Keywords

1 Introduction

In order to counter credit card fraud, a new card payment technology, ‘chip and pin’ is being introduced into the USA in 2015. This new card payment standard has been developed by EMVCo, whose member organizations are American Express, Discover, JCB, MasterCard, UnionPay and Visa [1]. At the same time that merchants are upgrading their payment terminals, consumers are being encouraged to upgrade their payment cards. The Aite Group estimates that 75 % of the credit cards in the USA will be chip enabled by the end of 2015 [2].

The EMV standard includes contactless payments using Near Field Communications (NFC). Many smartphones also have NFC capability, which enables them to replace the physical wallet with a ‘mobile wallet’ [3]. With the physical card, consumers are aware of the organizations that assure the security of the payment transaction, but with the mobile wallet, the ecosystem is more complex, with the addition of smartphone manufacturers, app developers and mobile network providers [4]. As examples, Google Wallet is a partnership with Sprint and Citi MasterCard [5] and Isis Mobile Wallet is a partnership between the US wireless companies, Verizon, T-Mobile and AT&T [5]. On 9 September 2014, Apple, who was the last major smartphone supplier not supporting NFC, added NFC to its iPhone 6, with the introduction of ApplePay [6].

In order to advance acceptance of the mobile wallet, companies that make up the extended ecosystem must consider additional investments in the infrastructure to overcome any adoption barriers, such as security concerns. In order to assist practitioners with their investment decision, our research question is how does perceived security influence consumers’ intentions to use a mobile wallet.

This paper is organized as follows. The next section is the literature review, in which we develop our research model. The third section is the research methods where we introduce the scales that will measure each construct. We analyze the results in the fourth section. In the fifth section we discus the results and include the limitations of the current research and suggestions for future research. Our conclusions are presented in the final section.

2 Literature Review

2.1 Technology Acceptance

TAM was originally designed by Davis [7] to help software designers design applications for an organization that would be adopted by its employees. It has since been applied to the adoption of personal computing, such as Internet shopping [8, 9] and mobile commerce [10, 11]. The model is parsimonious with two independent variables that predict intention to use: perceived usefulness (PU) and perceived ease of use (PEOU).

Designers of the wallet are emphasizing the usefulness of the mobile wallet when compared to the physical card. It is also easy to use by just waving the phone over the terminal. Our first two hypotheses are:

Hypothesis 1::

Perceived usefulness positively influences intention to use a mobile wallet.

Hypothesis 2::

Perceived ease of use positively influences intention to use a mobile wallet.

2.2 The Role of Perceived Security

Chellappa and Pavlou [12] define the security of a payment as ‘the flow of information originating from the right entity and reaching the intended party without being observed, altered or destroyed during transit and storage’. When making a payment, consumers’ prime concern is security [13, 14]. In a study by Linck et al. [15], consumers stated that their concerns were confidentiality, authentication, integrity, authorization and non-repudiation. Mobile payments decrease the sense of security because personal data is stored on the smartphone, which can be lost or stolen. We therefore add perceived security to our model:

Hypothesis 3::

Perceived security positively influences perceived usefulness.

Hypothesis 4::

Perceived security positively influences perceived ease of use.

2.3 Personal Innovativeness

In Roger’s study of the theory of diffusion and innovations [16], he found that innovations diffuse at different rates depending, amongst other characteristics, on the attitude of the individual. Although there is less information available from the trials and observations of the innovation, early adopters are willing to take more risks. They are more comfortable with uncertainty [17] and tend to seek out stimulating experiences [18]. Agarwal and Prasad captured this concept with the construct of personal innovativeness, which they defined ‘as the willingness to try out a new information technology’ [19]. We follow their practice and extend TAM by hypothesizing:

Hypothesis 5::

Personal innovativeness positively influences perceived usefulness.

Hypothesis 6::

Personal innovativeness positively influences perceived ease of use.

2.4 Perceived Security as a Mediating Variable

Innovators are willing to take risks [17], but when it comes to payments they still require that the transaction is processed securely. If the new mobile payment ecosystem is perceived to be less secure than the current means of payment, early adopters might hesitate until they are persuaded that more security is in place. Following Baron and Kenny’s description that mediation is the mechanism where ‘an active organism intervenes between stimulus and response’ [20], we hypothesize that perceived security is a mediating variable.

Hypothesis 7::

Perceived security mediates the influence of perceived usefulness on intention to use the mobile wallet.

Hypothesis 8::

Perceived security mediates the influence of perceived ease of use on intention to use the mobile wallet.

Fig. 1.
figure 1

Research model for acceptance of mobile wallet

Full size image

2.5 Research Model

The research model is shown in Fig. 1.

3 Research Methods

Data from an online survey was analysed using PLS. The survey questions used indicators for the constructs that were borrowed from the extant literature. For perceived ease of use, perceived usefulness and intention to use, we turned to Chandra et al. [21] whose study was focussed on the use of a mobile payment system. We adopted the scale for perceived security from D. Shin [3] and for ppersonal innovativeness we went to the original study by Agarwal and Prasad [22].

The content of the questionnaire was developed with the help of experts. Then tested against a small sample. Finally, with the help of a sampling company that had panels of consumers, we sent the survey to 800 participants in the United States who were eighteen years old or over and who owned a smartphone. 597 completed questionnaires were received and further analyzed with the help of Partial Least Squares (PLS) using the SmartPLS software.

Following the recommendations of Hair et al. [23], we first evaluated the measurement model for internal consistency by calculating Cronbach’s alpha and evaluating composite reliability. The convergence of indicators on their constructs was tested by calculating the average variance extracted. In addition, the Fornell-Larcker criterion was used to test the discriminant validity of all the constructs in the model.

After completing the analysis of the measurement model, we evaluated the structural model [23]. The coefficients of determination (R2) were calculated for all endogenous variables. For each path in the model, the size of the path coefficients were calculated and bootstrapping was used to determine their significance. f2 was calculated to measure the effect size for each construct. To test the role of perceived security as a mediating variable, we calculated the Variance Accounted For (VAF) factor, following Preacher’s method of multiplying the indirect effects [24].

4 Results

4.1 Descriptive Statistics

In the sample, there were 296 males (49.6 %) and 301 females (50.4 %). Almost half the sample (4 %) was between 18 and 40 years of age and remaining 52 % was 41 and above, with the oldest participant 75. The median length of ownership for those who possessed a smartphone was 3.5 years with 50 % having owned a smartphone for three years or more. Table 1 shows the ownership by type of phone.

Table 1. Ownership by type of phone
Full size table

4.2 The Measurement Model

The cross loadings of the measurement model were calculated by the SmartPLS software and the indicators were shown to be collinear. All correlation coefficients were greater than the threshold value of 0.708 [25]. By running a Bootstrap within SmartPLS with 5,000 samples using the replacement method, the t statistic for each cross loading was calculated and in every case, the significance was p < 0.001.

The internal consistency of each construct was assessed via Cronbach’s alpha [26], where values above 0.8 indicate reliability. The Average Variance Extracted (AVE) for each construct further confirmed the reliability of the model, where the AVE was above the guideline of 0.5. In addition, the Composite Reliability was above the guideline of 0.6 [25].

Discriminant validity was tested using the Fornell-Larcker score, where the AVE must be greater than the square of the correlations [27]. Table 2 compares the correlations with the square root of AVE (shown in italic bold along the diagonal).

Table 2. Values for Fornell Larcker test
Full size table

4.3 The Structural Model

The SmartPLS algorithm calculated the R2 measures for each endogenous variable and the path coefficients for each path within the model. R2 for intention to use was 0.670, which is considered moderate [28]. All hypotheses were supported. Results are shown in Fig. 2.

Fig. 2.
figure 2

Results of analysis of structural model

Full size image

The effect size was calculated in a series of steps, where each exogenous variable was removed from the model in turn and the new R squared calculated. The effect size is represented by f squared, where values between 0.02 and 0.14 are small, between 0.15 and 0.34 are medium and 0.35 and above are large [25]. Table 3 shows that PU has a large effect size and PEOU has a small effect size.

Table 3. Effect Size on Intention to Use
Full size table

We also evaluated the effect size of personal innovativeness and perceived security on the intervening variables, PEOU and PU. Personal innovativeness had a large effect and personal security had a medium effect. See Table 4.

Table 4. Effect size of PI and PS on PEOU and PU
Full size table

4.4 Intention to Use

In the questionnaire, participants were asked about their intention to use other features that would be enabled by a mobile wallet. See Table 5.

Table 5. Intention to use specific services enabled by the mobile wallet
Full size table

Participants indicated their preference to use their smartphone for the convenience of handling loyalty points and for managing receipts and coupons. Unlike payments, these are features that are enabled by the mobile wallet innovation.

4.5 Perceived Security as a Mediator

In order to evaluate the effect of perceived security as a mediating variable, we evaluated the indirect paths: personal innovativeness to perceived security to perceived ease of use; and personal innovativeness to perceived security to perceived usefulness. From running a bootstrap, all paths were significant with p > 0.001. We also ran the model without perceived security. The paths are illustrated in Figs. 3 and 4.

Fig. 3.
figure 3

Mediation of perceived security on the effect of personal innovativeness on perceived ease of use.

Full size image
Fig. 4.
figure 4

Mediation of perceived security on the effect of personal innovativeness on perceived usefulness.

Full size image

The indirect effect was calculated as the product of a and b [24]. The Variance Accounted For was derived from the formula

$$ {\text{VAF }} = {\text{ a }}*{\text{ b }}/ \, \left( {{\text{a }}*{\text{ b }} + {\text{ c}}}^{\prime} \right) $$

Table 6 shows the results. In both cases, the mediation is partial. VAF can have values from 0 % to 100 %, where 100 % represents full mediation. A value of 41 % is a moderate VAF level [23, 29, 30]. Our conclusion is that perceived security as a mediator has a moderate effect on perceived ease of use and a moderate effect on perceived usefulness, accounting for 41 % and 49 % of the variance respectively.

Table 6. Mediation - Variance Accounted For
Full size table

5 Discussion

PU has a large effect on intention to use, consistent with other studies of TAM [31, 32]. This also confirms Roger’s conclusion that adopters perceive a relative advantage of the innovation over the current process [16]. As Table 5 shows, consumers see advantages, over and above just the payment, such as the handling of loyalty points and the reduction of paper receipts.

In our empirical setting, the effect of PEOU on intention to use is small, which is consistent with past studies [31]. The explanation here is that using a smartphone to pay is a simple operation and does not differ very much from the use of the physical payment card. Consequently learning to use the mobile wallet is perceived to be easy.

Perceived security has a large effect on the intervening variables, PEOU and PU. Payment transactions contain sensitive data and consumers need to be assured that their account is debited the correct amount. This is no different than considerations around payments made by more traditional methods, but there is the complexity of additional parties who are involved in moving the transaction from the app on the smartphone to the financial institution.

Personal innovativeness has a medium effect on the intervening variables, PEOU and PU. We would expect consumers who are personally innovative to be the early adopters because they are more comfortable with new technology. They would be less inhibited by issues of ease of use and would be prepared to compromise on usefulness in order to adopt the mobile wallet sooner than others. But because payment transactions are involved, perceived security is still a concern and acts as a mediator. Those consumers who are more innovative are less likely to perceive security as a barrier to adoption.

5.1 Theoretical Contribution

Our theoretical contribution is to add to the theory that has extended TAM in the context of consumer acceptance of the mobile wallet. Past studies have evaluated personal innovativeness and perceived security, but they have not been combined in the same model with an evaluation of the mediation effect of perceived security. It is the challenge of the researcher to construct a parsimonious model that explains phenomena minimizing confounding effects [33]. Our model has only five constructs, but with the added path of mediation, it is able to explain which consumers are more likely to adopt the mobile wallet.

A further contribution to theory is the comparison of the influence of PU to that of PEOU. Meta-analysis of the TAM literature has indicated that PU has a stronger influence than PEOU [31] and we confirm these findings and agree with Geffen and Straub, who proposed that PEOU relates to the ‘intrinsic characteristics of the IT artefact…whilst PU is a response to user assessment of its extrinsic outcomes’ [34]. These results suggest that if the IT artefact is simple to use and its use is similar to current actions, PEOU has a small effect on intention to use.

5.2 Limitations and Future Research

Our sample was from a panel conducted by a professional organization experienced in conducting surveys with selected audiences. Panel members have voluntarily offered their services and receive some form of compensation for taking a survey. The results reflect the responses of the panel population, which may be different than the general population. A further limitation is that the research was conducted with residents of the USA, and the findings may not be applicable to other geographical or cultural groupings.

Our theoretical contribution lays the groundwork for future researchers. The model can be tested across a broader cross section of the general population. The data can be segmented to determine whether age and income are moderating factors. Given the dominance of PU as an influencing factor, future researchers could explore other antecedents of PU. In addition, perceived security has been the subject of past studies and more detailed research could seek to decompose this construct. Because offerings and infrastructure vary by country, research could be further extended by comparing acceptance in different countries.

6 Conclusion

The mobile wallet, defined as an app on the smartphone to be used for face-to-face payment transactions, is relatively new. For the USA, with more NFC-enabled terminals becoming available because of the impending chip and pin standard, the capability of making contactless payments at retail outlets is growing. Consequently there is an increased opportunity for the smartphone to initiate payment instead of a physical card.

The development of these mobile wallet apps depends upon providers investing further in the software and the infrastructure, and their decision to invest depends upon the acceptance by consumers of this new technology. In order to understand the factors that influence guests, we have applied the Technology Acceptance Model, which is a seminal theory for an individual’s acceptance of a new IT artefact. For payment transactions, security is a primary concern. We have added the construct of perceived security and investigated how it mediates personal innovativeness.

The results of our empirical study confirm that perceived usefulness is the most important influencing factor, and that personal innovativeness and perceived security are significant antecedents. A further contribution of this paper is how perceived security is a mediating factor, mediating personal innovativeness. Consumers are willing to use the mobile wallet if they perceive it to be secure. Practitioners are in a position to influence consumer acceptance of smartphone apps and, given the importance of security, they should emphasize the security of their solutions.