Study of Customer Satisfaction Towards E-Wallet Payment System in Bangladesh

The American Journal of Economics and Business Innovation (AJEBI) is a blind peer-reviewed international journal publishing article that emphasizes research, development, and application within the fields of business, economics, finance, marketing, management, business analytics, etc. The AJBEI reviews papers within approximately one month of submission and publishes accepted articles online immediately upon receiving the final versions. Payment systems have been enormously switched out by introducing a new dimension in fintech where e-wallets can be used in conjunction with mobile payment. The severe com petition of e-wallet services has forced providers where satisfaction is of prime concern. A total of 480 data were obtained from the respondents living in Dhaka city. The structure of this study was developed by approaching the TAM model, and Structural Equation Model ing was applied to examine all the hypotheses. Results revealed that technology self-efficacy is one of the exigent factors of satisfaction where a positive relationship exists in between. All the hypotheses were found to be significant except the relationship between perceived usefulness and satisfaction. This study has validated external variables to contribute to the existing theory based on the previous literature. Lastly, in order to promote the enhancement of the mobile payment system, proposals for developing e-payments were made to increase the degree of satisfaction.


INTRODUCTION
The electronic wallet, also known as an e-wallet, has widely been used to reduce cash carriage hassling. An E-wallet is a form of an electronic card that is used via a digital device such as a smartphone or computer to make transactions online. In order to make purchases, it requires a debit or credit card to be linked with e-wallet applications (Digital Wallet, 2019). Consumers could use the new digital payment system to pay utility bills and move money to pay for financial transactions. In addition to debit or credit cards, electronic cards allow customers to store information about their physical cards and bank account numbers in order to make certain payment steps (Ray, 2017). For multi-channel transactions such as consumer-to-consumer (C2C), consumer-to-business (C2B), consumer-to-machine and consumer-to-online, consumers may use mobile wallets (Lee, 2019). The digital wallet concept is similar to the concept of mobile banking and online banking. A personal folder containing customer details required by service providers to ensure the protection of mobile transactions could be regarded as an electronic wallet . E-wallet transaction does not just offer a degree of ease and speed and gives customers a level of satisfaction and protection in transactions elsewhere and at any moment (Liébana-Cabanillas et al., 2014). The use of e-wallets provides financial transactions, even on a small scale, that is very easy to execute (Punwatkar & Verghese, 2018). Due to the existence of many e-wallets, such as bKash, Rocket, iPay and Nagad, mobile payments are becoming common in Bangladesh. As the economy of Bangladesh has developed rapidly and strongly recently, demand for services is growing substantially, especially in the finance sector. In Bangladesh, cashless transactions have seen Keywords E-wallet, Customer satisfaction, TAM, Mobile payments, Bangladesh enormous growth. Since 2016, the country has witnessed tremendous growth in terms of digital transactions made by citizens. Since then, the transactions have been growing exponentially (The Daily Star, 2018). As digital financial transactions gain attention among the public, the Bangladesh Bank has granted a further payment service provider license to operate e-wallet facilities (Bangladesh Bank, 2020). When consumers become accustomed to performing online-based financial transactions, entrepreneurs sense the potential of business in the field and encourage entrepreneurs to take out licenses.A previous study was conducted by Uddin and Akhi (2014), explaining that the growth of electronic payment systems in Bangladesh was due to the implementation of internet banking. Mobile wallet usage has increased by replacing physical wallets, and cash-based transactions were shifted by cashless payments systems whereby people involved in transactions have shown their relative interest in digital payments. However, the behaviour among consumers regarding the use of e-wallets has not been found to be efficient as the technological ease of use is not the only factor that changes the behaviour of consumers (Amin et al., 2015). As the number of digital wallet service providers increases, consumers' satisfaction levels are therefore a matter of concern. A previous study by Amin et al. (2015) found that effortless technology is not only a reason to change behaviour rather, but it is also essential to understand gratification among people using technology. Additionally, the ability to use the particular technology is an important issue to opine. Self-confidence regarding the use of digital payment systems may satisfy consumers in order to attempt any behaviour. However, it is vital to consider information security and privacy because cases and issues of information infringement are growing worldwide. A lack of information security expertise leads a cyber-attacker to violate someone's privacy, i.e., identity theft, credit card fraud and cybercrime (Barrett-Maitland et al., 2016). This paper aims to examine the factors that affect satisfaction among Bangladeshi consumers regarding the use of e-wallets as a transaction method. Five variables such as technology self-efficacy (TSE), perceived ease of use (PEU), perceived usefulness (PU), perceived trust and payment security (PS) were elected to investigate the relationship with satisfaction. Five parts of the remaining paper cover literature review, study methods, results and findings and limitations, as well as future research directions.

LITERATURE REVIEW Extended Technology Acceptance Model (ETAM)
In order to understand the consumer's intention to use information system technology, several theories have been propounded. In various literature, the TAM model has been broadly studied (Davis, 1989), and it is actually the theory of information services that models how a specific technology is adopted and used by individuals (Dauda & Lee, 2015). The TAM model has been expanded by numerous researchers and applied to several different technologies, including e-learning (Cheung and Vogel, 2013;Al-Maroof and Al-Emran, 2018), mobile-based technology (Barry & Jan, 2018) and also short message services (SMS) advertising (Muk & Chung, 2015). According to Vijayasarathy (2004), TAM variables are ideally suited for a decision about embracing new technologies. TAM is regarded as a well-recognised expansion of academic research to study innovative technologies' acceptance and usage intention (Aydin & Burnaz, 2016). However, the original TAM variables may not accurately describe fundamental beliefs that influence consumers' attitudes towards technology use. In order to reinforce the model, researchers have further extended the TAM model by applying satisfaction to generate more clarity in the model. Phuong et al. (2020) studied the use of e-wallet payment methods where customer satisfaction was embedded to enhance the understanding of behaviour. Privacy and protection are some of the extended variables that have been shown to be useful in the use of modern technologies for behavioural purposes (Barry & Jan, 2018). Islam (2020) has examined the extended technology acceptance model (ETAM) to validate all the constructions to test satisfaction. However, the new ETAM model has adopted self-efficacy to portray the incorporation between self-efficacy and satisfaction regarding the use of new technology. Technological Self-efficacy Self-efficacy is viewed as one's judgment according to how well various courses of action can really be carried out with many unpredicted and challenging materials in different prospective circumstances (Bandura & Schunk, 1981). Earlier research focused more on the technical method to online learning, self-efficacy, e.g., computer self-efficacy (Jan, 2015;Pellas, 2014), internet self-efficacy (Lin et al., 2013;Kuo et al., 2014), information-seeking behaviour related self-efficacy (Tang & Tseng, 2013). According to the study of Islam (2020), self-efficacy was validated to describe the relationship with satisfaction. Satisfaction and self-efficacy were incorporated as intrinsic motivation attributes in ETAM. Several previous studies have scrutinised the relationship between self-efficacy and behavioural intention to use (Wu et al., 2008;Ahmad et al., 2010). Later, individual satisfaction was associated with self-efficacy (Huang, 2008). It is presumed that a connection exists between CSE and satisfaction through the mediating influence of the behavioural purpose of a person. Self-efficacy plays a significant role in terms of its impact on perceived ease of use (Madorin & Iwasiw, 1999). A multivariate analysis carried out by Mutahar et al. (2018) posited that mobile-based technology is indeed important among Yemeni bank clients, and self-efficacy was found to impact perceived ease of use and perceived usefulness significantly. Self-efficacy is also considered as self-confidence among people involved in mobilebased technology such as mobile banking. In the mobile technology context, it requires skill, knowledge and ability for operating particular mobile technology so that it can intervene easily (Singh & Srivastava, 2018). Multiple empirical evidence showed a causal link between selfefficacy and perceived ease of use (Luarn and Lin, 2005;Jeong and Yoon, 2013;Sripalawat et al., 2011). Therefore, the following hypotheses are proposed: H1: Technology self-efficacy has a positive impact on perceived usefulness. H2: Technology self-efficacy has a positive impact on customer satisfaction. H3: Technology self-efficacy has a positive impact on perceived ease of use.

Perceived Usefulness
Perceived usefulness refers to the degree to which a person assumes their performance will be improved by using a particular information system (Davis, 1989). Several previous studies have been conducted to portray the effect of perceived usefulness on satisfaction.  have studied mobile communication technologies through which mobile payment methods and mobile commerce websites have gained tremendous attention. The study found that the usefulness of mobile technology directly affects customer satisfaction. Phuong et al. (2020) have empirically investigated the continuous usage intention to use an e-wallet and found a significant association between perceived usefulness and satisfaction. E-government services through digital platforms have also received a matter of concern whereby satisfaction is derived from perceived usefulness (Sachan, Kumar, & Kumar, 2018). Mobile-based technology has been improved and widespread throughout the Middle Eastern region like Bahrain. A quantitative analysis postulates that using mobile technology such as e-wallets motivates users to continue using through satisfaction (AlKubaisi & Naser, Am. J. Econ. Bus. Innov. 1(1) 1-10, 2022 2020). A similar study also classified the continuance intention to use e-wallets, but the most important factor in the new era of competitive markets is uttered to be consumer satisfaction. Based on the discussion above, the following hypothesis is approached: H4: Perceived usefulness has a positive impact on customer satisfaction Perceived Ease of Use PEOU has been considered as the primary construct for assessing and analysing user acceptance of a specific technology, drawing from the literature on information technology (IT). A significant motivating factor for the use of technology by consumers is PEOU (Revels, Tojib, & Tsarenko, 2010). PEOU typically relates to the perception of users of whether it would require a mental effort on their part to complete a specific technical task (Fishbein and Ajzen, 1980;Rouibah et al., 2011). According to , the PEOU for mobile networks would directly enhance personal innovation. Al-Maroof and Al-Emran (2018) performed research on undergraduate students who believe that it is easy and userfriendly to use web service technology, thus significantly impacting overall usefulness. The recent development of mobile-based technology, such as e-wallets, has gained tremendous attention in Malaysian markets. Empirical research on e-wallet usage behaviour postulates that the easier mobile technology is to use, the more it can be used in many forms or for practical reasons (Karim et al., 2020). Wasiul et al. (2020) provided empirical evidence of smart device applications' usage intention and found that PEOU and PU are positively associated. However, determining continuous usage, users' satisfaction with using an e-wallet is central to signifying the relationship between PEOU and continuation (Phuong et al., 2020). Mobile phones in the globalised market are pretty remarkable communication devices based on which usability and their user interface are the phenomena of satisfaction . Based on the above disputation, the following hypotheses are formed: H5: Perceived ease of use has a positive impact on perceived usefulness. H6: Perceived ease of use has a positive impact on consumer satisfaction.

Security System
Security is an important aspect when financial transactions are conducted through digital platforms. Therefore, this could become one of the main barriers to the implementation of electronic wallets, as personal or financial data could be revealed and used for fraudulent activities. Security was identified by Kalakota & Whinston (1997) as a threat involving an event linked to data theft, destruction, data alteration, fraud, breach and violence. Consumers may refuse to use mobile-based technology unless it is trustworthy because e-payments require additional privacy and security features (Gitau & Nzuki, 2014). Online transactions could increase the perceived risk of clients exposing private information. Therefore, in the e-wallet payment process, payment security features greatly affect consumer trust Shao and Yin, 2019;Phuong et al., 2020). Users and business environments embrace highly secured systems, as business units' provision of facilities and procedures depend on easy access and satisfactory payment infrastructure operations (Roozbahani, Hojjati, & Azad, 2015). Based on that, the following hypotheses are proposed: H7: Security system has a positive impact on Perceived trust. H8: Security system has a positive impact on consumer satisfaction.

Perceived Trust
Trust has been commonly exercised in prior studies, which defines the service holder's belief in the service provider in order to ensure the safety assessment of the use of the relevant information system (Komiak & Benbasat, 2004). Trust has previously been recognised as an essential factor in terms of mobile payment and shopping usage (Wang and Lin, 2016;Chong, 2013). Based on previous research, trust and satisfaction are two closely related constructs (Liébana-Cabanillas., 2013;Zhu et al., 2014). Trust has previously been examined in the context of mobile technology and found that effective mobile web systems are required to be trustworthy and thus build customer satisfaction . A later study by Phuong et al. (2020) have highlighted the positive impact of trust on the continuance intention of using an e-wallet, but trust has also been examined to observe customer satisfaction with e-payments (Sutia et al., 2020). Based on that, current research proposed a hypothesis for e-wallet usage satisfaction: H9: Perceived trust has a positive impact on customer satisfaction Figure 1 displays the conceptual framework of this study.

METHODOLOGY
The target population of this current study were smartphone users in Bangladesh. Respondents were selected for questionnaire distribution using the purposive sampling technique. A total of 480 questionnaires were distributed through social media and face-to-face surveys in order to get at least 385 respondents, as per the suggestion of Glenn (1992). After the data screening and cleaning process, 402 valid data remain, representing 83.75% of the total questionnaire distributed. Four items from prior research (Islam, 2020) have been adapted to measure technology self-efficacy (TSE). Five items to measure perceived usefulness (PU), four items for perceived ease of use (PEU) and four items for security system (SS) were taken from Barry and Jan (2018). Meanwhile, perceived trust (PT), which consists of three items, were adapted from Al-Sharafi et al. (2016), and satisfaction (SAT) was adopted with three items and then adapted from Ghani et al. (2017). For measuring TSE, PU, PEOU, SS, PT and SAT levels, a five-point Likert scale, starting from 'strongly disagree' (1) to 'strongly agree' (5), was used. The structural equation modelling was also used to test the causal model and analyse the reliability and validity of the measurement model; confirmatory factor analysis (CFA) was conducted (Harrington, 2009). To validate the measurement items of the proposed model, twenty-five (25) sets of questionnaires were distributed to the users of smartphones living in Dhaka city for the purpose of the pilot test, and afterwards, a reliability test was undertaken, resulting in a total Cronbach alpha of >0.60.  Table 1 displays the descriptive data where most of the participants are male (57.96%) and female, found to be 42.04%. In terms of age, the respondents were between 21-30, the most with the number of 196 respondents. However, considering income among respondents, most of them earn below BDT 20000. Respondents then were asked which e-wallet they use the most; bKash was found to be the most useful financial service with 55.22%, followed by Rocket at 26.87%.

Reliability Statistics
As suggested, the minimum threshold for Cronbach's alpha value is 0.60 (Nunnally, 1978). The reliability statistics of the figures range between .884 and .952, surpassing the minimum threshold. The overall value of 23 items' reliability was obtained at .923, Specifying acceptable consistency.

Exploratory Factor Analysis (EFA)
EFA is one of the most widely used statistical methods mainly used to simplify data by identifying interrelationships within a variable set on the basis of strong correlations (Kline, 2013). KMO value obtained 0.866 and p-value for Barlett's Test of Sphericity obtained .000, where it is essential to follow the minimum value of >0.60 for KMO and p-value (p<0.05) is required (Tabachnick & Fidell, 2007). The total variance in the six factors explained was 83.539. The Eigenvalue is generally used to measure the number of variables, and a cut-off value of 1 is usually used to define factors that are dependent on Eigenvalues. Accordingly, the eigenvalues for the six factors were 8.604, 3.411, 2.793, 1.767, 1.524 and 1.114.

Confirmatory Factor Analysis (CFA)
In order to test the hypotheses, the measurement model was tested first for goodness-of-fit indices and for important and appropriate factor loadings. The hypothesised measurement model (Figure 2) was assessed using AMOS and presumed the association of five latent constructions. According to the modification indices given by AMOS, no indicators were withdrawn from the measurement model as the loads of each item were 0.7 or higher. Hair et al. (2010) designated that any loading value of less than 0.5 shall be considered insignificant. Table 2 displays the overall fit model to ensure that the model suits well. The five fit indices (good-of-fit) were evaluated. According to Hair et al. (2010), the value of CFI is suggested to be more than 0.90, and RMSEA should be less than 0.08 in order to be well-suited. For ᵡ2/df, the value below three was considered to be fine and appropriate. The 23-item scale indicates that the model suits well and is appropriate for the agreed minimum thresholds. Convergence and discriminatory validity have been carried out in order to test the measurement model. Composite reliability was measured for the assessment of convergence validity. In order to achieve good composite reliability, a predetermined threshold value of 0.7 was suggested by Chin (1988). In addition, the average variance extracted (AVE) value should be at least 0.5 or higher, as recommended by Hair et al. (2006). Present AVE-calculated study giving a good sign. Standardised loads of each item were also used to measure AVE, where the total of the squared standardised loads was divided by the number of indicators. AVE was found to be between 0.659 and 0.875, which was higher than the recommended threshold of 0.5. The paper also demonstrates CR samples ranged from 0.810 to 0.965, which surpassed the given threshold of 0.7. Therefore, it would be necessary to show, in an attempt to uphold discriminatory validity, that the steps that should not be related are not probably connected (Hair et al., 2013). Table 3 shows the high validity of discriminants as it illustrates that each factor's square AVE exceeds all its correlations with the other factors (Fornell & Larcker, 1981).

Structural Model
Structural equation modelling is a popular method, particularly for multivariate analyses. It has been used for marketing , business , and finance studies (Salleh et al., 2020). This study consisted of nine (9) hypotheses which were tested to identify the causal pathways. The results demonstrated good fit indices of the model approach. In this case, the normed Chi-square (ᵡ2/df= 2.315). RMSEA is 0.078, which is below the cut-off level of 0.08, and CFI 0.936>0.90. Based on the hypothesised structural model, the results of hypothesis testing are listed in Table  4. Eight of the nine hypotheses were tested and found statistically significant (p < 0.05), and the remaining hypothesis was not found to be significant. We have evaluated the R2 value. In explaining perceived ease of use, technology self-efficacy accounted for 8.8 percent of variance (R2= 0.088) and 31.4 percent of variance (R2= 0.314) in illustrating perceived usefulness. Meanwhile, 45.1 percent of the variance (R2= 0.451) was accounted for by technology self-efficacy, perceived ease of use and perceived usefulness in describing customer satisfaction with e-wallet systems. Lastly, perceived trust was explained by 8.2 percent (R2=0.082) of the variance of the security system. Results of nine hypotheses were evaluated, from which H1 was found to be significant. It signifies the positive relationship between technology self-efficacy and perceived usefulness which was supported based on β = 0.324; SE = 0.057; CR = 4.764. Similarly, H2 and H3 were supported on the basis of the β value, showing 0.147 respectively for H2 and 0.296 for H3. H2 also indicates the importance of SE= 0.056; CR= 3.245, and significant P, which showed that self-efficacy of technology had a positive and significant effect on satisfaction. For H3, where SE= 0.049; CR= 4.010, H3 was also helped by demonstrating that there is a positive relationship between the self-efficacy of technology and the perceived ease of use. H4, on the other hand, indicates that there was no significant relationship between perceived usefulness and satisfaction. The hypothesis (H4) was not accepted with a value of β = 0.104; SE = 0.073 and CR = 1.506 and a p-value greater than 0.05. Based on the p-value (p<0.05), H5 was found to be statistically significant, which suggests a positive relationship between perceived ease of use and perceived usefulness. H6 is identified with a positive relationship between perceived ease of use and satisfaction. The outcome of H6 is therefore supported on the basis of β= 0.248; SE=0.092, and CR=3.641. Similarly, H7, H8 and H9 are supported on the basis of the above criteria, where a C.R. value greater than ± 1.96 (C. R.= > ± 1.96) is needed to be considered significant for a particular element (Hair et al., 2010).

DISCUSSION
This study proposes a model for customer satisfaction to use e-wallets on the basis of the technology acceptance model. In addition, to explore the applicability of TAM from a Bangladeshi point of view in the satisfaction of e-wallets, the current study also investigated those factors, for instance, technology self-efficacy, perceived usefulness, perceived ease of use, system security, perceived trust and satisfaction towards the use of e-wallet payment system. Firstly, hypothesised (H1) model of technology selfefficacy on perceived usefulness was examined. The study found a positive association between TSE and PU. Since mobile technology has widely been spread throughout the Middle Eastern regions, their IT usage capabilities enhanced the understanding of its usefulness. The current study is in line with the previous findings of Mutahar et al. (2018). The study then examined the relationship between technology self-efficacy and perceived usefulness (H2). The hypothesis was found to be significant based on the model validation of Islam (2020), where self-efficacy on satisfaction stands validated. In terms of technology self-efficacy on perceived ease of use (H3), it was found to have a positive relationship in between, thus, in line with the previous study (Luarn and Lin, 2005;Jeong and Yoon, 2013;Sripalawat et al., 2011). It is noteworthy that throughout the analysis, PU was found to have no major influence on SAT (H4), while PU was validated as a major factor influencing the intention of users to implement different technologies in both the original and the extended TAM. It can be asserted that since the majority of respondents are young adults (between 21 and 30 years of age), Mobile technology is becoming more and more useful for the younger generation, so they are now willing to explore the ease of use because they no longer want to prioritise its performance over the easy use of mobile technology. However, a significant relationship was found between PEOU and PU, and the finding (H5) was consistent with the prior study (Phuong et al., 2020;Wasiul et al., 2020). Similarly, H6 was also supported by explaining that a positive relationship was associated between PEOU and SAT; thus, the finding was congruent with . The study then examined the security system on perceived trust (H7), which signifies the existence of a positive relationship in between. Consumers perceive that the more the system is secured, the more it is trustworthy. Prior research by Roozbahani et al. (2015) revealed the satisfaction level of Iranian consumers toward e-payment where system security was found to have the strongest determinant; thus, H8 was supported in current research based on the prior study. Lastly, perceived trust is a factor that every consumer wants, at least for safety purposes. Sutia et al. (2020), in their non-parametric study, exposed that trust is an essential part of every cashless transaction based on which satisfaction level is determined. So, the current hypothesis (H9) is also in line with the previous study where perceived trust was found to influence customer satisfaction positively.

IMPLICATIONS
The current study has two implications, namely theoretical and managerial. Firstly, this study provided tremendous support to the TAM in the sense of customer satisfaction with the usage of e-payment through e-wallets. These present results also contribute to the established research body by analysing the impact of user satisfaction. In addition, the model of the current study was validated by explaining previous related literature. The new validation of the extended TAM model rectifies the prior study model by approaching the technological sense where cognitive skills are required to operate a particular system. For E-wallet providers, this research has many practical consequences. Service providers should concentrate on building the trust of their customers and obtaining their satisfaction on the basis of our findings. For operators of Bangladeshi e-payment apps, the study offers few management implications. This analysis will assist the team involved in the management and administration of the service to follow the best policy to gain access to broader customers. On the other hand, building confidence by providing its customers with suitable services will lead an organisation to achieve a comparative edge. In addition, the main factors that decide whether to attempt or to refuse are satisfaction among customers.

LIMITATIONS AND FUTURE RESEARCH
The study has a few limitations as well. First of all, researchers only selected respondents from the Dhaka city area, which may not be considered the best representation of Bangladesh; thus, the study should be carried out other than in Dhaka city as the number of e-payment services are spreading nationwide. Secondly, respondents were mostly chosen from the university area where students' "voluntary participation" assisted in completing the survey, but few responses were unexpected. It may be a reason for haste or any urgency of the respondents. Thirdly, researchers must take precautions when selecting samples. A little formal education is a precautionary step that must be observed. However, the current study has adopted and followed the variables which were previously suggested by TAM, TAM2 and UTAUT models with other additional factors. Thus, additional predictors are highly recommended as the rapid switch in mobile phone technology has taken part. As per the suggestion of Phuong et al. (2020), self-efficacy has been employed in the current study and explored by conducting factor analysis to validate extending the TAM model. Future research could adopt performance expectancy (Zhou, 2014) to evaluate customer satisfaction which could be influenced by other social factors and other social influences. However, excessive usage of e-wallets perhaps intensifies consumers' expectations of what discounts or bonuses and promotions to give out. Consequently, the perceived benefit should be selected to determine how significantly these affect the continuation or user satisfaction with e-wallet payment systems.