Who signs up for NFC mobile payment services? Mobile network operator subscribers in Germany

Accepted Manuscript Who signs up for NFC mobile payment services? Mobile network operator subscribers in Germany Torsten J. Gerpott, Phil Meinert PII: DOI: Reference:

S1567-4223(17)30008-X http://dx.doi.org/10.1016/j.elerap.2017.03.002 ELERAP 704

To appear in:

Electronic Commerce Research and Applications

Received Date: Revised Date: Accepted Date:

25 October 2016 10 March 2017 11 March 2017

Please cite this article as: T.J. Gerpott, P. Meinert, Who signs up for NFC mobile payment services? Mobile network operator subscribers in Germany, Electronic Commerce Research and Applications (2017), doi: http://dx.doi.org/ 10.1016/j.elerap.2017.03.002

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Manuscript ID: ECRA-D-16-00341R4 / ELERAP 704

Torsten J. Gerpott & Phil Meinert

Who signs up for NFC mobile payment services? Mobile network operator subscribers in Germany



Torsten J. Gerpott Chair of Strategic and Telecommunications Management Mercator School of Management University of Duisburg-Essen Lotharstr. 65 47057 Duisburg, Germany Phone +49 203 379 3109, Fax +49 203 379 2656 E-mail: [email protected]

Phil Meinert Chair of Strategic and Telecommunications Management Mercator School of Management University of Duisburg-Essen Lotharstr. 65 47057 Duisburg, Germany E-mail: [email protected]

March 10,2017



Corresponding author.

NFC mobile payment use

– II –

10th March 2017 (ECRA-D-16-00341R4)

Who signs up for NFC mobile payment services? Mobile network operator subscribers in Germany Abstract Near field communication (NFC) mobile payment systems (MPS) enable customers to use their smartphones for over-the-counter payments instead of cash or magnetic strip cards. To date, worldwide the use of NFC MPS is still a niche application but, at the same time, it is growing quickly. Research exploring how NFC MPS users differ from non-users is scarce and mostly limited to future behavioral intentions instead of actual use behaviors. Therefore, the present study explores how 677 real NFC MPS users of the German business unit of a large mobile network operator differ from 677 non-users in terms of socio-demographic and mobile-device characteristics, mobile communication behaviors and the availability of merchants with NFC terminals in the neighborhood of their place of residence. NFC MPS users were likely to be male and older subscribers who were equipped with a larger-screen smartphone, generated higher mobile Internet traffic, held a music streaming service subscription, paid lower monthly bills for mobile communication services and lived in a residential area with a higher number of merchant NFC terminals. Furthermore, we shed light on how use predictors are related to the timing of subscribers’ NFC MPS use starts. Early users tended to be equipped with a highly-priced smartphone whose display screen was nevertheless smaller, used a music streaming service and generated lower monthly communication service bills. Implications for practitioners and scholars are drawn from the results.

Keywords:

Proximity payment systems; near field communication (NFC); mobile payment systems; use of innovations; Germany.

NFC payment adoption

1.

–1–

10th March 2017 (ECRA-D-16-00341R4)

Introduction

Since the introduction of the first payment systems making use of cellular mobile communication network elements and devices in the early 1990s, a plethora of such offerings has emerged on the market. Most of the early mobile payment system (MPS) variants required a radio connection between the customers’ devices and their mobile network operator (MNO) in order to receive or send a text message, a voice call or other data, respectively. This requirement is unfortunate because establishing a radio link is not always possible (due to network coverage gaps) and may consume considerable time. Therefore, new so-called proximity-based MPS solutions were developed. They enable MNO customers to use their device to pay without a radio connection to a mobile communication network. One variant of this class of MPS are near field communication (NFC) solutions (Ovum, 2016; see Figure 1 below). They trigger payments by simply placing an NFC-enabled device at a distance of 10 cm or less to a seller’s NFC terminal at a point of sale (POS) as long as the transaction value is below a country-specific threshold.1 Many MNOs offer NFC MPS in collaboration with a financial institution. Accordingly, MNOs act as an intermediary which provides banks with access to potential NFC payment service users. Typically, MNOs receive a fixed and/or a usagedependent monthly fee per adopter. Therefore, running a NFC MPS is a potential source of new revenues for MNOs. In addition, the spread of NFC payment services among an MNO’s subscriber base may prevent its customers from churning (Reuver et al., 2015). Worldwide, by end-2014 about 2 million smartphone owners had used their device at least once a month to make contactless NFC in-store payments at retail outlets. One year later this number already amounted to more than 31 million (Deloitte, 2015). Various market analysts agree that the proliferation of NFC MPS will continue to grow rapidly in the near future (e.g., Juniper Research, 2016; Ovum, 2016; Trendforce, 2016). However, to date, the use of MPS in general and NFC MPS in particular is still a niche application (Ovum, 2016). For Germany, which is the country setting of the present study, several reports also predict a speedy diffusion of NFC payment (e.g., Herrmann, 2016; PWC, 2014, 2016; Verifone, 2016). But in stark contrast to the “bright” forecasts, by the end of 2015 less than 4% of the 82 million people living in Germany had used their mobile devices at least once for an NFC-based mobile payment (Deloitte, 2016). Three country-specific reasons can be taken to explain prior low adoption rates: Firstly, at the end of 2015 merely 6.5% of the total of 108 million active mobile communication devices in Germany were NFC-enabled (Verifone, 2016). Secondly, in late 1

This hurdle varies from country to country (e.g., in Germany it amounts to EUR 25 during the period covered in the present empirical study). In case that the value of a transaction exceeds the threshold, payers have to additionally confirm the money flow by entering a four-digit pin code.

NFC mobile payment use

–2–

10th March 2017 (ECRA-D-16-00341R4)

2015 only 10% (= 80,000) of all POS terminals in Germany were able to process payments via NFC (Bitkom, 2016). Thirdly, a majority of 53% of retail consumers in Germany still preferred cash to alternative means of payment (Bundesbank, 2015). This environment creates substantial challenges for NFC MPS suppliers in promoting the take-up of such offerings. Therefore, it is pivotal for NFC MPS providers to adequately identify consumers who are most likely inclined to start using NFC payment services as well as their counterparts who require extraordinary measures to lure them into MPS subscriptions. Against this background, it comes as a surprise that no earlier empirical work has explored how MNO customers who have “really” adopted an NFC MPS offer of their carrier (i.e., are characterized by observable use behaviors instead of stated intentions to use/behave) differ with respect to objectively recordable personal characteristics, which are anyway at hand in an MNO’s data warehouse. Therefore, the present study contributes toward closing this research gap by examining NFC payment adoption in a sample of 1,354 postpaid MNO subscribers residing in Germany. Adoption is captured as a two-facet phenomenon. Its first facet entails MNO subscribers’ (yes or no) decision to start using the NFC MPS in a study period from July 2014 to May 2015. The second facet concerns the timing of adopters’ use start during the study period (i.e., whether the first use occurred earlier or later). For both criteria we examine the extent to which socio-demographic, device- and usage-related characteristics as well as supply-side factors are able to predict them. The remainder of this article is divided into five sections. The next section introduces variants of MPS. Section 3 reviews the relevant literature to develop hypotheses and research questions. Section 4 describes the empirical methods. Section 5 presents the empirical results. The last section highlights implications for both MMO practitioners in charge of improving the acceptance of NFC MPS among their employer’s customers and management scholars in the field of innovative MPS. 2.

Background: Variants of mobile payment systems

To adequately discuss the literature on MPS adoption, an overview of available MPS solutions is helpful to better understand nuances regarding the underlying communication technologies, the category of goods which can be paid through an MPS and the processes buyers have to go through to settle payments (Au and Kauffman, 2008; Dahlberg et al., 2015; Dahlberg et al., 2008; Dennehy and Sammon, 2015; Patel et al., 2015; Slade et al., 2013; Taylor, 2016).

NFC mobile payment use

–3–

10th March 2017 (ECRA-D-16-00341R4)

As shown in Figure 1, MPS can be classified according to the communication technology used to link a device to the infrastructure of a financial institution. One group of MPS (remote payment systems) enables customers to pay for digitized or physical goods at a “virtual” or a “real” POS through connecting their device to (long-range) cellular radio or (mid-range) Wifi networks. The second group of MPS (proximity payment systems) is limited to processing payments for (mostly physical) goods at a “real” POS in brick and mortar stores via shortrange communication links. Remote payment systems (see left part of Figure 1) are particularly suited for paying in online transaction contexts and on market platforms such as the Apple Store, eBay or Google Play with money transfers being processed via received or sent short message service (SMS) text, voice calls or Internet protocol-based data packets transmitted over cellular infrastructures of MNOs. Besides, payments can be processed via fixed line infrastructures (i.e., Wifi networks) which provide access to the Internet. MNO network-based payment solutions which also revert to MNOs’ billing systems (i.e., Pay by call, pay by SMS, direct carrier billing) are typically limited to purchases of digitized goods in an online context (Gerpott and Meinert, 2016a). In contrast, other mobile Internet-based payment systems such as one-click bank transfers2 and third-party billing (e.g., Alipay or PayPal) are designed to pay for both tangible

Figure 1 Structuring of mobile payment systemsa Mobile payment systems

Remote payment systems • Via “established“ MCS – Pay by call – Pay by SMS • Via mobile Internet – Direct carrier billing – One-click bank transfer – Third-party billing

2

Proximity payment systems • Via BLE/Beacons • Via infrared • Via code scanning – Barcode – QR code • Via NFC – NFC bank card/label – NFC SIM/SD card – NFC smartphone

of the present study Although MPSFocus overlap significantly with (one-click) bank transfer technologies the latter are considerably more comprehensive than MPS which typically do not cover a broad range of additional banking applications BLE = Bluetooth low energy. MCSthe = MPS Mobile communication = Near field is only (e.g.,a)management of giro accounts). Hence, extensive literature services. on mobileNFC banking adoption communication. QRpresent = Quickstudy response. = Secure digitalbanking memory. SIM =see Subscriber marginally relevant for the (for a SD review of mobile research Bagadia idenand Bansal, 2016). tity module. SMS = Short message service.

NFC mobile payment use

–4–

10th March 2017 (ECRA-D-16-00341R4)

and intangible goods (Guo and Bouwman, 2016). A common advantage of remote MPS is that they do not require specific devices but can be accessed with the help of any customer premise equipment that is capable of linking to cellular radio networks. Proximity payment systems (see right part of Figure 1) contactlessly transmit transaction data from the device of MNO customers to a POS terminal in “real” stores via various short-range communication technologies such as bluetooth low energy, code scanning, infrared and NFC. Code scanning requires camera phones and POS terminal optical scanning functionalities. A two-dimensional (bar- or quick response) code is either generated by a POS terminal and scanned with a consumer’s camera phone or vice versa (Liébana-Cabanillas et al., 2015). NFC MPS transfer (bank) data over high frequency magnetic fields. They differ with respect to the storage location of a customer’s bank information. Some systems store it on the microchip of bank-issued NFC (credit or debit) cards or NFC labels. Others require a smartphone which keeps the information in a so-called secure element. This can be designed as hardware component embedded in the device of MNO subscribers (e.g., Apple Pay). Alternatively, it can be located in a cloud storage unit only accessible with a token that is deposited in an internal storage unit of the subscribers’ handsets (e.g., Android Pay). Thirdly, it can be part of the subscriber identity module (SIM) or secure digital (SD) memory card, respectively, which is inserted into the MPS users’ handset (Coskun et al., 2013; Ondrus, 2015; Urien and Aghina, 2016). The latter case corresponds to the MNO’s NFC payment system under study (see Figure 1). A general feature of most proximity MPS is that they do not run on all types of mobile devices but imply application-specific hard- and/or software components. In conclusion, MPS variants differ with the regard to the sales situations they are designed to handle (i.e., online transactions or purchases at “real” POS). Therefore, empirical studies, which fail to specify the type of MPS they are examining, are difficult to interpret. Consequently, our subsequent research review focuses on investigations of the adoption of proximity MPS and excludes the vast literature on MPS in general and on remote MPS in particular. 3.

Literature review and research hypotheses

3.1

Review of extant NFC payment use research

Proximity MPS studies either examine MNO customers’ decision to start using proximity payment systems in general or NFC payment in particular. Table 1 profiles pertinent prior empirical publications. In terms of theoretical frameworks, the large majority of earlier work applies a narrow set of lenses, namely theory of reasoned action (TRA; Fishbein and Ajzen, 1975), technology ac-

NFC mobile payment use

–5–

10th March 2017 (ECRA-D-16-00341R4)

ceptance model (TAM; Davis, 1989), diffusion of innovation theory (DOI; Rogers, 2003) and unified theory of acceptance and use of technology (UTAUT; Venkatesh et al., 2003). These conceptual frameworks emphasize MNO customers’ behavioral intention to use NFC MPS in the future. This criterion does not necessarily converge with actual adoption behaviors. In addition, regardless of whether the focus is on proximity MPS in general or on NFC MPS in particular, prior investigations are typically based on (paper- or web-based) questionnaires or (face-to-face or telephone) interviews, respectively and ask participants to report their intentions to use proximity/NFC MPS. Regrettably, such purely questionnaire-based research designs suffer from severe common method bias: Relationships between predictor and criterion variables are likely to be distorted by “artificial covariance ... that would not otherwise exist at the same level in real-life settings.” (Podsakoff et al., 2003, p. 881) Prior studies generally incorporate MNO customers’ perceptions of paying via NFC (e.g., in terms of compatibility, usability or security) as potential predictors of NFC MPS use measures. The strength of a focus on customer perceptions is that it helps to understand motives for using NFC MPS. However, perception studies are less strong when it comes to derive suggestions for MNO practitioners because they can not revert to perceptional variables at the individual customer level but only to a limited set of objective individual characteristics stored in their customer data files. Therefore, a few pieces of research have recently started to include such non-perceptual variables in attempts to predict NFC MPS use (e.g., Leong et al., 2013; Liu et al., 2013; Shaw, 2015b). A final shortcoming of prior proximity MPS use research is that many studies rely on small, convenience samples (e.g., Apanasevic, 2013; Ferreira et al., 2014; Silic et al., 2014). In summary, a plethora of prior empirical investigations on proximity MPS adoption at the individual customer level have dealt with intentions to use MPS in the future and have priTable 1 Profile of general proximity and NFC MPS use studies Author(s)

Type of MPS under study Samplea

Dependent variableb Key use correlatesc

Apanasevic (2013)

NFC

OAS (11; six FUI countries)d

Balachandran and Tan (2015)

NFC

OAS (487; Malaysia)

Ceipidor et al. (2012)

Proximity in general

OAS (1,001; FUI Italy)

FUI

• Lack of uniform technological standards (–) • Network externalities (–) • Availability of value-added services (+) • Information about MPS (+) • Lack of security (–) • Lack perceived advantage (–)

NFC mobile payment use

–6–

10th March 2017 (ECRA-D-16-00341R4)

Cocosila and Trabelsi (2016

NFC

OAS (289; Canada)

Ferreira et al. (2014)

NFC

OAS (30; Por- FUI tugal)

Jenkins and Ophoff (2016) Kerviler et al. (2016)

NFC

OSS (331; FUI South Africa) OAS (363; FUI France)

Lee et al. (2015)

NFC

Leong et al. (2013) Li et al. (2014)

NFC

Luna et al. (2016)

NFC

OAS (191; Spain)

Makki et al. (2016) Moroni et al. (2015)

NFC

OSS (412; FUI USA) OAS (1,051; FUI Italy)

Proximity in general

NFC

NFC

FUI

OAS (206; FUI South Korea) OSS (262; FUI Malaysia) OAS (377; FUI China) FUI

Morosan and NFC DeFranco (2016a)

RAS (794; USA)

FUI

Morosan and NFC DeFranco (2016b)

RAS (347; USA)

FUI

Ooi and Tan (2016) Ozturk (2016)

NFC

OAS (459; Malaysia) OAS (305; USA)

FUI

Pal et al. (2015)

NFC

FUI

Pham et al. (2015)

NFC

OAS (270; Thailand OAS (402; China)

NFC

FUI

FUI

• Preference for established payment methods (+) • Utilitarian and enjoyment values (+) • Psychological and privacy risks (–) • Complexity (–) • Lack of availability at POS terminals (–) • Performance (+) • Security (+) • Perceived value (+) • Utilitarian, hedonic and social benefits (+) • Financial and privacy risks (–) • Ease of use (+) • Benefits (+) • Ease of use (+) • Usefulness (+) • Ease of use (+) • Compatibility (+) • MPS knowledge (+) • Compatibility (+) • Attitude toward NFC MPS use (+) • Social norms (+) • Level of innovativeness (+) • Self-efficacy (+) • Compatibility with needs (+) • Habits (+) • Lifestyle (+) • Performance expectancy (+) • Hedonic motivation (+) • Habit (+) • Hedonic motivation (+) • Performance expectancy (+) • Privacy concerns (–) • Trust (+) • Ease of use (+) • Ease of use (+) • Usefulness (+) • Risk (–) • Ease of use (+) • Usefulness (+) • Compatibility (+) • Usefulness (+) • Risk (–)

NFC mobile payment use

–7–

10th March 2017 (ECRA-D-16-00341R4)

Sam et al. (2014) NFC

OSS (200; China)

FUI

Shaw (2014)

Proximity in general

OSS (284; Canada)

FUI

Shaw (2015a)

Proximity in general Proximity in general NFC

OAS (597; USA) OAS (411; USA) OAS (196; China) OAS (296; South Korea) OAS (585; South Korea) OAS (7; France OAS (244; United Kingdom) OAS (156; Malaysia)

FUI

OAS (418; Malaysia)

FUI

Shaw (2015b) Shen et al. (2017) Shin (2009)

Proximity in general NFC

Shin and Lee (2014) Silic et al. (2014) NFC Slade et al. (2015)

NFC

Tan et al. (2014)

NFC

Tang et al. (2014)

NFC

a) b) c) d)

PUB-S

• Ease of use (+) • Usefulness (+) • Personal innovativeness (+) • Usefulness (+) • Trust (+) • Informal learning (+) • Ease of use (+) • Usefulness (+) • Effort expectancy (+) • Performance expectancy (+) • Initial trust (+) • Compatibility (+) • FUI (+)

FUI

• Usefulness (+)

FUI

• Security (+)

FUI

• Performance expectancy (+) • Habit (+) • Hedonic motivation (+) • Personal innovativeness (+) • Ease of use (+) • Usefulness (+) • Effort expectancy (+) • Facilitating conditions (+) • Hedonic motivation (+)

FUI FUI

FUI

OAS = Opportunity adult sample. OSS = Opportunity student sample. RAS = Random adult sample. The first entry in brackets indicates the sample size, the second entry is the country in which data was collected. FUI = Future use intention. PUB-S = Past use behavior (self-reported). Sign in brackets indicates direction of correlation. Finland, France, Italy, Netherlands, Norway and United Kingdom.

marily employed perceptual measures as predictors of initial MPS takeover plans. Against this background, the present work contributes to the literature on NFC MPS adoption by exploring associations between actual behavioral MPS use measures and a set of objectively observable customer-related characteristics.

3.2

Research hypotheses and questions

The collaborating MNO enabled us to extract a total of seven socio-demographic and mobile communication services (MCS) usage-related individual characteristics from its customer data base. Furthermore, two variables were construed from the device type recorded for each subscriber by the MNO by combining this information with data collected from the websites of

NFC mobile payment use

–8–

10th March 2017 (ECRA-D-16-00341R4)

mobile phone manufacturers. Finally, a tenth variable was derived with the help of information collected from the online data base of the credit card company offering the studied NFC MPS. Subsequently, we develop hypotheses on likely associations between each of the ten variables and the probability that an MNO customer actually uses the carrier’s NFC payment system. The first two characteristics available were a customer’s gender and age. Some NFC MPS adoption studies found no significant correlations between MNO customers’ gender and NFC payment use (e.g., Leong et al., 2013; Shin, 2009; Tan et al., 2014). However, the majority of prior work suggests that male subscribers are more prone to use their handset for over-thecounter payments than their female counterparts (e.g., Ceipidor et al., 2012; Ferreira et al., 2014; Henry et al., 2015; Oh et al., 2014). There are two explanations for this association: Firstly, males are usually more technology-savvy than females (Tan et al., 2014). Secondly, males have less security concerns in using their device for MPS (Jenkins and Ophoff, 2016; Liu et al., 2013). Therefore, our first hypothesis (H) is: H1:

Male MNO subscribers are more likely to use NFC MPS than female customers.

With regard to the age variable, there is evidence suggesting that younger MNO subscribers are more inclined to use NFC MPS than older customers (e.g., Accenture, 2015; Ceipidor et al., 2012; Ferreira et al., 2014; Oh et al., 2014). An explanation for this observation is that age is positively related to fraud and privacy concerns in the context of MPS (Liu et al., 2013). Put differently, older customers are more likely to use more established payment methods such as cash or magnetic strip cards because they believe that these methods have strong security advantages (Bundesbank, 2015; Trütsch, 2016). Hence, we pose: H2:

Older MNO subscribers are less likely to use NFC MPS than younger customers.

Two device-related characteristics, which were at hand, were the screen size of a customer’s smartphone and its normal stand-alone retail sales price (when bought without a postpaid MCS contract). Studies on correlations between the screen size of MNO customers’ handsets and mobile services use contain mixed results. One the one hand, some studies suggest that larger screens increase the usability of the device (Carter and Yeo, 2016; Raptis et al., 2013). On the other hand, large displays can make it more challenging for customers to handle complex tasks on their equipment with only one hand (Xiong and Muraki, 2016) and to conveniently carry it while being on the move (Kim et al., 2011). This may deter MNO customers to use a (proximity) MPS such as NFC payment because its handling may be perceived as bur-

NFC mobile payment use

–9–

10th March 2017 (ECRA-D-16-00341R4)

densome on a large device. However, most empirical studies indicate that larger device screens coincide with a higher likelihood that the device owner uses advanced mobile data applications (e.g., Ericsson, 2015; Gerpott et al., 2013; Hsiao and Chen, 2015; Kim and Sundar, 2014). Therefore, we surmise: H3:

The screen size of MNO subscribers’ handsets is positively related to the likelihood of NFC MPS use.

The second device variable available was its average retail sales price. It may be taken as a proxy of the technical performance level linked to key device components (e.g., battery capacity, processor speed or memory size). In general, a higher retail price level is indicative of more advanced device capabilities. Hence, subscribers using less expensive devices may believe that their handset is restricted in its functionalities. As a consequence, they may shy away from trying advanced mobile data services such as NFC MPS (e.g. Kim, et al., 2015; Shin, 2015). Therefore, prices which MNO customers paid for their device and NFC MPS use should be positively correlated. However, Hsiao and Chen (2015) found no significant correlation between the price levels of MNO customers’ handset and their NFC MPS use. Unfortunately, their study only reports simple bivariate correlations between self-report measures and thus does not address whether results are different in case that multivariate methods and objective use measures are applied. Therefore, notwithstanding the Hsiao and Chen (2015) investigation, we expect: H4:

The price level of MNO subscribers’ handsets is positively related to the likelihood of NFC MPS use.

The collaborating MNO provided us with information on five variables reflecting subscribers’ MCS use volumes and revenues. Earlier work incorporating various MCS use metrics mainly detected positive relationships between mobile service consumption levels and use of advanced MCS (e.g., Gerpott, 2015; Gerpott and Meinert, 2016b; Ghose and Han, 2011). According to Wei (2008, p. 39), an “activation effect” or a “positive .. spillover” (Xu et al., 2014, p. 109) results from the consumption of one class of cellular services to the use of other mobile service categories. Therefore, the take-up likelihood of NFC MPS should increase with the consumption levels of various other mobile services. Thus, we posit: H5a: MNO subscribers’ mobile voice use level is positively related to the likelihood of NFC MPS use. H5b: MNO subscribers’ SMS use level is positively related to the likelihood of NFC MPS use. H5c: MNO subscribers’ mobile Internet use level is positively related to the likelihood of NFC MPS use.

NFC mobile payment use

– 10 –

10th March 2017 (ECRA-D-16-00341R4)

H5d: MNO customers’ subscription to a music streaming offer is positively related to the likelihood of NFC MPS use. H5e: MNO subscribers’ monthly bill for mobile services is positively related to the likelihood of NFC MPS use. Since MNO customers can switch handsets relatively quickly, a more persistent core bottleneck for MPS market success may be the availability of merchants supporting a MPS. For proximity services such as NFC, availability is mainly determined by the number of retail stores equipped with NFC-enabled POS terminals in the vicinity of the subscribers’ place of residence since people tend to make most of their smaller purchases in the neighborhood of their residence (Google, 2014; Nielsen, 2016). Therefore, a higher number of NFC-enabled POS terminals in the residential area of MNO customers is likely to raise their propensity of using NFC MPS. Correspondingly, we propose: H6:

The number of NFC terminals in MNO subscribers’ residential area is positively related to the likelihood of NFC MPS use.

The preceding hypotheses have treated NFC MPS use or adoption as a dichotomous yes-or-no phenomenon. However, following Rogers (2003, p. 299) there are also “many important differences between earlier and later adopters of innovations in (1) socioeconomic status, (2) personality variables, and (3) communication behavior.” Accordingly, Pal et al. (2015), who examined NFC MPS adoption in a sample of 270 MNO customers equipped with credit cards, detected that early and late adopters differ in terms of “various [personality] factors that ultimately can affect the adoption of NFC payment system” (Pal et al., 2015, p. 23). Consequently, the timing in the sense of the initial use start of an NFC MPS at an earlier or later date may be correlated with the variables covered in the preceding study hypotheses. However, to the best of our knowledge, no study has explored associations between each of the ten study variables and the timing of NFC MPS use. Therefore, we raise the following research question (RQ): RQ1: How are socio-demographic, device- and usage-related characteristics as well as NFC merchant availability related to the timing of NFC MPS use? 4.

Methods

4.1

Sample generation

The data set underlying the empirical analysis was gained from the customer information and billing system of the German subsidiary of a large MNO running cellular networks in about 15 countries. The firm controls a nationwide 2G cellular network in Germany since the early

NFC mobile payment use

– 11 –

10th March 2017 (ECRA-D-16-00341R4)

1990s, a 3G network for 15 years and is currently completing the roll-out of a 4G/LTE infrastructure in this country. The MNO supported our research by granting us access to data of an anonymous random sample of 264,679 residential postpaid subscribers who were customers of the company at least between June 2014 and May 2015 and who had not cancelled their contract during these 12 months (= the study period). For each subscriber the MNO provided monthly data on whether (s)he was subscribed to the NFC payment system or not, the use of mobile voice, SMS and mobile Internet in general and of a music streaming service in particular along with basic demographics and device entries as of 31st May 2015. Within the study period, only subscribers whose handsets supported both NFC and ran the operating system (OS) Android were technically eligible for the MNO’s NFC MPS. Therefore, we excluded subjects from the initial data set who were not equipped with an NFCenabled handset in each study month. Furthermore, subscribers whose handset ran an OS other than Android were discarded. In addition, we dropped subscribers who had already signed up for the NFC MPS before July 2014.3 Accordingly, a small number of pilot users, who mainly were employees of the collaborating MNO and who evaluated the service before it was launched on the German market, were dropped. The application of the three selection criteria resulted in an adjusted sample of 219,441 subscribers. 677 of the remaining subjects (= 0.3%) were registered users of the carrier’s NFC payment system in at least one month between July 2014 and May 2015.4 To prevent that the statistical results may be severely distorted by unequal sizes of the NFC MPS user and non-user groups we derived a random sample of 677 individuals (= size of the user subgroup) from the subgroup of 218,764 NFC MPS non-users.5 As a result, the final study sample encompasses 1,354 MNO subscribers.

4.2

Variable measurements

NFC payment use criteria, socio-demographic characteristics as well as variables reflecting a subscriber’s consumption pattern of mobile communication and Internet protocol-based services were obtained from log data of the MNO’s billing and network management systems. Measures of the two device-related characteristics and of the number of NFC terminals in a

3

4

5

The data does not allow a distinction between subjects who had already used the NFC MPS before June 2014 and those who started using it in June 2014. Hence, a positive MPS subscription entry for June 2014 may stem from a sign on in June 2014 or in an earlier month preceding the study’s time window. Therefore, we deleted 33 subjects who were recorded as NFC MPS users already in June 2014. The MNO charges subscribers a monthly registration fee of 0.99 EUR for its NFC MPS, unless their annual transaction revenues exceed a threshold of 600 EUR. A technical discussion of the properties and advantages of the under-sampling approach can be found in Amin et al. (2016) and Kotsiantis et al. (2006).

NFC mobile payment use

– 12 –

10th March 2017 (ECRA-D-16-00341R4)

participant’s residential area were generated by the researchers based on publicly available information from the device manufacturers’ and credit card company’s websites. 4.2.1 NFC payment use criteria The study’s focal dependent variables are NFC MPS use and timing of NFC MPS use. Both variables are based on the MNO’s monthly recorded status information for each participant on whether (s)he holds an NFC MPS subscription or not. The first study criterion, NFC MPS use, reflects the yes-or-no decision of the individual customer to sign up for the MNO’s NFC MPS at least for one month of the study period. The binary variable scores 1 for 677 subscribers who registered for the NFC payment system for the first time between July 2014 to May 2015. None of the 677 users stopped their initial subscription at least until the end of May 2015. The second study criterion, timing of NFC MPS use, is the total number of months for which an NFC MPS user was signed up for the service from July 2014 to May 2015. Higher (lower) variable values correspond to an earlier (later) use decision. The minimum variable value amounts to 1 and reveals that a subscriber’s positive use started in May 2015. The maximum value of 11 indicates that a person’s subscription began in July 2014. As shown in Figure 2, by the end of November 2014 (= study month #5) 210 subscribers (= 31% of the total number of consumers starting to use the service until May 2015) had signed up for the NFC payment system. The highest share (16.1%) of new NFC MPS users was recorded for the 11th study month (i.e., May 2015). 4.2.2 Potential antecedents of NFC payment use This section outlines the measurement of two socio-demographic, two device-related and five usage-related customer characteristics as well as the availability of NFC terminals at a subscriber’s residential area.

NFC mobile payment use

– 13 –

10th March 2017 (ECRA-D-16-00341R4)

Figure 2 Share of NFC payment service users at the end of each month in the study period (n = 677)a Cumulative share of users

100.0%

100% 90%

83.9%

80%

72.2%

70%

60.4% 53.8%

60% 50% 40% 30% 20% 10%

38.7% 31.0% 23.2% 17.3% 11.4% 6.8% 0.0% Study month

0% 0

1

2

3

4

5

6

7

8

9

10

11

a) Period # 0 = June 2014. Period # 11 = May 2015. NFC = Near field communication.

The two socio-demographic variables, which were at hand from the MNO’s customer files, are gender and age. 65.7% of the subjects in the sample were male (see variable #1 in Tables 2 and 3). 51.4% of the subscribers were older than 40 years (see variable #2 in Table 2). The mean age in the sample was 40.9 years (SD = 13.1). The median age amounted to 41 years (see also variable #2 in Table 3). Gender and age distributions in the study sample deviate significantly from the adult population in Germany such that women and people older than 50 years were underrepresented (Statistisches Bundesamt, 2016). However, the sample’s gender and age distributions resemble those of other samples in recent studies on NFC MPS use in Germany (e.g., Bitkom, 2014; IP, 2016). The MNO recorded the device type of the customers at the end of the study period. Based on this information, two device-related characteristics were constructed from data published on the websites of various device manufacturers, namely, device screen size and its price. The screen size of a customer’s device is measured by the diagonal length of its display screen. 58.9% of the subscribers were equipped with a handset whose screen size was 5 inches or larger (see variable #3 in Table 2). The mean screen size amounted to 4.9 inches (median = 5; SD = 0.45; see variable #3 in Table 3). Based on market research on the stand-alone retail sales prices of the most common handsets in our sample, we found that a cut-off value of 650

NFC mobile payment use

– 14 –

10th March 2017 (ECRA-D-16-00341R4)

EUR was best suited for separating high-end devices (e.g., Samsung Galaxy S5, Samsung Galaxy Note 4) from less sophisticated devices (e.g., Sony Xperia M2, LG Electronics G3). Accordingly, we captured retail price level of a subscriber’s device as a dichotomous variable. It is coded as 1 for subscribers whose device was priced at 650 EUR or more when launched on the German market (= premium device), and 0 otherwise (= basic device). 51.0% of the subscribers in the sample were equipped with a premium device (see variable #4 in Tables 2 and 3).

Table 2 Sample characteristics (n = 1,354) Characteristics

Number (n)

Percentage (%)

1. Gender – Male – Female

889 465

65.7 34.3

2. Age – 18-20 years – 21-30 years – 31-40 years – 41-50 years – 51-60 years – > 60 years

31 336 305 324 264 94

2.3 24.8 22.5 23.9 19.5 7.0

3. Device screen size – < 4.0 inches – 4.0-4.5 inches – 4.5-4.9 inches – 5.0-5.4 inches – ≥ 5.5 inches

5 286 266 658 139

0.4 21.1 19.6 48.6 10.3

Characteristics 4. Device price levela – High (“premium”) – Low (“basic“)

Number (n) 690 664

Percentage (%) 51.0 49.0

5. Music streaming service subscription – Yes 110 8.1 – No 1,244 91.9 6. Number of NFC terminals in own residential areab – 0-3 terminals 668 – 4-6 terminals 354 – 7-9 terminals 176 – ≥ 10 terminals 156

49.3 26.2 13.0 11.5

a) Device price level at market entry of smartphone  650 EUR = Premium smartphone. Device price level < 650 EUR = Basic smartphone. b) Residential area = Self-reported five-digit postal code area in which an MNO subscriber had its main residence on May 31, 2015. NFC = Near field communication.

In addition, we obtained five indicators capturing facets of subscribers’ use of MCS. Firstly, in line with earlier work (Ghose and Han, 2011; Grzybowski and Pereira, 2008), mobile voice use intensity is measured by the average monthly number of mobile voice connections a subscriber initiated in the study’s time window. For NFC MPS users, arithmetic means were calculated by summing up the monthly outgoing mobile voice connections in the period from the 1st June 2014 to the end of the month before customers started to use NFC MPS and dividing it by the number of months that had passed from June 2014 to the first use month. For NFC MPS non-users, average calculations refer to the whole study period. This measurement approach tries to control for contaminating reinforcement effects of NFC MPS subscription on

NFC mobile payment use

– 15 –

10th March 2017 (ECRA-D-16-00341R4)

MCS use intensities. The sample mean (median) number of mobile voice connections amounted to 51.6 (34.5) (SD = 53.3). Secondly, as suggested by prior investigations (Gerpott, 2015, Gerpott et al., 2014), SMS use intensity is gauged by the average number of SMS a subscriber sent during a month. Analogous to the procedure just explained, we computed a customer’s average monthly number of SMS sent either for the pre-MPS-adoption time window or for the full study period. In our sample, the average monthly number of SMS sent was 31.3 (median = 10.1; SD = 96.4). Thirdly, similar to Gerpott (2011) and Ghose and Han (2011), mobile Internet use intensity was metered by the Internet protocol down- and upload traffic over cellular networks a subscriber generated during a month in megabytes. The time window for computing this mean was derived as just described for the preceding two MCS use measures. The sample’s mean monthly mobile Internet data volume was 860.1 megabytes (median = 547.2; SD = 1036.0). Fourthly, music streaming service subscription was measured as a dichotomous variable coded 1 for subjects who were subscribed to the MNO’s music streaming offer for at least one month in the study period and 0 otherwise. The monthly subscription fee allowed subscribers unlimited music consumption both over fixed-line Wifi networks and cellular infrastructures.6 In the sample, until May 2015, 8.1% of the sample members had signed up for the MNO’s music streaming offer. Finally, we obtained the mean total amount subscribers paid for their mobile service consumption in each study month. The time frames for the average revenue measure of MPS users and non-users was construed as explained above. The sample mean (median) of subscribers’ monthly invoice amount for mobile services was 41.9 (38.3) EUR (SD = 19.0).

6

The MNO excludes the Internet protocol traffic subscribers generate through music streaming service use from their overall mobile Internet data volume measurements. Stated differently, the traffic subscribers create through music streaming is not included in their mobile Internet data volume.

Table 3 Descriptive statistics and bivariate correlations of predictors

NFC payment adoption D-16-00341R4)

Variables

Descriptive statisticsa M SD

1.

1. Male gender = truec

0.657

–

054*

020

–

2. Age (years) 3. Device screen size (inch) 4. Premium device = trued

0.475

40.936 13.051 [41.000]

2.

3.

Correlationsb 5. 6.

4.

229***

112***

–098*** –031

4.879 [5.000]

0.448

041

–030

0.510

0.500

–054

020 –202

– 16 –

608***

– 487***

***

084

*

– 104

**

5. Average monthly number of MV connections (ln)

3.482 [3.492]

0.965

–025

6. Average monthly number of SMS sent (ln)

2.439 [2.398]

1.507

–105** –238*** –026

7. Average monthly MI data volume (ln, MB)

5.995 [6.118]

1.206

052

–390***

238***

8. Music streaming service subscription = truec

0.081

0.273

080*

–249***

019

9. Average monthly invoice amount for mobile services (ln, EUR)

3.660 [3.665]

0.421

064+

–134***

280***

182***

10. Number of NFC terminals in own residential area

4.735 [4.000]

4.835

014

–018

010

014

–005

031

–134***

10th March 2017 (ECRA-

7.

8.

9.

141

141

10.

103***

–189*** –239*** –416*** –199*** –153*** 129*** –013

370***

129***

239*** –022

– 399***

000 267

***

–

115***

***

046+

253***

027

***

–003

360

***

087

179***

044

225***

070*

283***

461***

009 071*

382

387***

223***

–114**

080*

075*

308***

–

134***

385***

256***

444***

119**

–

110**

011

328***

107**

–019

055*

–

–002

068+

–021

–008 –

a) Statistics for socio-demographic variables # 1 to 2, device-related variables # 3 to 4, the contract-related variable # 8 and the geo-demographic variable # 10 refer to the end of the study period (May 31, 2015). Variables # 5 to 7 and variable # 9 are intraindividual monthly averages from 1st June 2014 to the end of the month before customers started to use NFC MPS (for users) or May 2015 (for non-users), respectively. n = 1,354. M = Mean. SD = Standard deviation. Figures in squared brackets below mean = Median. b) Leading decimals are omitted for correlation coefficients (e.g., 054 = 0.054). Figures above the main diagonal are Pearson product-moment correlations (r) in the full sample (n = 1,354). Coefficients below the main diagonal are Pearson correlations in the subsample of 677 payment service adopters. MB = Megabyte. MI = Mobile Internet. MV = Mobile voice. NFC = Near field communication. SMS = Short message service. Significance levels are flagged as follows: +p ≤ 0.10; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 (twotailed). c) Binary variable with 1 = true and 0 = otherwise. d) Binary variable is coded 1 for device retail price of smartphone ≥ 650 EUR when launched in Germany and 0 otherwise.

NFC payment adoption

– 17 –

10th March 2017 (ECRA-D-16-00341R4)

In the present sample, the distributions of the four metrically scaled use intensity measures (i.e., mobile voice, SMS and mobile Internet volumes, invoice amount) were highly skewed by a small group of “heavy users”. This result is in line with many earlier studies (e.g., Gerpott, 2015; Peng et al., 2013). In light of these variable distributions, we took the natural logarithm of the original values to avoid a distortion of the results by non-normal value distributions.7 Descriptive statistics of the transformed measures are reported in Table 3 (see variables #5 to #7 and variable #9). Finally, we combined information on a subscriber’s place of residence from the MNO’s files with data on the availability of NFC terminals taken from the website of the credit card company which supported the MNO in running the NFC MPS to get the number of NFC terminals in a subscriber’s resident area at the postal code level. This variable captures the number of merchants equipped with NFC-enabled POS terminals in the postal code area of a subscriber’s home address on 31st May 2015. 50.7% of the subscribers lived in areas with 4 merchants or more who accepted NFC payments (see variable #10 in Table 2). The average (median) number of merchant NFC terminals in a subscriber’s vicinity was 4.7 (4.0; SD = 4.8; see variable #10 in Table 3). Table 3 displays bivariate Pearson product-moment correlations (r) of the ten independent variables. The highest coefficient in the correlation matrix is 0.608 (r34) but most of the remaining coefficients are considerably below 0.400. Furthermore, in the multivariate analysis explaining the timing of NFC MPS use the variance inflation factors (VIF) are beneath a value of 1.65 (see also footnote c in Table 4). This is far below maximum cut-off thresholds of 5 to 10 suggested in the literature (Hair et al., 2014) as being indicative of severe multicollinearity problems. Consequently, multicollinearity among the predictors selected for the final analysis does not pose a validity threat to the present study’s results.

4.3

Statistical procedure

The first study criterion captures whether subscribers had started to use the MNO’s MPS within the period from July 2014 to May 2015 or not. The second criterion is based on counts of the number of months in which each of the 1,354 subjects was registered to the MNO’s NFC payment system. Across the two criteria, potential variable values range from 0 to 11 [months] with 50% of the subjects scoring 0 on the binary use measure. Thus, our criteria value distribution is highly left-censored. For this type of (interrelated) dependent measures, cen7

Given that ln(0) is not defined and ln(0 < x < 1) is negative, we increased the original values by 1 before taking their natural logarithms.

NFC mobile payment use

– 18 –

10th March 2017 (ECRA-D-16-00341R4)

sored regression analysis models for count data prevent severe distortions of the parameter estimates (Greene, 2012; Hilbe, 2008; Wooldridge, 2016). According to Greene (2012), the methods appropriate for explaining the type of criteria relevant in this study are zero-inflation and hurdle models. Zero-inflation models simultaneously incorporate both binary and count variables for estimating zero counts. For situations in which “only one [of the two] process[es] generates zeros”, Hilbe (2008, p. 174) suggests to prefer a two-step hurdle model which strictly “separates the modeling of zeros from the modeling of counts” over a single-step zero-inflation approach. In the study sample, a value of zero for the binary use criterion implies that a subscriber was not registered for the NFC payment system within the whole study period. Put differently, individuals with zero counts do not qualify for being assigned values of at least 1 on the timing criterion. Accordingly, we ran a binary Probit regression to contrast NFC MPS users and nonusers in the whole sample of 1,354 subjects. Subsequently, we computed linear ordinary least squares (OLS) and negative binomial regressions to explain the point in time at which the 677 subjects who scored 1 on the NFC use criterion first signed up for the NFC MPS offer. All calculations were made with the relevant regression procedures implemented in the Stata 12.1 software package. 5.

Empirical results

Table 4 shows unstandardized coefficients of the Probit regression with the binary adoption measure as the criterion and the OLS regression as well as the negative binomial count regressions, each with adoption timing as the dependent variable. The Probit analysis reveals seven statistically significant (p ≤ 0.05) coefficients (see predictors #1-3, #7-10 in column “Probit” of Table 4). The Probit likelihood ratio χ2 amounts to 480.54 (df = 10; p ≤ 0.001). Hence, the overall fit of the Probit model with the data makes it reasonable to test the study hypotheses. H1 states that male MNO subscribers are more likely to use NFC MPS than their female counterparts. The Probit regression yields a significantly (p ≤ 0.001) positive association between male gender and NFC MPS use. The likelihood of NFC MPS adoption among males is 34.1%

NFC mobile payment use

– 19 –

10th March 2017 (ECRA-D-16-00341R4)

Table 4 Regression models for NFC MPS usea

Predictors

Criterion/model NFC MPS use Timing of NFC MPS use b Probit Linear OLSc NBRc

1. Male gender = trued

0.854 (0.341)*** [0.084]

–0.082 [0.307]

–0.012 (–0.085) [0.043]

2. Age (years)

0.011 (0.004)*** [0.003]

0.008 [0.011]

0.001 (0.008) [0.002]

3. Device screen size (inch)

1.292 (0.515)*** [0.139]

–1.066** [0.370]

–0.156 (–1.088)** [0.054]

4. Premium device = truee

–0.036 (–0.014) [0.099]

1.389*** [0.270]

0.203 (1.414)*** [0.040]

5. Average monthly number of MV connections (ln)

0.026 (0.010) [0.045]

0.014 [0.139]

0.002 (0.012) [0.020]

6. Average monthly number of SMS sent (ln)

–0.027 (–0.011) [0.027]

–0.045 [0.086]

–0.006 (–0.042) [0.012]

7. Average monthly MI data volume (ln, MB)

0.170 (0.068)*** [0.046]

0.050 [0.145]

0.007 (0.049) [0.021]

8. Music streaming service subscription = trued 0.981 (0.391)*** [0.185]

0.809* [0.331]

0.117 (0.812)* [0.046]

9. Average monthly invoice amount for mobile services (ln, EUR)

–0.316 (–0.126)** [0.114]

–1.115*** [0.278]

–0.158 (–1.102)*** [0.039]

10. Number of NFC terminals in own residential area

0.021 (0.008)* [0.009]

0.034+ [0.020]

0.005 (0.032)+ [0.003]

14.886***

3.079***

Intercept Log Likelihood Likelihood ratio χ 2 (df = 10)

–7.365*** –698.25 480.54***

–1,701.77 50.54***

–1,739.92 46.53***

a) The table shows unstandardized coefficients of the two-step approach described in section 4.3 encompassing firstly a Probit and secondly a linear OLS or negative binomial regression, respectively. Values in round brackets are marginal effects evaluated at the predictor means. Robust (Eicker-Huber-White) standard errors are given in squared brackets below the unstandardized coefficients. Significance tests for model coefficients follow the Wald procedure with robust standard errors. NBR = Negative binomial regression. OLS = Ordinary least squares. The remaining abbreviations are explained in footnote b of Table 2. b) Non-users = 0. Users = 1. The Probit regression is run in the total sample (n = 1,354). The classification accuracy of the Probit regression amounts to 73.3%. c) The criterion in the linear ordinary least squares and negative binomial regressions is the number of months, in which an MNO customer was signed up for the operator's NFC MPS. Linear OLS and NBR analyses are limited to the subsample of 677 NFC MPS users. Variance inflation factors of the ten predictors range from a minimum of 1.03 to a maximum of 1.65. d) Binary variable with 1 = true and 0 = otherwise. e) Binary variable is coded 1 for device price level at market entry of smartphone ≥ 650 EUR and 0 for device price level < 650 EUR. Significance levels are flagged as follows: +p ≤ 0.10; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 (two-tailed).

NFC mobile payment use

– 20 –

10th March 2017 (ECRA-D-16-00341R4)

higher than among females if all other independent variables are set at the mean (see marginal effect of predictor #1 in column “Probit” of Table 4).8 Hence, our findings support H1. H2 posits that older subscribers are less likely to use NFC MPS than younger customers, probably due to higher security concerns. Our findings show a significantly (p ≤ 0.001) positive relationship between customer age and NFC MPS use. According to the marginal effect estimate of predictor #2 in column 2 of Table 4, a one year increase in age coincides with an 0.4% higher probability of NFC MPS adoption. Thus, H2 is rejected. A cause for this result may be that the relationship between age and NFC MPS use is not linear but inversely Ushaped. In fact, 24.4% of the 677 users were not more than 30 years old, whereas the corresponding share among the 677 non-users was 29.8% (i.e., 5.4 percentage points higher). Conversely, 50.8% of the users and 42.1% of the non-users were aged between 31 and 50 (difference = 8.7 percentage points). Lastly, the share of users who were at least 51 years old amounted to 24.8% which is 3.3 percentage points lower than the relevant proportion of 28.1% among non-users. Hence, our study supports the conclusion that people in their 30s and 40s are most inclined to use NFC MPS. Perhaps they earn more than individuals below this age range and are less rigid or loss averse than customers who are older than 50 years. H3 assumes that the screen size of MNO subscribers’ handsets is positively related to the likelihood of NFC MPS use. According to the Probit analysis, device screen size and NFC MPS use are significantly (p ≤ 0.001) positively associated. Evaluated at the mean, the marginal statistical effect of a screen size raise by one inch on the likelihood of NFC MPS adoption is 51.5% (see predictor #3 in column “Probit” of Table 4). This is in line with H3. H4 predicts that subscribers with a higher-priced smartphone are more likely to use NFC MPS than their counterparts without such a device. As can be taken from Table 4, subscribers who are equipped with a premium smartphone are not more prone to use the NFC MPS than consumers with a less expensive device (see predictor #4 in column “Probit” of Table 4). However, if the original device price is used instead of the transformed binary price level measure, device price is positively related to the NFC MPS use at p < 0.01. Given these mixed results, H4 can be neither clearly supported nor refuted. Hence, future research should take a closer look at customer characteristics associated with the purchase of more expensive smartphones (e.g., openness for innovative services, conspicuous consumption needs) to better understand which customer attitudes are actually captured by this variable. 8

For all independent variables the (non-linear) marginal effects according to the Probit analysis of the binary adoption criterion do not deviate strongly from (constant) marginal effects estimates obtained in an OLS regression. The latter is available from the authors on request.

NFC mobile payment use

– 21 –

10th March 2017 (ECRA-D-16-00341R4)

H5a and H5b suggest that MNO subscribers’ mobile voice as well as SMS use levels are positively related to NFC MPS use. The Probit regression uncovers no significant associations between subscribers’ NFC MPS use and their average monthly number of mobile voice connections on the one hand as well as their average monthly number of SMS sent on the other (see predictors #5 and #6 in column “Probit” of Table 4). Put differently, “heavy users” of established MCS such as mobile voice calls and SMS do not have a higher (or lower) propensity to use an innovative NFC MPS offer of their MNO than their counterparts with a “light” MCS consumption. Hence, H5a and H5b receive no support. Furthermore, H5c and H5d claim that subscribers’ mobile Internet use level and their subscription to a music streaming offer are positively related to NFC MPS use. Our data set contains a significantly (p ≤ 0.001) positive association for both explanatory variables. An increase in monthly mobile data usage by 1 megabyte goes along with a 2.5% raise (= e-1 x 0.068) in the NFC MPS adoption likelihood. The probability of being an NFC MPS user is 39.1% higher if a customer has subscribed to the operator’s music streaming offer (see marginal effects of predictors #7 and #8 in column “Probit” of Table 4). Thus, H5c and H5d receive support. Moreover, H5e claims that MNO customers’ monthly bill for mobile services is positively related to NFC MPS use likelihood. According to the Probit coefficient in Table 4, subscribers’ average monthly invoice amount for mobile services is significantly (p ≤ 0.01) negatively related to NFC MPS use. A 1 EUR uplift in the average monthly invoice amount translates into a decrease of the likelihood of adopting the NFC MPS by 4.6% (= e-1 x –0.126; see marginal effect of predictor #9 in column “Probit” of Table 4). Therefore, we reject H5e. One explanation for the missing support of H5e is that lower (higher) average monthly invoice amounts could reflect higher (lower) cost consciousness of MNO subscribers. Since there is evidence suggesting that NFC MPS support consumers in keeping tight control over their various expenditures (e.g., PWC, 2016; Total System Services, 2016), such systems are particularly suited to meet the needs of cost-conscious consumers with low average monthly invoice amounts for MCS. Therefore, future work should explore whether the average monthly invoice amount of MCS customers actually is a valid proxy for their cost-consciousness, which in turn drives their inclination to use NFC MPS. H6 states that the number of NFC terminals in the residential area of a MNO subscriber positively affects the likelihood of NFC MPS use. The Probit regression supports this proposition: Number of NFC terminals in the residential area of a subscriber is significantly (p ≤ 0.05) positively related to the likelihood of using NFC MPS at least once. According to the Probit marginal effect estimate, an additional NFC terminal in a customer’s neighborhood elevates the probability of NFC MPS adoption by 0.8% with the remaining explanatory variables set at

NFC mobile payment use

– 22 –

10th March 2017 (ECRA-D-16-00341R4)

their mean (see marginal effect of predictor #10 in column “Probit” of Table 4). Hence, H6 is supported. As explained in section 4.3, we ran linear OLS and negative binomial regressions of the timing of NFC MPS use on the ten predictors under study in the subsample of MPS users to address our research question. Table 4 indicates that each of the two regressions contains five significant (p ≤ 0.10) predictors of NFC MPS use timing (see predictors #3-4, 8-10 in columns “Linear OLS” and “NBR”). Five variables achieve no statistical significance (see predictors #1-2, #5-7 in columns “Linear OLS” and “NBR”). A likelihood ratio test (Wooldridge, 2016, pp. 529-530) reveals that the linear OLS regression better fits the data than the negative binomial regression.9 However, the sign and magnitude of OLS coefficients and negative binomial regression marginal effects are not materially different. Hence, the results of both regressions suggest similar answers to our research question. The price level of subscribers’ handsets and subscription to the music streaming offer both are significantly (p ≤ 0.05) positively associated with the timing of NFC MPS use.10 Customers equipped with a premium smartphone (a music streaming subscription) adopt the NFC MPS 1.41 (0.81) months earlier than their counterparts with a non-premium appliance (without a music streaming ticket; see marginal effects of predictors #4 and #8 in NBR column of Table 4)). These results are be taken to suggest that consumers who are ready to pay a higher price for their smartphone and who are already users of other new offers (mobile music streaming) are also more interested in innovative services such as NFC MPS and therefore are among the early users of such services. Furthermore, the screen size of subscribers’ handsets and their monthly average invoice amount for mobile services are significantly (p ≤ 0.01) negatively related to the timing of NFC MPS use. The negative regressions suggest that an increase of the screen size by one inch delays the timing of the first NFC MPS use by about 1.07 to 1.09 months. Furthermore, a monthly bill raise of 1 EUR is equivalent to an adoption delay of roughly 0.41 months (= e-1 x 9

10

Likelihood (L) ratio = –2 x [ ln L (Negative binomial regression) – ln L (Linear OLS regression) ] = –2 x [ – 1,739.92 – (–1,701.77) ] = 76.30. The critical value at the 99.99th percentile of the χ 2-distribution with 10 degrees of freedom is 35.56. This implies that the data fit of linear OLS regression is significantly better than that of the negative binomial regression. The statistical effect of handset price level is somewhat attenuated if the original values of the predictor #4 “device price level” instead of its dichotomized values are entered into the regressions explaining the timing criterion. The same holds for the predictor #3 (device screen size; see next paragraph). The reason for this attenuation is that the correlation between the untransformed device price and device price level is higher, which implies that the estimates of the two predictors may be more influenced by multicollinearity bias. However, the tendency of the results concerning predictors #3 and #4 remains unaffected by replacing our dichotomous price measure in Table 4 with its non-transformed value.

NFC mobile payment use

– 23 –

10th March 2017 (ECRA-D-16-00341R4)

–1.115 and e-1 x –1.102) (see marginal effects of predictors #3 and #9 in columns “Linear OLS” and “NBR” of Table 4). A reason for the negative association between screen size and MPS use timing may simply be that devices available at the beginning of our study period were equipped with smaller screens than those which were launched in the last month of our time window. Furthermore, MPS users with low bill amounts may be smart spenders who care about their expenses and therefore are more ready to test fancy NFC MPS at an earlier point in time. Finally, the number of NFC terminals operated by merchants in the subscriber’s home area is a marginally significant (p ≤ 0.10) predictor of the timing of NFC MPS use: The regression estimates indicate that ten additional NFC terminals accelerate the NFC MPS adoption by about 0.3 months (see marginal effect of predictor #10 in columns “Linear OLS” and “NBR” of Table 4). This result suggests that an extension of the points of sales accepting NFC mobile payments does not exert an exceptionally powerful but still noteworthy influence on consumers’ usefulness ratings of NFC MPS offers. We ran several robustness checks of the preceding results. Firstly, we repeated the sample generation procedure for NFC MPS non-users 10 times to test whether the composition of the subsample of NFC MPS non-users affects our conclusions. Secondly, we excluded sample members whose variable values of the four metrically scaled predictors deviated from the sample mean by at least two standard deviations to explore the extent to which outliers distort the findings. Thirdly, we computed a binary Logit instead of a binary Probit regression for the dichotomous NFC MPS use criterion. Fourthly, we applied a one-step zero-inflation model performing the binary and count regressions simultaneously (see section 4.3). Fifthly, we used bootstrapped instead of robust (Eicker-Huber-White) standard error estimates. Each of the five method modifications produced results that were not materially different from those presented in Table 4. Therefore, it is not very probable that the outcomes of our statistical analysis are an artifact of the methods underlying the baseline results. 6.

Discussion and implications

This paper analyzes correlates (1) of the basic decision to use NFC MPS in a sample of 1,354 residential postpaid subscribers of an MNO in Germany and (2) of the timing of the use start among 677 MPS adopters in a 12-months period. A key factor differentiating the present study is that we focused on “real” NFC MPS use behavior of MNO customers whereas many prior investigations looked at behavioral use intentions which may or may not lead to actual use of a MPS. We found that the likelihood of NFC MPS use was higher for subscribers who

NFC mobile payment use

– 24 –

10th March 2017 (ECRA-D-16-00341R4)

were male, older, lived in an area with an above average number of merchants with NFC terminals, generated a large MI data volume, opted for a music streaming subscription, had a device with a larger screen diagonal and generated a lower monthly mobile service amount. Early MPS users differed from laggards who subscribed to the MPS towards the end of the study period in that the former also voted in favor of a music subscription, paid a lower monthly invoice amount for MCS and had a device with a relatively modest screen diagonal which was nevertheless sold at a premium retail price when launched on the German market. The practical implications which can be drawn from our findings for NFC MPS marketing policies depend on whether an MNO favors a “procyclical approach” (Slade et al. 2015; Shin and Lee, 2014) which addresses people who most easily convinced of the benefits of NFC MPS and whose profile therefore resembles that of the MPS users in the present sample. Alternatively, MNOs may find it more efficient to follow an “anticyclical approach” which targets customers who are least likely to use MPS anyway on their own initiative. In this case, according to our results, they have to extract individuals from their customer base who are female, younger and generate an above average revenue although they are equipped with a smartphone with a relative small screen diagonal. Regardless of which of the two approaches operators choose, to target at potential addressees of marketing measures, MNO practitioners must adequately cope with the methodological problem, which stems from the fact that after the launch of a new service the number of its non-users by far exceeds the number of its users. Against this background, our study suggests that the application of an oversampling technique to solve this class imbalance problem is a viable approach to target potential (NFC MPS) users more accurately with campaigns promoting service adoption. Following the identification of the customer segment for NFC MPS marketing measures MNOs have to take a second, even more intricate step. It involves the development of measures to communicate NFC MPS key advantages to the chosen target segment and to improve the benefit-cost-balance of its MPS offerings. Such improvements may be achieved by offering access to NFC MPS free of change (at least for a significant trial period) or by setting a low minimum transaction/sales threshold which needs be passed to get an MPS subscription fee waiver (cf. Moroni et al., 2015). We suggest that such pricing-related incentives should work well in the NFC MPS case because in the present sample MCS users tended to be “smart shoppers” who were able to keep their monthly MCS spendings relatively low. Although such marketing measures may increase the MPS take-up rate, they do not solve a more fundamental problem which most NFC MPS are facing today and which explains the

NFC mobile payment use

– 25 –

10th March 2017 (ECRA-D-16-00341R4)

low take-up rate of such systems (0.3% in our sample): Present NFC MPS do not have a strong comparative advantage over other already well-established electronic payment systems. Hence, before fine-tuning marketing campaigns for a sub-optimal service MNOs may be better advised to take strategic steps changing the technical and commercial architecture of NFC MPS in a way, which ensures that such systems deliver a real-value-added compared to other payment methods. Additional value could be generated by integrating the NFC MPS of an MNO with other applications such as e-ticketing for public transport or customer loyalty cards/programs. Finally, MNOs may be backed in their efforts to increase the acceptance of NFC MPS by public regulatory measures, which oblige or subsidize (large) merchants to install NFC terminals at their outlets. Conceptually, the present study contributes to the literature on MPS use in particular and on MCS use in general at least in two ways. Firstly, our results indicated that objective customer characteristics are effective in separating MPS users from non-users. This is important because prior MPS acceptance research tended to focus on customer perceptions (usefulness, ease of use etc.) of MPS which are frequently not available in situations in which MNOs design measures to promote MPS use. Secondly, our outcomes support theoretical positions of others such as Wei (2008) and Kerviler et al. (2016). They argue that the use of one category of MCS results in an “activation” or “spill-over” effect with respect to the use of other MCS categories. In line with this reasons we find that subscribers who are heavy MI users and subscribers of a music streaming service (see variables #7 and #8 in Table 4) are also more inclined to try an NFC MPS offer. While the present research contributes to the literature on NFC MPS use, it is not without limitations. A first weakness lies in the composition of the study sample. It includes residential postpaid subscribers of a single MNO in one country, i.e., Germany, who were equipped with an Android smartphone. Furthermore, it purposely oversampled MPS users. Therefore, the extent to which our findings can be generalized to business and prepaid subscribers, customers of other MNOs or to people with a device running other OS (e.g., Apple iOS) is uncertain. Therefore, more work is needed which examines customers of further MNOs in Germany and in various countries as well as subscribers using a smartphone running other OS than Android. This work could also explore the extent to which the performance of the study variables in predicting NFC MPS adoption and use timing is affected by strategies of mobile network operators to explicitly position its NFC MPS as a unique service which differentiates it from its competitors.

NFC mobile payment use

– 26 –

10th March 2017 (ECRA-D-16-00341R4)

A second limitation is the narrow range of predictors, which were extractable from the MNO’s billing and customer systems. Socio-demographics were limited to age, gender and place of residence. Additional demographic data such as monthly income, educational level, family status etc. were not at hand. In addition, for privacy protection reasons, the MNO did not permit to send a questionnaire to customers in the data set. Accordingly, we were unable to capture customer perceptions of popular constructs such as ease of NFC MPS use, fraud concerns or contextual conditions such as purchase needs. Hence, additional work is justified which incorporates a broader range of objectively measurable and perceptual characteristics of MNO customers in predicting NFC MPS use. A third shortcoming is that the overall investigation period of one year is relatively short. Diffusion theory (Rogers, 2003) suggests that the proliferation of innovative complex services frequently extends to longer time periods. Thus, the moderate performance of our predictors in explaining the point in time at which an MPS customer started the use (i.e., differences between “early” and “late” users) may be partly due the restricted span of time covered in the present study. Hence, it would be desirable that future truly longitudinal work looks at the MPS use start and change dynamics over a period of several years. Such longitudinal studies could also move beyond the correlational nature of our research design to find out to what extent the predictors covered here are causally related to NFC MPS adoption and the timing of its usage. A final constraint relates to the criterion measures. They are based on the MNO’s recorded monthly status information for each customer on whether (s)he holds an NFC MPS subscription or not. Although a subscription to the MNO’s NFC payment system was subject to an annual fee (see footnote 4), it is not sure that subscribers who MPS-based signed up for the MNO’s NFC MPS actually made an NFC payment. MPS-based transaction revenues are likely to be a more accurate measure the intensity of NFC MPS use. Therefore, future studies should capture NFC MPS use by its transaction amounts.

References Accenture, 2015. When it comes to payments today, the customer rules. Dublin. Retrieved from . Amin, A., Anwar, S., Adnan, A., Nazad, M., Howard, N., Qadir, J., Hawalah, A., Hussain, A., 2016. Comparing oversampling techniques to handle the class imbalance problem: A customer churn prediction case study. IEEE Access 4, 7940-7957.

NFC mobile payment use

– 27 –

10th March 2017 (ECRA-D-16-00341R4)

Apanasevic, T., 2013. Factors influencing the slow rate of penetration of NFC mobile payment in Western Europe. Proceedings of the 12th International Conference on Mobile Business, Berlin, AIS. Retrieved from . Au, Y.A., Kauffman, R.J., 2008. The economics of mobile payments: Understanding stakeholder issues for an emerging financial technology application. Electron. Commer. Res. Appl. 7, 141-164. Bagadia, P., Bansal, A., 2016. Risk perception and adoption of mobile banking services: A review. IUP J. Inform. Technol. 12(1), 52-71. Balachandran, D., Tan, G.-W., 2015. Regression modelling of predicting NFC mobile payment adoption in Malaysia. Int. J. of Model. Oper. Manag. 5, 100-116. Bitkom, 2014. Jeder Dritte zahlt lieber ohne Bargeld. Berlin. Retrieved from . Bitkom, 2016. Bezahlen mit dem Smartphone funktioniert, aber kaum jemand weiß wie. Berlin. Retrieved from . Bundesbank, 2015. Zahlungsverhalten in Deutschland 2014. Frankfurt. Retrieved from . Carter, S., Yeo, A.C., 2016. Mobile apps usage by Malaysian business undergraduates and postgraduates. Internet Res. 26, 733-757. Ceipidor, U.B., Medaglia, C.M., Volpi, A.O., Moroni, A., Sposato, S., 2012. A survey about user experience improvement in mobile proximity payment. Proceedings of the 4th International Workshop on Near Field Communication, Helsinki, IEEE. Retrieved from . Cocosila, M., Trabelsi, H. (2016). An integrated value-risk investigation of contactless mobile payment adoption. Electron. Commer. Res. Appl. 20, 159-170. Coskun, V., Ozdenizci, B., Ok, K., 2013. A survey on near field communication (NFC) technology. Wireless Pers. Commun. 71, 2259-2294. Dahlberg, T., Guo, J., Ondrus, J., 2015. A critical review of mobile payment research. Electron. Commer. Res. Appl. 14, 265-284. Dahlberg, T., Mallat, N., Ondrus, J., Zmijewska, A., 2008. Past, present and future of mobile payments research: A literature review. Electron. Commer. Res. Appl. 7, 165-181. Davis, F.D., 1989. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quart. 13, 319-340. Deloitte (2015). Contactless mobile payments (finally) gain momentum. London. Retrieved from . Deloitte (2016). Mobile Payment: Deutschland Schlusslicht beim mobilen Bezahlen. Düsseldorf. Retrieved from . Dennehy, D., Sammon, D., 2015. Trends in mobile payments research: A literature review. J. Innovat. Manag. 3(1), 49-61.

NFC mobile payment use

– 28 –

10th March 2017 (ECRA-D-16-00341R4)

Ericsson, 2015. Screen size matters. Stockholm. Retrieved from . Ferreira, M.C., Cunha, J.F. e, José, R., Rodrigues, H., Monteiro, M.P., Ribeiro, C., 2014. Evaluation of an integrated mobile payment, ticketing and couponing solution based on NFC. Proceedings of the 2nd World Conference on Information Systems and Technologies, Funchal, IEEE, pp. 165-174. Fishbein, M., Ajzen, I., 1975. Belief, attitude, intention and behavior: An introduction to theory and research. Reading, Addison-Wesley. Gerpott, T.J., 2015. SMS use intensity changes in the age of ubiquitous mobile Internet access – A two-level investigation of residential mobile communications customers in Germany. Telemat. Inform. 32, 809-822. Gerpott, T.J., Meinert, P., 2016a. Correlates of using the billing system of a mobile network operator to pay for goods and services. Inform. Syst. Front. 18, 1265-1283. Gerpott, T.J., Meinert, P., 2016b. The impact of mobile Internet usage on mobile voice calling behavior: A two-level analysis of residential mobile communications customers in Germany. Telecomm. Policy 40, 62-76. Gerpott, T.J., Thomas, S., Weichert, M., 2013. Characteristics and mobile Internet use intensity of consumers with different types of advanced handsets: An exploratory study of iPhone, Android and other web-enabled mobile users in Germany. Telecomm. Policy 37, 357371. Gerpott, T.J., Thomas, S., Weichert, M. 2014. Usage of established and novel mobile communication services: Substitutional, independent or complementary? An empirical study of residential mobile network operator customers in Germany. Inform. Syst. Front. 16, 491507. Ghose, A., Han, S.P., 2011. An empirical analysis of user content generation and usage behavior on the mobile Internet. Manage. Sci. 57, 1671-1691. Google, 2014. The 3 new realities of local retail. Mountain View. Retrieved from . Greene, W.H., 2012. Econometric analysis (7th ed.). New York, Pearson. Grzybowski, L., Pereira, P., 2008. The complementarity between calls and messages in mobile telephony. Inf. Econ. Policy 20, 279-287. Guo, J., Bouwman, H., 2016. An ecosystem view on third party mobile payment providers: A case study of Alipay. Info 18(5), 56-78. Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E., 2014. Multivariate data analysis (7th ed.). Harlow, Pearson. Henry, C.S., Huynh, K.P., Shen, Q.R., 2015. 2013 Methods-of-payment survey results. Ottawa. Retrieved from . Herrmann, E., 2016. Mobile Payment in Deutschland: Bezahlen mit dem Smartphone. Berlin. Retrieved from . Hilbe, J.M., 2008. Negative binomial regression. New York, Cambridge University Press. Hsiao, M.-H., Chen, L.-C., 2015. Smart phone demand: An empirical study on relationships between phone handset, Internet access and mobile services. Telemat. Inform. 32, 158.

NFC mobile payment use

– 29 –

10th March 2017 (ECRA-D-16-00341R4)

IP, 2016. IP Trendline-Begragung zu Mobile Payment. Cologne. Retrieved from . Jenkins, P., Ophoff, J., 2016. Factors influencing the intention to adopt NFC mobile payments – A South African perspective. Proceeding of the 9th International Conference on Information Resources Management, Cape Town, AIS. Retrieved from . Juniper Research, 2016. Mobile & Online Ticketing: Deep Dive Strategy & Competition 2016-2020. Basingstoke. Retrieved from . Kerviler, G. de, Demoulin, N.T., Zidda, P., 2016. Adoption of in-store mobile payment: Are risk and convenience the only drivers? J. Retailing Consum. Serv. 31, 334-344. Kim, K.J., Sundar, S.S., 2014. Does screen size matter for smartphones? Utilitarian and hedonic effects of screen size on smartphone adoption. Cyberpsychol. Behav. Soc. Netw. 17, 466-473. Kim, K.J., Sundar, S.S., Park, E., 2011. The effects of screen-size and communication modality on psychology of mobile device users. Proceedings of the 29th Conference on Human Factors in Computing Systems, Vancouver, ACM. Retrieved from . Kim, M., Chang, Y., Park, M.-C., Lee, J., 2015. The effect of service interactivity on the satisfaction and the loyalty of smartphone users. Telemat. Inform. 32, 949-960. Kotsiantis, S., Kanellopoulos, D., Pintelas, P., 2006. Handling imbalanced datasets: A review. GESTS Int. T. Comput. Sci. Eng. 30(1), 25-36. Lee, J.-B., Lee, S.-B., Park, C., 2015. A study on the individual and environmental factors affecting mobile payment service acceptance – A focus on NFC-based payment services. Inform. 18, 1185-1196. Leong, L.-Y., Hew, T.-S., Tan, G.W., Ooi, K.-B., 2013. Predicting the determinants of the NFC-enabled mobile credit card acceptance: A neural networks approach. Expert Syst. Appl. 40, 5604-5620. Li, H., Liu, Y., Heikkilä, J., 2014. Understanding the factors driving NFC-enabled mobile payment adoption: An empirical investigation. Proceedings of the 18th Pacific Asia Conference on Information Systems, Chengdu, AIS. Retrieved from . Liébana-Cabanillas, F., Luna, I.R. de, Montoro-Ríos, F.J., 2015. User behaviour in QR mobile payment system: The QR payment acceptance model. Technol. Anal. Strateg. 27, 10311049. Liu, Y., Kostakos, V., Deng, S., 2013. Risks of using NFC mobile payment: Investigating the moderating effect of demographic attributes. Proceedings of the 15th International Conference on Electronic Commerce, Turku, TUCS. Retrieved from . Luna, I.R. de, Montoro-Ríos, F., Liébana-Cabanillas, F., 2016. Determinants of the intention to use NFC technology as a payment system: An acceptance model approach. Inf. Syst. EBus. Manag. 14, 293-314.

NFC mobile payment use

– 30 –

10th March 2017 (ECRA-D-16-00341R4)

Makki, A.M., Ozturk, A.B., Singh, D., 2016. Role of risk, self-efficacy, and innovativeness on behavioral intentions for mobile payment systems in the restaurant industry. J. Foodservice Bus. Res. 19, 454-473. Moroni, A., Talamo, M., Dimitri, A., 2015. Adoption factors of NFC mobile proximity payments in Italy. Proceedings of the 17th International Conference on Human-Computer Interaction with Mobile Devices and Services, Copenhagen, ACM. Retrieved from . Morosan, C., DeFranco, A., 2016a. It’s about time: Revisiting UTAUT2 to examine consumers’ intentions to use NFC mobile payments in hotels. Int. J. Hosp. Manag. 53, 17-29. Morosan, S., DeFranco, A., 2016b. Investigating American iPhone users’ intention to use NFC mobile payments in hotels. Proceedings of the 23rd ENTER eTourism Conference, Bilbao, IFITT, pp. 427-440. Nielsen, 2016. Think smaller for big growth. New York. Retrieved from . Oh, J.-S., Park, C.-U., Lee, S.-B., 2014. NFC-based mobile payment service adoption and diffusion. Journal of Convergence, 5(2), 8-14. Ondrus, J., 2015. Clashing over the NFC secure element for platform leadership in the mobile payment ecosystem. Proceedings of the 17th International Conference on Electronic Commerce, Seoul, ICEC. Retrieved from . Ooi, K.-B., Tan, G.W., 2016. Mobile technology acceptance model: An investigation using mobile users to explore smartphone credit card. Expert Syst. Appl. 59, 33-46. Ovum, 2016. Global mobile proximity payment users to surpass 1 billion by 2019. London. Retrieved from . Ozturk, A.B., 2016. Customer acceptance of cashless payment systems in the hospitality industry. Int. J. Contemp. Hosp. M. 28, 801-817. Pal, D., Vanijja, V., Papasratorn, B., 2015. An empirical analysis towards the adoption of NFC mobile payment system by the end user. Procedia Comput. Sci. 69, 13-25. Patel, R., Kunche, A., Mishra, N., Bhaiyat, Z., Joshi, R., 2015. Comparative review of existing mobile payment systems. Int. J. Appl. Eng. Res. 10, 16873-16884. Peng, J., Quan, J., Zhang, S., 2013. Mobile phone customer retention strategies and Chinese e-commerce. Electron. Commer. Res. Appl. 12, 321-327. Pham, T.-T., Ho, J.C., 2015. The effects of product-related, personal-related factors and attractiveness of alternatives on consumer adoption of NFC-based mobile payments. Technol. Soc. 43, 159-172. Podsakoff, P.M., MacKenzie, S.B., Lee, J.-Y., Podsakoff, N.P., 2003. Common method bias in behavioral research: A critical review of the literature and recommended remedies. J. Appl. Psychol. 88, 879-903.

NFC mobile payment use

– 31 –

10th March 2017 (ECRA-D-16-00341R4)

PWC, 2014. Mobile Payment in Deutschland 2020. Frankfurt. Retrieved from . PWC, 2016. Mobile Payment: Repräsentative Bevölkerungsumfrage 2016. Frankfurt. Retrieved from . Raptis, D., Tselios, N., Kjeldskov, J., Skov, M., 2013. Does size matter? Investigating the impact of mobile phone screen size on users’ perceived usability, effectiveness and efficiency. Proceedings of the 15th International Conference on Human-Computer Interaction with Mobile Devices and Services, Munich, ACM. Retrieved from . Reuver, M. de, Verschuur, E., Nikayin, F., Cerpa, N., Bouwman, H., 2015. Collective action for mobile payment platforms: A case study on collaboration issues between banks and telecom operators. Electron. Commer. Res. Appl. 14, 331-344. Rogers, E.M., 2003. Diffusion of innovations (5th ed.). New York, Free Press. Sam, K.M., Chatwin, C.R., Zhang, J.X., 2014. Adoption of near field communication for mobile payment: Evidence from Macau. Proceedings of the 9th International Conference on Industrial Engineering and Engineering Management, Malaysia, IEEE. Retrieved from . Serif, T., Ghinea, G., 2008. Mobile information access in the real world: A story of three wireless devices. Comput. Human Behav. 24, 1385-1403. Shaw, N., 2014. The mediating influence of trust in the adoption of the mobile wallet. J. Retailing Consum. Serv. 21, 449-459. Shaw, N., 2015a. The mediating role of perceived security: An empirical study of mobile wallet adoption in USA. Proceedings of the 2nd International Conference on HumanComputer Interaction in Business, Los Angeles, HCI International, pp. 358-369. Shaw, N., 2015b. Younger persons are more likely to adopt the mobile wallet than older persons, or are they? The moderating role of age. Proceedings of the 21st Americas Conference on Information Systems, Puerto Rico, AIS. Retrieved from . Shen, H., Liu, P., Yi, S., 2017. Trust transfer mechanism and intention on accepting NFC mobile payment: An empirical research. Proceedings of the 7th International Conference on Human Factors and System Interactions, Florida, AHFE, pp. 363-376. Shin, D.-H., 2009. Towards an understanding of the consumer acceptance of mobile wallet. Comput. Human Behav. 25, 1343-1354. Shin, D.-H., 2015. Effect of the customer experience on satisfaction with smartphones: Assessing smart satisfaction index with partial least squares. Telecomm. Policy 39, 627-641. Shin, S., Lee, W.-J., 2014. The effects of technology readiness and technology acceptance on NFC mobile payment services in Korea. J. Appl. Bus. Res. 30, 1615-1626. Silic, M., Back, A., Ruf, C., 2014. Mobile contactless payments adoption challenge in the complex network actor ecosystem. Proceedings of the 27th Bled eConference, Bled, University of Maribor. Retrieved from .

NFC mobile payment use

– 32 –

10th March 2017 (ECRA-D-16-00341R4)

Slade, E., Williams, M., Dwivedi, Y., 2013. Mobile payment adoption: Classification and review of the extant literature. Market. Rev. 13, 167-190. Slade, E., Williams, M., Dwivedi, Y., Piercy, N., 2015. Exploring consumer adoption of proximity mobile payments. J. Strat. Market. 23, 209-223. Statistisches Bundesamt, 2016. Bevölkerung (Zensus): Deutschland, Stichtag, Geschlecht, Altersgruppen. Wiesbaden. Retrieved from . Tan, G.W., Ooi, K.-B., Chong, S.-C., Hew, T.-S., 2014. NFC mobile credit card: The next frontier of mobile payment? Telemat. Inform. 31, 292-307. Tang, C.Y., Lai, C.C., Law, C.W., Liew, M.C., Phua, V.V., 2014. Examining key determinants of mobile wallet adoption intention in Malaysia: An empirical study using the unified theory of acceptance and use of technology 2 model. Int. J. Model. Oper. Manag. 4, 248265. Taylor, E., 2016. Mobile payment technologies in retail: A review of potential benefits and risks. Int. J. Retail Distrib. Manag. 44, 159-177. Total System Services (2016). 2015 German consumer mobile payment study. Columbus, GA. Retrieved from . Trendforce, 2016. TrendForce says global mobile payment market to reach US$620 billion in 2016 with Apple and Samsung staking large claims in the ecosystem. Taipei. Retrieved from . Trütsch, T., 2016. The impact of mobile payment on payment choice. Financ. Mark. Portf. Manag. 30, 299-336. Urien, P., Aghina, X., 2016. Secure mobile payments based on cloud services: Concepts and experiments. Proceedings of the 2nd International Conference on Intelligent Data and Security, New York, IEEE. Retrieved from . Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D., 2003. User acceptance of information technology: Toward a unified view. MIS Quart. 27, 425-478. Verifone, 2016. Schnell, sicher, digital: Zukunftsmarkt kontaktlos. Bad Hersfeld. Retrieved from . Wei, R., 2008. Motivations for using the mobile phone for mass communications and entertainment. Telemat. Inform. 25, 36-46. Wooldridge, J.M., 2016. Introductory econometrics (6th ed.). Cincinnati, Cengage Learning. Xiong, J., Muraki, S., 2016. Effects of age, thumb length and screen size on thumb movement coverage on smartphone touchscreens. Int. J. Ind. Ergon. 53, 140-148. Xu, J., Forman, C., Kim, J.B., Ittersum, K. van, 2014. News media channels: Complements or substitutes? Evidence from mobile phone usage. J. Mark. 78(4), 97-112.

NFC mobile payment use

– 33 –

10th March 2017 (ECRA-D-16-00341R4)

Manuscript ID: ECRA-D-16-00341R4

Electronic

Commerce

Research

and

Applications

submission

(October 25 and November 2, 2016; January 19, 2017; March 10, 2017 [R3 & R4])

Who signs up for NFC mobile payment services? Mobile network operator subscribers in Germany

Highlights 

Explores customer use of the near field communication (NFC) payment system of a mobile network operator (MNO)



Distinguishes between basic NFC payment use and timing of this use as dependent adoption facets



Considers socio-demographic, smartphone-, communication behavior- and location-related variables as antecedents of dependent variables



Analyzes system-captured behavioral data for 1,354 postpaid residential subscribers of an MNO in Germany