, institutional-based trust can impact positively on clients’ satisfaction and loyalty in the M-insurance industry. We therefore postulate that:

Service quality (SERQUAL) is defined as a condition where a delivered service is able to meet the expectation or satisfaction of a client. In a generic perspective, it is defined as clients’ overall disposition about the relative superiority or otherwise of a service provider (Edvardsson, 2005; Hubbert, 1995). Parasuraman, Zeithaml, & Berry (1985) expressed SERQUAL as the difference between clients’ perceptions (P) and their expectations (E) about a rendered service. There is a battery of evidence that suggests that SERQUAL is the main driver and precursor of clients’ satisfaction and loyalty. It anchors clients’ satisfaction and sets the pace for WOM communications, while encouraging post-purchase intentions. Perceived SERQUAL creates a congenial atmosphere for the assurance of value proposition, security and trust. Thus, clients are loyal to organisations that are perceived to offer high quality services. The reverse also holds. (Zehir & Narcıkara, 2016; Minjoon & Serglo, 2016; Chandra, Hafni, Chandra, Purwati, & Chandra, 2019).

Empirically, clients’ satisfaction and its resultant customer loyalty as a result of perceived SERQUAL lead to an upsurge in market share through customer retention and attraction of new clients (Adnan et al., 2015; Trabelsi-Zoghlami, Berraies, & Yahia, 2018; Acrand, Prom Tep, Brun, & Rajaobelina, 2017; Thakur, 2016). Thus, SERQUAL dimensions are arguably one of the most researched areas in service management discipline. The SERQUAL model identified five gaps (the 5-GAP model) that cause unsuccessful service delivery. Service quality is mostly assessed through the SERQUAL model dimensions – reliability, assurance, tangibles, empathy, responsiveness (Parasuraman, Zeithaml, & Berry, 1996; Zhou, 2013). Delivering quality service to clients is a way by which service providers inculcate trust in their clientele and it is evidenced that service providers who provide quality services are perceived to be trustworthy (Cronin & Taylor, 1992; Kandampully & Butler, 2001). By extension, we propose, therefore, that:

Clients’ satisfaction connotes the successive disposition of clients’ adoption behaviour relative to a service delivery or a product offering, expressed by a continual relationship or interactions with service providers. Oliver (1980) opines that it is the building block of post-adoption accomplishments. Other seminal studies also defined satisfaction as a person’s perception – either satisfied or otherwise achieved through the comparison of a person’s perceived performance of a service or with his/her expectations. Satisfied customers express it in a form of loyalty to the service providers. Conversely, dissatisfied customers refrain from future purchases and customer referrals. This adversely affects future service utility of service providers as disgruntled customers divert to other service providers (Kim, Shin, & Lee, 2009; Suki, 2014).

Clients’ satisfaction is touted, theoretically, as the result of a subjective evaluation between what is perceived and what is expected for post-purchase adoptions. Clients’ satisfaction triggers post-purchase intention, positive WOM referrals, retention drive and consequently cements clients loyalty to a product or a service (Cronin & Taylor, 1992; Parasuraman, Zeithaml, & Berry, 1985). A myriad of studies have examined the inextricable link between satisfaction and loyalty. For instance, Peng et al. (2013) established that satisfaction is one of the main underpinning factors of client retention and loyalty in the Chinese mobile phone and e-commerce industry. On the basis of these assertions, we hypothesise that:

Corporate identity is the impression created by business entities in the ‘public eye’. In other words, it is the image of the creator of one’s product/service. It encompasses a collection of a multi-sensuous variables that marketing practitioners utilise as a conduit to communicate visualised information relative to branding of products or services to the consuming public. Theoretically, the fundamental goal of corporate communications is to foster and enhance strong corporate identity. It is aimed at creating and preserving corporations’ identity or image (reputation) in consonance with the business’s objectives, mission and vision, and also to position the brand on the mind of consumers (Ang, 2014). It rejuvenates clients’ confidence and decreases their perception of risk when examining the quality and performance of a service or a product offering (Keh & Xie, 2009).

A corporation’s name and colours, product/service design, public relations, slogan, advertising, architectural design, décor, logo and physical appearance of frontline staff are some of the areas of attention whenever corporations want to create a formidable corporate identity (Dowling, 1993). The corporate identity of a firm or a corporation can only be described as strong or weak. Strong corporate identity promotes competitive advantage and fosters a bond of trust between consumers and business entities, which in turn bolsters clients’ satisfaction, loyalty and improves return on investment and vice versa (Keh & Xie, 2009; Roberts & Dowling, 2002).

On the basis of the corporate identity of a business entity, clients are able to infer the likelihood of achieving their value proposition as assured (Chen, 2010; Hsu, Chang, Chu, & Lee, 2014). According to Calvo-Porral and Lévy-Mangin (2015), corporate identity remains a filter that influences the operation of organisations and the perceptions of consumers are anchored on it. Several studies have espoused and garnered support for the direct link between satisfaction and corporate identity (Helm, Eggert, & Garnefeld, 2010; He & Lai, 2014; Keh & Xie, 2009; Loureiro & Kastenholz, 2011; Richard & Zhang, 2012). We therefore, argue that:

The proposed research model (see Figure 1) thus sheds graphical light on both the direct and the indirect relationships among the constructs of the study.

The conceptual framework (Figure 1) consists of five constructs. Clients’ loyalty is the response (dependent) variable; SERQUAL and the mediating variables (corporate identity, clients’ satisfaction and institutional-based trust) serve as the independent/explanatory variables within the model. According to the model, CL within the M-insurance ecosystem is impinged upon by the collective behaviour of SERQUAL, corporate identity, clients’ satisfaction and institutional-based trust, all others things being equal.

The sex distribution of the 334, respondents after a preliminary data analysis was found to be 192 (57.2%) for males and 142 (42.5%) for females. The respondents’ age distribution was obtained as 28 (8.4%) for those who were ≥50 (8.4%), and 79 (23.7%), 143 (42.8%), 84 (25.1%), respectively, for the age brackets ‘40–49’, ‘30–39’ and ‘20–29’ (see Table 3).

In order to reduce data dimensionality, and also to create suitable dimensions to address the problem of the study, EDA and CFA were conducted to test for reliability, model adequacy, convergent and/or discriminant validity of the measurement model and goodness-of-fit for model fit measurements. The p-value of the chi-square statistic for Bartlett’s test (see Table 4) was 0.000 (p < 0.05), implying that the observed variables of the five-factor model were related, and therefore, fit for factor analysis (FA).

Structural equation modelling (SEM) was conducted in order to examine the presence of causality among the variables of the five-factor model. The maximum likelihood estimation (MLE) with promax rotation was used in grouping the variables under few unrelated factors; and the robustness of our proposed model and the psychometric properties of the scale were evaluated accordingly (Kline, 2016). The FA output produced a Kaiser–Meyer–Olkin (KMO) of 0.912 (see Table 4) which exceeded the recommended threshold of ≥ 0.7, for model adequacy (Nunnally & Bernstein, 1994). Another indication of model adequacy can be inferred from the extraction of communalities (see Table 5). All the reflective factor loadings of the pattern matrix were > 0.3 (Thompson, 2004). Model adequacy in SEM simply means how well a hypothesised model ‘fit’ or adequately describes the sample data.

Reliability is defined as the consistency of scores on an instrument given repeatedly to the same set of individuals (Haertel, 2006; Iacobucci & Duchachek, 2003). It reflects the internal consistency of the entire scale measurement. Accordingly, a Cronbach’s alpha point estimate for internal consistency measurement was obtained as 0.923 (see Table 6) and the deleterious effect of each item of our five-factor model exceeded the recommended threshold (> 0.7) (Finch, Immekus, & French, 2016; Nunnally & Bernstein, 1994).

In order to obtain greater information about the population parameter relative to internal consistency, we constructed a 95% confidence interval (Koning & Franses, 2003) (see Table 7). We are, therefore, 95% confident that the actual reliability of the population from which our sample was obtained is between 0.920 and 0.942.

The validity of an instrument involves the evaluation of the extent to which obtained scores represent the intended measured traits. As evidence of the convergence validity of the five-factor model, all the factor loadings of the pattern matrix (see Table 5) were > 0.5 and were within the recommended threshold of ≥ 0.5. Our results also (see Figures 2 and 3) give a pictorial evidence of convergent and discriminant validity of the five-factor model. Furthermore, all the standardised regression estimates on the path diagram (see Figures 2 and 3) were above the recommended threshold of > 0.7, affirming the convergent validity of our model (Kane, 2006; Zumbo, 2007).

The discriminant validity of our model was assessed graphically and numerically (see Figures 2–4 and Tables 5 and 8). There were no cross-loadings among the multi-indicator variables and also each of the off-diagonal values did not exceed the recommended threshold of > 0.7 (Raykov & Marcoulides, 2011).

The standardised estimates among the unobserved variables were all below the recommended value of < 0.8 (Raykov & Marcoulides, 2011) (see Figures 3 and 4). The results obtained from both the EDA and the CFA for testing model adequacy, reliability, convergent and discriminant validity overwhelmingly supported our model.

As empirically evidenced (Adnan et al., 2015; McKnight & Chervany, 2002; Pavlou & Gefen, 2004; Thakur, 2016), we found a significant positive association between institutional-based trust and clients’ satisfaction (b = 0.43; p < 0.001); and also, a similar association was found between institutional-based trust and CL (b = 0.43; p < 0.001), given support for H₁ and H₂.

Again, SERQUAL was identified to have a considerable positive impact on clients’ satisfaction (b = 0.30; p < 0.001); Institutional-based trust (b = 0.58; p < 0.001); and corporate identity (b = 0.43; p < 0.001); as demonstrated by Calvo-Porral and Lévy-Mangin (2015); Chen (2010); Hsu et al. (2014); and thus, providing support for H₃, H₄ and H₅. Similarly, H₇, H₈ and H₉ were supported because of the fact that we found corporate identity to correlate positively with clients’ satisfaction (b = 0.25; p < 0.001); institutional-based trust (b = 0.36; p < 0.001); and CL (b = 0.49; p < 0.001) confirming the studies conducted by Calvo-Porral and Lévy-Mangin (2015); He and Lai (2014); Richard and Zhang (2012).

Furthermore, as espoused by prior research (Trabelsi-Zoghlami et al., 2018; Acrand et al., 2017; Peng et al., 2013), we found an evidence of a positive association between clients’ satisfaction and CL (b = 0.24; p < 0.001), and thus, provided support for H₄ (see Table 9 and Figures 2–4). We used eleven (11) goodness-of-fit statistics namely, c²/df, p, incremental fit index (IFI), Tucker–Lewis index (TLI), standardised root mean square residual (SRMR), goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), parsimony goodness-of-fit index (PGFI), comparative fit index (CFI), root mean square of approximation (RMSEA) and normed fit index (NFI) to verify whether our hypothesised model correctly fits the sampled data, and also to affirm the model’s predictive robustness (see Table 10).

The chi-square value (2.149) fell within the recommended value of < 3.0; RMSEA value (0.059) was within the acceptable threshold of < 0.8; SRMR value (0.028) was within the recommended value of < 0.08; and the GFI value (0.873) was also above the recommended value of > 0.8 (Finch et al., 2016; Lee, 2007; Thompson, 2004).

The other goodness-of-fit measures – AGFI (0.846), PGFI (0.814), CFI (0.953), IFI (0.953), NFI (0.919) and TLI (0.947) were all within the recommended thresholds (> 0.8, > 0.8, > 0.9, ≥ 0.9, > 0.9 and ≥ 0.9, respectively) (Byrne, 2001). Based on the model fit measures obtained, we can confidently conclude that the structural model correctly fits the sampled data and affirms its predictive robustness (see Table 10).

The study was intended to explore the motivating factors of CL within the M-insurance industry. It contributes to service management literature; however, it is not without limitations. Firstly, the study was cross-sectional in nature, and thus, centred on some selected M-insurance service points in Accra, Ghana. It is, therefore, suggested that future studies should move beyond cross-sectional to longitudinal studies, and also broaden the scope of the study. Secondly, the study used the constructs – SERQUAL, institutional-based trust, corporate identity and satisfaction as the basis of examining CL. There may be other factors that impact clients’ satisfaction within the M-insurance industry. Future research should seek to explore these factors.