Mobile Financial Services as the Next Silver Bullet?
The role of mobile money usage in financial inclusion
Master Thesis
MSc Economic Development & Globalization
Name: Eline Timmer1
Student Number: S2944944 Supervisor: dr. A. Minasyan Date: January 5, 2020
Abstract
Mobile phone-based financial innovations, such as mobile money, have the potential to reach the global unbanked poor and provide a pathway to formal financial inclusion. This thesis extends the existing literature, on the relationship between mobile money and financial inclusion, by making use of a panel of 87 countries using the Global Findex survey for the years 2014 and 2017. The analysis with random and fixed effect specifications, provides evidence that mobile money is negatively related to financial inclusion. Only in Sub-Saharan Africa does mobile money complement formal banking services. The initial low level of financial inclusion, enabling expansion of mobile money through leapfrogging existing deficient formal banking services, is the mechanism behind the positive relationship for Sub-Saharan Africa. Policymakers are advised to enhance mobile money access, however also integrate formal financial services into the mobile money product, in the undertaking of increasing global financial inclusion.
Table of Contents
1. Introduction 1
2. Conceptual Framework 4
3. Data and Methodology 8
3.1 Data 8
3.1.1 Dependent variable - Index of financial inclusion 8
3.1.2 Main variable of interest - Mobile money 9
1. Introduction
In today’s digitalised world, the increased use of mobile phones has given birth to mobile phone-based financial innovations, such as mobile money, which enables financial transactions to anyone owning a mobile phone. In the past decade, mobile money has been booming; with over a billion registered users and close to $2 billion in daily transactions, the services keep expanding and now reach customers across 95 countries (GSMA, 2020). Globally, 1.7 billion adults are excluded from formal financial services of which a disproportionate share live from less than $1.90 a day (Demirguc-Kunt et al., 2018). Irregularity of income of the unbanked, high costs and low density of banking services and an inadequate product range are rationales behind financial exclusion (Ahmad et al., 2020). Banks are discouraged to serve low-income clients because of asymmetric information, high operating costs of physical infrastructure and investment uncertainty (Ahmad et al., 2020; Demirguc-Kunt et al., 2018; Mbiti & Weil, 2015). Research shows that financial inclusion contributes to economic growth, income equality and poverty reduction through the access of funds, growth of capital, consumption smoothing and risk reduction (e.g. Beck et al., 2007; Chinoda & Kwenda, 2019; Demir et al., 2020; Ozili, 2018). Moreover, increased financial access facilitates involvement in economic activities, resilience and efficiency (Kim et al., 2018). Consequently, financial inclusion has a prominent role in policy, such as the UN Sustainable Development Goals (UNCDF, 2020a).
Simultaneously, around two-thirds of the world’s unbanked population owns a mobile phone and evidence exists of a strong association between mobile phone diffusion and financial inclusion (e.g. Chinoda & Kwenda, 2019; Lenka & Barik, 2018). Mobile financial innovations, such as mobile money, could offer solutions to the two major problems for banking exclusion: the price of traditional banking services (Mbiti & Weil, 2015; Jack & Suri, 2011) and its inaccessibility (Munyegera & Matsumoto, 2018). Mobile money is cheap, fast, accessible2, reliable and secure and is offered through a wide physical network of mobile money agents3. Although the efficiency of the services might discourage users to bypass the formal financial sector (Yenkey et al., 2015), barriers to formal financial inclusion are lifted through familiarization with financial products and the use of bank-integrated services (Demirguc-Kunt et al., 2018).
2Generally, registration is done with a SIM card and identification at a local mobile money agent. Cash is deposited at an agent in exchange for electronic money and payments are made by sending a pin secured text message with the phone number or account ID of the recipient (IMF, 2019).
Since its first launch in 2001, the global number of mobile money services has exploded to 290 services in 20194. Today, Sub-Saharan Africa is at the forefront of mobile money uptake, drives around two-thirds of the global transactions and the mobile money market is characterised by success stories such as Econet EcoCash in Zimbabwe and Safaricom M-Pesa in Kenya (GSMA, 2020). Nevertheless, mobile money is a global phenomenon. Figure 1 shows Asia’s strong contribution to the growth in transaction volume and active mobile money accounts in 2019. In almost all other regions uptake is also intensifying, demonstrating the increased demand for digital financial transactions around the world.
Figure 1. Growth in mobile money transaction volume and active accounts, by continent (2019)
Source: Global System for Mobile Communications Association (2020).
The potential transformative ability of mobile money is extensive and reported as a pathway to formal financial institutions for the unbanked (Aron, 2018). Nevertheless, findings of the empirical literature are mixed. While in countries Kenya and Eswatini mobile money usage is a supporting instrument for bank account ownership, (formal) savings and loan acquisition (Mbiti & Weil, 2015; Myeni et al., 2020; Ouma et al., 2017), in Uganda and the SADC5 region mobile money serves mainly as an alternative to informal and formal banking services (Fanta et al., 2016; Munyegera & Matsumoto, 2018). The only global cross-country study, by Demir et al. (2020), reveals that FinTech6
4 See Appendix 1 for the evolution of global mobile money services, 2001-2019. 5 The Southern African Development Community (SADC).
is an important driver for financial inclusion.
Despite evolving research on mobile money and its development outcomes, there are still opportunities for improvements and extension of current literature. First, although some articles show that mobile money can ease the way to financial inclusion (e.g. Gosavi, 2018; Mbiti & Weil, 2015), results are often based on cross-sectional data and limited empirical research exists on the explicit relationship between mobile money and financial inclusion. Mobile financial services are incorporated in strategies of multilateral organisations, such as the UN, the World Bank and the IMF, to transform economic prospects of many by a more inclusive financial system (Barajas et al., 2020; UNCDF, 2020b; World Bank, 2018). Yet, mobile money should not automatically be considered as ‘the silver bullet’ to fight financial inclusion and empirical research is crucial in the recognition and promotion of the service. Currently, the narrow focus of existing research on certain countries and, successful, services as well as the context-specific, country-level nature of the data limits generalisation of the outcomes and the implications for financial inclusion in other contexts.
Therefore, the main research question of this thesis which I will empirically analyse is: ‘Is mobile money usage positively related to financial inclusion levels?’. The geographical reach and generalisability of results are extended by using panel data on 87 countries over the years 2014 and 2017. I use panel random effects and fixed effects models as the methods for estimation. The empirical findings show that mobile money serves as a substitute for formal financial services, as characterised by a negative relationship between mobile money and financial inclusion. Only in Sub-Saharan African countries is the relationship positive for which the mechanism of impact is the initial low level of financial inclusion; mobile money enhances financial inclusion because it leapfrogs over the existing, exclusive formal banking infrastructure. Therefore, mobile money access should be facilitated, though should not automatically be considered ‘the answer’ to advance financial inclusion. A type of service with more bank involvement, with formal finance integrated in the mobile money product through partnerships, is proposed.
2. Conceptual Framework
The fight for the expansion of financial inclusion is defined as “a process that ensures the ease of access, availability and usage of the formal financial system for all members of an economy” (Sarma & Pais, 2011, p.1). Access to necessary financial services, such as savings- and risk management products, is important for disadvantaged and low-income social groups (Sekantsi & Motelle, 2016; Sinclair, 2001). The theoretical underpinning of the relationship between mobile money and financial inclusion is the premise that a large part of the global unbanked owns a mobile phone. The development of cost-effective mobile-based innovations, such as mobile money, generates the potential for extending the reach of traditional banking systems to the unbanked (Beck et al., 2015). Mobile money is broadly described as a financial transaction, using a mobile phone, without the prerequisite of an account at a formal financial institution7 (Nan et al., 2020). Mobile money is offered by a mobile network operator independent of the traditional banking sector. By contrast, mobile banking is provided by banks or other financial institutions and offers access to traditional banking products and services through a mobile phone8 (Ahmad et al., 2020). Mobile financial services have, in most definitions, not been counted as part of financial inclusion but are rather seen as a pathway to formal inclusion (Aron, 2018).
A handful of studies offer empirical insights into the relationship between mobile money and the (in)ability to enhance formal financial inclusion. In the empirical literature, financial inclusion is measured through several characteristics, among which are financial behaviour, such as saving and borrowing (e.g. Demir et al., 2020; Mbiti & Weil, 2015) and formal account ownership (Gosavi, 2018a; Myeni et al., 2020). Mbiti and Weil (2015) use panel data from 2006 and 2009, based on financial access surveys in Kenya, to find that using mobile money service M-Pesa increases the demand for banking products and the probability of being banked. No significant relationship is found regarding (in)formal loans. The fixed effects regression, with sub-location instrumental variable, also reveals that the service significantly decreases the use of informal saving mechanisms such as ROSCAs9 and using secret hiding places. Therefore, the study provides evidence that M-Pesa is used as a complementary tool to formal finance. A similar argument is made by Ouma et al. (2017) who suggest that individuals who use mobile financial services are more likely to save than those who do not. In their study, also based on survey level data from Kenya for 2013, the authors show that both
7Customer can transfer money, make and receive payments, do in-store purchases, pay bills and, at some operators, save or take out small loans (Ahmad et al., 2020).
8 Payment services such as Apple Pay and Alipay, which are linked to a banking product, are a form of mobile banking but not of mobile money services (GSMA, 2020).
access and the amount saved through basic mobile phone savings, as well as bank integrated mobile phone savings, are enhanced by the use of mobile financial services. The analysis shows that there is a role for mobile financial services in savings mobilization. However, no conclusions can be made about the relationship with financial inclusion; in the empirical analysis, the authors do not distinguish between the two types of mobile phone savings which makes it impossible to identify the impact of either type. Munyegera and Matsumoto (2018) establish that for 820 rural households in Uganda, the likelihood and value of saving, borrowing and receiving remittances increases with usage of mobile money. Probit, Tobit and Logit regression models are used to analyse cross-sectional data. The authors find that the credit channel is used as an alternative to formal banking since the probability of usage is largest for the informal sector. Although it is clear from the analysis that mobile money use increases access to finance, no inferences can be made on the nature of this financial access. The authors fail to distinguish between formal and informal ways of saving and therefore their argument that mobile money has great potential to boost financial access is not substantiated. Nevertheless, a mechanism behind the observed effect is found which is the reduction in the distance to service points for mobile money, in the form of mobile money agents.
The difference between formal and informal saving is important to distinguish when one wants to make an argument for mobile money regarding financial inclusion or access. A study by Myeni et al. (2020) do allow for conclusions on the relationship with financial inclusion; the authors explicitly measure financial inclusion by using a binary variable capturing bank account ownership in Eswatini in 2014. Evidence, although based on cross-sectional data, shows that individuals who use mobile money are more likely to have a bank account at a formal financial institution. Hence, mobile money is used as an accelerator for financial inclusion rather than a substitute. The authors propose that the response of local banks to the increased use in mobile money, in the form of partnerships, could be the reason behind this positive relationship with account ownership, however, no empirical evidence exists. Not only can mobile money enhance financial inclusion at the household level, Gosavi (2018) also finds that SME10 that use mobile money use formal financial institutions more intensively. Business-level data from Tanzania, Kenya, Zambia and Uganda on the access to finance (bank account and loan, bank account or no bank account) for the year 2013, shows that firms who use mobile money are more likely to acquire loans and lines of credit.
In contrast, Fanta et al. (2016) demonstrate that mobile money is used to substitute for formal finance. In their cross-country study based on survey data from 11 countries across the SADC region, access to ATMs and both informal and formal bank account ownership are inversely related to mobile
money adoption. Consistent with Mbiti and Weil (2015), Fanta et al., (2016) find that mobile money ownership decreases the use of informal finance. Nevertheless, there is a lack of consistency among surveys, across countries, and the years the survey was conducted which likely affects the outcomes of the study.
Overall, recent empirical evidence shows that mobile money users tend to save more. However, whether mobile money stimulates, or substitutes for, formal finance, varies across countries and is likely influenced by the nature of the data. The results of these studies are not always convincing in explaining the relationship between mobile money and financial inclusion; the majority of these studies is based on context-specific, single-country, survey data for which the external validity and translation into general policy is questionable. Moreover, apart from Mbiti and Weil (2015), all studies use cross-sectional data which provides only a one-year data observation. Cross-sectional data does not allow for causal inferences, change over time cannot be analysed and time-invariant unobserved heterogeneity is not controlled for. Another limitation is the narrow focus on certain countries or regions, and products such as Kenya, Eastern Africa and M-Pesa respectively11. Very few studies have examined countries from other regions or across different continents, since data is limited. Nevertheless, Demir et al. (2020), do investigate the relationship between FinTech and financial inclusion for a panel of 140 countries for the years 2011, 2014 and 2017. Findings reveal that FinTech is an important driver of formal account ownership, formal savings and formal credit. To my knowledge, the paper by Demir et al. (2020) is the only paper testing the relationship while using a panel of countries from different continents or regions. However, in the analysis only one general effect is measured and no distinction is made between countries with different levels of income or world regions. This makes the results very general and any policy implications hard to infer. Moreover, the use of pooled OLS for estimation in combination with very few control variables could be sources of bias caused through unobserved heterogeneity. Lastly, the variables used to measure financial inclusion do not include a financial access variable; financial access is found to be an important factor for financial inclusion due to factors of convenience, trust and advice (Gosavi, 2018; Sarma & Pais, 2011). In sum, when wanting to observe more general, macro-level trends regarding mobile money and financial inclusion and write policy implications for countries with specific characteristics or from certain regions, there is a need for more global-level research with multiple year data. Hence, I extend the research by Demir et al. (2020) to examine the relationship between
mobile money usage and financial inclusion while using panel level data and a comprehensive financial inclusion index. Moreover, I use random and fixed effects specifications to analyse the panel data. Overall, empirical findings on the relationship between mobile money and formal finance from single- and cross-country studies suggest a positive relationship. This leads to the main research question and corresponding testable hypothesis:
H1: An increase in the usage of mobile money is positively associated with financial inclusion. Additionally, mobile money is aspired to be especially effective in developing countries for societal groups for which formal financial services are either unsuitable, inaccessible or unaffordable. Mobile money services are a means of averting this traditional banking infrastructure with its high costs and high risks. In these, mostly poor and remote areas where financial inclusion is often low, mobile money can complement, and serve as a pathway to, formal financial services. The level of financial inclusion varies widely across countries and income levels, yet income is found to be a significant predictor of financial inclusion (see e.g. Evans, 2016; Lenka & Barik, 2018; Park & Mercado, 2015; Sarma & Pais, 2011). Hence, I propose:
H2: The positive relationship between mobile money usage and financial inclusion is more
pronounced in countries with low income levels.
Lastly, Sub-Saharan Africa is the leading region when it comes to the number of active mobile money accounts and the corresponding transaction volume and value. Sub-Saharan Africa drives around two-thirds of the global mobile money transactions (GSMA, 2020). Moreover, a plethora of research exists on mobile money adoption of which the majority of research is conducted in Sub-Saharan African countries. Mobile money supports a wide range of socio-economic factors, particularly in the Sub-Saharan African region. Therefore, I hypothesize:
H3: The positive relationship between mobile money usage and financial inclusion is more
3. Data and Methodology
3.1 Data
To measure the main variable of interest, ‘mobile money’, data from the Global Financial Inclusion Database (Findex) is used. The Global Findex indicators are drawn from nationally representative surveys from over 150,000 randomly selected adults over 140 countries. Indicators on individual financial behaviour, such as the use and access of both formal and informal financial services and financial technology, provide insight into how adults make payments, borrow, save and manage risk. Data for constructing the Index of Financial Inclusion is obtained from the IMF Financial Access surveys, World Development Indicators of the World Bank and the Global Findex12. The sample consists of 87 countries for the available years 2014 and 2017.
3.1.1 Dependent variable - Index of financial inclusion
The inclusiveness of the financial sector is measured by the Index of Financial Inclusion (hereafter IFI) as developed by Sarma and Pais (2011). This financial inclusion index is widely used in the financial inclusion literature (e.g. Chinoda & Kwenda, 2019; Della Peruta, 2018; Lenka & Barik, 2018). Some authors have used one or a few variables to measure financial inclusion, such as account ownership, the number of ATMs in 1000 km² or formal savings (e.g. Andrianaivo & Kpodar, 2012; Demir et al., 2020; Evans, 2016; Ouma et al., 2017). Yet, this provides only partial information on the comprehensiveness of the complex financial system and hence an index is more appropriate. The IFI is a multidimensional index, using the three basic dimensions of financial inclusion, namely accessibility, availability and usage, to measure the extent of inclusion13. For the two indices that constitute the availability dimension, a weighted average is used giving two-third weight to the bank branch index and one-third weight for the ATM index. Next, the three different dimensions are also given a weight, namely; 1 for the accessibility index, 0.5 for the availability index and 0.5 for the usage index (Sarma & Pais, 2011). The final IFI provides a single index measure to capture three key dimensions of financial inclusion and takes a value between 0 and 1. A value of 1 corresponds to complete financial inclusion and 0 to complete financial exclusion14.
12 For a complete overview of the data sources for all variables, including control variables, see Appendix 2. 13 Appendix 2 shows the corresponding definition and measurement type for each dimension.
3.1.2 Main variable of interest - Mobile money
Mobile money is measured by the indicator ‘mobile phone used to pay bills’. This Global Financial Inclusion indicator from the Global Findex database of the World Bank is defined as: “the percentage of respondents who report using a mobile phone to pay bills in the past 12 months (% age 15+)” (Demirgüç-Kunt et al., 2017). This is the most widely used measure for the use of mobile money services and in line with recent literature (Asongu et al., 2018; Asongu & Odhiambo, 2018; Demir et al., 2020).
3.1.3 Control variables
To preclude the omission of relevant variables, the model also controls for a set of other relevant factors that influence financial inclusion at the country level. In line with recent financial inclusion literature these control variables are: GNI per capita (e.g. Allen, 2012; Della Peruta, 2018; Evans, 2018; Sarma & Pais, 2011); education level (Della Peruta, 2018; Grohmann et al., 2018; Lenka & Barik, 2018; Park & Mercado, 2015); rural population (Asuming et al., 2019; Evans, 2018; Lenka & Barik, 2018); internet access (Asuming et al., 2019; Della Peruta, 2018; Sarma & Pais, 2011); rule of
law (Evans, 2018; Park & Mercado, 2015; Rojas-Suarez, 2016); and population (Evans, 2016;
Gutierrez & Singh, 2013).
With a higher GNI per capita, one is more likely to be included in the formal financial sector (Evans, 2016; Sarma & Pais, 2011; Park & Mercado, 2015). Education level is a significant factor for the involvement in the formal financial system since it teaches the basic understanding of financial concepts, inter alia, financial literacy. Next, rural population rates tend to be associated with lower levels of financial inclusion since people in rural areas are more distant to financial centres (Asuming et al., 2019; Sarma & Pais, 2011). In contrast, secure internet services, connectivity and information spreading enhances financial inclusion and is the foundation for a digital and rapidly developing economy (Evans, 2016; Sarma & Pais, 2011). A good rule of law enables the enforcement of financial contracts, strong legal rights, increases the willingness of banks to lend and gives creditors and debtors more incentives to trust financial institutions (Rojas-Suarez, 2016). Lastly, a larger population indicates a larger market size and thus higher financial access and inclusion (Park & Mercado, 2015).
3.1.4 Sample description
in 2017 and the Republic of Korea in 2017 respectively. Mobile money usage and financial inclusion levels are presented by income-groups in Figure 2. The figure shows data insights concerning Hypothesis 2 and shows that the IFI is highest in richer countries. Contrastingly, this does not seem to be the case for the usage of mobile money; although the mean value is highest for high-income countries, it is the low- and lower-middle-income countries that have the second and third highest mean value of mobile money usage respectively. Nevertheless, the value for high-income countries is still higher than the low-income and lower-middle-income combined. Overall, the usage of mobile money in the sample is relatively low with a mean of 4.183 (percentage of respondents who report using a mobile phone to pay bills in the past 12 months). Figure 2 also reveals that the diffusion of mobile money differs significantly across income groups where the large variation in the usage is also indicated by the large standard deviation of 5.894. Values are clustered; many countries have a value between 0-10% and mobile money usage values of 0% are found for countries in both 2014 and 2017. The highest value of mobile money usage is observed for Kenya; the usage of mobile money exceeds 37% and is rationalised by the rapid spread of mobile services by provider M-Pesa since 2007 (Achmad et al., 2020). There has been strong growth in the usage of mobile money between 2014 and 2017 in all income classes. The percentage change in the usage has been highest in upper-middle-income countries where the average usage increased from 1.117 to 4.261 between 2014 and 2017, which is about a 280 percentage change. The second-highest increase is depicted by low-income countries15. For the sample as a whole, the usage of mobile money in 2017 was three times as high compared to 2014, which demonstrates the intensification of the use of mobile money in countries of all income classifications.
The sample is characterized by a high mean value for GNI per capita; the mean value of $12.884 is just above the dividing line between upper-middle-income and high-income countries16 and therefore corresponds to the classification of ‘high income’ countries. Moreover, the sample has a moderate level of education, the majority lives in urban areas and internet access is just below upper-middle-income standards. The mean percentile rank for the rule of law is exactly at the midpoint of 50.865 which represents moderate quality and confidence of institutions. The difference in percentile rank for the rule of law between Afghanistan (4.327) and Finland (100) shows one of the large contrasts in the sample with highest and, close to, lowest values worldwide. Lastly, the mean population is high, nevertheless, large heterogeneity exists. The large variation between the minimum
and maximum values for the control variables exhibits the large development contrast across the countries in the sample worldwide.
Table 1. Descriptive Statistics
Variable Obs. Mean Std. Dev. Min Max
IFI 174 38.527 11.793 19.794 73.812 Mobile money 174 4.183 5.894 0 37.105 GNI (thousands) 174 12.884 17.012 0.470 104.560 Education 174 8.853 2.650 2.800 13 Rural 174 37.430 20.632 0 83.033 Internet 174 12.954 12.465 0 42.826 Law 174 50.865 26.494 4.327 100 Population (millions) 174 64.608 203.513 0.435 1386.395
Note: The table reports the descriptive statistics of the variables used in the estimations. Obs. stands for number of observations and Std. Dev. for standard deviation of each variable for a panel of 87 countries for the years 2014 and 2017. Mean is the arithmetic average of each variable; Min is the minimum value of each variable; Max is the maximum value of each variable. The dependent variable is IFI which is the index of financial inclusion. The independent variable is mobile money which is mobile money usage. Control variables are: GNI, the GNI per capita; Education the average years of schooling; Rural the % of the population living in rural areas; Internet the number of broadband subscriptions, Law the rule of law and Population the total population in the country.
Figure 2. Financial inclusion and mobile money, by income classification
3.1.5 Data exploration
A check for potential outliers that behave atypical or create a significant association is done based on checking the values in a residual versus fitted plot, presented in Figure 3. The figure shows that the residuals are randomly scattered around the centreline of zero; there are no distinct patterns and the assumption of linearity is reasonable. However, two residuals stand out from the basic random pattern of residuals which are the residuals for Turkey. Nonetheless, excluding Turkey from the regression analysis changes neither the estimates nor significance of the estimates and therefore the outliers are not excluded from the sample.
Figure 3. Residuals versus fitted values plot for outlier check
Note: The horizontal line reflects the estimated regression line, y=0, where the residuals are equal to the
predicted or fitted values.
0.181 indicating a weak positive relationship between the two variables. While the correlation is only weak, this is in line with the theory described in Section 2; increased mobile money usage is expected to be positively associated with financial inclusion levels. Control variables, which have a strong positive correlation with the financial inclusion index, are education, internet subscriptions and the rule of law, which is as expected. The moderate negative correlation between IFI and the variable rural population is likewise as predicted. No conclusion can yet be made based on exploring the data via a correlation matrix as it only shows the unconditional relationships between variables.
Table 2. Correlation Matrix
Variables (1) (2) (3) (4) (5) (6) (7) (8) (1) IFI 1.000 (2) Mobile money 0.181 1.000 (3) GNI 0.490 0.447 1.000 (4) Education 0.685 0.255 0.578 1.000 (5) Rural -0.552 -0.137 -0.577 -0.590 1.000 (6) Internet 0.712 0.390 0.758 0.755 -0.607 1.000 (7) Law 0.629 0.404 0.743 0.646 -0.527 0.777 1.000 (8) Population -0.095 -0.006 -0.106 -0.158 0.164 -0.068 -0.068 1.000 Note: This table reports the unconditional correlation between two variables for all variables used in the estimation.
Figure 4. Unconditional relationship between the IFI and mobile money
Note: The grey line represents the linear fit or the line of best fit that expresses the relationship between the
data points.
3.2 Empirical strategy
In this study, I aim to quantify the relationship between mobile money use and financial inclusion. The baseline model for the study is specified as follows:
𝐼𝐹𝐼𝑖,𝑡 = 𝛽0+ 𝛽1𝑚𝑜𝑏𝑖𝑙𝑒𝑚𝑜𝑛𝑒𝑦𝑖,𝑡+ 𝑥′𝑖,𝑡𝛽2+ 𝑢𝑖,𝑡 (1) where IFI is the Index of Financial Inclusion which measures the level of financial inclusion,
mobile money denotes the usage of mobile money and x' serves as a set of control variables commonly
used in the financial inclusion literature. u is an error term with assumed zero mean and variance equal to one. Lastly, i and t are indicators for individual countries and time.
higher efficiency. Next, the Breusch Pagan test is performed to test the existence of the unique individual-specific error components in the random effects model. The null hypothesis states that the variances across all entities are zero (i.e. no panel effect). I reject the hypothesis that there are no random error components and, therefore, I favour the random effects model over pooled ordinary least squares (POLS) estimation.
Albeit the outcome of the Hausman test, not rejecting the null hypothesis does not mean accepting it; likely, the test does not have sufficient statistical power to distinguish between zero and small correlation reliably (Clark & Linzer, 2015). Furthermore, I have concerns about the appropriateness of the strong assumption underlying the random effects model; it is in general not plausible to make pronounced conclusions about the correlation of the (unobserved) predictors and the error term. Not controlling for all time-invariant variables in the random effects model would mean the predictor variables could be correlated with the error term and estimated coefficients are biased. Moreover, the underlying reasoning of my research question is based on the variation of the explanatory and predictor variables within a country instead of the variation across countries. It is rather mobile money usage within a country, which I am interested in and expect to be associated with financial inclusion instead of the differences in mobile money usage across countries. Therefore, based on reasoning and the shortcoming of possible omitted variable bias in the random effects model, the fixed effects model will be used in addition to the random effects model in the empirical analysis. The coefficients of the fixed-effects model cannot be biased due to omitted time-invariant characteristics since the model controls for all country-specific time-invariant heterogeneity. Unobserved time-invariant cross-country differences are controlled for; hence, what is left to explain is the within-country variation. The fixed effects model is specified as follows:
4. Empirical Results
4.1 Main resultsTable 3 shows the results of the regression as characterised by the random effects model17. Three different model specifications are reported corresponding to the three hypotheses created in Section 2. First, column (1) shows the baseline specification as characterised by equation (1) which tests the main research question of this thesis; is an increase in the uptake of mobile money positively related to the level of financial inclusion? Next, in column (2), I examine whether the hypothesised association varies across countries with different income levels. Therefore, I re-estimate the baseline model of column (1) and I include a dummy variable for income classifications and its interaction with the variable mobile money. Here the income classifications are interpreted in relation to the reference or base group (for which the dummy variable is equal to zero) which is, in this case, the high-income countries. Lastly, column (3) is a test of Hypothesis 3, which includes an interaction term between mobile money and the continent Sub-Saharan Africa. The coefficient of the interaction term is interpreted as the additional effect for Sub-Saharan Africa, relative to the rest of the world. I will analyse the outcomes of the random effects columns (1-3) before comparing the regression results by the fixed effects model (4-6) presented in Table 4.
In Table 3, the estimates from the random effects specifications indicate that in column (1)
mobile money is not a significant predictor of financial inclusion. In column (2) the results of the
interaction by income classification show that there is significant variability in the relationship across income groups. A one percentage point increase in mobile money creates a marginal increase of 0.232 in financial inclusion for low-income countries compared to high-income countries, statistically significant at the ten per cent level. For lower-middle-income countries, the association is also positive; a one-percentage-point increase in mobile money usage is significantly expected to increase financial inclusion levels by 0.313 compared to high-income countries. For upper-middle-income countries, the relationship is positive but not statistically different from high-income countries. In sum, relative to high-income countries, the relationship between mobile money and financial inclusion is positive for all other income classifications and significant for lower-middle and low-income countries. Hence, the low-income context matters showing that the relationship is positive in income countries relative to high-income countries. This result is intuitive given that
income economies have considerably lower financial inclusion levels than high-income groups, and so the increase in financial inclusion would be greater given an increase in mobile money. Moreover, the findings might reflect the digital transformation of banks and branches, and the move towards cashless societies in high-income countries which reduces the availability component of financial inclusion (Nguyen, 2014). Overall, Hypothesis 2 is not satisfied; although there is a positive relationship for low-income countries, this positive relationship is stronger for lower-middle-income countries.
Table 3. The relationship between Mobile Money and Financial Inclusion - Random Effects Model
Note: This table reports the regression results for the relationship between mobile money and financial inclusion by using the panel random effects model specified in equation (1). The dependent variable is the Financial Inclusion Index. The explanatory variable is Mobile money. Control variables are GNI = Gross National Product per capita; Education = mean years of schooling; Rural = rural population (% of total population); Internet = fixed broadband subscriptions per 100 people; Law = Rule of law; Population = total population. These variables are further described in Appendix 2. SSA = Sub-Saharan Africa countries. Low income, Lower middle income and Upper middle income reflect income groups. Robust standard errors clustered at the country level in parentheses: *** p<0.01, ** p<0.05, * p<0.1
Appendix 6 gives a more in-depth outlook of the relationship between mobile money and financial inclusion by income classification, regarding results in column (2). For high-income countries, the direction of the relationship is clearly negative for the countries Finland, Norway and Sweden (see Appendix 6.1). Reason for the reduction is found in the calculation of the IFI and is characterised by the declining number of commercial bank branches and ATMs in these countries. In
(1) (2) (3)
VARIABLES Baseline Model Income Classification Sub Saharan Africa
Interaction Mobile Money -0.143 -0.243*** -0.222** (0.089) (0.092) (0.091) GNI (log) -0.038 0.443 -0.591 (1.003) (1.248) (1.003) Education 1.913*** 1.830*** 1.654*** (0.341) (0.350) (0.369) Rural -0.076 -0.078 -0.076 (0.050) (0.047) (0.048) Internet 0.249** 0.287** 0.313*** (0.104) (0.108) (0.106) Law 0.040 0.051* 0.049 (0.036) (0.04) (0.037) Population 0.002 0.001 0.001 (0.007) (0.007) (0.007) Low income 0.232** (0.105)
Lower middle income 0.313***
(0.119)
Upper middle income 0.183
(0.199)
SSA * Mobile Money 0.325***
(0.104)
Constant 19.970** 14.294 26.670***
(8.673) (12.350) (9.177)
R-squared 0.577 0.597 0.599
upper-middle-income countries, there are no notable in- or decreases in the IFI between 2014 and 2017, except for Turkey and Costa Rica (see Appendix 6.2). These relationships are not significantly different from high-income countries. The positive direction of the relationship for lower-middle-income countries, as seen in Appendix 6.3, is mainly driven by Cote d’Ivoire, Kenya and Ghana. In these countries, the number of bank accounts and physical infrastructure (bank branches and ATMs) has been increasing resulting in rising IFIs. In low-income countries, the use of mobile money has increased considerably from 2014 to 2017 which resulted in slight increases in the IFI (see Appendix 6.4).
In Table 4, I present the results of the fixed effect regression analysis, as specified by equation (2). The results show how much the dependent variable changes, on average, in response to the variation in the independent variables within the countries. In column (1), mobile money is negatively associated with financial inclusion, significant at the ten per cent level. Therefore, Hypothesis 1 does not hold based on the results of column (1). These results are in contrast with Demir et al. (2020) who state that FinTech is an important driver for financial inclusion. Differences in the measurement of financial inclusion could potentially drive the contrasting results; Demir et al. (2020) do not include physical infrastructure in their measurement. Appendix 6.2 shows that in some high-income countries, where the majority of the population has access to formal financial services and an account, financial inclusion is decreasing over time. This is potentially explained by banks, which are closing down bank branches to reduce costs, whilst more customers are using online banking services (Bhattacharya, 2016).
Column (2) shows that, once I control for country-specific time-invariant factors by adding country fixed effects, the significance for the income classifications, as in the random effects model, disappears. Although the estimates are positive for all income classifications, I no longer find significant within-country variation for the different income groups. Therefore, Hypothesis 2 does not hold based on these results. When changes within one country are measured, while controlling for time-invariant country-specific effects, there is no significant relationship by income classification. However, when changes between countries are measured, as in the random effects model, there is a significant relationship for income classifications. Therefore, there is significant variation across countries, but insignificant variation over time.
countries over time and there is evidence that Hypothesis 3 holds. There is a positive correlation, and thus complementary relationship, between mobile money and financial inclusion in Sub-Saharan African countries. This is in line with the empirical literature where mobile money is found to increase the demand for banking products (such as savings), the probability of being banked and bank account ownership in Sub-Saharan African countries (see e.g. Mbiti & Weil, 2015; Myeni et al., 2020; Ouma et al., 2017). A possible mechanism of this relationship for Sub-Saharan African countries is presented in Section 4.2.
Table 4. The relationship between Mobile Money and Financial Inclusion - Fixed Effects Model
Note: This table reports the regression results for the relationship between mobile money and financial inclusion by using the panel fixed effects model specified in equation (2). The dependent variable is the Financial Inclusion Index. The explanatory variable is Mobile money. Control variables are GNI = Gross National Product per capita; Education = mean years of schooling; Rural = rural population (% of total population); Internet = fixed broadband subscriptions per 100 people; Law = Rule of law; Population = total population. These variables are further described in Appendix 2. SSA = Sub-Saharan Africa countries. Low income, Lower middle income and Upper middle income reflect income classifications. Robust standard errors clustered at the country level in parentheses: *** p<0.01, ** p<0.05, * p<0.1
4.2 Mechanisms
To find out why mobile money is successful in enhancing financial inclusion in Sub-Saharan African countries, I add additional control variables to the regression that may represent the intermediate mechanism by which the independent variable influences the dependent variable. Table 5 presents the estimation results with different model specifications. Column (1) shows the regression results for the relationship without control variables. In column (2) the baseline model with control variables,
(1) (2) (3)
VARIABLES Baseline Model Income Classification Sub Saharan Africa
Interaction Mobile Money -0.180* -0.239** -0.238** (0.092) (0.098) (0.093) GNI (log) -0.418 -0.558 -1.078 (1.403) (1.545) (1.387) Education 2.917*** 2.846*** 2.871*** (0.822) (0.864) (0.871) Rural -0.211 -0.080 -0.133 (0.261) (0.273) (0.255) Internet 0.118 0.136 0.213 (0.130) (0.154) (0.138) Law 0.041 0.045 0.045 (0.045) (0.046) (0.046) Population 0.075*** 0.061** 0.079*** (0.023) (0.030) (0.025) Low income 0.168 (0.129)
Lower middle income 0.223
(0.143)
Upper middle income 0.167
(0.182)
SSA * Mobile Money 0.267**
(0.109)
Constant 16.421 11.241 17.963
(14.577) (15.250) (14.930)
R-squared 0.313 0.371 0.367
as specified by equation (2), is presented. Next, in column (3) and (4) I add control variables that are the focus of the mechanism analysis.
In column (3) I include the variable ‘Mobile cellular subscriptions per 100 people’ which represents the number of mobile telephone subscribers. Ownership of a mobile phone subscription is a prerequisite for mobile money usage and the fast penetration of mobile phone ownership in Sub-Saharan African countries could be a potential mechanism through which mobile money usage influences financial inclusion. Most Sub-Saharan African countries have known rapid growth in mobile cellular subscriptions; the growth between 2003-2018 in Sub-Saharan African countries have been 3.8 times higher compared to the world average18. In column (3) the mobile subscriptions coefficient is positive and significant. When adding the mobile subscriptions variable, the coefficient for the interaction between mobile money and Sub-Saharan African countries decreases, though only slightly and remains significant at the five per cent level. Hence, mobile subscriptions are not considered a sound mechanism of impact.
In column (4) I postulate that the mechanism of impact in Sub-Saharan African countries is the initial level of financial inclusion. In developed countries, where most people are banked, it takes longer to adapt to new technologies since the infrastructure for traditional technology is already in place and functioning. Because in Sub-Saharan African countries most people are not banked to begin with, it is easier to step in when new technologies are presented. Therefore, new technologies are leapfrogging over existing formal banking services, especially when these technologies do not require large-scale, expensive infrastructure projects to set the service up (Aron, 2018; Bar & Galperin, 2007). Hence, the model in column (4) controls for the initial level of financial inclusion by using the variable ‘number of deposit accounts with commercial banks per 1000 adults’. When adding the variable bank
account in column (4), the coefficient for the interaction between mobile money and Sub-Saharan
African countries turns negative and is no longer significant. The decrease from 0.267 in column (2) to -0.012 in column (4) shows that, once bank account ownership is controlled for, I find that mobile money no longer plays a significant role in explaining financial inclusion in Sub-Saharan African countries. Therefore, part of the reason mobile money is positively correlated with financial inclusion in Sub-Saharan African countries, is that those who live in Sub-Saharan African countries are initially less likely to be financially included. The rapid spread of mobile money in Sub-Saharan African countries solves problems, which arise from the drawbacks of conventional banking, through leapfrogging over the existing deficient formal banking infrastructure (Aron, 2018). The low level of
financial inclusion in combination with low adoption costs of new technologies19 accelerates the speed of innovation and uptake in Sub-Saharan African countries (Andrés et al., 2010).
Table 5. The relationship between Mobile Money and Financial Inclusion - Mechanism testing
Note: This table reports the regression results for the relationship between mobile money and financial inclusion by using the panel fixed effects model specified in equation (2). The dependent variable is the Financial Inclusion Index. The explanatory variable is Mobile money. SSA = Sub-Saharan Africa countries. Control variables are GNI = Gross National Product per capita; Education = mean years of schooling; Rural = rural population (% of total population); Internet = fixed broadband subscriptions per 100 people; Law = Rule of law; Population = total population; Mobile Subscriptions = mobile cellular subscriptions per 100 people; Bank Account = Number of deposit accounts with commercial banks per 1000 adults. These variables are further described in Appendix 2. Robust standard errors clustered at the country level in parentheses: *** p<0.01, ** p<0.05, * p<0.
19 Technological innovations are generally early adopted by high-income countries, the speed of innovation is higher in low-income countries since those countries benefit from lower adoption costs (Andrés et al. 2010).
(1) (2) (3) (4)
VARIABLES Uncontrolled
Model
Baseline Model Mobile Subscriptions Variable Bank Account Variable Mobile Money -0.235** -0.238** -0.236** -0.099*** (0.093) (0.093) (0.091) (0.029)
SSA* Mobile Money 0.206** 0.267** 0.234** -0.012
5. Conclusion
In this thesis, I hypothesized that the use of mobile money is positively associated with financial inclusion and additionally that this association is more pronounced for countries with lower income levels and for Sub-Saharan African countries. Using panel data covering 87 countries in 2014 and 2017, I find evidence of the contrary; mobile money decreases the level of financial inclusion for the sample as a whole. Analysis by income classification shows positive significant variation across countries in low- and lower-middle-income countries relative to high-income countries. Nonetheless, mobile money is positively associated with financial inclusion in Sub-Saharan African countries. I find that the mechanism of this positive relationship is the initial level of financial inclusion reflecting the ‘leapfrogging’ over formal banking services by using mobile money. Mobile money works because it provides an accessible, cheap and reliable alternative to the exclusive formal banking infrastructure that already exists and provides a pathway to formal finance where financial inclusion levels are low. These findings contribute to the existing cross-country literature on the role of mobile money in promoting financial inclusion. Although mobile money serves as a complementary service to the formal financial sector in Sub-Saharan African countries, the results do not validate mobile money to be ‘the answer’ to the global financial exclusion of around 1.7 billion adults and the solution for global poverty. Despite the fact that the lack of formal banking services in lower-income countries might have encouraged financial resilience and fast growth of mobile money services, for the sample as a whole mobile money serves as a substitute for formal finance. The question remains whether mobile money has the ability to substitute formal financial services because access to finance, such as savings- products and credit, play a critical role for small businesses, mitigating risks and ultimately economic development.
money becomes available at the global level. For future research, I recommend replicating the study using more recent and extensive data to capture the current development and status of mobile money. Additionally, exploration of data on mobile money usage from the population with the lowest quantiles of income would provide for better insight into the potential of mobile money for the unbanked and allows for better policy recommendations. Moreover, using a revised definition of financial inclusion could allow for the inclusion of more present-day digital developments and transformations of the banking sector. Therefore, future research should focus on an index that gives less weight to physical infrastructures, such as ATMs, and better captures recent developments towards online banking and societies that are more cashless.
References
Ahmad, A. H., Green, C., & Jiang, F. (2020). Mobile money, financial inclusion and development: A review with reference to African experience. Journal of Economic Surveys, 34(4), 753-792. Allen, F. (2012). Trends in financial innovation and their welfare impact: an overview. European
Financial Management, 18(4), 493-514.
Andrés, L., Cuberes, D., Diouf, M. A., & Serebrisky, T. (2010). Diffusion of the Internet: A Cross-country Analysis. Telecommunications Policy 34 (5–6), 323–340.
Andrianaivo, M., & Kpodar, K. (2012). Mobile phones, financial inclusion, and growth. Review of
Economics and Institutions, 3(2), 30.
Aron, J. (2018). Mobile money and the economy: a review of the evidence. The World Bank Research
Observer, 33(2), 135-188.
Asongu, S. A., & Odhiambo, N. M. (2018). Human development thresholds for inclusive mobile banking in developing countries. African Journal of Science, Technology, Innovation and
Development, 10(6), 735-744.
Asongu, S. A., Nwachukwu, J. C., & Orim, S. M. I. (2018). Mobile phones, institutional quality and entrepreneurship in Sub-Saharan Africa. Technological Forecasting and Social Change, 131, 183-203.
Asuming, P. O., Osei-Agyei, L. G., & Mohammed, J. I. (2019). Financial inclusion in sub-Saharan Africa: Recent trends and determinants. Journal of African Business, 20(1), 112-134.
Bar, F. & Galperin, H. (2007) Wireless communication and development: micro and macro linkages.
Information Technologies and International Development, 3(2), 1–2.
Barajas, A., Beck, T., Belhaj, M., & Naceur, S. B. (2020). Financial inclusion: what have we learned so far? What do we have to learn? International Monetary Fund Working Papers,
20(157), 1-51.
Beck, T., Demirgüç-Kunt, A., & Levine, R. (2007). Finance, inequality and the poor. Journal of
economic growth, 12(1), 27-49.
Beck, T., Senbet, L., & Simbanegavi, W. (2015). Financial inclusion and innovation in Africa: An overview. Journal of African Economies, 24, 3-11.
Bhattacharya, S. G. (2016, September 5). The problem with closing bank branches, International Finance. https://internationalfinance.com/the-problem-with-closing-bank-branches/
Chinoda, T., & Kwenda, F. (2019). Do mobile phones, economic growth, bank competition and stability matter for financial inclusion in Africa?. Cogent Economics & Finance, 7(1), 1622180. Clark, T. S., & Linzer, D. A. (2015). Should I use fixed or random effects. Political Science Research
and Methods, 3(2), 399-408.
Della Peruta, M. (2018). Adoption of mobile money and financial inclusion: a macroeconomic approach through cluster analysis. Economics of Innovation and New Technology, 27(2), 154-173.
Demirguc-Kunt, A., Klapper, L., Singer, D., Ansar, S., & Hess, J. (2018). The Global Findex
Database 2017: Measuring financial inclusion and the fintech revolution. The World Bank.
https://globalfindex.worldbank.org/sites/globalfindex/files/201804/2017%20Findex%20full%2 0report_0.pdf
Evans, O. (2016). Determinants of financial inclusion in Africa: A dynamic panel data approach.
University of Mauritius Research Journal, 22.
Evans, O. (2018). Connecting the poor: the internet, mobile phones and financial inclusion in Africa. Digital Policy, Regulation and Governance, 20(6), 568-581
Fanta, A. B., Mutsonziwa, K., Goosen, R., Emanuel, M., & Kettles, N. (2016). The role of mobile
money in financial inclusion in the SADC region. FinMarkTrust.
Gosavi, A. (2018). Can mobile money help firms mitigate the problem of access to finance in Eastern sub-Saharan Africa?. Journal of African Business, 19(3), 343-360.
Grohmann, A., Klühs, T., & Menkhoff, L. (2018). Does financial literacy improve financial inclusion? Cross country evidence. World Development, 111, 84-96.
GSMA (2020). State of the Industry Report on Mobile Money, 2019.
GSMA. https://www.gsma.com/sotir/wp-content/uploads/2020/03/GSMA-State-of-the-Industry-Report-on-Mobile-Money-2019-Full-Report.pdf
Gutierrez, E., & Singh, S. (2013). What regulatory frameworks are more conducive to mobile banking? Empirical evidence from Findex data. Policy Research Working Paper; No.
6652. World Bank, Washington, DC.
IMF (2019). “Mobile Money Note 2019”. International Monetary Fund. http://fdg.extcc.com/sites/default/files/users/user381/Mobile%20Money%20Note%2020 19%20(1).pdf
Jack, W. and T. Suri (2011), ‘Mobile money: The economics of M-PESA’. Working Paper 16721,
NBER, Cambridge, MA
Kim, M., Zoo, H., Lee, H., & Kang, J. (2018). Mobile financial services, financial inclusion, and development: A systematic review of academic literature. The Electronic Journal of Information
Systems in Developing Countries, 84(5), e12044.
Lenka, S. K., & Barik, R. (2018). Has expansion of mobile phone and internet use spurred financial inclusion in the SAARC countries?. Financial Innovation, 4(1), 5.
Van Loo, Rory (2018). "Making Innovation More Competitive: The Case of Fintech". UCLA Law
Review. 65 (1): 232.
Mbiti, I., & Weil, D. N. (2015). Mobile banking: The impact of M-Pesa in Kenya. University of
Chicago Press, 3, 247-293
Miriri, D. (2019, April 29). Kenya's Equity Bank, Safaricom team up to expand digital banking. Thomson Reuters. https://www.reuters.com/article/us-kenya-equity-group-hol-safaricom-idUSKCN1S51IC
Myeni, S., Makate, M., & Mahonye, N. (2020). Does mobile money promote financial inclusion in Eswatini?. International Journal of Social Economics, 47(6), 693-709
Nan, W., Zhu, X., & Lynne Markus, M. (2020). What we know and don't know about the socioeconomic impacts of mobile money in Sub‐Saharan Africa: A systematic literature review. The Electronic Journal of Information Systems in Developing Countries, e12155. Nguyen, H. L. Q. (2014). Do bank branches still matter? The effect of closings on local economic
outcomes. Department of Economics, Massachusetts Institute of Technology, Cambridge, MA. Ouma, S. A., Odongo, T. M., & Were, M. (2017). Mobile financial services and financial inclusion:
Is it a boon for savings mobilization?. Review of development finance, 7(1), 29-35.
Ozili, P. K. (2018). Impact of digital finance on financial inclusion and stability. Borsa Istanbul
Review, 18(4), 329-340.
Park, C. Y., & Mercado, R. (2015). Financial inclusion, poverty, and income inequality in developing Asia. Asian Development Bank Economics Working Paper Series, (426).
Rojas-Suarez, L. (2016). Financial inclusion in Latin America: Facts and obstacles. Center for Global
Development Working Paper, (439).
Sarma, M., & Pais, J. (2011). Financial inclusion and development. Journal of international
development, 23(5), 613-628.
Sekantsi, L. P., & Motelle, S. I. (2016). The financial inclusion Conundrum in Lesotho: Is mobile money the missing piece in the puzzle. Journal of Financial Research, 8(3), 138-192.
Sinclair, S. P. (2001). Financial Exclusion: An Introductory Survey Centre for Research into Socially Inclusive Services. Edinburgh College of Art, Heriot Watt University, Edinburgh.
UNCDF (2020a). Financial inclusion and the SDGs, Unlocking Public and Private Finance for the
Poor. UNCDF. https://www.uncdf.org/financial-inclusion-and-the-sdgs
UNCDF (2020b). Leaving no one behind in the digital era. UN Capital Development Fund. file:///X:/My%20Downloads/25.08.20UNCDF_Inclusive_Digital_Economy_EN.pdf World Bank (2018, October 1). UFA2020 Overview: Universal Financial Access by 2020. World
Bank. https://www.worldbank.org/en/topic/financialinclusion/brief/achieving-universal-financial-access-by-2020
Appendix
Appendix 1. Evolution of global mobile services, 2001-2019
Appendix 2. Specification of Variables and Sources
Variable Definition and Measurement Source
Panel A: Financial Inclusion Indicators
Accessibility Number of deposit accounts with commercial banks (no per 1000 adults)
IMF Financial Access Survey & Global Findex
Availability Number of commercial bank branches per 100000 adults
Number of ATMS per 100000 adults
IMF Financial Access Survey
Usage Domestic credit to private sector (as % GDP) World Development Indicators (World Bank)
Panel B: Mobile Money Indicator
Mobile Money The percentage of respondents who report using a mobile phone to pay bills in the past 12 months (% age 15+)
Global Findex
Panel C: Control Variables
Level of Income GNI per capita, PPP World Development
Indicators (World Bank)
Education Level Mean years of total schooling across all education levels (years)
UNDP Human Development Report
Rural Population Rural population (% of total population) World Development Indicators (World Bank) IT Infrastructure: Internet Access Fixed broadband subscriptions (per 100 people) World Development
Indicators (World Bank) Institutional Quality Rule of law: percentile rank Worldwide Governance
Indicators (World Bank)
Population Population, total World Development
Indicators (World Bank) Mobile Subscriptions Mobile cellular subscriptions (per 100 people) World Development
Indicators (World Bank) Bank account Number of deposit accounts with commercial
banks( per 1000 adults)
IMF Financial Access Survey
Appendix 3. Sample Characteristics
Continent
Income classification Low Income Lower
middle Income Upper middle income High Income Total
East Asia & Pacific 0 9 7 6 22
Europe & Central Asia 0 11 9 34 54
Latin America & Caribbean 0 9 15 8 32
Middle East & North Africa 0 7 5 8 20
South Asia 4 6 0 0 10
Sub-Saharan Africa 13 17 6 0 36
Total 17 59 42 56 174
Appendix 4. Index of Financial Inclusion and Rank
Country IFI in 2014 Rank in 2014 IFI in 2017 Rank in 2017
Afghanistan 19.844 88 19.794 88 Algeria 26.227 69 27.047 72 Argentina 33.808 56 38.659 43 Armenia 41.283 33 44.076 25 Austria 43.957 26 41.912 30 Bangladesh 30.838 65 32.506 62 Belgium 67.497 2 68.383 2 Benin 22.249 82 22.534 86 Bolivia 34.732 51 38.650 44 Bosnia and Herzegovina 41.791 31 40.717 36 Botswana 32.144 61 32.490 63 Brazil 37.363 44 36.891 51 Cameroon 21.881 85 22.682 85 Chile 51.145 10 52.625 8 China 25.115 74 24.744 77 Colombia 40.468 34 43.002 27 Congo, Rep. 21.788 86 23.277 83 Costa Rica 42.589 29 45.315 21 Cote d'Ivoire 23.424 81 23.722 82 Croatia 41.337 32 37.892 46 Czech Republic 50.447 13 44.391 24 Dominican Republic 31.260 64 32.400 64 Ecuador 31.703 63 32.784 61
Egypt, Arab Rep. 26.215 70 28.574 68 El Salvador 34.220 52 36.289 53
Finland 47.871 20 35.599 56 Georgia 45.204 22 50.498 11 Ghana 25.184 73 27.795 71 Greece 63.166 4 64.566 5 Guatemala 43.123 27 41.822 31 Honduras 37.100 45 37.844 47 Hungary 37.677 43 36.500 52 India 38.727 40 43.392 26 Indonesia 35.350 50 40.966 35 Ireland 53.934 8 49.561 14 Israel 40.208 35 41.394 33 Italy 38.750 39 39292 40 Japan 66.582 3 66.632 3 Jordan 33.727 57 33.830 59 Kenya 32.293 60 35.609 55 Korea, Rep. 72.955 1 73.812 1 Kosovo20 39.107 38 36.969 50 Kuwait 40.052 37 41.526 32 Kyrgyz Republic 25.723 72 27.812 70 Latvia 48.701 17 46.291 20 Lebanon 40.087 36 39.273 41 Madagascar 20.336 87 20.869 87 Malaysia 47.982 19 46.852 18 Malta 56.827 7 61.317 6 Mauritania 23.437 80 24.317 78 Mauritius 50.897 12 49.109 15 Mexico 34.031 54 35.188 57
Uruguay 32.848 59 33.996 58 Uzbekistan 33.863 55 37.754 48
Vietnam 32.868 58 32.283 65
West Bank and Gaza 34.074 53 37.152 49
Zambia 23.821 79 23.805 81
Zimbabwe 24.758 76 25.251 75 Note: The index of financial inclusion for all countries in the sample as calculated by Sarma & Pais (2011). Based on the author’s calculations.
Appendix 5. Increase in mobile money usage, by income classification
Mobile money 2014 Mobile money 2017 Percentage change
Low income 1.453 5.230 259.945%
Lower middle income 1.411 4.090 189.865%
Upper middle income 1.117 4.261 281.468%
High income 3.493 10.637 204.523%
World 2.011 6.355 216.012%
Appendix 6. Scatterplot Mobile Money and IFI
Appendix 6.1. Scatterplot Mobile Money and IFI - High-income countries
Appendix 6.3. Scatterplot Mobile Money and IFI - Lower-middle-income countries
Appendix 7. Mobile money between 2014 and 2017
Appendix 8. Mobile cellular subscriptions 2003-2018
Source: International Telecommunication Union (2020).
4.960 76.159 22.221 104.067 0 20 40 60 80 100 120 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
