Abstract
Sankaran, R., & Chakraborty, S. (2021). Factors impacting mobile banking in India: Empirical approach extending UTAUT2 with perceived value and trust. IIM Kozhikode Society & Management Review. Advance online publication.
Ahead of Print article withdrawn by publisher.
Due to an administrative error, this article was accidentally published Online First and in Volume 11, Issue 1 with different DOIs.
The correct and citable version of the article remains:
Sankaran, R., & Chakraborty, S. (2022). Factors impacting mobile banking in India: Empirical approach extending UTAUT2 with perceived value and trust. IIM Kozhikode Society & Management Review, 11(1), 7–24.
Introduction
Digital transaction is the future and this is a well-known fact around the world. A revolution is happening right now in India, the way consumers shop, and the move towards digital money. The ubiquitous nature of mobile devices is inevitable today in people’s lives. Mobile phone subscribers in India are greater than 1.176b (TRAI, 2019) with a teledensity of 89 per cent (or the population in possession of a mobile phone) and 478m mobile internet users as of June-2018 (IMAI, 2017). World Bank reports that the number of unbanked who possess a mobile phone in India is greater than 50 per cent, and people with a savings bank account only doubled in 2017 when compared to 2011 (Demirguc-Kunt et al., 2018).
India’s move towards becoming a digital economy projects a tremendous growth inspite of an estimated 80 per cent cash transactions (ACI, 2018). Mobile technology has a far-reaching impact on mobile financial services (MFS), be it payment or the banking industry worldwide (Gupta, 2013), and has no temporal and spatial constraints for financial transaction using mobile devices. The transformation of new services using mobile technology has improved the lives of millions in an emerging economy (Barrett et al., 2015).
Mobile banking (mBanking) is part of mobile financial services delivering application-based banking services through mobile phones and is used as a channel for offering financial services (Barnes & Corbitt, 2003). In the context of this study, mBanking can be defined as a service provided by banks, mobile network operators and financial organizations for conducting financial transactions (opening fixed deposit, paying utility bills and transferring funds to another bank account) and non-financial transactions (checking account balance) using a mobile phone of the consumer (Shaikh & Karjaluoto, 2015). In India, mBanking services are facilitated through an app or through National Unified Unstructured Supplementary Service Data (USSD) platform, which allows mobile phone users to access banking services on their feature phones. mBanking can be viewed as an alternative banking platform. Studies have posited challenges affecting mBanking adoption due to the involvement of third-party with outsourced functions like call centre (Tam & Oliveira, 2016), third-party guarantees (Kim et al., 2009), vulnerability to interceptions (Kim et al., 2009; Zhou et al., 2010).
In India, in spite of 473 banks permitted by Reserve Bank of India (RBI) for providing mBanking services (RBI, 2019a), digital payment stands out for its low use (Demirguc-Kunt et al., 2018). India is an emerging economy and digital payment is still in a nascent stage despite several initiatives being taken by Government of India to promote digital payments in the country such as incentivizing digitization of payments with RBI waiving off the transaction fee charges on using the preferred digital routes for funds transfer (RBI, 2019b). Demonetization of 500 and 1000 currency notes in India during the month of November 2016 helped fast-track digital payments (Sankaran & Chakraborty, 2020), bypassing the personal computer revolution entirely and leapfrogging to the digital payment ecosystem. A digital mindset was created with the advancements in mobile technology (4G), availability of low-cost smartphones and affordable data plans.
Technology acceptance model (TAM) and Unified Theory of Acceptance and Use of Technology (UTAUT) are mature and influential theories, which authors have recommended to advance to the next level of conceptualization (Benbasat & Barki, 2007; Venkatesh, 2015). Combining theoretical models or factors that influence consumer’s intent to adopt information and communication technologies (ICT) and MFS can establish a more comprehensive view than any of the theoretical models can offer (Oliveira et al., 2014; Pavlou & Fygenson, 2006). Though UTAUT2 was empirically tested, several authors (Burton-Jones & Straub Jr., 2006; Venkatesh et al., 2012) have emphasized that future research would include more structural elements related to users.
The developers of the conceptual model of perceived value (PERVAL) empirically tested the constructs to study consumer durable goods (Sweeney & Soutar, 2001). Despite the inclusion of PERVAL components to study mBanking worldwide, it has been discussed only in a few studies (Berraies et al., 2017). Also, only a few studies have included the effect of moderation variables (Oliveira et al., 2014; Venkatesh et al., 2016) in order to understand the strength of structural relationships due to the complexity in interaction effects (Strasheim, 2014).
Several empirical studies have highlighted the reasons for technology adoption (Alalwan et al., 2017; Baptista & Oliveira, 2015, 2017; Berraies et al., 2017; Oliveira et al., 2014) as performance expectancy (PE), effort expectancy (EE), facilitating conditions (FC), perceived value and trust. Similarly, the reasons for resistance to adopting mBanking are pointed out as perception of cost, risk, relative advantage and complexity (Cruz et al., 2010); tradition and image barriers affect usage, value and risk barriers (Chaouali & Souiden, 2019); information and guidance offered by a bank followed by usage, image, value and risk barriers (Laukkanen & Kiviniemi, 2010); and investigating risk factors of non-users (Akturan & Tezcan, 2012).
Despite the inherent benefits of the dispersion of mobile phones and digital payments, there is a reluctance amongst consumers to adopt mobile payments (Sankaran & Chakraborty, 2020; Thakur & Srivastava, 2014). Moreover, there is a scarcity of research on the factors influencing mobile banking adoption (Makanyeza, 2017) among the urban population in developing countries. This research envisages to examine the factors for mBanking adoption due to its low use (only one digital payment reported to have been made in 2017 by one-third of account owners in the country) (Demirguc-Kunt et al., 2018), although urban areas have higher subscriber base of mobile phones (TRAI, 2019). This study aims at bridging these research gaps in the following ways: (a) extending the UTAUT2 model with social value (SV), monetary value (MV), emotional value (EV), quality value (QV) and trust; (b) examining the relationship of these factors and the moderation effect of gender on mBanking adoption in India.
The research undertakes to answer: (RQ1) What are the various factors that influence mBanking in India? (RQ2) Does perceived value affect the adoption of mBanking in India? (RQ3) Does trust influence consumers’ use of mBanking in India? (RQ4) Does the moderator gender have a causal relationship on mBanking adoption?
The research objectives are: (a) To examine the factors impacting mBanking use by consumers in India; (b) To predict a model for intention to use mBanking in India by various factors obtained.
Literature Review
The vast span of the literature indicates that models on technology adoption and use can be classified as: (a) foundational models that describe the theory and aspects of adoption of innovations (Aswani et al., 2018); (b) models extended by researchers by adding constructs to adapt to their research study (Natarajan et al., 2017); (c) combination of two or more models to arrive at a working model and empirically testing its significance, which offers a comprehensive view (Oliveira et al., 2014; Pavlou & Fygenson, 2006).
Technology Adoption Models
For over three decades, various models were adapted from information systems, sociology, psychology and theories of human behaviour, which predict technology adoption and use.
Origin of UTAUT
Looking upstream, the conceptualization of the Theory of Reasoned Action (TRA; Fishbein & Ajzen, 1975) is from the domain of social psychology principles that attitude is a function of (a) strength of belief about an act and (b) the evaluative aspect of these beliefs. TAM is an adaption from TRA with perceived ease of use (PEOU) and perceived usefulness (PU) as the antecedents of behavioural intent (BI) and actual system use (Davis, 1989).
Unified Theory of Acceptance and Use of Technology (UTAUT)
UTAUT was formulated as a unified model derived from the theoretical integration of elements across earlier models (TRA and TAM) and was empirically tested (Venkatesh et al., 2003). Four core constructs (out of seven) played a major role as determinants of BI. These constructs are effort expectancy (EE), social influence (SI), performance expectancy (PE) and facilitating conditions (FC). UTAUT combines the concept of PU as PE and PEOU as EE derived from TAM.
Extended Unified Theory of Acceptance and Use of Technology (UTAUT2)
UTAUT2 model is an extension of UTAUT with the addition of three endogenous constructs: price value (PV), habit (HT) and hedonic motivation (HM) (Venkatesh et al., 2012). The empirical test of this model indicates R2 results, showing an improvement (as compared to UTAUT) from 56 to 74 per cent in BI. By realizing the benefits of technology more than the cost (Venkatesh et al., 2012), it implies a positive effect on technology use. UTAUT and TAM were empirically tested from the context of technology (email usage in the organization, various industries, including telecom services, banking and financial services). Since the introduction of TAM, UTAUT2 is the first technology adoption model in the past three decades that the authors have empirically tested for consumer use.
Well-established theories of information systems (IS) (Moore & Benbasat, 1991) measure the perception of an individual and the role of intention as a predictor of behaviour (Ajzen, 1991; Taylor & Todd, 1995; Venkatesh et al., 2003). From the perspective of this study, BI means predicting intention to use or to make mBanking transactions.
Trust
Trust is defined as ‘the perceived credibility and benevolence of a target of trust‘ (Doney & Cannon, 1997). This definition is relevant in the mBanking context. A consumer faces some degree of risk while making a financial transaction. A consumer expects a reliable (credible) mobile app that the service provider can offer in the consumers’ best interests (benevolent). Trust is used in various branches of psychology and sociology and is critical to strengthen customer relationship (Lewicki, 2006). UTAUT2 model was tested with and without the trust construct (Alalwan et al., 2017) to verify the predictive power of the model. The model was able to predict 65 per cent for BI with trust and 59 per cent without trust, with an inference that trust contributes to more power in predicting BI together with UTAUT2 constructs. Another study (Chong, 2013b) extended TAM by incorporating trust to determine factors influencing m-commerce adoption. The aspects of trust and security (Kim et al., 2010) have been detailed in the e-payment systems. An integrated model was used to study trust with the TAM model on the online shopping behaviour of students (Gefen et al., 2003).
Perceived Value (PERVAL)
Extant literature yields theories that show different aspects of PERVAL components, since the meaning of value for the customers varies a lot (Zeithaml, 1988). PERVAL is described with four unique value constructs, namely, performance/quality value, emotional value, monetary value and social value (Sweeney & Soutar, 2001). PERVAL was replicated and validated in the UK and USA (Walsh et al., 2014) and was extended with the additional constructs of etrust, esatisfaction and eloyalty on mBanking (Berraies et al., 2017). Further, Peng et al. (2019) examined the relationship between PERVAL and purchase intention on social e-commerce sales promotion.
Consumer Behaviour Related Studies on MFS
This study elucidates several aspects that are useful to understand the key influencing factors that impact mBanking by consumers. Shaw and Sergueeva (2019) modified the UTAUT2 model with constructs such as perceived value and perceived privacy concerns to determine the factors for mobile commerce. TAM continues to be a popular model used for empirical analysis of adoption of technology. There are studies that have extended TAM with trust by comparing the factors for mBanking adoption in Brazil and USA (Malaquias et al., 2018), non-users of mBanking in Turkey extending TAM with the diffusion of innovation (DoI) and perceived risk (Akturan & Tezcan, 2012).
An empirical study in Jordan explores the factors that influence BI and adoption of mBanking in which the key factors were found to be BI and FC (Alalwan et al., 2017). Examining the structural model with and without trust, R2 was found to improve by 6 per cent with the inclusion of the construct trust. Hence, the results indicate trust to be the prime factor in determining the factors that influence mBanking adoption in Jordan. An empirical test conducted in Mozambique (Baptista & Oliveira, 2015) explains the factors affecting mBanking by using an extended model of UTAUT2 with Hofstede cultural moderators. The study (Boonsiritomachai & Pitchayadejanant, 2019) identified the key factors in adopting mBanking to be FC and mobile banking expectancy (MBE), and hedonic motivation was used as a mediator. MBE construct was formulated by merging PE and EE due to high consistency measurement. A comprehensive study by Oliveira et al. (2014) combines three models, that is, UTAUT with task technology fit (TTF) and initial trust model (ITM), to study mBanking in Portugal, showing PE as the most important factor with no moderating factor for age and gender. The factors impacting the adoption of mBanking using TAM and DoI (Deb & Lomo-David, 2014) were used with cluster analysis to understand specific segment intentions and reduce overgeneralization in understanding customers. Additional studies related to MFS adoption in India along with the findings are listed in Appendix (Table A2).
Drawing from the extant literature, perceived value is conceptualized in this study based on PERVAL (Sweeney & Soutar, 2001) and its integration with UTAUT2 (Venkatesh et al., 2012) to develop a comprehensive model to study mBanking in the Indian context. Accordingly, in this study, it was decided to use BI as a dependent variable to predict the use of mBanking. The measure of individual intent is in sync with other studies (Sheppard et al., 1988; Venkatesh et al., 2012).
Moderator
As revealed by extant literature, gender difference (male/female) in decision making (Bem & Allen, 1974) constitutes social aspects that influence behaviour (Gefen et al., 2003), as men are task-oriented (Minton & Schneider, 1980). Researchers (Nysveen & Pedersen, 2005) have studied the role of gender in technology adoption. Task orientation is more in men than women and seems to be process oriented (Venkatesh & Morris, 2000). Male and female display different notions in several dimensions (Wolin & Korgaonkar, 2003). Several studies underscore that male and female exhibit differences in the measures of perceived trust, and also infer that females display reduced levels of trust during internet transactions (Nysveen & Pedersen, 2005).
The importance of moderators has been emphasized in various studies, which help explain the relationship between dependent and independent variables (Baron & Kenny, 1986; Walsh et al., 2008). Baptista and Oliveira (2015) excluded moderators’ age and gender from the analysis due to resulting complexity and only a small incremental improvement in BI variance. This study examines the moderator role of gender based on extant literature and its usage in various studies (Baptista & Oliveira, 2015; Berraies et al., 2017; Gupta, Dasgupta, & Gupta, 2008; Natarajan et al., 2017).
Conceptual Model and Hypotheses
The conceptual model and the working constructs are given in Figure 1.
Performance expectancy (PE) measures usefulness/helpfulness and seamless experience of consumers across various interactions in performing banking activities, and also the flexibility of the mBanking application (Venkatesh et al., 2012). In our research, the consumer perceives flexibility offered by the mBanking app to perform banking transactions free of temporal constraints (Hanafizadeh et al., 2014), mobility (Deb & Lomo-David, 2014), leading to accomplishing things quickly, which, in particular, is important to a consumer (Venkatesh et al., 2012). Thus, consumers would adopt mBanking if they find mBanking app services beneficial (Oliveira et al., 2014). In UTAUT, PE has a strong influence in predicting BI, but beta value is reduced in UTAUT2. Studies have found PE-BI to be a significant relationship (Deb & Lomo-David, 2014; Gupta, Manrai, & Goel, 2019; Hanafizadeh et al., 2014; Malaquias & Hwang, 2019; Oliveira et al., 2014). We hypothesize:
Effort expectancy (EE) measures whether using mBanking enhances the ease of use or convenience of the consumers (Sankaran & Chakraborty, 2020; Venkatesh et al., 2012). In the context of this study, EE is the convenience factor offered by the mBanking app in terms of extent of simplicity (Alalwan et al., 2017), user-friendly interface (Deb & Lomo-David, 2014) and ease of learning (Baptista & Oliveira, 2017; Zhou et al., 2010). There is a need to overcome the impediments in the process issues with EE, expected in the nascent stages of a new behaviour. Extant studies (Alalwan et al., 2017; Baptista & Oliveira, 2017; Giovanis et al., 2018; Gupta et al., 2019) have found a significant EE-BI relationship. We hypothesize:
Habit (HB) is a perception-based approach, also reported in psychology research (Limayem et al., 2007). Habit is a prior behaviour (Kim & Malhotra, 2005) and the extent to which people tend to use mBanking app automatically is because of learning (Limayem et al., 2007). Belief influences the behaviour of a person by making an impression that builds a strong habit. Habit embeds an intention that affects behaviour. In UTAUT2, habit is the strongest predictor of BI (Venkatesh et al., 2012). Studies (Baptista & Oliveira, 2015, 2017; Kwateng et al., 2018) have examined HB-BI relationship. We hypothesize:
Perceived value (PERVAL) was replicated and validated in UK and USA (Walsh et al., 2014) with regard to the relationship of mBanking and customers etrust, esatisfaction, eloyalty (Berraies et al., 2017) and social ecommerce (Peng et al., 2019). This study includes the PERVAL construct with the following four components:
Social value (SV) is defined as ‘the utility derived from the product’s ability to enhance social self-concept’ (Sweeney & Soutar, 2001). Though the UTAUT model (Venkatesh et al., 2003) has used social influence (SI), it is how an individual perceives others’ opinions to use the system. TRA and the theory of planned behaviour (TPB) refer to this construct as the subjective norm. Basically, the understanding here is that SV determines the extent of influence others’ opinion has on consumers (Alalwan et al., 2017; Gupta et al., 2019; Venkatesh et al., 2003; Zhou et al., 2010), for example, friends, family and peer groups, and provides the stimulus towards using mBanking apps (Baptista & Oliveira, 2015; Deb & Lomo-David, 2014; Venkatesh et al., 2012).
Monetary value (MV) is ‘the utility derived from the product due to the reduction of its perceived short-term and long-term costs’ (Sweeney & Soutar, 2001). UTAUT2 has used a construct price value, which is the variation between perceived benefits and perceived cost. The cost or the price value has an important role as a stimulus for intent to use (Venkatesh et al., 2012), as part of UTAUT2 and is a construct defined as the monetary value in PERVAL. In our research, MV is a cognitive trade-off (Kwateng et al., 2018; Sankaran & Chakraborty, 2020; Venkatesh et al., 2012), which means more cost savings or the ability to obtain a better product or service in exchange for the cost and benefits perceived by the customer, it can have an impact on the intended use of mBanking. MV, from a consumer’s perspective, is an important factor as the costs of smartphones and mobile data need to be borne by them (Alalwan et al., 2017; Baptista & Oliveira, 2015). We hypothesise:
Emotional value (EV) is ‘the utility derived from the feelings or affective states that a product generates’ (Sweeney & Soutar, 2001). Here, EV is an extent to which a consumer will have the pleasure to use the mBanking app. A few empirical research was conducted to study mBanking by extending the UTAUT2 model using gamification (GAM) (Baptista & Oliveira, 2017), resulting in a strong relationship between gamification (for hedonic oriented) and intention. Another study (Shareef et al., 2018) used the theoretical concept of GAM model in Bangladesh. The hedonism achieved with the usage of mobile phone (Baptista & Oliveira, 2017; Shareef et al., 2018) using the construct hedonic motivation or EV in this study will be a useful determinant of BI. A higher hedonism will lead to greater acceptance intention (Baptista & Oliveira, 2015, 2017).
Quality value (QV) is ‘the utility derived from the perceived quality and the expected performance of the product’ (Sweeney & Soutar, 2001). The importance of quality has been emphasized to affect the ROI and its elusiveness (Parasuraman et al., 1988). The evaluation of service quality is defined as a comparative process known as ‘disconfirmation paradigm’ (Oliver, 1980). Consumers compare the service they expect with technical (what), functional (manner) quality types (Gronroos, 1984), using a multi-stage model (Bolton & Drew, 1991), which reveals that service quality leads to service value (Parasuraman et al., 1988). From this study’s perspective, QV is the expected performance perceived by a consumer in using mBanking app.
Trust (TR) is a construct used in studies concerning financial transactions, be it electronic or mobile commerce/mBanking/mobile payment solutions (Sankaran & Chakraborty, 2020). Dahlberg et al. (2015) criticized that security and trust have been repeatedly used in all studies and cited these factors as a prerequisite and as principal theoretical contributions. However, another study (Baptista & Oliveira, 2015) did not include trust in the model, and it resulted in BI not being significant. From this study’s perspective, trust is built through safety mechanism (Gefen et al., 2003), and it reduces uncertainty (Giovanis et al., 2018) and the consumer’s willingness to depend (Alalwan et al., 2017) on mBanking transaction. In TAM meta-analysis (Wu et al., 2011), trust was identified as an important component used in various studies that found the results to be significant (Alalwan et al., 2017; Gefen et al., 2003; Giovanis et al., 2018; Wu et al., 2011). We hypothesize:
Moderation effect: The presence of moderation changes the relationship between the independent and the dependent variable, which is modelled as an interaction (Strasheim, 2014). This study examines the interaction of the dichotomous variable, gender (male and female). The perceived usefulness of mBanking is found to a greater extent in men than women (Nysveen & Pedersen, 2005), which can be attributed to their being more focused on the task and goal (Cruz et al., 2010). Some researchers (Garbarino & Strahilevitz, 2004) have shown that females are more apprehensive during a mobile transaction, while others (Koenig-Lewis et al., 2010; Laukkanen & Pasanen, 2008) have inferred that males are more likely to use mBanking. There is an influence of the gender moderator on BI, and we hypothesize:
Sampling Procedure
The primary data was collected using both online and offline mode (people coming out of the bank) or intercept sampling (Link, 2018), ensuring that the criteria is met (whether the respondents use mBanking). Offline mode was used to collect data from respondents, including those who were not connected to emails. Online respondents included those who were conducting m-banking transactions and were conversant with mobile (technological devices). They were recruited using emails (volunteer sampling), a non-probabilistic convenience sampling, which is an attractive option (Link, 2018). Convenience sampling has been chosen to draw cross-cultural and cross-regional samples (Zikmund et al., 2016) from various urban regions of India (north, east, west and south), including Bengaluru, Chennai, Delhi, Kolkata, Mumbai, Hyderabad and Ranchi.
Instrument Development
The survey instrument or questionnaire consists of two parts, namely demographic profile and scale items for constructs.
Scale items: The questionnaire includes four items per construct totalling 40 items for ten constructs, adopting a good practice of parsimony (Hair et al., 2010). The questionnaire (refer to Table A1) has been developed with the help of extant research (Gefen et al., 2003; Sweeney & Soutar, 2001; Venkatesh et al., 2012), complying with pre-validated measures. Scale items for UTAUT2 constructs EE, PE, HB, BI were adopted from Venkatesh et al. (2012), the construct PERVAL (comprising of SV, EV, QV, MV) from Sweeney and Soutar (2001) and trust from Doney and Cannon (1997) and Gefen et al. (2003) and were rephrased in the context of mBanking in India.
Measurement scale: The UTAUT2 extensively used in various contexts across various countries had satisfactory validity and reliability. A 7-point Likert scale (7=strongly agree to 1=strongly disagree) was used to measure the responses. The scale psychometric properties were empirically tested and validated from the mobile internet context by the developer of UTAUT2 (Venkatesh et al., 2012).
Face Validity and Content Validity
The validity was examined ensuring that the questions are clear, unambiguous and can be easily comprehended by respondents. It was validated with the help of a pilot study before beginning the actual data collection phase.
Data Collection
The primary data was collected from 466 respondents using both online and offline modes. Employing a digital mechanism (respondents using a computer and now a mobile phone) is a frequented approach used in an online survey and offers various advantages (Evans & Mathur, 2018).
Data validation checks were performed to assess plausibility of data to ensure consistency within the elements of the data set, which includes data integrity and validity of entries (valid data type and numerical range), no missing data, no duplicate entries, data validation rules and completeness checks on the content of column (gender includes male or female; age group; income; education level; city name is a valid name; use mBanking – Yes/No; and items with Likert scale are between 1 to 7).
Usable sample: Nine respondents were excluded from the dataset due to unengaged response, incomplete data and for not using mBanking, resulting in 457 usable responses. Data adequacy assumes that the sample size required falls in the range of five times to ten times the number of items (Hair et al., 2010). Four hundred and fifty-seven respondents meet the recommended stipulation of sample size (n>150) if a dataset has several high factor loading scores (>.8) (Guadagnoli & Velicer, 1988).
Data Analysis and Results
A two-stage methodology was adopted to test the measurement and structural model using IBM SPSS and AMOS (Anderson & Gerbing, 1988).
Descriptive Statistics
Out of the usable responses, 64 per cent were male and 36 per cent were female. The overall data collected is a true representation of various regions in India, which consists of 50 per cent students and 50 per cent others (working professionals and retired consumers), and accordingly the distribution of income (students less than INR 20k). The educational qualification of the respondents is bachelor and higher (99%).
Instrument Validity
Cronbach alpha, composite reliability, convergent validity, AVE and discriminant validity were used to test the validity and reliability of the survey instrument.
Non-response bias
Survey methods pose non-response bias and an independent sample t-test was used to compare the respondents (early wave and late wave) (Armstrong & Overton, 1977) with demographic variables (Sivathanu, 2018; Wang, 2012). The results of wave analysis indicated that non-response bias was not a serious concern in this study.
Normality
Normality for each indicator was tested using skewness-kurtosis in SPSS. The skewness values ranged from benign to 1.47 and are below the recommended value <3, and kurtosis values ranged from benign to 7.2 and are within recommended value <8 (Kline, 2015).
Reliability
Cronbach alpha for nine constructs resulted in a low correlation for one item QV4R (-.08). The suggested correlation value for each item should be greater than .4 (Leech et al., 2015). Cronbach alpha for QV improved to .86 with item QV4R deleted. The resulting value of Cronbach alpha (>.82) for all constructs PE, EE, HB, SV, QV, EV, MV, TR and BI indicates good internal consistency (Lance et al., 2006; Nunnally, 1978).
Measurement Model
KMO measure of sampling adequacy
The KMO value obtained was .94, which is well within the acceptable range (Kaiser & Rice, 1974). Bartlett’s test of sphericity under the null hypothesis (H0) follows chi-square distribution and the significance was found to be <.05, which indicates that correlation exists amongst the variables and, hence, factor analysis can be performed (Hair et al., 2010).
Factor analysis
Factor analysis was performed using Promax rotation, thereby extracting nine factors for the items. Subsequently, the items removed are HB4 due to low factor loading. The final rotated component matrix (Table 1) ascertains that the factor loading has been obtained (>.58 for all items) as recommended (Bagozzi & Yi, 1988) and distinctiveness of factors have been established.
Rotated Component Matrix
Rotated Component Matrix
(CFA) is used to test the measurement model (Anderson & Gerbing, 1988). Common latent factor (CLF) method was used to identify the common variance among all variables in this research model (Podsakoff et al., 2003) in AMOS. By comparing the standardized regression weights from the model to a model without the CLF, all were found to be greater than .7 (MacKenzie & Podsakoff, 2012). This result implies the absence of common method bias (CMB).
Discriminant validity
has been examined using the factor correlation matrix to identify the magnitude to which factors are distinct, and the correlations between factors are found to be not exceeding the suggested value .7 (Kline, 2015). The validity of the measurement model was performed in AMOS and the result was obtained (Table 2) for CR>.7, AVE>.5 and discriminant validity when doing CFA (Fornell & Larcker, 1981; Hair et al., 2010).
The diagonal values (given in bold) in the Table 2 represents the square root of AVEs, and was found to be greater than its off-diagonal values for each factor, which represents that the factors are able to discriminate among each other, indicating that they are distinct. In accordance with Fornell and Larcker (1981) criterion, square root of AVE of each construct must be greater than the inter-construct correlations, which is satisfied in Table 2. The values found in this study (closeness of diagonal values and off-diagonal values) are similar to other studies (Baptista & Oliveira, 2017; Chong, 2013a; Oliveira et al., 2014; Oliveira et al., 2016; Ramadan & Aita, 2018; Singh, Sinha, & Liebana-cabanillas, 2020; Sivathanu, 2018). Also, it can be noticed that the items of each construct have higher loading to its construct than on other construct and, hence, discriminant validity is established.
Convergent and Discriminant Validity
Convergent and Discriminant Validity
To summarize, the tests validate high reliability and internal consistency, which are necessary conditions for scales construct validity.
The overall model fit indices CMIN/df=2.50, GFI=.86, CFI=.94, RMSEA=.05 and SRMR=.05 were found to be satisfactory as per the recommended thresholds (CMIN/df<=3; GFI>.7; CFI>.9; RMSEA .05-.10; SRMR<.08) (Bentler, 1990; Browne & Cudeck, 1992; Hair et al., 2010; Hu & Bentler, 1999; Joreskog & Sorbom, 1993; Schreiber et al., 2006; Tanaka & Huba, 1985).
Structural Model
The latent model is detailed in Figure 2. The variance inflation factors (VIF) was examined on dependent variables and observed that no VIFs were found greater than 4 and are below the recommended threshold of 10 (Hair et al., 2010).
Path Model (Latent Variable)
Path Model (Latent Variable)
Testing of hypotheses: From test results (Figure 2), hypothesis H2 or EE-BI (β=.13, p<.001), H3 or HB-BI (β=.21, p<.001), H5 or MV-BI (β=.36, p<.001), H8 or TR-BI (β=.32, p<.001), and H6 or EV-BI (β=.10, p<.05) have a significant positive relationship, H7 or QV-BI (β=-.08, p<.05) has a significant negative relationship, and H1 or PE-BI (β=.05, p=.16) and H4 or SV-BI (β=-.03, p=.49) were found to be not significant. The total variance or R2 explained by the model is 78 per cent.
Analysis of moderator: The effect of moderation is based on the concept that the hypothesis is supported if the interaction is significant (Baron & Kenny, 1986; Hayes, 2017). The study found hypothesis H9b (EE x Gender) supported with the moderation effect stronger for males (coefficient -.18, p<.05) and hypothesis H9e (MV x Gender) supported with the moderation effect stronger for males (coefficient -.16, p<.01) as compared to females. As described in Table 3, other hypotheses (H9c, H9f, H9g and H9h) using gender as a moderator with HB-BI, EV-BI, QV-BI, and TR-BI were not significant.
Hypothesis Test Results
Based on results obtained, this study highlights the following research objectives: (a) examining the factors EE, HB, MV, EV, TR has a positive significant relationship with BI, whereas the factor QV has a negative significant relationship with BI and PE-BI, SV-BI was not significant; (b) these factors are considered important for consumers to adopt and to continue to use mBanking in India; (c) the moderator gender had a stronger effect on EE and MV for males. The path coefficients PE-BI was not significant and this result is similar to other studies (Boonsiritomachai & Pitchayadejanant, 2019; Kwateng et al., 2018; Verkijika, 2018). This implies that consumers find alternate ways of banking (internet banking, ATM, m-payments or cash) more advantageous. The study found a substantial impact of PE-BI (Oliveira et al., 2014) whereas PE-BI was found to be weak as compared to other constructs of UTAUT (Gupta et al., 2019). Effort expectancy is based on the expectations of easy to use (Venkatesh et al., 2012), easy to learn and essential with the realization of these expectations. The results obtained in the study had similar pattern observed in other studies with a significant positive relationship (Alalwan et al., 2017; Giovanis et al., 2018; Gupta et al., 2019), wherein consumers would advance towards forming a habit related to increased frequency of transactions to use mBanking (Baptista & Oliveira, 2017).
Habit is influenced by the ubiquitous nature of mobile phones. Though the results obtained for HB was significant and similar to other studies (Baptista & Oliveira, 2015, 2017; Kwateng et al., 2018), the beta value was quite low from which it can be inferred that (a) people at large use mobile phones for watching video content and as a networking sharing tool and (b) there are other considerations like non-monetary value that consumers could evaluate. The role of non-monetary costs could be an area to assess whether to use or not to use mBanking and may, at times, be concerned compared to monetary price. Habit was not included in the study (Alalwan et al., 2017) as mBanking adoption was sluggish and still considered a new technology in Jordan.
Trust is the second most important factor to influence consumers’ behavioural intent as revealed in studies by Alalwan et al. (2017), Giovanis et al. (2018) and Kwateng et al. (2018). The probable cause of this outcome is attributed to high penetration and use of mobile phones, making mBanking easy to use. The app developers should ensure that transactions are secure and no threat is imposed on consumers. Therefore, implementing trust mechanisms will affect the consumer’s usage of mBanking, while enabling multiple levels of security will provide trustworthiness and belief that safety mechanism is inbuilt (Gefen et al., 2003).
Monetary value was the most significant factor, having a positive relationship with behavioural intent, which implies that there are various intrinsic motivation factors that consumers consider critical to mBanking adoption in India. The influence of societal norms on the consumer is based on the popularity of the technology (Venkatesh et al., 2003, 2012) and recommendations, which influence the intention to start using mBanking. The results are in-line with other studies (Alalwan et al., 2017; Baabdullah et al., 2019; Baptista & Oliveira, 2015; Boonsiritomachai & Pitchayadejanant, 2019; Kwateng et al., 2018; Singh et al., 2018). The results indicate that consumers do not find social value, linked to the influence of friends and family, essential for adoption of mBanking. Consumer perceptions are formed by the service experiences according to the disconfirmation paradigm (Oliver, 1980) and service performed as expected increases perceived pleasure (EV), similar to research by Baptista & Oliveira (2015, 2017) and Shareef et al. (2018). The results of this study indicate that consumer considerations are critical for QV regarding the standard of quality and for mBanking to perform consistently (Sweeney & Soutar, 2001).
If a consumer can obtain their non-financial banking transactions (access to passbook) or perform banking transactions (make utility payments or transfer money to another account) via an mBanking app, they can save time and transportation costs to visit the bank. Our results agree with previous findings (Alalwan et al., 2017; Baptista & Oliveira, 2015; Kwateng et al., 2018; Venkatesh et al., 2012). These stimuli of MV would imply that better the perception a consumer has on the mBanking to help save money, more likely are they to adopt mBanking. MV could also relate to consumers using the technology; mBanking is free, implying a low barrier to start using it.
As mentioned earlier, the effect of moderation is based on the concept that the hypothesis is supported if the interaction is significant. This study found EE x Gender and MV x Gender supported and the moderation effect was stronger for males. This coincides with the rationale that the use of mobile phones for communication purposes holds importance in females while using mobile phones for transactions is a preference among men (Jackson et al., 2008). The introduction of biometric identification cards (known as Aadhar card) during 2014 by Government of India gave boost to account ownership among unbanked consumers and has helped narrow the gender gap (Demirguc-Kunt et al., 2018). Other hypotheses using gender as a moderator with TR-BI, EV-BI, QV-BI and HB-BI were not significant in-line with other studies (Ladhari & Leclerc, 2013). It can be inferred that male and female respondents perceive a similar level of service for the factors trust, habit and perceived value and is not affected by their BI for mBanking adoption in India. The possible explanation for the non-significant effect of moderator lie in that trust, EV, QV and habit components are considered to be of equal importance by both males and females. Hence, consumer gender differentiated approach is not required for mBanking firms to develop trust and value related strategies.
Conclusion and Recommendations
Theoretical Contribution
This study enriches the theoretical contribution in the following ways. First and foremost, it embraces alternative theories on mBanking use and adoption. The espousal of the UTAUT2 extending as part of the conceptual model in this study is a way forward in contributing to the literature and allows for future research to extend theories (Webster & Watson, 2002).
Second, the study has focused on PERVAL components and the results obtained are relevant in the context of mBanking. Furthermore, this study utilizes a comprehensive model to explore mBanking in which it considers new factors such as social value, emotional value, monetary value and quality value of PERVAL and the moderator of these relationships.
Managerial Implications
The study can be useful to banking organizations in formulating strategies not only to attract customers but also to continue to engage and retain them using these influencing factors to adopt mBanking. This is particularly relevant to everyday transactions using mBanking for a variety of utility services.
Numerous banking organizations are entering into this competitive market as MFS has undergone multi-faceted changes, thus paving the way for companies to augment their revenue sources, and they use various novel methods to retain consumers. With the advent of innovations in mobile technology, banks can effectively meet with the ever-growing needs of their customers. This study adds focus on trust to be at the helm for consumers to consider the country’s digital environment. It is time to watch what they do and not just what they say.
The study includes PERVAL components and the results reveal monetary value to be the most important factor for consumers to adopt mBanking. These factors will enable banks to develop services that will be perceived as high quality by consumers while using mBanking for both financial and non-financial transactions.
With trust being identified as the prime factor in this study, banks and banking organizations can use this to strategise their relationship marketing and win customer trust. The absence of trust might lead to the risk of private data being misused. This is crucial for banking transactions.
A World Bank report indicates that in the past year, less than 10 per cent of the account holders in India (an emerging economy) used mobile phone or internet to make a minimum of one financial transaction (Demirguc-Kunt et al., 2018). Hence, the factors identified in this study will enable banking organizations to explore the means of facilitating customers in adopting and using mBanking services.
In this era of monetization in India, this study will give significant insight for policymakers and will help banks and banking organizations to (a) identify the factors that consumers consider important to adopt and to continue to use the mBanking application for financial and non-financial transactions, and (b) strategize due to reduction in footfalls and reduction of opening of new bank branches and ATMs. The convenience factor is not only considered important for consumers, but it also results in reduced costs for vendors and organizations, as payments made in cash transactions need to be counted, stored and transported. Adopting digital payments and mobile banking options to accept cashless payments will save business time and costs associated with the processing of cash.
The factors identified will help banks and financial organizations to (a) view the channel as a cost-saving avenue which will eventually reduce the operational cost and time cost of the user (no need to physically visit the bank), thereby also reducing dependency on bank branch; (b) develop offerings by ensuring all banking services are available via mobile, negating the need to visit the bank; (c) scale up their business even after footfall reduction; (d) be cost-effective and stay efficient by automating the process and minimizing human error; and (e) retain customers, be connected to them and be competitive.
Limitations and Directions for Future Study
This study focused on the middle-class, educated Indian consumers owning a smartphone in the urban regions of India. It will be an important aspect to investigate the banking services using mobile phones for the transfer of funds, including those families not having savings accounts.
The results cannot be generalized to the whole country because convenience sampling was used in this study (Baabdullah et al., 2019; Neuman, 2011). Adequate care is required for generalizing the results to the whole population. Another consideration would be to utilize non-urban region to determine the factors considered important for using mBanking, contributing to the ‘Digital India’ initiative. This study focuses on one type of use of technology (mBanking). A comparison of various MFS (like mCommerce vs mBanking) can lead to interesting findings. Longitudinal surveys were not used, and hence future studies can focus on identifying the change in behaviour of respondents across time.
The study includes trust and PV components as an extension to UTAUT2. Digital payments are still at a nascent stage and are used less in India (Demirguc-Kunt et al., 2018), and with the government initiative, there is growth potential for mBanking (RBI, 2019c). It will become prudent for future researchers to include other factors like satisfaction level (post-use evaluation), loyalty and cost factors, which will gain prominence with the increase in the extent of mBanking penetration. Similarly, previous experience of consumer influences the trust factor for mBanking use, and this study includes respondents who have used mBanking app. Future studies can examine the effect of trust on mBanking use by both users and non-users.
Due to the growth in the number of banks approved to provide mBanking services in India (RBI, 2019a), there is a potential for each bank to offer mBanking services. Accordingly, this study will help banks and MFS organizations to strategize and consider these key factors while offering services to consumers. People will consider these factors important while the mobile technologies spread to the rural parts of the country as part of the ‘Digital India’ movement, and this study will be useful for banks while offering these services in the rural areas. Finally, keeping in view the emerging trend and tremendous growth of digitization in the country, future studies can dwell into the espousal model regarding customers’ satisfaction with the adoption of mobile technology services, as this research examines the consumer’s behavioural intent to adopt mBanking.