An Application of TAM and TRI on the Factors Affecting Internet Banking Adoption in Bangladesh

Abstract

This study assesses the Internet banking adoption tendency by existing bank customers of Bangladesh. Currently, almost all the leading banks in the country have implemented Internet banking platforms. However, the active user count remains relatively low and there hasn't been any conclusive research on the drivers and inhibitors of Internet banking. This study evaluates the reasons and quantitatively establishes the factors leading to the adoption and usage continuance of internet banking by existing bank customers. Responses from 460 bank account holders were collected via online questionnaires using a purposive sampling approach, and a core conceptual framework based on Technology Acceptance Model (TAM) and Technology Readiness Index (TRI) was used. The study concluded that internet banking adoption is significantly impacted by the ease of use, customer service, and technology familiarity. Similarly, customer satisfaction is affected by the perceived value and the perceived risk. Through regression analysis, it was found that usage continuance is 89% explained by adoption and customer satisfaction. Multi-group moderation showed significant impact by groups divided based on usage frequency, income level, and age. Perceived risk weakened the impact of perceived value and technology familiarity on usage adoption. Additionally, perceived risk reduced the impact of consumer satisfaction and usage continuance.

1. Introduction

Bangladesh is a very fast-growing country and the significance of strong banking is immense. Currently, 61 scheduled banks in Bangladesh operate under full control and supervision of Bangladesh Bank which is empowered to do so through the Bangladesh Bank Order, 1972 and Bank Company Act, 1991 (Bangladesh Bank, 2021). With the development of various mobile applications and websites, customers have the resources to make financial transactions with just a few clicks. Almost all the leading banks of Bangladesh have successfully converted their system to digital, which is very sophisticated and driven by a very strong cyber protection system. Hence, the banking system has evolved over time and slowly digital banking is replacing traditional banking.

From the perspective of sustainable development of a country, internet banking plays a very important role. To achieve a sustainable and strong financial system with high inclusion, internet banking adoption is a major contributor. For example, internet banking provides operational excellence and creates opportunities for financial inclusion by expanding access into hard-to-reach customer segments covering wide geographical areas. The customers of a bank also reap the benefits as internet banking ensures extended service hours, and timely service delivery, and also increases consumer trust by guaranteeing information availability and so on. Particularly during the COVID-19 pandemic, banks continued their operation via internet banking channels and provided undisrupted banking services to customers.

Even though it is now convenient for the customers to avail of almost all types of banking services with the help of digital platforms, the usage is still not up to par in Bangladesh. Despite all the efforts made by the banks to create awareness on their respective digital platforms, the conversion rate from traditional to digital banking systems is still low. The study seeks to draw a conclusion and give recommendations to the banking service providers by assessing the drivers and inhibitors of user adoption and customer satisfaction, along with their subsequent effect on the usage continuance of internet banking users in Bangladesh. For this, the study developed and tested a conceptual framework combining TAM and TRI models. The study collected responses from 460 bank account holders via online questionnaires. The study used the variables, namely, Value Perception, Ease of Use, Customer Service, Perceived Risk, and Technology Familiarity, and assessed their impact on Usage Adoption, Customer Satisfaction, and Usage Continuance.

The rest of the sections of the study are as follows. Section 2 consists of the literature review. Section 3 depicts the research design in detail. Section 4 describes the findings and analysis part. Section 5 discusses the practical implications and recommendations of the study. Section 6 describes the scope for future works and the conclusion.

The study achieved the following three outcomes, which add to the literature: (1) Validate the applicability of the TAM and TRI extended model to the banking industry in the context of internet banking (2) Establish a combination of factors that drive adoption, satisfaction, and thus usage continuance of internet banking (3) Ascertain factors moderating the combination mentioned earlier.

2. Literature Review

Recent research on the factors affecting internet banking adoption and its utilization among users mostly shows that the accessibility of the Internet, customer awareness, adaption to change, proper guidance to use internet banking services, trust and goodwill of banks, security concerns, and overall perceived value plays a major role to drive people towards adoption of internet banking services in Bangladesh (Khan, Chowdhury, Haque, Akter, & Ahsan, 2021; Rahaman, Luna, Kejing, Ping, & Taru, 2021; Shareef, Baabdullah, Dutta, Kumar, & Dwivedie, 2018). In Bangladesh, Rahaman et al. (2021) studied internet banking adoption using constructs of TAM, such as perceived usefulness and perceived ease-of-use and added three variables, such as trust, social influence, and perceived enjoyment. Researchers have also used a conceptual model based on TAM with some added aspects such as social image, perceived risk, and perceived trust, adopted from Muñoz-Leiva, Climent, and Liébana-Cabanillas (2017), and studied the adoption behavior of users of digital banking applications in Yogyakarta, Indonesia (Mufarih, Jayadi, & Sugandi, 2020). Another study researched the aspects influencing the intention of using digital banking in Vietnam by considering the constructs such as perceived usefulness and perceived ease-of-use from TAM along with two other variables perceived risk and trust (Nguyen, 2020).

Davis (1989) defined value perception as the strong mindset that consuming a particular product or service would improve one’s need fulfillment. While studying the conceptual model, value perception can also be defined as individuals’ own experience and importance of the internet banking services. Hong, Thong, and Tam (2006) noted that users’ value perception of a new product is a key indicator of satisfaction and continued usage intention.

Perceived ease of use can be defined as the degree to which a prospective adopter expects the new technology to be free from effort (Phillips, Calantone, & Lee, 1994; Davis, 1989). Perceived ease of use has a dual effect, both direct as well as indirect, on consumers’ intention to adopt certain products or services. The indirect effect on intention occurs through perceived usefulness because the easier a technology is to use, the more useful it can be (Venkatesh & Davis, 2000; Dabholkar, Thorpe, & Rentz, 1996; Davis, Bagozzi, & Warshaw, 1989). In studies related to internet banking, it has been found that a higher level of technological complexity corresponds to a lower level of customer’s perceived ease of use, hence lowering the individual behavioral intention to use Internet banking services (Ndubisi & Sinti, 2006; Gerrard, Cunningham, & Devlin, 2006).

Triandis (1977) proposed the theory of personal behavior, which hypothesizes that the utilization of new technology by an individual in a voluntary environment is influenced by their beliefs towards the new technology, social influences, experience or habit, their expected outcome of using the new technology, and facilitating conditions to use the new technology in a conducive environment. Thompson, Higgins, and Howell (1991) further established relationships among different determinants of various personal device usage behavior, such as computers, and found that social factors, technological complexity, job fit, and long-term consequences have a significant impact on user behavior. The measurement of Technology Familiarity requires basic knowledge of new technology, an individual’s awareness of the application of the new technology as well as the facilitating conditions such as demo, training, or user manual that a service provider is providing to the user. It has been found that perceived ease of use positively affects the perceived utility of the new technology and also the attitude towards technology adoption (Perez, Sanchez, Carnicer, & Timenez, 2004).

Perceived risk plays a critical role in affecting individual decision to accept or reject a new technology (Ivanova & Kim, 2022; Ndubisi & Sinti, 2006; Rotchanakitumnuai & Speece; 2003). Aldás-Manzano, Lassala-Navarré, Ruiz-Mafé, and Sanz-Blas (2009) defined perceived risk as the customers’ subjective expectation about financial and/or information loss due to devices and media, such as the Internet. Though it is very difficult to address risk objectively, in the context of internet banking services it is relatively easier to address key potential risk issues that may have an adverse impact on the behavioral intention of an individual to adopt or reject internet banking services (Wang, Wang, Lin, & Tang, 2003). There can be five dimensions of perceived risk associated with internet banking services. The first dimension is a security risk, related to consumers’ fear about the safety of their financial transactions over the Internet. Sathye (1999) in the study of internet banking adoption in Australia, Gerrard and Cunninghham (2003) in the study of adopters and non-adopters in Singapore, Cheng, Lama, and Yeung (2006) in a study of customers’ acceptance of internet banking in Hong Kong have found the security of financial transaction as a key determinant of internet banking adoption. The second dimension is a privacy risk, which reflects the customers’ worry about leakage of their personal details and/or account details to any third parties (Furnell & Karweni, 1999). The third dimension is performance risk. It is concerned with the Customer’s Perception that how well the system has the capability to perform their financial transaction through the Internet effectively and efficiently (Gerrard & Cunninghham, 2003). The fourth dimension is time loss risk i.e., loss of time spent by the customers in dealing with erroneous search and transactions process, filling required information, waiting for the response of apps or website confirmation, internet server and download speed, etc. (Littler & Melanthiou, 2006; Jayawardhena & Foley, 2000). The fifth and final dimension is related to the social risk i.e., risk related to the lack of human interaction and also risk related to the possibility of drawing in some unfavorable attention and response from society or family having negative attitude internet banking technology (Littler & Melanthiou, 2006; Suganthi, Balachandher, & Balachandran, 2001).

Kotler and Keller (2009) defined satisfaction as a person’s feeling of pleasure that results from comparing a product’s perceived performance or outcome with their expectations. Customer satisfaction is the consumer’s response to the evaluation of the perceived discrepancy between prior expectations and the actual performance of the product or service as perceived after its consumption (Tse & Wilton, 1988). It can also be defined as a positive feeling experienced by customers after consumption of a product or service. So, if internet banking services are considered as value additive, easy to use, and little or no exposure to risks, their satisfaction level with the new product or service will be lifter (Thong, Hong, &Tam, 2006). Earlier studies also showed that customer satisfaction is a key driving force for continued usage intention and positive brand attitude (Eriksson & Nilsson, 2007; Thong, Hong, &Tam, 2006).

3. Research Design and Method

3.1. Hypothesis Development

The study used appropriate statistical techniques to assess the effects of each of the constructs that have been linked in the conceptual framework, based on the hypotheses summarized in Table 1 below:

Table 1: Summarized Table of Hypotheses Development

To reiterate the strength of the framework, three moderating factors were incorporated: Usage Frequency, Income, and Age. This will allow a better understanding of the extent these moderating variables change the nature of the respective relationships between the hypothesized predictors and outcomes (Table 2).

Table 2: Hypotheses for Multi-Group Moderators: Usage Frequency, Income, and Age

3.2. Conceptual Framework

The drivers and inhibitors inspired the creation of a hybrid version of the TAM and TRI models. Such an extended version has been implemented in various industries and proved to be an effective tool for measuring user adoption (Lin, Juan, & Lin, 2020; Weng, Yang, Ho, & Su, 2018; Na, Heo, Han, Shin, & Roh, 2022). Technology Acceptance Model or TAM points to the causality between the perceived usefulness of an action or work, as well as the ease of use and the attitude and actual user adoption behavior (Davis, Bagozzi, & Warshaw, 1989). Technology Readiness Index or TRI (Parasuraman, 2000) talks about four factors: Optimism- the degree to which people have a positive view of technology, Innovativeness- the degree to which people are technological pioneers, Discomfort- the degree to which people perceive a lack of control and are overwhelmed by technology and lastly, insecurity- the degree to which people distrust technology and its ability to solve or ease banking needs. Anticipating the impact of these key constructs the proposed framework considers the effect of the perceived value, ease of use, risk, and technological know-how on customer satisfaction, adoption, and ultimately the continuance of use, etc. (Figure 1).

Figure 1: Conceptual Framework of the Study

3.3. Sampling and Data collection

An online survey was conducted on 460 citizens among which 366 were found to be Internet Banking users. In the case of describing the construct, multiple choice questions and a Likert Scale of 1–5 were used, assuming 1 as strongly disagree and 5 as strongly agree (Likert, 1932). As Internet Banking users are scattered throughout the country, the study used the Purposive Sampling method from each of the eight divisions in Bangladesh, spanning a variety of professions. The research deals with those bank account holders who are able to operate their account(s) themselves and surveyed participants from all socioeconomic classes to make the research more comprehensive and inclusive.

3.4. Data Analysis

The study performed the Confirmatory Factor Analysis (CFA) using IBM SPSS AMOS Version 23. It is a widely accepted tool to perform CFA, check the reliability and validity of a conceptual model (Thompson, 2004), and reject the measurement theory. CFA is a multivariate statistical procedure used to test how well the variables measured via the questionnaire represent the concepts being tested. The study also used IBM SPSS Statistics v.25 to analyze the data.

4. Results

4.1. Evaluation of Measurement Model

To validate the conceptual model, the study has used the Confirmatory Factor Analysis (CFA) (Stapleton, 1997). To measure the internal consistency of the variables used to measure the concepts, the study used Cronbach’s alpha, Composite Reliability. To measure the convergent validity, the study used factor loading and the average Variance extracted (Hair, Hult, Ringle, & Sarstedt, 2016). The average Variance extracted can also be used to confirm the discriminant validity. The study used Kaiser-Meyer-Olkin (KMO) measures of sampling adequacy with Barlett’s Test of Sphericity. The summary result of these two tests is presented in Table 3.

Table 3: Kaiser-Meyer-Olkin Measure and Bartlett’s Test of Sphericity

The Value of KMO was 0.937, which implies that the factor analysis is useful for this survey’s 366 internet banking users. Barlett’s test shows the Variance is true before running the further statistical tests (Snedecor & Cochran, 1989).

It is important to validate a conceptual model; a confirmatory factor analysis (internal consistency, convergent validity, and discriminant validity) was conducted.

4.2. Confirmatory Factor Analysis (CFA)

CFA is a multivariate statistical procedure used to test how well the variables, measured via the questionnaire, represent the concepts being tested. The variables are Value Perception, Ease of Use, Customer Service, Perceived Risk, Technology Familiarity, Usage Adoption, Customer Satisfaction, and Usage Continuance. These variables are depicted in Appendix A.

The CFA model, represented in Figure 2, shows the estimation of each of the variables tested with standardized regression weights as well as covariance among the concepts. High standardized regression weights are interpreted as a high correlation between the variables. Similarly, high covariance among the various concepts translates to a high level of correlation among them. However, the study wanted the concepts to be as mutually independent as possible.

Figure 2: CFA Estimation Using Standardized Regression Weights and Covariance

In Table 4, the fit indices of the model were given. These are all Chi-square (χ2) statistics and include Adjusted Goodness of Fit (AGFI), Comparative Fit Index (CFI), Root Mean Square Error (RMSEA), and the Minimum Discrepancy (CMIN/df) (Maat, Adnan, Abdullah, Ahmad, & Puteh, 2015). As shown in the table, the model meets all the respective threshold criteria, and so it can be inferred that this hypothesized model is a good enough fit.

Table 4: Fit Indices of the Model, as per Confirmatory Factor Analysis (CFA)

4.3. Validity and Reliability Test

A convergent validity test is conducted to check the same or similar constructs, which generally have a higher correlation between them. To determine the Convergent Validity, Factor Loading (FL), Average Variance Extracted (AVE), and Maximum Shared Variance (MSV) value was generated.

Factor Loading (FL) is conducted to get the correlation coefficient for the variables and factors, as shown in Figure 2. Ideally, the value should be between 0.4 to 0.7. This was achieved for all the constructs except for Technology Familiarity (0.267) and Customer Service (0.334) as shown in Appendix B.

The low value for these two constructs is most likely for the general customer’s lack of keeping up-to-date with the technological trends and in the case of Customer Service, due to interruption in online transactions using the switches used in the country (Kite & Whitley, 2018).

Average Variance Extracted (AVE) measures the amount of Variance that is captured by the construct in relation to the amount of Variance due to measurement error. The rule-of-thumb for AVE is greater than 0.50 (Fornell & Larcker, 1981), which was achieved in all the constructs except for Customer Service (0.49). It can be inferred that the lower value is due to the same problem as mentioned for the lower value in Factor Loading, which is the risk of a transaction failing is not as low as customers’ expectations (Aldás-Manzano, Lassala-Navarré, Ruiz-Mafé, and Sanz-Blas, 2009; Hair, Hult, Ringle, & Sarstedt, 2016).

Discriminant Validity determines whether the variables are truly distinct from others (Hair, Hult, Ringle, & Sarstedt, 2016). For having Convergent Validity results all permissible, the study ran discriminant validity by square rooting AVE and compared whether the value is greater than the Maximum Shared Value (MSV) (Hair, Black, Babin, Anderson, & Tatham, 2010). In the study five constructs (Value Perception, Ease of Use, Customer Service, Perceived Risk, and Usage Continuance) were found to pass this criterion, indicating that the adequacy of discriminant validity as well (Fornell & Larcker, 1981; Farrell, 2010) whereas, Technology Familiarity, Usage Adoption, and Customer Satisfaction, AVE value is less than the MSV value.

Reliability indicates how consistently the test can measure the construct characteristics. To conduct a Reliability test, Cronbach’s alpha (α) was measured. Cronbach’s alpha indicates the internal consistency by determining the close relatability of the items of the construct. The generally acceptable value of alpha is 0.7 to 0.8 for comparatively good levels with >0.8 indicating an excellent level, however, >0.95 is not necessarily good (Hulin, Netemeyer, & Cudeck, 2001; Ursachi, Horodnic, & Zait, 2015). All the values are in permissible limit 0.7 < α < 0.95. The composite Reliability (CR) value was determined using the component matrix value derived for each parameter. The permissible limit of CR value is between 0.7 to 0.95 (Hair, Hult, Ringle, & Sarstedt, 2016), which also complements the model in Appendix B.

4.4. Application of Structural Equation Model (SEM)

Structural Equation Modeling (SEM) suggested by Malhotra and Dash (2011) implies that with five or fewer constructs with each having at least three variables, a proper model fit can be done with a sample size of at least 100. The study conducted Structural Equation Model (SEM) on 366 Internet banking users to analyze multiple direct and indirect relationships between independent and dependent variables simultaneously (Hooper, Coughlan, & Mullen, 2006; Hair, Black, Babin, Anderson, & Tatham, 2010).

Using “Data Imputation” the exogenous variables were imputed onto an SPSS data file. Using these newly created variables, the model in Figure 3 was prepared. This SEM and its derivatives were used in all of the oncoming calculations, like path analysis.

Figure 3: High-Level Conversion of CFA to SEM Using Data Imputation

4.5. SEM Analysis Results

The Structural Equation Model (SEM) was used to test all the hypotheses from Table 1. Table 5 summarizes the findings of SEM and provides a clear decision on whether a chosen hypothesis is supported or not.

Table 5: Path Estimation Using SEM, Without Moderating Effect

The beta (β) estimate, also known as the regression coefficient is the degree to which change occurs in the dependent variable, for every 1-unit of change in the independent variable. A positive value indicates a positive change, whereas a negative value indicates a negative change. The Critical Ratio (CR) is the ratio of a respective estimate by its standard error. When CR > 1.96 for a regression weight, the estimated path parameter is significant, at a 5% Significance Level (Byrne, 2013).

The coefficient of determination (R²) shows how much the variability of one factor can be caused by its relationship with the other factor being tested. This metric presents an in-sample predictive power (Sarstedt & Mooi, 2014). A value of 0.777 means that 77.7% of Usage Adoption of Internet Banking can be explained by Value Perception, Ease of Use, Perceived Risk, Customer Service, and Technology Familiarity. The R² value is adjusted for predictors that are not statistically significant in the model and it’s a better model evaluator than just the R² value.

From the hypothesis testing endeavor, the study found three hypotheses that are not supported, and the rest of the thirteen were indeed supported at a 5% significance level. Usage Adoption is directly and positively impacted by the Ease of Use, Customer Satisfaction, and Technology Familiarity. It was’s negatively impacted by the Value Perception and Perceived Risk of using Internet Banking. Customer Satisfaction, on the other hand, was positively impacted by Usage Adoption, Perceived Risk, Customer Service, Technology Familiarity, and Usage Adoption. Usage Continuance, as hypothesized, is positively impacted by both Usage Adoption and Customer Satisfaction, with the highest R² value (~88.5%).

An interesting finding is that the hypothesized positive impact claims of Ease of Use on Customer Satisfaction were not supported, as per the calculations (with a P-value of 0.325). This indicates that the customers that were satisfied with using Internet Banking did not do so because of its easy-to-use features. Similarly, there is statistical evidence to support that the Perceived Value and the Risk Perception in Internet Banking did not have enough impact on Usage Adoption, but has a positive effect on Customer Satisfaction.

4.6. Testing for Multi Group Moderating Effect

To test the moderating interaction among the constructs, the study used the categorical variables: usage frequency, income level, and age. Multi-group moderation technique was used (Ketab, Sharif, Mehrabi, & Abdul Rahman, 2019). All three of the moderating variables were categorized into three groups: low, medium, and high, respectively. Path analysis was done using these grouped data. However, this technique involves another extra step of ensuring that the comparison model created using the groups is valid. Only when both the model comparison and the respective path analysis produce statistically significant results, is the tested hypothesis considered valid. Table 6(a), 6(b), and 6(c) summarize the findings after applying the multi-group moderation technique. It provides a clear decision on whether a chosen hypothesis was supported or not.

Table 6: Path Estimation and Model Comparison Using SEM, With Moderating Effect

4.7. Measuring the Interaction Effect using Dawson and Richter Method

To examine the moderating effect via a continuous variable, Baron and Kenny’s moderation analysis (Hayes, 2009) was used. The interaction is created by multiplying standardized values of the exogenous variables and the moderating variable after each were centered to a mean of zero, respectively. The moderator and these interactions together were used to measure the interaction effects among the relationship between the exogenous and endogenous variables. The SEM was introduced with these interaction variables and treated as exogenous variables. Upon path analysis, only the statistically significant paths were kept and the rest were truncated. The resultant SEM is summarized in Figure 4, and the Dawson and Richter (2006) plot is given in Figure 5.

Figure 4: Compilation of SEMs Used to Measure Interaction Effect by Perceived Risk

Figure 5: Dawson Richter Plot

All interactions plotted were significant in nature, however, their respective regression coefficients varied in weight and nature. Higher β estimate magnitude is characterized by higher weight in the moderation effect. A positive β estimate indicates a positive slope and the opposite for negative values.

Perceived Risk dampens the relationship between Usage Adoption-Technology familiarity (β = –0.051, p < ***), as indicated by the decreasing slope at high risk. The same is the case for customer satisfaction and user adoption (β = –0.089, p < ***). Perceived risk also weakly dampens the relationship between usage continuance and usage adoption (β = –0.022, p < ***). The study encountered some seemingly odd findings in the form of the relationships between usage adoption-perceived value (β = 0.042, p < ***) and customer satisfaction-perceived-value (β = 0.087, p < ***). In both cases, the higher values of user adoption and customer satisfaction are found in high-risk groups. This was characterized by the increasing slope in the high-risk groups, respectively.

5. Practical Implications and Recommendations

This study has scrutinized the Internet banking adoption tendency of existing customers using various statistical analyses. CFA was used to validate the conceptual framework and SEM to test the hypotheses. While unmoderated, it was found that Usage Adoption conforms to the hypotheses of being positively impacted by the Ease of Use, Customer Service, and Technology Familiarity. Customer Satisfaction, on the other hand is positively impacted by Perceived Value, Customer Service, Technology Familiarity, and Usage Adoption. It was negatively affected by the Perceived Risk. Usage Continuance was confirmed to be positively impacted by both Usage Adoption and Customer Satisfaction. While moderated (using multi-group interaction), it was found that usage frequency, income level, and age all moderated all the 21 different combinations of these IV-DV relationships. Interaction using the continuous variable Perceived Risk also produced intriguing results that showed damping effects in Adoption-Technology familiarity, Customer Satisfaction-Usage Adoption, and Usage Continuance-Usage Adoption.

So, for someone to adopt internet banking, they must find it easy to use the application, find the customer service acceptable and have experience using web or mobile-based applications (Lin, Juan, & Lin, 2020). A customer will likely be satisfied with using internet banking if the technology adds a certain level of value to their lives, does not find much risk in using the application and all the rest as needed to adopt the technology. If a customer adopts internet banking and is satisfied, they will continue using the technology. The usage frequency, income level, and age have been found to control how customers adapt, achieve satisfaction, and continue using the technology. The risk one perceives is another vital factor that negatively affects one’s chances to adopt, be satisfied and thus continue to use internet banking.

To substantiate the findings of this study, a Focus Group Discussion (FGD) was done with industry experts from top commercial banks in Bangladesh. The group included personnel who lead the departments or sub-departments that comprise Internet banking, with wide experience in the field. The subject matter experts unanimously agreed with the findings from Table 6. Furthermore, the FGD demystified the seemingly confusing results in Figure 5(a). The groups perceiving higher risks are the ones who adopt and are more satisfied with their internet banking. The underlying reason, agreed upon by the experts, is that the perceived risk is the opportunity cost the customers bear to obtain the ability to make online transactions as well as avail of any other associating facilities. Customers are aware that if their account credentials are compromised, they are highly likely to fall victim to fraud. This risk realisation is greater among users who find greater value in the internet banking services being used. Lastly, the dampening effect of perceived risk in Figures 4, 5(b), and 5(c)were also declared rational by the group.

The data and in-depth analysis point towards a few causal relationships and recommendations. Firstly, to get customers to adopt or start using internet banking, banks must make them perceive ease and work on customer service. Current marketing tactics involve witty and promotional videos elucidating the easy-to-use application interface. To keep a customer satisfied, banks need to again put emphasis on customer service and ensure that customers can do most of their transactional necessities via the internet banking application. They must also ensure positioning the technology as full-proof and thus risk-free. By guaranteeing seamless transactions and minimum transaction failures, banks will be able to build confidence among the customers. Thus, reliable switch technology needs to be implemented by the central bank is necessary to achieve better Customer Service and enhance Customer Experience. The study shows the risk of a transaction failing in customers’ minds is one of the reasons the banks are not meeting the customer’s service expectations. Lastly, banks could try appealing to certain age groups and income levels to better penetrate the market.

The following summarizes the author’s and industry experts’ recommendations for increased adoption of internet banking:

• “Trust” can be considered a primary construct as it is one of the major determinant for Internet Banking Adoption.

• Banks should focus on marketing activities, short videos on key Internet Banking features, and adding more value-added services to enrich Internet banking and reduce silent attrition.

• Branch banking should promote and support internet banking more rigorously as the more internet banking usage will increase, the better cost-efficient the bank will become, and the more the banks can reduce operating costs and utilize the low per transaction cost channel to cater to larger demography.

• Investment in Internet banking is a must from central banks to all individual banks for more robust dispute-free transactions, quicker resolution, and hassle-free transactions.

• Uniform account numbers for all banks, and interoperable Internet banking to curtail the dispute by traditional channels such as National Payment Switch Bangladesh (NPSB).

• An easy onboarding process to decrease customer attrition is a must for better penetration into the market.

Educating customers about digital banking and basic security measures about not sharing PINs, passwords, OTPs, etc. should be considered an additional Corporate Social Responsibilities (CSR) for banks to protect customers’ interests and gain their trust and confidence in the Internet banking.

6. Conclusion

This research seeks to understand why and how bank account holders use internet banking. As per literature and market experience, a comprehensive framework was built which laid the foundation of this research. The framework was judged valid and upon in-depth analysis using various statistical tests, conclusions could be made about the extent to which the hypothesized factors affected customer satisfaction, internet banking adoption, and finally, the continuation of use. It was shown that Internet banking is positively impacted by both Usage Adoption and Customer Satisfaction, with the highest R², i.e., 89%. This means that Usage Continuance is 89% explained by the Usage Adoption and the Customer Satisfaction concepts.

Although there is still a portion of the market that remains untapped, there are people within this demographic, willing to avail of Internet Banking. This presents a new prospect of analysis among non-adopters on the factors affecting the non-adoption of internet banking. Working with additional endogenous and exogenous factors on top of the existing framework will build on its robustness.

The study was done using the purposive sampling method; however, due to the COVID-19 pandemic, it struggled to reach deep into many factions of the population, like remittance earners, garments workers, etc. The study mostly focuses on the key drivers and inhibitors towards adoption and, ultimately, the continuance of the use. For future research, the age group of 25 to 50 can be further grouped to find out the relation between age and adoption.

Appendix A

Appendix B