1. Introduction

The penetration of mobile services in the world increased up to 97% in 2015. Only four regions in the world have a coverage rate of less than 100%: South Asia (77%), Africa (82%), and Central America (88%), while Central and Eastern Europe concentrate the highest adoption rate at 139%, along with South America and its 121% adoption rate (^{Rivero, 2015}). Despite the increase in mobile phone usage, mobile banking has not grown at the same rate. In Latin America, Colombia has led mobile banking use with a 58% adoption rate (^{Latinia, 2017}).

Concerning digital literacy, mobile phones make it much more comfortable because they allow people to go online at any time and place, thereby reaching unbanked rural areas, and avoiding geographical barriers and installation costs. According to ^{Rivero (2015)}, by 2020, mobile phones will be responsible for 80% of the world’s banking market. Therefore, Latin American economies are challenged to reduce inequality by improving microenterprises’ financial inclusion through mobile banking since just 13% have access to financial services compared to 94% in developed countries.

Thus, the need to design and distribute financial services microentrepreneurs at the Bottom of the Pyramid (BOP) can afford, becomes relevant (^{Pankomera and Van Greunen, 2018}). The BOP represents the economically weaker segment of the world’s population with a daily per capita income of US$2 or less (^{Prahalad and Hammond, 2002}). In Colombia, 7.20% of the population lives in extreme poverty (^{DANE, 2018}).

Studies on mobile banking in the context of microenterprises are still rudimentary. Furthermore, no studies have been conducted in Colombia. Moreover, according to ^{Shaikh and Karjaluoto (2015)}, two of the main variables in the context of mobile banking have not been tested in terms of TPB extended to relative advantage and perceived risk. Hence, this study aims to assess the impact of those factors that affect microentrepreneurs’ at the BOP intention to adopt mobile banking.

To that end, we use a sample of microentrepreneurs in the Inclusive and Opportunities Territories (TIOS), which comprise geographical areas characterized by critical indicators concerning poverty, violence, and insecurity. Thus, it becomes essential to identify the factors that affect mobile banking adoption to increase financial inclusion and reduce poverty indexes at the BOP, which represents a new market opportunity for the banking sector (^{Kansal, 2016}). Hence, this study helps extend the knowledge about the factors that influence the intention to adopt mobile banking and thus improve banking strategies.

In this regard, the Theory of Planned Behavior (TPB) is one of the best-suited models on technological adoption to explain individuals’ behavior when performing a given action in those contexts where they lack full control and are conditioned by other non-motivational factors associated with the availability of specific requirements, knowledge, skills, and resources (^{Ajzen, 1991}; ^{Venkatesh, Morris, Davis, and Davis, 2003}; ^{Santos, Veiga, and Souza, 2011}). According to ^{Hasan, Lowe, and Petrovici (2019)}, given the limitations that BOP consumers face, TPB is a good predictor. Moreover, two other independent variables, namely, relative advantage and perceived risk, explain differences in individuals’ behavior in the context of adopting new technologies (^{Brown, Cajee, Davies, and Stroebel, 2003}).

Our findings confirm that attitude, subjective norms, behavior control, and relative advantage have a positive effect on mobile banking adoption, while the perceived risk did not. Consequently, this study helps to strengthen the literature on the adoption of technology, pointing out managerial aspects to implement successful strategies towards the adoption of mobile banking that increase the financial inclusion of microenterprises at the BOP.

First, this paper presents the literature review, the research model, and the hypotheses. Secondly, we move on to explain the methodology used, the results, and there will be a discussion thereof. Finally, we examine the conclusions, limitations, and future lines of research.

2. Literature review

Mobile banking (M-Banking) is an application of mobile commerce (M-commerce) (^{Lee and Chung, 2009}). According to ^{Shaikh and Karjaluoto (2015)}, mobile banking dates back to the late 1990s when the German company Paybox launched its first service in collaboration with the Deutsche Bank. Initially, it was tested in Germany, Spain, Sweden, Austria, and the United Kingdom, and Kenya was the first to introduce this M-Pesa text-based service in 2007 regarding developing countries. M-banking also can improve the quality of life of underserved populations (^{Hassan and Wood, 2020}).

Researchers are interested in explaining how users’ attitudes and intentions affect mobile banking appropriation. The most widely used and recognized theories on the study of technology adoption are the Technology Acceptance Model (TAM), the Theory of Reasoned Action (TRA), the Theory of Planned Behavior (TPB), and the Unified Theory of Technology Acceptance and Use (UTAUT), which integrates eight models and theories (^{Venkatesh et al., 2003}). Another model used to a lesser extent is the perspective of the Diffusion of Innovation (^{Rogers, 1995}).

TPB extends TRA by adding the perceived behavioral control construct, understood as the ease or difficulty in implementing the behavior (^{Ajzen, 1991}; ^{Mathieson, 1991}; ^{Taylor and Todd, 1995}). TPB points out that behavior is a direct function of intention and the perception of control over that behavior. The intention is formed by attitude, which reflects the favorable or unfavorable feeling towards the carrying out thereof. The subjective norms reflect the perception held by individuals’ relevant points of reference, which influence the decision of whether or not to implement a behavior. In contrast, the perception of control reflects the internal and external perceptions that constrain such behavior (^{Taylor and Todd, 1995}).

Most studies on technology adoption have focused on TAM (^{Shaikh and Karjaluoto, 2015}). However, in the study of individual behavior, TPB is much broader than the TAM for predicting intention and behavior (^{Ajzen, 1991}; ^{Davis, 1989}). TPB’s constructs have been tested by different researchers to predict different individuals’ adoption of mobile banking, i.e., bank users, university students, young people, bank customers (^{Priya, Gandhi, and Shaikh, 2018}; ^{Elhajjar and Ouaida, 2019}; ^{Danyali, 2018}; ^{Shankar and Kumari, 2016}; ^{Narteh, Mahmoud, and Amoh, 2017}; ^{Lu, Tzeng, Cheng, and Hsu, 2015}; ^{Verma, and Sinha, 2018}; ^{Woodson, Alcantara, and Do Nascimento,2019}). According to ^{Baishya and Samalia (2019)}, studies on technology adoption at the BOP are scarce.

Systematic empirical research on the adoption of innovation at the bottom of the pyramid has begun to develop (^{Hasan et al., 2019}). The studies on mobile banking adoption at the BOP are consumer-oriented and use different models to predict adoption intent. For example, ^{Kansal (2016)} applied the TAM model to determine the factors that affect mobile banking adoption by families at the bottom of the Indian pyramid. On the other hand, ^{Hassan and Wood (2020)} used TAM, social influence, and cultural variables to determine mobile banking adoption at the BOP in Egypt and the US.

The first study that used TPB to predict mobile banking adoption at the BPO was that by ^{Hasan et al. (2019)}, which determined the factors that affect the adoption of mobile consumer banking at the bottom of the pyramid. Their study employed the Consumer Acceptance Model of Technology (CAT), Diffusion of Innovations (DOI), Technology Acceptance Model (TAM), Theory of Planned Behavior (TPB), Theory of reasoned action (TRA), Value-based adoption model (VAM) models. The TPB model explained intent at 27%, and the three predictors had a positive effect on the intention of adopting mobile banking. In contrast, the relative advantage variable did not directly influence intention. Nevertheless, there was no evidence of studies on the adoption of mobile banking among microentrepreneurs’ context at the BOP.

3. Research methodology

This research is quantitative and makes part of a project aimed at microentrepreneurs who took part in the TIOS (Territories of Inclusion and Opportunities), an intervention program that advocates for the generation of conditions of equality. The research found that 95% of 180 microentrepreneurs own a mobile phone; however, only 5% of that number had adopted mobile banking. This finding is akin to the study by ^{Brown et al. (2003)} in South Africa, where 91% of the participants possessed a cellphone, but only 6% had embraced mobile banking. For that study, 95% of microentrepreneurs who had not adopted mobile banking were selected. The participants’ age ranged between 35 and 54 years (66%), and 58% were women. Regarding their educational level, 45% had completed high school, and 33% had technical studies.

Concerning microenterprises (i.e., >10 employees), 42% belonged to the commercial sector, 41% to industry, 16% to services, and 1% were in the agricultural sector. In terms of monthly sales, 75% of microentrepreneurs reported sales under 970 dollars a month, and 34% were one-employee companies. It should be noted that banking entities face microentrepreneurs’ apathy since 53% do not own banking products. The other 47% held products such as a savings account (28%), a loan (16%), a credit card (11%), or a checking account (7%). Moreover, 39% of microentrepreneurs do not visit banks, given their businesses’ lack of formality and their preference for cash.

The measuring instrument was built drawing from different academic papers. We used the ^{Aboelmaged and Gebba (2013)} scale to measure intention, while the scale for attitude was drawn from ^{Pattansheti, Kamble, Dhume, and Raut (2016)}. The subjective norms scale was taken from ^{Khasawneh and Irshaidat (2017)}. For the control over behavior, we employed ^{Luarn and Lin’s scale (2005)}. Finally, to measure relative advantage and perceived risk, the scale used by ^{Al-Jabri and Sohail (2012)} was chosen. We also employed a Likert scale using 1 to express total disagreement and 10 to show total agreement to measure the concepts.

A pre-test was administered upon microentrepreneurs to assess the instrument’s validity and adjust wording following the recommendations of ^{Podsakoff, MacKenzie, Lee, and Podsakoff (2003)} to prevent common method bias due to the use of one informant. In the next stage, to confirm the questionnaire’s reliability, its internal consistency was measured through Cronbach’s Alpha. The intention to adopt mobile banking, subjective norms and behavioral control scored above 0.8. As for attitude, relative advantage, and perceived risk, these were above 0.9, thus confirming the scales’ converging reliability (^{Hair, Black, Babin, and Anderson, 2010}).

Another method used to assess validity and reliability is factorial analysis, a data reduction multivariate analysis technique that seeks the minimum number of dimensions capable of explaining the maximum amount of information in the data (^{Hair et al., 2010}). The convergent and discriminant validity of the measurement scales were analyzed. Table 1 below shows the results of the factorial analysis performed using LISRELL 8.8.

All concepts’ items have a significant standardized factorial load above 0.5 (the lowest t-value is 0.67), which is evidence of convergent validity (^{Hair et al., 2010}). ^{Bagozzi and Yi (2012)} point out that the Scale’s Composite Reliability (SCR) should be above 0.7. For the proposed model, subjective norms and behavioral control came above 0.8, while for intention, attitude, relative advantage, and perceived risk, their values were above 0.9.

Regarding the factorial loads, as represented by t-values, all items scored above 2.56. The individual standardized coefficients of each concept surpassed 0.5, which is the minimum recommended value by ^{Hair et al. (2010)}, thus proving convergent validity. Another reliability index is the Average Variance Extracted (AVE), which according to ^{Fornell and Larcker (1981)} “is a ratio between the variance that is captured by a factor i and the total variance due to that factor’s measurement error.” AVE values must be above 0.5. Our lowest AVE value was 0.6 for subjective norms.

The confirmatory factorial analysis results, using the maximum likelihood method, suggest that all of the model’s goodness of fit indexes beat their respective typical acceptable levels as recommended by ^{Hair et al. (2010)} (IFC=0.97; NFI=0.94, NNFI=0.97; SRMR=0.058). RMSEA came at 0.08, which matched the suggested value for a sample smaller than 250 (Hair et al., 2010).

Likewise, this research needed to study the possible relationship or behavior between the model variables, for which we used a correlation matrix. The results show that attitude, subjective norms, control over behavior, and relative advantage, correlate positively and significantly at a level of 0.01. Nevertheless, these values are not met for perceived risk because of its r=0.1, which was not significant.

Notwithstanding the foregoing, as stated by ^{Hernández, Fernández, and Baptista (2014)}, a cause-and-effect notion can be established theoretically; however, the test did not admit causality. Therefore, the hypotheses should be tested through the multiple regression model, which, according to ^{Hair et al. (2010)}, is used to analyze the relationship between a single dependent variable and several independent variables. A hierarchical method was thence used, giving way to two models: TPB and TPB extended to relative advantage and perceived risk.

Table 2 shows the outcome, displaying the predictability of the TPB model regarding microentrepreneurs’ at the bottom of the pyramid intention to adopt mobile banking. The variable that best explains their intention to adopt mobile banking is the attitude (b=0.387, p<0.01). Simultaneously, behavioral control proved to be the second-best explanatory variable (b=0.275, p<0.01). Finally, the least explanatory variable in the TPB model proposed by ^{Ajzen and Fishbein (1980)} is the subjective norms (b=0.217, p<0.1).

In general, the TPB model in the context under study had a confidence level of 99% and an error probability of 1%, i.e., it is highly significant (level of 0.01) in explaining the intention to adopt mobile banking. In this regard, the hypotheses proposed in the study were tested, showing that attitude as much as subjective norms and behavioral control have a positive influence on the intention to adopt mobile banking.

Concerning the second model, when introducing the relative advantage variable, it became highly significant regarding the intention to adopt mobile banking (b= 0.309, p<0.01), thereby supporting hypothesis 4. Then, attitude follows (b=0.295 p<0.01) and with behavioral control (b=0.247, p<0.05). On the other hand, when perceived risk was introduced in the model, its effect (b=-0.044) was not significant. Consequently, it is not significant to the model despite its validity and reliability; hence, hypothesis 5 goes unsupported. In the second model and context, the subjective norms construct would not explain the intent of adoption. These results suggest that subjective norms do not influence the intention to embrace mobile banking in the presence of relative advantage. According to ^{Püschel et al. (2010)}, for a non-adopter of mobile banking, the relative advantage is an effective factor, while the effect of subjective norms is minimal. Moreover, as ^{Mwangi and Brown (2015)} found, microentrepreneurs’ decision criteria to adopt mobile banking are endogenous (improved banking processes) instead of exogenous, i.e., they do not stem from other types of influences such as partners, suppliers or customers. This would explain why subjective norms are not influencing factors in adopting mobile banking in the presence of relative advantage.

The results obtained from the TPB model confirmed that attitude has a positive effect on the adoption of mobile banking. As a favorable attitude towards this technology grows, microentrepreneurs’ intention to adopt it also grows. This finding agrees with the prior study by ^{Shaikh and Karjaluoto (2015)}.

Likewise, subjective norms positively affect the adoption of mobile banking. Five dimensions make up this variable: competitors, social factors, suppliers, customers, and government, of which competitors and government showed less impact on the variable. In contrast, the first three dimensions exerted more substantial influence, especially if any of these three parties, which microentrepreneurs regard as relevant, recommend using mobile banking; then, their likelihood of adopting it increases. ^{Wu and Chen (2005)} and ^{Nasco et al. (2008)} also produced these results. According to ^{Baishya and Samalia (2019)}, since the BOP people are less educated, they would be more influenced by social factors than their knowledge of the technology. It should be noted that subjective norms have a more significant effect in the early stages of technological adoption (^{Nasco et al., 2008}).

Behavioral control was found to affect the adoption intention positively, which matches the findings of several other authors (^{Püschel et al., 2010}; ^{Lu et al., 2015}). Firstly, this variable defines whether microentrepreneurs perceive mobile banking as easy to use; if so, they would be more likely to adopt it. Similarly, if their confidence in their ability to operate mobile banking increases, so does their intention to adopt it.

For the second model, the relative advantage was found to affect the intention to adopt mobile banking positively. According to ^{Taylor and Todd (1995)}, when microentrepreneurs perceive that mobile banking offers an advantage and benefits over previous finance-managing methods, their intention to adopt it grows. On the other hand, subjective norms lose their predictive power in the presence of relative advantage because in the context of microentrepreneurs, according to ^{Mwangi and Brown (2015)}, adopting mobile banking depends more on improved financial processes than it does on exogenous demands.

Finally, perceived risk has been identified at the managerial and academic levels as one of the main barriers to mobile banking (^{BBVA, 2015}). This study rejected such a hypothesis because it did not explain microentrepreneurs’ at the BOP intention to adopt it. This finding agrees with ^{Hassan and Wood (2020)}. They found that adapting to m-banking, as an innovative technology, is driven more by trust in the technology and its provider, than by uncertainty regarding the technology itself. Moreover, ^{Yadav et al. (2015)} found that perceived risk does not affect online banking adoption. These authors argue that the lack of knowledge about the technology does not show the risk that operating it could unleash.

4. Conclusion

The growth of mobile services worldwide and domestically has been prominent. In this vein, mobile phones offer great opportunities compared to other technologies when providing mobile banking services to the bottom of the pyramid. However, the acceptance of mobile banking has been slow. Worldwide only 8.6% of bank customers use mobile banking services. For this reason, this research’s main objective was to determine how the factors in the Theory of Planned Behavior and its extended version, including relative advantage and perceived risk, affect microentrepreneurs’ at the bottom of the pyramid intention of adopting mobile banking.

For this purpose, we used a sample of microentrepreneurs characterized by vulnerable conditions. The results obtained confirm that attitude, subjective norms, behavioral control, and relative advantage, affect microentrepreneurs’ at the BOP intention of adopting mobile banking. At the same time, the perceived risk has no bearing on the intention of adoption. Thus, BOP microentrepreneurs have a favorable attitude towards mobile banking, and they also believe that it can improve efficiency and effectiveness in the administration of their finances. They are more likely to trust and embrace mobile banking if recommended by customers, suppliers, family, and friends. BOP microentrepreneurs do not perceive risks associated with using mobile banking; therefore, their adoption thereof in the short term is not restricted.

Although mobile phones’ penetration rate is high, mobile banking adoption in the context of microentrepreneurs at the BOP is shallow. This study identified the determining factors on which the financial sector must focus to increase financial inclusion and literacy. The perceived risk does not explain the intention of adopting mobile banking since those factors are more oriented to technology knowledge and literacy. Therefore, self-efficacy is the second most significant predictor of mobile banking adoption. Likewise, it was demonstrated that the influence of clients, suppliers, and family or friends has a positive impact on adoption. However, the relative advantage is more critical because microentrepreneurs seek new technologies to improve their current financial processes.

In this regard, banks and mobile service providers can implement strategies to increase microentrepreneurs’ at the BOP mobile banking usage rates by focusing on these key factors. Similarly, developing countries’ governments should harness the potential of mobile banking and reduce the costs of financial services for financially excluded people in order to expedite the financial-deepening agenda to reduce income inequality and poverty. Currently, academic studies that include the TPB to analyze mobile banking adoption by microentrepreneurs are embryonic and incipient. Therefore, our findings demonstrate the usefulness of the TPB model to predict the intention of mobile banking adoption in Latin American countries (Mohr, Sengupta y Slater, 2010).

5. Limitations and future research

This empirical study is not limitation-free. The sample is homogeneous because it concentrates on microentrepreneurs at the bottom of the pyramid regarded as non-adopters of mobile banking. Future studies should extend these studies to other contexts that involve multiple economic sectors, sales volumes, banking penetration rates, level of education, number of employees, and diverse regions. Following the study by ^{Grandon and Mykytyn (2004)}, it would be interesting to analyze mobile baking’s adoption intent by segmented economic sectors. Perhaps socio-demographic variables have a more significant impact on the adoption of mobile banking.

Moreover, future studies should assess the adoption intent from the standpoints of users and non-users of financial products, as has been approached by ^{Riemenschneider and McKinney (2002)}. Comparative studies could also be conducted to examine differences in the adoption processes between different banking channels such as online banking, telephone banking, and mobile banking to understand why users choose a particular channel over others. Such a study could also reveal the transactions made in each channel (^{Brown et al., 2003}).

Similarly, the TPB model should cover key constructs such as personal innovation (^{Pattansheti et al., 2016}), the cost of financial services (^{Luarn and Lin, 2005}; ^{Mwangi and Brown, 2015}), complexity (^{Narteh et al., 2017}; ^{Lu et al., 2015}), compatibility (^{Shaikh and Karjaluoto, 2015}; ^{Püschel et al., 2010}), and enabling conditions (^{Lu et al., 2015}). This empirical study could be a reference for other contexts and Latin American countries, and help lead microentrepreneurs to adopt mobile banking. However, according to ^{Brown et al. (2003)}, although there are many similarities among microentrepreneurs in these countries, there are many differences based on technological infrastructure, business practices, government, regulatory constraints, and cultural differences, which might need to be accounted for in each study. For this reason, specific scenarios using TPB are needed to contextualize the phenomenon and delve deeper into a growing topic, as is mobile banking.