Implication of mobile money adoption for financial behaviour in Cameroon: a gendered analysis Open Access

Sub-Saharan Africa (SSA) has experienced significant growth in financial inclusion over the past decade, largely driven by the widespread adoption of mobile money (Demirgüç-Kunt et al., 2022). According to the World Bank's Global Findex Database 2021, despite a significant increase in overall financial inclusion in SSA, nearly half of the adult population still lacks access to formal financial accounts. Unfortunately, women are disproportionately affected by this issue, with only 49% of women having access to formal accounts, compared to 61% of men (World Bank, 2022). A substantial proportion of this population continues to rely on informal mechanisms for saving and borrowing, such as rotating savings and credit associations (ROSCAs), moneylenders, and family networks (Allen et al., 2016; Beck et al., 2007).

Some of the informal saving and borrowing mechanisms includes saving and borrowing from a group of friends and relatives such njangi groups or stokvels where individuals meet and contribute a certain amount of money in a rotatory system, saving “under a mattress”, saving in jewels or livestock (Steinert et al., 2018; Ky et al., 2017). Nevertheless, these informal saving techniques provide an insurance which is perceived to be risky and incomplete based on the idea that it does not provide access to immediate liquidity if individuals/households wish to deal with unexpected events like health and agricultural shocks that are common in developing countries settings. The absence of safe and effective means of saving in developing countries is expected to lead to insufficient saving. The insufficient saving and insufficient access to credit when negative shocks are experienced, make it not easy for individuals to buffer the negative shock effects. As such, the availability of effective means of saving and borrowing can help individuals smoothen their consumption easily.

However, the advert of mobile technologies has changed the financial environment in most sub-Saharan countries coupled with the fast expansion of telecommunication networks and the increase in the proportion of individuals having mobile phones even in rural areas have offered a new and cheap platform that helps the resource poor individuals to pool and share resources in an effective manner (Tabetando and Matsumoto, 2019). Mobile money can be perceived as the simplest form of fintech (financial technology) which is a payment system that allows individuals to carry out financial services through mobile phones. This has been perceived as an important means of dealing with the developmental issues in developing countries by smoothing financial flow (Seng, 2021). One of the most of acclaim of the developmental potential of mobile money is supported by M-Pesa – agent-assisted, mobile-phone-based, person-to-person payment and money transfer system – in Kenya (Bateman et al., 2019). Saving and borrowing via mobile money account is perceived to be more efficient and convenient compared to saving and borrowing in the formal financial establishment like the bank. Indeed, in the last decade, the number of mobile money operators have increased enormously both in the urban and rural settings.

The mechanisms via which mobile money is expected to influence saving and borrowing is based on several transmission channels. Firstly, a great proportion of individuals do not have access to formal financial institutions and the possibility of saving and borrowing via mobile money is expected to increase saving when the experience a positive shock and borrowing when experience a negative shock by individuals using the mobile money services. Secondly, mobile money can affect the saving and borrowing behaviour of individual using mobile money services in that, it can serve as a safe and cost effect technique of storage and therefore creates an incentive to save. Thirdly, Mobile money services can reduce the costs of conducting financial transactions, such as sending and receiving money. This may encourage people to save and borrow more frequently, as they do not have to incur the costs associated with traditional banking services. For example, individuals may be more likely to save or borrow small amounts of money, which may not have been cost-effective with traditional banking services. According to Jack and Suri (2014), the low money transfer cost might increase the likelihood of individuals establishing mutual insurance groups. The insurance effect is expected to reduce the propensity to save and to borrow (Yang and Choi, 2007). As such, based on these possible mechanisms, the question of whether the use of mobile money will increase or decrease savings and borrowing remain an empirical question.

Beyond these mechanisms, gender remains a critical dimension in understanding the impact of mobile money on financial behaviour. Women continue to represent a significant proportion of the financially excluded population in developing countries (Demirgüç-Kunt et al., 2018; Wanjala, 2014). Research on gender disparities is important based on the idea that resource poor women are expected to succeed in business via access to savings and credit, and empowerment. According to Adaba and Ayoung (2017), the new financial technologies like mobile money services in developing countries has provided an opportunity for the unbanked individuals especially women to get access to financial services like saving and borrowing. Specifically, the use of mobile money in conducting saving and borrowing without needing at link to bank account has spread very fast across developing countries. As such, the widespread acknowledgement that increasing financial inclusion via mobile devices – and among poor women in particular can help enhance broader economic developmental objectives (Koomson et al., 2021; Suri and Jack, 2016).

There is growing empirical evidence of the implications of mobile money usage for outcomes like savings and borrowing. Using Tanzania data, Naito et al. (2021) show that the use of mobile money services increases the likelihood of saving and receiving remittances, while it reduces the likelihood of saving in less liquid assets such as livestock. Still in Tanzania, Riley (2018) reveals that households using mobile money services are likely to experience a decrease in per capita consumption after an aggregate shock. Jack and Suri (2014) also provide insight of consumption smoothing linked to using mobile money services in Kenya. Ky et al. (2017) equally provided evidence using Burkina Faso data that using mobile money services is expected to increase the propensity of disadvantaged groups such as rural dwellers, female and less educated persons to save for health emergencies. Batista and Vicente (2020) using experimental data in Mozambique find remunerated mobile money saving account increased mobile saving only when interest was paid. Munyegera and Matsumoto (2016) reveal that, in Uganda, users of mobile money services receive remittances more frequently and have higher consumption expenditure compared to non-users of mobile money services.

In this context, the main objective of this paper is: to examine the implications of mobile money adoption for saving and borrowing behaviour in Cameroon, both overall and by gender. Specifically, the paper aims: (1) to assess gender disparities in the probability of mobile money adoption and (2) to evaluate the impact of mobile money adoption on the likelihood of saving and borrowing, both overall and across gender lines. These objectives are guided by two hypotheses: Other things being equal: (1) men have a higher propensity of adopting mobile money services compared to women; and (2) the impact of mobile money on saving and borrowing is higher amongst men compared to women. Our empirical strategy is based on the endogenous switching probit regression (ESP) model and the FinScope survey Cameroon, 2017 Dataset. The advantage of the endogenous switching probit model is that it addresses simultaneously endogeneity of mobile money adoption and selection bias resulting from unobserved confounders and structural differences between users and non-users of mobile money services. We find that men are 5.6% more likely to adopt mobile money services compared to women. The impact evaluation results further reveal that mobile money adoption has a positive impact on saving and borrowing behaviour in Cameroon. The underlying mechanism of an increase in saving borrowing is due to the reduction in transactions cost provided by mobile money services. We equally observed that the impact of mobile money adoption on saving and borrowing is higher amongst men compared to women.

The rest of the paper is organized as follows: Section 2 reviews the background information on mobile money in Cameroon. Section 3 dwells on the empirical strategy and data description. Section 4 focuses on the empirical results, and conclusion and policy implications are sketched in Section 5.

MTN and Orange were the pioneers of mobile money technology in Cameroon, with their services officially launched in 2012 (Business in Cameroon, 2015). Currently, five active mobile money providers compete nationally: MTN Mobile Money, Orange Money, Express Union Mobile Money, Nexttel Possa, and YUP. Among these, MTN Mobile Money and Orange Money dominate the market and are the leading mobile network operators in the country (GSMA, 2021). The Global System Mobile Association (GSMA) Mobile Connectivity Index 2019 reported that 96% of Cameroon population is covered by mobile Global System for Mobile communication (GSM) network (2G and above). Thus, making Cameroon a high candidate to use Mobile Money in order to accelerate financial inclusion. In Cameroon, holding a mobile money account significantly influences the facilitation of access to certain financial services for households. All these mobile money providers make it easy for account holders to save money at their best convenience, to withdraw their savings whenever need arises and to send and receive money all over the country.

The percentage of adults holding a bank account increased from about 15% in 2011 to 52% in 2021. This increase is reflected from the sharp increase in mobile money account ownership from about 15% to 42% between 2017 and 2021 (International Monetary Fund, 2023). It was also revealed by the ministry of Finance (MINFI) in a FinScope survey in 2017 that about 29% of those aged 15 and over use mobile money service while the mobile money penetration rate was 76%. Considered as the economic powerhouse of the CEMAC, Cameroon accounted for 64.8% of the overall accounts active during the period, thus the leader in that market. The report further explains that, as far as the number of transactions is concerned, payment service providers operating in Cameroon accounted for 73.13% of the numbers recorded in the CEMAC community. The International Monetary Fund (2023) equally indicated that registered number of mobile money accounts for Cameroon increase sharply from 181,577 to 15, 649, 570 between 2011 and 2020.

It is fair to say that mobile money technology has disrupted the financial sector in Cameroon and fueled the social economic transformation in all regions of the country. As evidence, the international monetary Fund (IMF, 2020) reported that the total value of annual mobile money transactions as a percentage of GDP have increased from 0.08% in 2013 to 4.50% in 2016, to 30.24% in 2018 and then to about 55% in 2020. This likely reflects the sharp increase in the number of transactions per year for 1,000 adults, from about 700 in 2014 to about 70,000 in 2020. In particular, the number of transactions per active account has grown from about 5 per year in 2014 to about 126 on average in 2019 (Financial Access Survey, 2019). In terms of taxes, a 0.2% tax on the transfer and withdrawal of money through mobile money wallets came into force from January 1st 2022. In other words, people using money transfer platforms incur additional fees of 0.2% when sending and 0.2% when withdrawing as of January 2022. These taxes, expected to add more liquidity to state coffers, are in addition to existing charges on mobile money transactions in Cameroon.

In order to further provide more insights on the impact of mobile money usage on financial behaviour (saving and borrowing), we employ the endogenous switching probit (ESP) model proposed by Aakvik et al. (2005). To assess the effect of mobile money on financial behaviour (saving and borrowing), the conceptual financial behaviour generating function can be written as:

Where FB denotes the individual financial behavour (saving and borrowing) which takes the value 1 if the individual saves (borrows) and 0, otherwise. $MM$ is the mobile money indicator (which is a dummy equal to 1 if the individual uses mobile money services and 0, otherwise). $X$ is a vector other explanatory variables that are expected to determine the individual savings and borrowing behaviour. From Equation 1, the following econometric model can be derived as:

Where γ captures the effect of mobile money on financial behaviour (saving or borrowing). Based on the idea that the outcome variable financial behaviour (⁠$FB$⁠) is binary in nature, γ can be estimated by making use of the traditional probit model under two assumptions which are; the normality of the error term (ε) and the exogeneity of the independent variable of interest mobile money (MM). While the normality of the error term can be easily considered, the exogeneity of the of mobile money dummy, cannot be easily guaranteed. Indeed, the variable of interest-mobile money adoption is likely to be endogenous due to potential reverse causality. The endogenous switching probit model is more plausible based on the idea that problem of selectivity bias and endogeneity can be simultaneously resolved by appealing to the endogenous switching probit model in an explicit fashion (Wirba and Baye, 2020). As such, the advantage of the endogenous switching regression model over other techniques such as the PSM is based on the idea that it simultaneously accounts for selectivity bias and endogeneity

Assume that the individual decision to use mobile money services sorts individual into two regimes-mobile money users and mobile money non-users in the form:

The outcome equations for the two regimes are given as:

The observed financial behaviour (⁠$FBj$⁠) is defined as:

Where $FB1j*$ and $FB0j*$ are the latent variables (probability of saving and borrowing) that determine the observed binary outcome variables $FB1j$ and $FB0j$⁠. Regime 1 denotes individuals that are using mobile money services (MM users) while Regime 0 denotes individuals that are not using mobile money services (MM non-users). $X1$ and $X0$ denote the vectors of explanatory variables expected to influence the respective regime equations; Z represents the vector of correlates expected to affect the usage of mobile money services (⁠$MMj$⁠). Although the vectors–Z and X can overlap, at least one variable in Z should not appear in X to properly identify the outcome equations. In the present study, the exclusion restriction variable that is expected to determine $MMj$ but not directly correlated to the outcome variable $FBj$ is the non-self-cluster proportion on mobile money. The non-self-cluster identification strategy is based on the idea of social interactions to compute a non-self-cluster proportion of mobile money adoption. This identification strategy has been used by Epo et al. (2023), Rahman and Mishra (2020), Baye et al. (2020) and Wirba et al. (2021). The basic idea behind the nonself cluster identification strategy is that the average neighbourhood behavioural tendencies through the imitation and emulation channels is likely to influence the decision of a individuals to adopt or not to mobile money services, but this may not directly influence its borrowing or savings, except through mobile money adoption.

The error terms $ν$⁠, $ε1$ and $ε0$ are assumed to have a trivariate normal distribution, with mean vector zero and follow the covariance matrix:

Where $ρ0$ and $ρ1$ are the correlations between $ε0$ and $ν$⁠, and $ε1$ and $ν$ respectively. $ρ10$ denotes the correlation between $ε0$ and $ε1$⁠. $FB1j$ and $FB0j$ are mutually exclusive and therefore we assume the $ρ10=1$ since it is not estimable.

Where $Φ2$ denotes the cumulative function of a bivariate normal distribution while $wj$ represents the optional weight for individual j. After estimating the model's parameters, the different effects (i.e. ATT, ATU and ATE) of mobile money on financial behaviour can be obtained respectively as:

The effect of the mobile money usage (treatment) on the treated (mobile money users), or the expected effect of the treatment on individuals with observed characteristics x who are using the mobile money services (TT):

Meanwhile the treatment on the untreated (TU) which represents the expected effect of the mobile money usage on persons with observed characteristics x who are not using mobile money services:

The treatment effect (TE), which is the expected effect of mobile money usage (MM) for the person with observed characteristics x randomly drawn from the whole population of users and non-users:

The average treatment effects (ATT, ATU, and ATE) are computed by taking the average over the corresponding sub samples.

As such, the ATT is computed as:

Meanwhile the average treatment on the untreated (ATU) is given as:

The average treatment effect (ATE) is given as:

Where $NT$ denotes the total number of individuals using mobile money services (i.e. $MMj=1$⁠, the number of users) in Equation 10 meanwhile in Equation 11, $NT$ represents the number of individuals not using mobile money services (i.e. $MMj=0$ , the number of non-users). In Equation 12, $NT$ denotes the full sample (users and non-users of mobile money).

This study makes use of the FinScope Consumer surveys undertaken by FinMark Trust, a non-profit organisation based in South Africa. The survey was conducted in Cameroon in 2017 with the support of the Cameroon National Institute of Statistics (NIS). The FinScope methodology uses a sampling procedure that ensures minimum acceptable levels for national, urban/rural and regional reliable estimates with acceptable margins of error. A multi-stage sampling methodology is applied which entails selection of enumeration areas from recent census or population estimates using probability proportional to size followed by the selection of households as well as the selection of one adult in the selected household using a Kish Grid. Based on face-to-face interviews, a total of 6,826 individuals were surveyed in Cameroon. The data contains detailed information about the demographic and socio-economic characteristics of the individuals, and their access to and use of formal, informal and mobile money/digital financial services. In addition, there are important variables in the data collection, such as financial literacy, financial capability, income, and employment. One of the strengths of this data is its comprehensive information regarding all financial products (that is, credit, transfers and remittances, savings, risk and insurance and mobile money). Information regarding variables of interest is captured by the FinScope questionnaire. The demographic section captures information on the level of education, gender, age, marital status and region of residence of a respondent. Table 1 provides the descriptive Statistics for the variables used in the analysis.

Table 2 presents the mean comparison of individual characteristics with respect to their, mobile money adoption status. Results indicate that the proportion of mobile money users with savings is about 77.3% meanwhile for mobile money non-users, the proportion of individuals with savings is about 48.7. Regarding borrowing behaviour, we equally observed that the percentage is higher for mobile money users (28%) compared to non-users of mobile money (22%). This is a preliminary indication that mobile money users have higher propensity to save and to borrow. Findings also reveal that, a higher percentage of mobile money users are men, while a higher percentage of the non-users of mobile money are women. We also observe that, on average, individuals using mobile money services are about 5 years younger than non-users of mobile Money services. Amongst the mobile money users, the percentage of married individuals is about 48.2% meanwhile amongst the non-users of mobile money, the percentage of married persons is about 53.8%.

On the average, only about 4% of mobile money users have no educational level meanwhile up to about 30% of the non-users have no educational level. Also, 14.5% of mobile money users have primary level of education, while up to about 33.2% of the non-users of mobile money have primary education. In terms of secondary education, 59.8% of mobile money users and 35% of non-users of mobile money have this level of education. On the average, 21.7% of those using mobile money services have a tertiary level of education, while only 1.8% of those not using mobile money services have this level of education. These results indicate that individuals with at least a secondary school person is more likely to have access to financial services. Furthermore, 78% of the mobile money users are urban dwellers, while only 32.2% of the non-users of mobile money are urban dwellers. Averagely, 33.2% of the mobile money users are self-employed while about 48.3% of the non-users of mobile money services are self-employed. We equally observed that the percentage of anglophones is 15.3 and 22.6% respectively for mobile money users and non-users.

Regarding the variable mobile phone ownership, findings indicate that amongst the mobile money users the percentage of mobile phones owners is about 93.7% meanwhile amongst non-users of mobile money services, the percentage of mobile phone owners is about 66.8%. Indeed, the descriptive statistics reveal significant differences in in characteristics between users and non-users of mobile money services.

In Table 3, we present the determinants of the decision to use mobile money in Cameroon using the probit model. Results indicate that, with respect to women, being a man increases the likelihood of using mobile money by 6.3%. This is a revelation that gender disparities exist in mobile money usage in Cameroon, which leads us to reject the null hypothesis of there is no gender differences in the propensity to adopt mobile money services in Cameroon.

This disparity is consistent with human capital theory (Becker, 1964), which posits those differences in education and skills—often lower among women in Sub-Saharan Africa—can lead to unequal outcomes in the adoption of new technologies. Lower digital and financial literacy among women restricts their ability to navigate mobile money platforms, which are often perceived as complex financial tools (Demirgüç-Kunt et al., 2018). Also, Lundberg and Pollak (1996) suggests that limited decision-making power within households may reduce women's autonomy to adopt and use financial technologies, particularly in patriarchal societies. Supply-side constraints also contribute to this gap. The transaction cost theory (Williamson, 1981) helps explain how high costs of access—such as lack of mobile phones, inadequate identification documents, or distant mobile money agents—disproportionately affect women, especially in rural areas. This result aligns with findings by Suri and Jack (2016) in Kenya, who observed that men were significantly more likely than women to use mobile money, largely due to differences in education, income, and ownership of mobile phones. Similarly, the Global Findex database (Demirgüç-Kunt et al., 2018) shows that the gender gap in mobile money usage persists across many developing countries, underscoring the systemic nature of the issue.

For all samples, the age of the individual relates with the probability of using mobile money up to a certain age threshold, beyond which, it starts decreasing with the probability of using mobile money services. The concave behaviour of the age depicts that when Cameroonian are young and active, they are likely to be interested in using mobile money services, but as they advance in age, the likelihood of using mobile money services tend to decrease. This reflects the idea that young and middle-aged individuals, who are more economically active, are more likely to adopt mobile money to manage income and transactions. This finding aligns with the life-cycle hypothesis (Modigliani and Brumberg, 1955), which links financial behaviour to stages of economic activity over the lifespan. Younger users are also more exposed to technology and quicker to adopt innovations. In contrast, older individuals may face barriers such as lower digital literacy, reduced trust in new platforms, and declining need for mobile financial services (Demirgüç-Kunt et al., 2018).

The marginal effect indicates that for the overall sample and female sample, secondary and tertiary educational levels relate positively with the likelihood of using mobile money services for all samples. The degree of correlation intensifies with the level of education. Particularly, for the pooled sample, compared to the uneducated individuals, those with secondary education are having about 18.7% higher probability of using mobile money serves, meanwhile, those with tertiary education are having about 40.7% higher likelihood of using mobile money services. Specifically, for the male subsample, men with primary education are having a probability of 15.5% higher of using mobile money services, compared to the uneducated men, meanwhile, men with secondary education are having about 26.9% higher probability of using mobile money services. Equally, men with tertiary education are having about a 51.5% higher likelihood of using mobile money. For the female subsample, women with primary education, there is no significant difference in probability of using mobile money, compared to the uneducated women, meanwhile, women with secondary education are having about 14.2% higher probability of using mobile money. Equally, women with tertiary education are having about 34% higher likelihood of using mobile money services compared to uneducated women.

Compare with persons dwelling in the rural arears in the overall sample, individuals in the urban areas are having about 19.1% higher probability of using mobile money services. For the male subsample, men residing in urban areas are likely to have a likelihood of about 21.3% higher than rural dwellers. The marginal effect results of the female subsample equally indicate that urban women are expected to have a probability of about 17% of using mobile money higher than the rural women. Compare with francophones, anglophone are expected to have lower probability of using mobile money services for all samples.

In order to properly identify the selection equation, we follow the same logic as the one used by Epo et al. (2023), Rahman and Mishra (2020), and Wirba et al. (2021). The idea is that households or individuals in a cluster are exposed to similar external factors that influence their decision to adopt mobile money services, and by using the cluster proportion as an instrument, the study aims to correct for potential endogeneity issues. The finding that non-self-cluster proportion of mobile money adoption has a positive significant effect on the decision to adopt mobile money services is consistent with the theory of network effects, which suggests that individuals are more likely to adopt a new technology or behavior if they see others in their social network doing the same. The estimated effect indicates that individuals in clusters with a higher proportion of mobile money adoption are more likely to adopt mobile money services themselves.

Table 4 displays the estimates of the Endogenous Switching Probit (ESP) regression model, which involves the determinants of saving and borrowing behaviour in Cameroon and the selection (mobile money) equations. According to Lokshin and Sajaia (2004), the Wald test or the likelihood ratio test in the framework of the endogenous switching regression helps to evaluate the hypothesis of joint independence of the three equations, as well as the covariance coefficients providing information on self-selection. Result of the Likelihood ratio test reveals that the three equations (the selection equation and the two regime equations) are jointly dependent for both saving and borrowing, thus revealing evidence of endogeneity that needs to be addressed both in the saving and borrowing models. Findings display clear differences across the two regimes for the two outcomes-saving and borrowing. We observed that at 5% level of significance, there exist no gender-based disparity in terms of savings and borrowing amongst mobile money users and amongst mobile money non-users. This suggests that once individuals gain access to financial services—whether digital or traditional—men and women tend to exhibit similar financial behaviour.

This outcome is consistent with the unitary model of household behaviour (Becker, 1965), which assumes joint decision-making and pooled resources within households, thereby minimizing gender-specific differences in saving and borrowing patterns. Among mobile money users, the absence of gender disparity may reflect the equalizing effect of digital financial platforms, which deliver uniform services irrespective of gender and help reduce traditional barriers faced by women in accessing financial institutions (Demirgüç-Kunt et al., 2018). Furthermore, it implies that mobile money may serve as a tool for promoting financial inclusion by easing access to savings and credit for both men and women. In the case of non-users, the lack of gender difference could stem from common socio-economic constraints—such as low income or rural location—that limit financial engagement for both genders equally.

Results also reveal that the age of the individual only significantly relates with the propensity to saving. Particularly at 5% level of significance, age depicts a linear relationship with likelihood to save among mobile money adopters meanwhile among non-users of mobile money services, age depicts an inverted u-shaped relationship with the probability to save.

Concerning the marital status of an individual, results show that amongst non-users of mobile money, married persons are expected to have higher probability of saving compared to their unmarried counterparts. In terms of borrowing, we observed that only for the non-users of mobile money services, married persons are likely to have higher probability of borrowing compared to their unmarried counterparts. Results further suggest that for mobile money users, only individual with primary education are likely to have higher probability of saving compared to their uneducated counterparts. Meanwhile amongst non-users of mobile money services, persons with primary and secondary levels of education, are expected to have higher likelihood of saving compared to uneducated persons and only persons with primary education are likely to have a higher propensity to borrow. For non-users of mobile money services, persons dwelling in urban areas are likely to have lower propensity of saving while both users and non-users individuals residing in urban are likely to have lower probability of borrowing in Cameroon. We equally observed that anglophones are more likely to save compared to the francophones for both users and non-users of mobile money services.

Table 5 presents the results of the estimated average treatment effects of mobile money on saving and borrowing behaviour in Cameroon. Using the post estimation command, we predicted the average treatment effect on the treated (ATT), average treatment effect on the untreated (ATU) and the average treatment effect (ATE). The results indicate that the average treatment effects on the treated (ATT) are approximately 0.5665 and 0.2639 respectively for saving and borrowing. This is an indication that amongst mobile money users, the actual use mobile money services have led to about 56.7 and 26.4% more probability of saving and borrowing respectively compared to non-users of mobile money services in the counterfactual scenario. These results are in line with Naito et al. (2021) who indicated that mobile money adoption increases the probability of saving in liquid asset and decreases the probability of saving in less liquid assets such as livestock.

The overall findings of the average treatment effects on the untreated (ATU) reveal amongst the non-users of mobile money services, if the nonusers of mobile money that actually did not use the mobile money services would have had about 41.6 and 35.5% more likelihood of carrying out savings and borrowing respectively if they had used the mobile money services. As such, they would improve their probability of saving and borrowing if they were to use the mobile money services. ATE, as expected, is between ATT and ATU. Its positive estimate which represents treatment effect for the overall target population in the study (treated and untreated subjects together) also reveals that mobile money users would have probabilities of 45.9 and 33.1% higher in saving and borrowing if they are using the mobile money services.

The increase in the probability of saving and borrowing among mobile money users can be explained through several key mechanisms. First, mobile money expands access to financial services for those excluded from formal institutions, allowing users to save and borrow, particularly during financial shocks (Jack and Suri, 2014). Second, mobile money provides a low-cost, secure storage option, incentivizing saving by reducing the high transaction fees associated with traditional banking (Naito et al., 2021). Third, by lowering transaction costs, mobile money encourages more frequent saving and borrowing, especially for small amounts, which would otherwise be uneconomical with traditional banks. Lastly, mobile money helps with risk mitigation, offering quick access to funds in emergencies, thus increasing borrowing behavior during financial shocks (Yang and Choi, 2007).

Table 5 also provides a gendered analysis of the impact of mobile money saving and borrowing behaviour in Cameroon. These findings were obtained by implementing the analysis separately for the male and female subsamples. The detail endogenous switching probit results are presented in  Appendix 1 and  Appendix 2 respectively for the male subsample and female subsample. Amongst the male subsample, the ATT indicates that mobile money users that are men would have experienced a reduction of about 69.5 and 22.5% in the likelihood of saving and borrowing respectively if they were not using the mobile money services. Meanwhile female users of mobile money would have witnessed reductions of about 59.2 and 21.2% in the probability of saving and borrowing respectively if the women were not using the mobile money. Regarding the ATU estimate, the gendered analysis results reveal that men not using mobile money services would have increased their likelihood of saving and borrowing by 50.4 and 38.4% respectively if they were using the mobile money services meanwhile women not using the mobile money services would have increased their probability of saving and borrowing by 37.1 and 23.3% respectively if they were making use of the mobile money services. The ATU estimate reveals that the impact of mobile money adoption on saving and borrowing is higher men compared to women in Cameroon. The ATE estimates indicate that the male adopters of mobile money services are likely to experience an increased the likelihood of saving and borrowing by approximately 56.1 and 33.6% respectively. Meanwhile, female users of mobile money services are expected to enjoy an increase of 42.6 and 22.7% respectively in the probability of saving and borrowing. Based on the ATE which the whole sample estimate, the impact of mobile money usage is likely to be higher amongst men compared to women. This leads to the rejection of the null hypothesis that the impact of mobile money usage on saving and borrowing is the same for men and women.

The higher impact of mobile money usage on men compared to women can be attributed to several factors. First, men generally have greater access to formal financial institutions, such as bank accounts and credit services, which means that mobile money adoption may act as a more significant supplement to their existing financial practices, leading to a larger increase in their saving and borrowing behaviors. In contrast, women, particularly in developing economies, often face greater financial exclusion due to limited access to banking, financial literacy, or credit, which may limit the transformative effect of mobile money on their financial behaviours (Demirgüç-Kunt et al., 2018). Additionally, socio-cultural factors such as gender norms and roles may restrict women's financial autonomy, further reducing the scope for mobile money to make as large an impact on their saving and borrowing habits compared to men (Suri and Jack, 2016). Therefore, while mobile money provides significant benefits to both genders, men's financial behaviour may be more responsive to its adoption due to fewer pre-existing constraints.

The objective of this study was to examine the implications of mobile money adoption for saving and borrowing behaviour in Cameroon, both overall and by gender. Specifically, we assessed whether gender disparities exist in mobile money adoption and evaluated the impact of mobile money usage on the probability of saving and borrowing, both in the general population and across gender lines. Results indicate that compared to women, men are more likely to use mobile money services by 6.3%. The endogenous switching probit results revealed that being a mobile money user, increased the probability of saving and borrowing in the order of 45.9 and 33.1% respectively. Following these results, mobile money adoption yields significant improvement in the saving and borrowing behaviour of Cameroonians.

The mechanisms through which mobile technology can transform the economies of sub-Saharan Africa have not been well researched. The government and other policy makers like the Central Bank have put in place several policies in order to increase the supply of money services across Cameroon. Nevertheless, in spite of these measures taken, the mobile money adoption is still relatively low compared to many other countries. Indeed, the descriptive statistics reveals that percentage of mobile money adopters is about 28.7%. The relative adoption of mobile money services is likely to be due to existing inconsistencies of legal and regulatory framework of electronic money system in Cameroon. Therefore, considering consistent policies and regulatory reforms on mobile money services and providers can foster mobile money penetration as such improve the financial inclusiveness of most of the Cameroonians. Specifically, the government is expected to increase access to mobile money services. This can be achieved via the expansion of the of electronic money issuers and retailer agents. The expansion is expected to enhance competitiveness in the financial system and therefore decrease the transaction costs and increase efficiency. The government can contribute to this expansion of the of electronic money issuers and retailer agents by establishing partnerships between governments and mobile money issuers for the payments of salaries to civil servants and the payment of taxes. Indeed, the involvement of governments in the mobile money system can boost the confidence of Cameroonians to use the financial technology. Also, policymakers could consider initiatives to empower women to take control of their own finances, such as providing access to financial education and resources, and creating opportunities for women to own businesses and assets.