Determining Factors of Digital Wallet Usage

This research aims to examine the effect of perceived usefulness, perceived risk, and trust on the usage intention of digital wallets among college students. Respondents in this study consisted of 138 undergraduate students of the Faculty of Economics and Business, Tarumanagara University, who use ShopeePay. The primary data used for the study were obtained through questionnaires distributed online, in which to select respondents a purposive sampling method was applied. Then the collected data is processed with the analysis technique of Structural Equation Modeling (SEM), using SmartPLS software. The results showed that perceived usefulness and trust positively influenced usage intention. Meanwhile, the perceived risk did not affect usage intention.


INTRODUCTION
In this 21 st century, all kinds of human activities can not be separated from technological interference. With technology, activities that were initially done traditionally and took a long time now can be completed easily in a short time.
Nowadays technological advances are growing rapidly. The emergence of an instant lifestyle phenomenon in society has triggered various innovations, especially in the fields of finance and information technology.
The developments resulting from the two combinations are then presented into financial technology or what is also known as fintech. (Sangwan et al., 2020) said fintech promotions in the form of cashback, so of course, many people from various professions and ages (teenagers to adults) are tempted to become users of this digital wallet. From what has been stated, this matter makes the researcher interested in choosing ShopeePay and further research its users, to understand specifically the reasons that influence the intention to use ShopeePay as a payment instrument.

THEORETICAL REVIEW
Technology Acceptance Model (TAM). TAM was designed by Davis in 1989, is a behavior model that is often used to explain and predict how the acceptance of the adoption of the latest technology in individuals (Davis, 1989). TAM is the derivative of the expansion of the Theory of Reasoned Action (TRA) introduced by Fishbein and Ajzen in 1975. TRA assumes that a person's behavior is influenced by behavioral intention, then behavior intention is influenced by subjective norm and attitude.
Based on Figure 1, TAM represents when the real use of technology is based on perceived usefulness and perceived ease of use, which is mediated by attitudes towards the use of the technology. Then from this model, it can be concluded, if a new technology is easy to use, easy to access, or easy to implement, then individuals will be happy to use that technology. In addition, if the use of new technology has uses that can provide more benefits to its users, then an individual will never hesitate to adopt the technology.  (Davis, 1989) in Matikiti, Mpinganjira, & Roberts-Lombard, (2018) The relationship between TAM and this research is that TAM describes what is behind the use of technology. TAM explains that technology adoption is influenced by two main factors, namely perceived usefulness and perceived ease of use, in which the perceived usefulness is one of the variables used in this study.
Unified Theory of Acceptance and Use of Technology (UTAUT). UTAUT is a theory of acceptance and application of technology designed by (Venkatesh et al., 2003). UTAUT was chosen because this theory comes from a combination of eight models and theories of individual acceptance of technology which has existed before so that UTAUT has been tested to be superior to its predecessors (Venkatesh et al., 2003).
The eight models and theories are Theory of Reasoned Action (TRA), Technology Acceptance Model (TAM), Model of PC Utilization (MPCU), Motivational Model (MM), Theory of Planned Behavior (TPB), Combined Technology Acceptance Model, and Theory of Planned Behavior (C-TAM-TPB), Innovation Diffusion Theory (IDT), the last is Social Cognitive Theory (SCT). The relationship of the eight theories is shown in Figure 2.
Through a study conducted by (Venkatesh et al., 2003) found that the constructs of performance expectations, effort expectations, social influence, and facilitation conditions had an important role as direct determinants of behavioral intentions and usage behavior, while the rest were insignificant.
The relevance of UTAUT and this research is to understand if the behavior of acceptance and use of technology, starts from the intention and behavioral interest. However, as the times and technology develop, this interest can also be influenced by other factors such as perceived usefulness, perceived risk, and trust.

Figure 2. Unified Theory of Acceptance and Use of Technology
Source: Venkatesh et al., (2003) in Williams, Rana, & Dwivedi, (2015) The Relationship between Perceived Usefulness and Usage Intention. Every technology that is created has a use. The more uses that can be given from technology, the more it will motivate users to use the technology. According to (Almarashdeh and Alsmadi, 2017), perceived usefulness refers to the advantages or benefits that will be obtained by users from using a particular service or application.
Meanwhile, (Sondakh, 2017) defines perceived usefulness as the user's perception of how far the use of service systems can improve performance. Not much different, (Phonthanukitithaworn et al., 2015) define the perceived usefulness by the extent to which a person believes that the use of a service will improve performance and productivity. Based on the definitions that have been described, the researcher concludes the perception of usefulness as an individual's view and assessment of the benefits that will be obtained from the use of a service. In this study, the indicators used to measure perceived usefulness come from TAM (Normalini, 2019), namely performance, productivity, and effectiveness.
According to the previous explanation about TAM, it can be seen that the perceived usefulness is one of the important factors in the acceptance of technology. This model is supported by three studies with samples originating entirely from India. The results were found that perceived usefulness has a significant effect on the intention to use digital payments (Kumar et al., 2018;Patel, 2016;Upadhyay and Jahanyan, 2016). Respondents from one of the studies above felt that this payment method could increase their efficiency.
Research on fintech conducted on respondents from China, Spain, and America also gave similar results. Analysis showed that perceived usefulness positively influences usage intention (Flavian et al., 2020;Shiau et al., 2020). The findings of these studies certainly emphasize that if financial services such as fintech can provide more utility, the usage intention will also be higher.
However, different results were reported by a study conducted in Thailand. The empirical findings suggest that the use of digital payments is not influenced by perceived usefulness (Phonthanukitithaworn et al., 2015). There are no additional benefits that users feel from using the system is suspected to be the reason behind it.

The Relationship between Perceived Risk and Usage Intention.
Even though fintech provides so many benefits, of course, this does not guarantee that its users will be protected from various uncertainties, losses, and other risks. (Bashir and Madhavaiah, 2015) define risk perception as the uncertainty that the user can experience losses when the user cannot predict the consequences of using a service. These losses are in the form of financial, social, privacy and performance losses.
On the other hand, according to (Chen, 2017), risk perception can be interpreted as a potential loss when involved in using online services in pursuing the desired output. Meanwhile, (Yang et al., 2015) describe the perception of risk by how a person perceives the opportunity for losses that may occur due to uncertainty.
From the various descriptions above, the researcher defines risk perceptions into views and assessments of negative impacts (in the form of losses, damages, and losses) that have the potential to befall the individual, as a consequence of the decision to use a system. To measure perceived risk, indicators based on (Featherman and Pavlou, 2003); and (Lim, 2003) research can be used, namely financial risk, performance risk. Privacy risk, time risk, and psychological risk.
Perceived risk is individual subjective assumptions about the bad effects that will be obtained, because of the actions that have been taken. Uncertainty about what will happen can cause perceived risk in someone. Although only a few uncertainties are faced, it can still reduce the usage intention.
Not much different from this study, (Koenig et al., 2015) used the TAM and UTAUT2 theory, to be analyzed and expanded by adding perceived risk variables in their research. From this study, it was found that perceived risk had a significant negative effect on the intention to adopt mobile payment. (Hoa et al., 2019); and (Sobti, 2019) researched digital financial services. Those two studies confirm that the use of digital payment is significantly influenced by perceived risk. Similar results were obtained from the study of (Gupta et al., 2017) which is applied to mobile banking services. But, research conducted by (Islam et al., 2020) in Pakistan revealed different findings, namely that the intention to use digital payment methods is not influenced by perceived risk. These findings have shown that people in these areas in adopting digital payments do not care about perceived risk.
The relationship between Trust and Usage Intention. The credibility of financial services is one of the things that potential users highlight. If the service can be trusted, then potential users don't need to think twice about using it. Because the more trustworthy the service is, the more it will attract the intentions and interests of potential users.
Referring to the research of (Devlin et al., 2015), trust is the overall attitude and belief about the beliefs that are held. In the research, the indicators applied to measure trust are based on research by (Sekhon et al., 2014) (Matemba and Li, 2018) argue that trust is a person's psychological condition in receiving consequences, as a result of positive expectations on the behavior and intention of the service provider.
Thus, the researcher concluded that trust is an individual's willingness to be open, to have expectations, and to believe in the system used. The measurement of trust in this study uses indicators of capability, integrity and stability, care, and moral values (Sekhon et al., 2014).
Trust is an important element for managing a relationship. With trust, someone will feel confident to rely on others. In this case, digital wallet service providers must be smart in keeping the trust that has been built, so that they can attract potential user intention, and retain permanent users.
Research conducted by (Barkhordari et al., 2017); and (Wang et al., 2019) have proven that trust has a positive effect on the intention to use services such as fintech. Then (Koksal, 2016); (Wang et al., 2015) have tested mobile banking regarding adoption intentions. The results of the study state that trust has a positive and significant effect on usage intention.
On the other hand, research by (Taufan, and Yuwono, 2019) has shown conflicting results. Their analysis shows that the intention to use a digital wallet is not influenced by trust. By researching mobile banking, (Susanto et al., 2016) also obtained the same results, namely that trust does not have an impact on usage intentions.

Hypothesis
In the TAM model, acceptance of technology is indicated by the actual (real) use of the system. This usage begins with the attitude of use, which is shown through the user's intention. Similar to the UTAUT theory, usage behavior begins with Intentional behavior that appears in individuals. Generally, use leads to continuous or temporary use. The use of digital wallets continuously or permanently by users is the main goal for any digital wallet service provider. Therefore, to achieve this goal, the developer should understand what are the determining factors that influence and can increase the intention to use the product.
Perceived usefulness is one of the factors from the external side that can affect the usage intention. Perceived usefulness can be explained as assumptions about the added value that will be received. If fintech can fulfill a person's perceived usefulness such as the realization of effectiveness and efficiency, then this will certainly create a positive impression on the service, so that the usage intention will also increase.
The next factor that comes from the external side is the perceived risk. Simply put, perceived risk is described as thinking about the loss that will be obtained. In adopting technological devices, users should be aware of the risks they will face. These risks can be in the form of mistakes in transferring money, misuse of data due to stolen data, and so on. The more risks that will be obtained, it will give rise negative the risk perception, and reduce the usage intention of the service.
Furthermore, there is trust as a determinant of usage intention from the internal side. In the context of fintech, trust is an individual's willingness to believe in and place their hopes on a service provider. If the service provider has a well-known reputation, of course, users will trust and have a positive impression of fintech, so that users are motivated to adopt the fintech service.
The model in this study, as shown in Figure 3, explains how the relationship between perceived usefulness, perceived risk, and trust with usage intention The hypothesis in this study can be formulated as follows; 1. There is an effect of perceived usefulness on usage intention, 2. There is an effect of perceived risk on the usage intention, and 3. There is an effect of trust on the usage intention.

METHODS
Based on the objectives, this research can be categorized as explanative research, using a quantitative approach. This study uses primary data which is collected through a survey method, by distributing online questionnaires in the form of Google form.
The population of this study consisted of undergraduate students of the Faculty of Economics and Business (FEB) Tarumanagara University (Untar) who are ShopeePay users. The sampling technique applied was non-probability sampling, namely purposive sampling.
Primary data that has been collected then will be analyzed using Structural Equation Modeling (SEM) analysis techniques and processed with the help of SmartPLS version 3.3.2 software. Where SmartPLS is used to test the relationship, influence, and feasibility of research instruments (Nuringsih et al., 2019). Data analysis in this research consists of measurement model test (validity and reliability test), structural model test (coefficient of determination test, predictive relevance test, and effect size test), Goodness of Fit (GoF) test, and hypothesis testing (path coefficient, t-statistic, and p-value).

RESULTS
From 138 samples met the criteria for this study. Then the data is tested for validity (Convergent Validity and Discriminant Validity) and reliability to ensure the quality of data from respondents (Nuringsih et al., 2020).
It appears from Table 1 that each indicator in this study is valid because it passes the convergent validity test. This is because the AVE value for each variable, namely the variable use intention, perceived usefulness, trustworthiness, and perceived risk is greater than 0.5. An indicator can be said to be valid if the AVE value is equal to or more than 0.5 (Hair et al., 2020). Judging from Table 2 about the cross-loading test above, it can be concluded that each indicator of each variable is considered valid so that all indicators can be used in this study. Each indicator is declared to have passed the discriminant validity test if the cross-loading value of each construct is greater than the other constructs (Wong, 2016). The indicator is considered reliable if both Cronbach's alpha and composite reliability values are higher than 0.7 (Sani and Wiliani, 2019). When viewed from Table 3, the value of Cronbach's alpha and composite reliability for each latent variable has exceeded the value of 0.7. After the validity and reliability tests, which are the measurement model tests, then perform the structural model tests (coefficient of determination, predictive relevance, and effect size tests), Goodness of Fit tests, and hypothesis tests (path coefficient, t-statistic, and p-value).

Measurement Model Testing (Outer Model Evaluation).
Evaluation of the measurement model is carried out to test the level of validity and reliability as has been done previously. This study consisted of four variables, namely, perceived usefulness, perceived risk, trust, and usage intention. In the eyes of consumers, the Perceive Risk aspect is more dominated by the PR5 indicator. Some consumers are still worried about using ShopeePay because the use of ShopeePay is related to the use of technology that may be something new for some consumers.
The smallest contribution to Perceive Risk is given by the PR4 indicator. Consumers think that using ShopeePay is relatively easy and doesn't take up too much time. Consumer trust in ShopeePay lies in ShopeePay's ability to provide honest services. This indicator contributed 0.851. Besides honest service, consumer trust in ShopeePay is manifested by concern and the payment methods used. Of the 5 indicators, consumers feel ShopeePay's reliability is still relatively low, so this needs ShopeePay's attention.

Usage Intention
Referring to Figure 7, the researcher concludes that the various change of the perceived usefulness variable is due to the UI1 indicator with a value of 0.816; UI2 indicator with a value of 0.760; UI3 with a value of 0.839; and UI4 with a value of 0.781.

Structural Model Testing (Inner Model Evaluation).
After evaluating the measurement model, the next step is to analyze and evaluate the structural model in this study. The relationship between exogenous variables (perceived usefulness, perceived risk, and trust) and endogenous variables (usage intention) is shown in Figure 8.  The coefficient of determination test is useful for knowing how much the contribution of the independent variable to the dependent variable in the study (Luis and Nuryasman, 2020). Based on Table 4, it is known that the R 2 value is 0.443, which indicates that the usage intention variable can be explained by the perceived usefulness, perceived risk, and trust variables of 44.3% and the remaining 55.7% can be explained by other variables not studied. Then based on the classification of the R 2 value according to (Hair et al., 2020); and (Leguina, 2015), where R 2 is grouped into weak (0.25), moderate (0.5), and strong (0.7), it can be concluded that this study is weak because the value is below 0.5. (Khairunnisa et al., 2020) explain that Predictive relevance is used to test how the value of observations is the result of the model. Obtained the Q 2 value of the usage intention variable is 0.267, which is greater than 0. So that the research model is considered predictive relevance, which means that perceived usefulness, perceived risk, and trust (as exogenous variables) can be used to predict usage intention (as endogenous variables) well, amounting to 0.267. Referring to Table 5, it is known that the F 2 value of perceived usefulness is 0.134, which means that perceived usefulness has little effect on usage intention; the F 2 value of the perceived risk is 0,000, which means that the perceived risk does not affect the usage intention; and the F 2 value of trust is 0.154, which means that the trust has a moderate effect on the usage intention. Based on the results of the manual calculation above, it can be concluded that this research model has stability with the values that are categorized as large because the GoF value is 0.5415 higher than 0.36. Hypothesis testing is a test used to determine the strength of the influence or relationship between exogenous variables and endogenous variables, which can be seen through the path coefficient, t-statistic value, and p-value (Nuryasman and Suryaman, 2018).
Based on table 6, the value of the path coefficients from the perceived usefulness of the usage intention is 0.358, which means that the perceived usefulness has a strong relationship and a positive direction to the usage intention. Furthermore, based on the perceived usefulness of the usage intention, the t-statistic value was 3,781 and was greater than 1.96; with a p-value of 0.000, which means that it is smaller than 0.005. So it can be concluded that H1 in this research hypothesis is accepted and has a significant effect.
The path coefficient value for perceived risk on usage intention is -0.004, which can be interpreted that perceived risk has a weak effect and a negative direction on usage intention. Then the perceived risk of the usage intention has a t-statistic value of 0.042, which means less than 1.96; with a p-value of 0.966, which means that it is greater than 0.05. Therefore, the researcher concludes that H2 is rejected because it does not meet the requirements of the acceptance of a hypothesis.
The path coefficient value for trust on usage intention is 0.379. This means that trust has a strong relationship and a positive direction to the usage intention variable. Then the trust of the usage intention has a t-statistic of 3.868 which is greater than 1.96 and has a pvalue of 0.000 where the value is less than 0.05. Thus it can be concluded that H3 is accepted and has a significant effect. The results of hypothesis testing are shown in Table 7 Table 7. Hypothesis Testing Results

H1
There is an effect of perceived usefulness on the intention to use the ShopeePay digital wallet. Accepted

H2
There is an effect of perceived risk on the intention to use the ShopeePay digital wallet. Rejected

H3
There is an effect of trust on the intention to use the ShopeePay digital wallet. Accepted

DISCUSSION
The Effect of Perceived Usefulness on the Usage Intention of Digital Wallet. The first hypothesis testing shows that perceived usefulness has a positive and significant effect on usage intention. This is supported by the perceived usefulness indicator that most influences the usage intention, namely ShopeePay will maximize the productivity of users' life. The statement is accompanied by the reason that with ShopeePay a person can do various kinds of financial activities such as making online and offline payments, purchasing digital products, and storing and transferring funds. The results of this study are similar to the study of (Pertiwi et al., 2020) which confirms that the intention to use digital wallets is positively and significantly influenced by perceived usability. In this study, it is known that the perceived usefulness has attracted the public's intention to switch to the non-cash transaction method.

The Effect of Perceived Risk on the Usage Intention of Digital Wallet.
The results of the second hypothesis test in this study indicate that the perceived risk has a negative and insignificant effect on the usage intention. The indicator of the perceived risk that most influences the usage intention is that transactions using ShopeePay make users confused and worried. The reason is that not everyone can easily understand how ShopeePay works, which can lead to confusion. Also, there is the possibility of feeling worried as a result of making mistakes when using ShopeePay. Confusion and worry are forms of stress that lead to psychological problems.
The results of this study are similar to the study conducted by (Chakraborty and Mitra, 2018), namely that perceived risk does not have a significant effect on the usage intention of digital wallets.
The negative effect means that if the user has a high perceived risk of fintech, then it can influence the users to stop using the service. Therefore, service providers should play an active role in overcoming these perceived risks and can provide security guarantees to their users.
The Effect of Trust on the Usage Intention of Digital Wallet. The results of testing the third hypothesis state that trust has a positive and significant effect on usage intention. The indicator of the trust that most influences the usage intention is the user's trust that ShopeePay is an honest financial service. One of ShopeePay's honest actions is shown by the availability of history regarding its users' financial activities. So that users can freely monitor any financial activities that have been carried out, if there is an irregularity in the transaction process, the user can find out about the problem and can report the problem to ShopeePay.
The results of testing this hypothesis are in line with the analysis of (Stewart and Jürjens, 2018). The findings show that user trust is very influential on the intention of fintech adoption.

CONCLUSION
Based on the research results it can be concluded; 1. There is an influence of Perceived Usefulness on Usage Intention, 2. There is no influence of Perceived Risk on Usage Intention, and 3. There is an influence of Trust on Usage Intention. This research suggests several things such as the following; 1. Perceived usefulness has a positive influence on usage intentions, so it is recommended that ShopeePay developers increase their users' perceived usefulness. One of the ways to do this is by making the service operation easier so that various ages can use and experience the benefits of using ShopeePay, 2. Although the results of this study indicate that perceived risk does not affect usage intention, the researcher suggests that ShopeePay developers continue to ensure the safety of ShopeePay users, so that users avoid various potential losses, 3. Trust has a positive effect on usage intention, so researchers suggest that ShopeePay developers can consistently maintain the trust of their users, to maintain the user's usage intention, 4. For the research to be more developed, it is suggested that further research should involve other variables that influence the intention of using fintech. Because the R 2 value in this study was 44.3%, so the remaining 55.7% can be explained by other variables not examined in this study, and 5. The researcher also suggests further research to expand the area of reach of the sample so that the research results better reflect the characteristics of the population and the actual conditions.