Mobile money: substitute or complement of informal and formal financial mechanisms? Evidence from Burkina Faso1 Serge Ky2, Clovis Rugemintwari and Alain Sauviat Université de Limoges, LAPE, 5 rue Félix Eboué, 87031 Limoges Cedex, France

Preliminary draft: Please do not quote without permission from the authors

Abstract. Using a survey data conducted between May and June 2014 in Burkina Faso, we explore the impact of mobile money as a substitute or complement of informal and formal financial instruments. The results show that mobile money account may act as a substitute of informal deposits mechanisms. Furthermore, we find that for individuals participating in informal mechanisms, mobile money may increase their likelihood to make deposits using bank account. We also find that the use of mobile money increases the likelihood to make deposits using bank and credit union accounts for disadvantaged groups especially, female, those with irregular income and less educated. In further investigations, we discuss mechanisms underlying these results. Our findings show the cost, illiquidity and low quality related to formal financial instruments, and the risk, illiquidity and the low quality of informal mechanisms as factors that may lead individuals to use mobile money account to make their deposits. Given the low access to formal finance in developing countries, our results taken together show how the increasing adoption of mobile technologies may be an opportunity towards financial inclusion. (JEL Classification C83, D14, G21, G23, O12) Keywords: mobile money, savings, financial inclusion, sub-Saharan Africa, Burkina Faso 1

An earlier version of the paper was presented at the LAPE PhD seminars. We are grateful to Philippe Rous, Céline Meslier, Emmanuelle Nys, and Leo Indra Wardhana for their helpful comments and suggestions. We also wish to thank Dénis Ouédraogo and Jean-Paul Ouoba from the Institut National de la Statistique et de la Démographie du Burkina Faso for providing us data on access to financial services in Burkina Faso. All remaining errors are naturally ours. 2 Corresponding Author. Tel: + 33 5 55 14 92 51. E-mail address : [email protected]

1

1. Introduction Well-functioning financial systems serve a vital purpose, offering savings, credit, payment and risk management products to people with a wide range of needs (Demirgüç-Kunt and Klapper 2012). The financial services needs of lower-income people in Africa, which have long been excluded from the formal finance, have receive more attention from governments, international organizations and even bank institutions. Several initiatives have been implemented such as microfinance and post offices or credit unions as suppliers of basic bank services to increase people access to formal finance. It is argued that the benefit of formal finance for poor households is that it helps build household resilience by enabling household anticipate, adapt to and/or recover from the effects of shocks in a manner that protects their livelihoods, reduces chronic vulnerability and facilitates growth (Gash and Gray, 2015). However, access to formal financial institutions remained very low in Sub-Saharan Africa. The low network of formal financial institutions and the location of retails outlets in urban area, the importance of population density are some factors among others that explained why poor people living far from financial institutions are less likely to access and use formal financial services (Allen et al., 2014; Beck, Demirgüç-Kunt and Honohan, 2009; Demirgüç-Kunt and Klapper, 2012; Dermish et al., 2012). For developing countries, financial development and deepening are likely to have important implications for economic growth and the effectiveness of monetary policy (Carpenter and Jensen, 2002). There is a massive effort underway to harness technology to overcome barriers that prevent poor people to access banking services. In this way, branchless banking3, especially mobile money is particularly prominent to potentially improve financial inclusion. Mobile money services are qualified as “semi-formal” as providers of this type of financial services are sometimes not registered and in general not supervised by the financial sector supervisors (De Koker and Jentzsch, 2013; Shem, Misati and Njoroge, 2012). In this regard, the development challenge in promoting formal financial access lies in the design of financial technology that can be accessible to meet the need of the unbanked population. In this article, we empirically test the hypothesis that the use of mobile money may act as a complement or a substitute of informal and/or formal financial instruments. Formal financial

3

Branchless banking refers to new distribution channels that allow financial institutions and other commercial actors to offer financial services outside traditional bank premises (Dermish et al. 2012).

2

instruments refer to bank accounts and credit union accounts while informal instruments refer to rotating savings and credit associations (ROSCA), savings at home or under mattress, with a neighbor or in livestock. In developing countries, individuals usually integrate a variety of deposit instruments into their deposits/savings portfolios to meet their financial needs including informal and formal instruments. However, informal deposit mechanisms despite meeting some specific goals of individuals remain risky as it refer to financial services that are rendered outside of the scope of the formally regulated and supervised financial services sector (De Koker and Jentzsch, 2013). The existence of an extensive population involving in informal deposit mechanisms represents an important opportunity for formal financial intermediaries. In this regard, microfinance institutions as well as credit unions, post banks and cooperatives play a prominent role in Burkina Faso by providing individuals with access to savings and credit and also contribute to the reduction of poverty (Thieba 2013). The development strategies of microfinance institutions entail the reduction of the gap in access to formal financial services between urban and rural areas with the priority to reach first women. Thus, microfinance institutions appear as a substitute of bank institutions particularly for people without access to basic formal financial services and individuals located in remote areas. The formal financial services are subject to laws, regulations and prudential supervision and provide some additional advantages for both individuals and economy. Promoting inclusive finance is critical to enhance efficiency and welfare by facilitating access to an appropriate formal financial services by poor individuals (Sarma and Pais, 2011). It is well known that financial development or inclusive financial system requires a transition from the informal to the formal finance. New technologies can particularly help to solve problems arising from weak institutional infrastructure (Klein and Mayer, 2011). Therefore, mobile money can play this key role by providing avenues for secure and safe deposit practices and by facilitating access and usage of a whole range of formal financial services. The innovation of mobile technology refers to the use of mobile phone to perform financial transactions as well as remittances, pay bills, purchase goods and services, and also allows individuals to store value through cash in and cash out functions. In developing countries where access to financial services are very limited the adoption of mobile money appears as an alternative to formal financial services for unbanked especially disadvantaged individuals.

3

Mobile money not only reduces transactions costs but also greatly increases individuals convenience for cash deposits and withdrawals, and minimize the need of costly physical infrastructure as well as branch networks (Kendall, Schiff, and Smadja 2014). There are many actors in partnership with licensed banks involved in the supply of mobile money services such as mobile operators (M-Pesa in Kenya and Tanzania), M-payment or electronic money issuer (WIZZIT in South Africa, Eko in India) (Mas 2009). This is similar in Burkina Faso where the providers of mobile money services include licensed commercial banks4 in partnership with mobile operators (M-Ligdi and Mobi-Cash), and electronic money (e-money) issuers (Inovapay). The success of mobile money also relies on retail networks or mobile money agents that interact with mobile money providers and also guarantee the conversion of cash into electronic money and vice versa for customers. Thus, the widespread of mobile money agents is hence essential for mobile money users to have convenient access to cash in/out options, and particularly held sufficient liquidity or e-money to ensure the efficiency of the conversion between e-money and cash. A substantial amount of research mainly focuses on the potential implications of mobile money for financial development in developing countries but little investigate empirically its impact on existing informal and/or formal financial services. Dermish et al. (2012) argues that branchless banking takes the advantage of increasingly mobile networks to bring banking services into every day retails stores, thereby alleviating the lack of banking infrastructure in the area where poor people live and work. Mobile money has the potential to enhance the relationship between banks and their clients as customers can guide banks to what their needs are. This may in return allow banks to provide them the right formal financial products (Mas, 2012). Indeed, mobile money allows the excluded from formal financial system to perform financial transactions relatively cheaply, securely and reliably (Demirguc-kunt and Klapper, 2012; Dermish et al., 2012; Mbiti and Weil, 2011). In this perspective, mobile money can be considered as a stepping stone to formal financial services by increasing the likelihood of individuals to use formal deposit accounts. Mbiti and Weil (2011) show that M-PESA adoption in Kenya reduces monetary and security costs of transferring money compared to traditional tools of money transfers such as Western Union, MoneyGram or transport companies. They also show how M4

The Central Bank (BCEAO) provides agreement to only financial institutions (commercial banks, e-money issuers) and supervised e-money activities. The compensation of e-money is necessarily held by the commercial banks.

4

PESA serves as a storage of value by decreasing the use of informal saving mechanisms such as ROSCA. Morawczynski (2009) in a fieldwork in Kenya describes how M-PESA acts as a complement rather than a substitute to other deposit/saving mechanisms such as banks, home bank and ROSCA. In the same vein, Morawczynski and Pickens (2009) also argue analytically that individuals use M-PESA as a substitute for informal methods of savings, especially keeping money at home. By contrast, De Koker and Jentzsch (2013) who use household surveys conducted in eight African countries find that holding a bank account is not negatively associated with the probability of using informal finance. More specifically, they show that an increase in the access to formal financial services including usage of mobile banking for receipt of salary or income payments has not resulted in a reduction of usage of informal financial services such as membership of cooperatives or informal employment. Our paper contributes to the existing literature in two main ways. First, we examine the potential of mobile money to enhance formal financial access as a channel that brings out individuals from informal to formal deposit mechanisms. Specifically, we test the capacity of mobile money to increase the likelihood of individuals participating in informal financial mechanisms to use formal financial instruments to make deposits, bank and credit union accounts more precisely. Second, it is well documented that the unbanked are more likely to be individuals with low and irregular incomes, those who live in rural area far from formal financial institutions, and socially excluded like female and less educated. As these categories of individuals are more likely to participate in informal deposit mechanisms, we analyze whether mobile money increases their likelihood to use formal deposit instruments. Hence, we test empirically the effect of the use of mobile money on their likelihood to make deposit in formal institutions as well as bank account and credit union account. Deposits or savings can help smooth low and irregular income patterns and meet individuals spending objectives such as school fees and health expenses. For people who make deposits through informal methods, money security can become a challenge and the need for the safety offered by formal financial institutions becomes stronger. As mobile money provides individuals with an electronic account that allows them to free deposit money as savings, it can be seen as a stable springboard from which to begin the path to formal financial inclusion.

5

The paper is organized as follows. In section 2 we present the research framework. Section 3 provides background on mobile money adoption and financial access in Burkina Faso, and also describes our data collection and summary statistics. Section 4 displays our methodology and this with the econometric results in section 5. In section 6 we discuss the potential mechanisms that support our results and we conclude in section 7.

2. Research questions According to the existent literature, mobile money can foster financial inclusion by improving population access to basic formal financial services. The changes may not only stem from the way individuals conduct their financial transactions such as remittances but also in their choices of deposit methods. Nevertheless, these effects may also depend on individuals’ socioeconomic characteristics (level and type of income, location, gender, and education level) and their perception of the myriad financial instruments available.

2.1.The technology of mobile money and usage of formal and informal saving instruments People save or make deposits for different reasons such as insurance against emergencies, investment, social obligation and derive ingenious, often costly, saving mechanisms. Individuals are very strategic when cultivating their savings portfolios as a wide range of deposit instruments ranked from informal, semi-formal to formal financial instruments are available to them (Carpenter and Jensen, 2002; Morawczynski, 2009; Robinson and Wright, 2001). The new technology of mobile money that is considered as a new channel to provide financial services may play a key role in the choice of deposit instruments that people made. In fact, Porteous (2006) distinguishes two ways through which mobile money affects the usage of existent financial services called the additive model indicating the usage of mobile phone as another channel to access an existing formal account and the transformative model that entails the use of mobile phone by the excluded from the formal financial system to access formal financial products (Demombynes and Thegeya, 2012; Porteous, 2006). It is important to understand and

6

identify the behavior of individuals toward different deposit instruments in order to formulate policies for enhancing formal financial inclusion and reduce poverty.

2.1.1. Additive model of mobile money (Complementarity) A strong formal financial system is essential to provide individuals with deposit instruments that allow them to smooth income and consumption over time and make efficient investments in health, education and business. Formal financial services refer generally to some basic services including lending facilities, savings facilities, payment and remittances services as it represent an important source of income for many individuals in developing countries. The institutions involving in formal financial sector differ from countries and consist of banks, post banks, credit union, and insurance companies and are subject to laws, regulations and prudential supervision (Demirgüç-kunt and Levine 2008; De Koker and Jentzsch 2013; Pande et al. 2012). The agreement between formal financial institutions and their customers are typically governed by formal written contracts, often in the form of standard agreements and are, at least theoretically, enforceable in court. The additive model of mobile money that refers to the access to an existent formal financial account especially bank account via mobile phone beyond basic money storage and transfers imply that individuals may use mobile money in addition to traditional financial services. It is well known that individuals with high and regular income, located in urban area where banks are concentrated, male and high educated individuals are more likely to access formal institutions compared to disadvantaged people with low and irregular income, located in rural area, female and less educated (Allan, Massu, and Svarer 2013; Demirgüç-Kunt, Klapper, and Singer 2013; Karlan, Ratan, and Zinman 2014; Ky, Rugemintwari, and Sauviat 2015; Morawczynski 2009; Triki and Faye 2013). In our case, the model of mobile money allows already banked individuals to make some transactions between their mobile money account and their banking account which may increase their likelihood to make deposit in formal financial institutions. In this context mobile money account appears to serve as a complement of formal deposit account especially for individuals with high and regular income, located in urban area, male and high educated individuals. However, this impact may lessen for deposit in credit union institutions, as there is no link between the mobile money account and credit union account. 7

Moreover, credit union institutions are used as an alternative to bank institutions for disadvantaged individuals or those in remote areas. Then, the use mobile money may appear as a complement of credit union institutions. Morawczynski (2009) in a fieldwork shows that some individuals may not use mobile money as a deposit/saving account because they already access and use other deposit mechanisms that meet their needs. In addition, some banked individuals find mobile money account not appropriate for big deposit/savings, and others may want to build a relationship with the bank institution to access credit in the future. Thus, it effects may be lesser or null on the behavior of individuals who already have access to formal deposit account. Several studies document that accessing formal financial instruments help individuals to make more deposits and securely, increased their productive investment, private expenditures and allow them to build a relationship with banks to access credit in the future (Dermish et al., 2012; Dupas and Robinson, 2013; Morawczynski, 2009; Shem, Misati and Njoroge, 2012). Therefore, increasing individuals’ access to mobile money may help individuals build strong resilience by providing them with more secure means of deposits and enhance their livelihoods strategies. The additive model can be extended to the use mobile money account in addition to informal saving mechanisms. In fact, several studies show that individuals use a combination of a variety of deposit mechanisms to manage their income and to meet their financial needs (Carpenter and Jensen 2002; Gash and Gray 2015; Kendall 2010; De Koker and Jentzsch 2013; Morawczynski 2009). The usage of mobile money as a deposit instrument seems to depend on the degree of commitment it provides. As informal deposit mechanisms are illiquid, individuals participating in informal financial mechanisms may additionally use mobile money that is more liquid and seems to be convenient for short term deposit and appropriate to face unpredictable life events. Therefore, one may consider mobile money as a complement of informal financial mechanisms that are illiquid and may be appropriate for long term deposit.

2.1.2. Transformative model of mobile money (Substitutability) Less than a quarter of adults in Sub-Saharan of Africa have access to formal financial services (International Finance Corporation, 2013). The frequent reasons cited for not using formal financial services include the lack of enough money, the fixed fees and high costs of opening and maintaining accounts, distance and insufficient documentation (Beck, Demirguc8

Kunt and Martinez Peria, 2008; Demirgüç-Kunt and Klapper, 2012; Honohan and Beck, 2007; Kendall, 2010; Mas, 2010). Thus, disadvantaged individuals who do not access formal financial services are more likely to be poor with unpredictable income, located far from formal financial institutions, female and with low level of education. In fact, poor people have access, if any, only to informal financial services to manage their finances. For instance, they stashes cash at home or under matrass, leaving money with a trusted neighbor, loaning funds to relatives or build assets as buying livestock, jewels. Some individuals pay deposit collectors to collect their deposits or deposit their money with local money-lenders. Others voluntarily form groups of saving such as rotating saving and credit associations (ROSCA) that meet at regular intervals and that allow members living or working near each other to lend their deposit to each other on a rotating basis and where each member makes sure that the other make deposit or contribute to the pot. Indeed, it is argued that even individuals who already access and use formal financial services continues to participate in informal mechanisms (De Koker and Jentzsch, 2013). However, these diversity of informal deposit mechanisms are very risky, illiquid and inappropriate. In the case of participating in ROSCA as the order of receipt is typically determined in advance, it is difficult to get deposits when more than one member has specific needs. The connection between depositors may renders the saving group vulnerable to shocks affecting local region or workers causing members default and threaten the solvency of the saving group. In addition, many rotating saving groups are functioning outside the scope of the formally regulated and supervised financial system. They involve local tradition as well as mutual trust that members reciprocally place in each other and the agreements underlying these mechanisms are generally verbal and in the case of breach of the agreement, enforcement is informal. In this situation, the transformative model of mobile money entails the usage of mobile money account rather than informal mechanisms because it appears to have the potential to solve these problems. Therefore, we assume that disadvantaged individuals may use mobile money as a substitute of informal deposit mechanisms because mobile money is personal, allows individuals to access a safe deposit without a required minimum balance, to perform financial transactions at relative low costs and more securely. By giving banks and financial services providers a cheap way to outsource cash handling, deposit and withdrawal transactions, mobile money can allow to serve individuals at lower costs,

9

securely and nearby them (Kendall et al., 2011). In fact, as the issuers of mobile money services are licensed banks, there is a strong link to the traditional financial system (Ramada-Sarasola, 2012). Thus, the mobile money is fully embedded within the traditional banking services sector although the mobile money account is managed by a third party; usually mobile network operators. As in developing countries the formal financial sector such as banks and credit unions remains underdeveloped and linked to a small network of branches, individuals may use mobile money account as a deposit instrument even though it is initially designed for transfers and payments and does not pay interest. In this vein, we suppose that mobile technology can also appear as a substitute of formal deposit instruments especially bank and credit union accounts for disadvantaged individuals. Individuals in less developed countries, who experience frequent economic need a secure and convenient way to accumulate sums of cash to invest in their micro-enterprises, pay for school fees, or finance major life and also need to hold “precautionary” deposits to ensure that unexpected shocks like health emergency or job loss do not push them deeper in poverty (Radcliffe and Voorhies, 2012). The rapid adoption of mobile money and its relative advantages compared to the informal and formal financial mechanisms raise the following questions that our paper aims to investigate:

1) Do individuals use mobile money account as a substitute or a complement of existing informal and formal deposit instruments? 2) Can mobile money increase access to formal financial deposit instruments (bank and/or credit union accounts)? To what extent excluded people may benefit from this innovation of mobile money technology?

10

3. Financial access strands and data collection 3.1.Background on mobile money, formal and informal savings in Burkina Faso Burkina Faso is one of the poorest countries in Sub-Saharan Africa, with a GDP per capita at around 761 USD and with 44.6 percent of the population living on less than $2/day international poverty line. Access to formal financial services, as in most low-income countries in developing countries, remains limited. In fact, formal, semi-formal and informal financial mechanisms coexist in the country. While growing, the formal and semi-formal5 system remains largely dominated by the informal sector as most of the population access financial services from it (Gash and Gray, 2015). The formal financial sector is still in its infancy and comprised of 13 commercial banks, and 4 financial society including insurance, lending and leasing institutions. The network of bank branches consists of around 244 branches and 305 ATM (BCEAO 2014) are concentrated urban areas and more interested by individuals with high and regular income than poor individuals as it is costly to collect their small and irregular income through physical infrastructure (Dermish et al. 2012). In this context, microfinance institutions or decentralized financial system (DFS) including credit union, post office and cooperatives play an important role in providing financial services such as saving accounts, loans, insurances and financial transactions including payments, pensions and money transfers. They dominate the banking sector in mobilizing individual savings and reaching

the excluded population from the banking sector especially small/medium enterprises and disadvantaged individuals with tools of deposit and facilities to access credit (Gash and Gray 2015; Nair and Kloeppinger-Todd 2007; Thieba 2013). There is around 64 decentralized financial system6 with 285 main agencies and 349 sub-agencies throughout the country (AP/SFD-BF, 2014). All these formal financial institutions (banks and decentralized financial institutions) are monitored and supervised by the Central Bank (BCEAO) and the Ministry of Economics and Finances through State Treasury. According to the Global Financial Inclusion Database7 (2015), while 51% the population report saved in the past years, around 13% have an account at a formal

5

Semi-formal or non-formal system includes informal savings groups that are supported by non-governmental organizations. 6 These data include RCPB (le Réseau des Caisses Populaires du Burkina Faso), a credit union that provides formal financial services that cover all the 45 provinces of the country with 185 counters in 2013. For more details, see the site web: www.rcpb.bf. 7 The data are collected for the year 2014 by the Global Financial Inclusion Database.

11

financial institutions. It also reports that 9% of the population save in a formal financial institutions compared to 18% that save using a savings club or a person outside the family. These reports illustrate the predominance of the informal savings mechanisms in the country. Financial exclusion can be defined as a process serving as an obstacle to a social category of the population to have access to formal financial services (Conroy 2005; Mohan 2006). It reflects the lack of access of disadvantaged people to appropriate services and formal financial products at low costs, safely and securely. The failure or inequality of access to formal financial services occurs in the country because the informal sector, although risky, remained the main channel through which the majority of population made deposits. The informal deposit/saving mechanisms consist of deposits at home, accumulate jewelry or livestock, pay deposit collectors, deposit money with local money-lender and/or participating in a rotating savings and credit associations (ROSCAs). The predominance of informal mechanisms in collecting deposits is an important indicator of the inefficiency of formal finance institutions to reach the financial needs of the population. Consideration such as gender, geographic isolation or low population density, documentation, and the costs and fees of formal financial services play an important role in explaining the low access in the country, Burkina Faso, with limited formal financial sector development. To promote financial inclusion in the countries members of WAEMU, then in Burkina Faso, the BCEAO cheer several initiatives on the issue of electronic money to take advantage of the opportunities of new technologies. These initiatives include internet banking, prepaid card, and in particular the mobile money to increase competitiveness in the banking sector. Mobile money have the advantage to lower costs of transactions, to be accessible anywhere throughout the country and benefit from a large network of mobile money agents for cash in/out services. Thus, the Central Bank allows the entrance in the banking system of new players such as issuers of electronic money and especially mobile network operators that can offer mobile money services in partnership with banks. Since the launch of mobile money, the takeoff remained modest around 5% of the adult population and can be explain by the idea of Mas and Porteous (2015) that the usage of new platforms as mobile money can accelerate dramatically but rapid takeoff may not be the norm and overcoming customer caution and resistance to change will take patience and experimentation. Hence, access to and usage of mobile money may help fill the gap in the access to more formal deposit instruments.

12

Table 1. Comparison of formal and informal mechanisms and mobile money (in Burkina Faso) Motivations or reasons

Formal Institutions (Banks and DFS)

Access

Literacy required. No one is excluded, no formal restrictions.

Risk

Formal deposit guarantee prohibited. Some episodes of failure remained in the mind of individuals

Transactions costs, fees

High costs of transactions. Withdrawals are charged according to the amount but deposits are free of charge

Liquidity

Aside long distance, funds may be withdrawn and deposited at agencies

Quality

Account is personal, deposits decision are made on individual basis. Can be used for predictable purposes and in a less manner for unpredictable events

Mobile Money

Informal mechanisms

No education level required, but knowledge on the use of mobile phone is necessary. No restrictions

Group exist everywhere, but are often stratified by gender, ethnic grouping, religion or social status. There may be screening, so groups do not take on risky members. For instance, those with irregular incomes might be excluded. No education level required.

Formal deposit guarantee prohibited. All the cash in exchange for electronic money is held by the bank. Sometimes network problem Relative small fees. Withdrawals, money transfers and payments, are charged (relatively low) but deposits are free of charge Funds may be withdrawn and deposited at the near mobile money agents or respective banks agencies/Constraint to agent cash or e-money shortage Account is personal, deposits decision are made on individual basis. Can be used for unpredictable purposes

Members are subject to disband without warning; they are also subject to covariant risks, then large shocks could cause default of the group. No fees. Contributions and receiving the pot are made on the basis of meeting). But, there is some high fees to move ahead in the queue. Typically, order of payout is fixed. It is difficult for individuals to move ahead in the queue (when more than one member face emergencies) Basis on social relationship, share problem with others. Development of community-level solidarity, Social pressure to make regular deposits. Can be used for predictable purpose and in a lesser manner for unpredictable events

Source: Authors’ analysis. DFS refers to Decentralized Financial System or Microfinance institutions.

3.2.Data collection and summary statistics Data collection We apply our hand-collected data from our survey8 that we designed and conducted in May 2014 that consist of 500 randomly selected individuals across the central region in Burkina Faso. The target population include 50% of users and 50% of non-users of mobile money and allow us to capture the impact of mobile money on the choice of deposit/saving instrument made by users compared to non-users. 8

For more details about the survey and data collection, see (Ky, Rugemintwari and Sauviat, 2015)

13

Summary statistics In our sample, descriptive analysis (Table 2) of the choice of savings instruments9 reveals that among individuals that report saved, 46% using credit union, 44% participate in informal saving mechanisms, 39% used bank account and 17% used mobile money account. Regarding individuals that report saved, there is about 49% of female and 51% of male, 52% of rural and 48% of urban, 43% of less educated and 57% of high educated, and 46% of low income and 54% of high income. Considering the motivations of savings, the report shows that individuals saved first for health emergencies with 82%, follow by 49% for developing an activity, 46% for a decrease of income, 33% for education expenditures, 24% for ceremonies (wedding, funerals) and 14% to reimburse a loan. Among individuals that participate in informal mechanisms, about 66% are female, 61% are located in urban area and 83% have a paid activity. Looking at, respectively, individuals saving with bank and those saving with credit union, 42% against 52% are female, 43% against 53% are rural and 82% against 87% have paid activity. When saving for a specific purpose, for instance health emergencies, 97% of individuals used mobile money, 94% used banks, 91% use credit union and 87% used informal methods. This is coherent with the fact that individuals combine a variety of saving instruments to meet their financial objectives. Regarding the sample of individuals that saved using the mobile money, the data shows that about 49% are female, 56% are married, 27% have low education level, 41% are located in rural area and 61% have an income more than 50 000 F CFA. The majority of them, around 63% also used bank to saved, 51% used in informal savings methods and 47% used credit union account to save. The mobile money account is also used to save for all the cited motivations with 97% for health emergencies and 7% to reimburse a loan.

9

We also report in Table A.8. in Appendix, descriptive statistics on the choice of deposit instruments.

14

Table 2. Data sample characteristics.

Full sample Gender Female Male Marital situation Married Single Person in charge Age < 30 >= 30 Education level Less than secondary education level At least secondary education level Living place Rural Urban Occupation / employment status Paid activity Unpaid activity (include student) Income level and type Income ranging from 10 000 to 50 000 FCFA Income more than 50 000 FCFA Irregular income Regular income

Sample of individuals that report with with with Informal with Bank Credit Mobile mechanisms Union Money

Full sample

Mobile money user

Saved

/

50,37%

91,60%

39,01%

46,91%

17,28%

44,69%

49,38% 50,62%

49,02% 50,98%

48,79% 51,21%

41,77% 58,23%

51,58% 48,42%

48,57% 51,43%

65,75% 34,25%

48,40% 50,86% 52,10%

54,90% 45,10% 52,45%

50,67% 48,79% 54,18%

57,59% 42,41% 65,82%

53,68% 45,79% 62,63%

55,71% 44,29% 54,29%

40,88% 58,56% 42,54%

50,62% 49,14%

48,53% 51,47%

48,25% 51,48%

39,87% 59,49%

46,84% 52,63%

48,57% 51,43%

59,67% 39,78%

41,73%

36,27%

42,59%

19,62%

41,58%

27,14%

50,83%

57,53%

63,73%

56,60%

79,11%

57,37%

72,86%

49,17%

52,10% 47,90%

59,31% 40,69%

52,02% 47,98%

43,04% 56,96%

52,63% 47,37%

41,43% 58,57%

38,67% 61,33%

80,99% 15,56%

77,45% 18,14%

83,56% 16,17%

82,28% 16,46%

87,37% 11,58%

71,43% 20,00%

82,87% 12,71%

48,64%

43,63%

46,09%

27,22%

39,47%

35,71%

62,43%

51,36% 47,90% 51,60%

56,37% 50,00% 49,51%

53,91% 53,10% 46,36%

72,78% 31,01% 68,35%

60,53% 56,84% 43,16%

64,29% 40,00% 58,57%

37,57% 49,72% 49,72%

Usage of mobile technology Mobile phone user 99,26% 99,02% 99,46% 99,37% 100,00% 50,37% 49,60% 57,59% 52,11% MM user MM non user 49,63% 50,40% 42,41% 47,89% Usage of saving instruments Formal 85,93% 93,14% 93,80% 39,01% 44,61% 42,59% 39,47% Bank Credit Union 46,91% 48,53% 51,21% 47,47% 17,28% 34,31% 18,87% 27,85% 17,37% MM Informal 44,69% 38,24% 48,79% 34,18% 49,47% Motivations Education 33,33% 35,29% 36,39% 50,00% 31,58% 49,14% 44,61% 53,37% 43,04% 61,05% Develop activity Repay loan 14,07% 10,29% 15,36% 10,13% 22,63% 46,17% 45,10% 50,40% 62,03% 53,16% Decrease of income Health emergencies 81,98% 85,29% 89,22% 93,67% 90,53% 24,20% 23,53% 26,42% 37,97% 31,05% Ceremonies (wedding, funerals) Source: Authors’ analysis of the survey data collected in May 2014 in Burkina Faso. Throughout, F CFA (Franc of the Community) refers to the local currency. The exchange rate during the survey period was about 500 F CFA = $1 US.

99,45% 43,09% 56,91%

62,86% 47,14%

29,83% 51,93% 19,89%

51,43% 37,14% 21,55% 40,00% 59,12% 7,14% 18,78% 78,57% 54,70% 97,14% 87,29% 42,86% 35,36% African Financial

15

4. Model specification We examine the impact of mobile money usage on individuals choice of saving instruments by using a logistic model specify as follow: PROByi  1  1   2 MMuseri   3 X i  where



(1)

is the cumulative distribution function of logistic distribution.

In the model (1), yi stands for our dependent variable that characterizes individuals’ choice of deposit instruments to capture the effective saving instruments used. It is dummy variable that alternatively stands for: deposit using formal instrument, deposit using bank account, deposit using credit union account, deposit using mobile money account and deposit through informal mechanisms especially participating in a ROSCA. These dependent variables, except deposit using formal instrument, are measured using the following questions asked to respondents that saving: “During the past 12 months, did you make deposit using bank account?”; “using credit union account?”; “using mobile money account?”; “participating in ROSCA?” All these variables are dummies and each variable takes the value one if respondent reports YES, and zero otherwise. MMuseri is the independent variable of interest that stands for the use of mobile money. It is a binary variable that takes the value one if the respondent indicates using mobile money, and zero otherwise. X i is a vector of control variables (age, marital situation, location, gender, occupation or employment status, person in charge, education level, level and type of income). In Table 3 we report definitions of variables along with some summary statistics. If both users and non-users of mobile money do not differ in terms of their likelihood to make deposit using informal and formal (bank and credit union) financial instruments, the coefficient  2 should not be statistically different from zero. If mobile money users have more likelihood to make deposit using informal of formal financial instruments than non-users, the coefficient  2 should be positive and statistically different from zero implying that mobile money account act as a complement of informal or formal deposit instruments. If mobile money users have less likelihood to make deposit using informal of formal financial instruments than non-users, then the coefficient  2 should be negative and significantly different from zero reflecting that mobile money account act as a substitute of informal or formal financial instrument used for money deposit. 16

Table 3. Definition of variables. Variable Mobile money user (MM user) Deposit in formal institutions

Definition Reply to the question: Do you use mobile money services? Encoded as yes = 1, no = 0 Indicate when respondents deposit their money using formal institutions, encoded as (save using banks and/or credit unions) = 1, others = 0

Obs.

Mean

405

0.5

402

0.7

Deposit in bank account

Reply to the question: During the past 12 months, did you deposit your money using bank account? Encoded as yes = 1, no = 0

402

0.4

Deposit in credit union account

Reply to the question: During the past 12 months, did you deposit your money using credit union account? Encoded as yes = 1, no = 0

402

0.5

401

0.4

176

0.9

190

0.9

404

30.55

405

0.51

405

0.48

401

0.53

402

2.67

405

0.52

391

0.84

Income

Indicate the monthly income of respondent, encoded as Less than 10 000 FCFA = 1, 10 000 to 50 000 FCFA = 2, 50 000 to 150 000 FCFA = 3, 150 000 to 300 000 FCFA = 4, 300 000 to 500 000 FCFA = 5, More than 500 000 FCFA = 6

405

2.61

Irregular income

Indicate the type of income of respondent, encoded as Irregular = 1, Regular = 0

403

0.48

Deposit in mobile money account Deposit using informal mechanisms Participating in informal mechanisms

Reply to the question: During the past 12 months, did you deposit your money using your mobile money account? Encoded as yes = 1, no = 0 Reply to the question: During the past 12 months, did you deposit your money using informal mechanisms (participating in a ROSCA)? Encoded as yes = 1, no = 0 Indicate when respondents report participating in informal mechanisms, encoded as (savings group, saving at home and/or with a neighbor) = 1, others = 0

Individuals characteristics Age Male Married Person in charge Education Rural Occupation

Indicate the age of respondent Indicate the gender of respondent, Encoded as Male = 1, Female = 0 Indicate the marital situation of respondent, Encoded as Married = 1, Single = 0 Indicate if the respondent has or not dependent, Encoded as Having dependent = 1, otherwise = 0 Indicate the education level of respondent, Encoded as Illiterate = 1, Primary = 2, Secondary = 3, University =4 Indicate the location of respondent, Encoded as Rural = 1, Urban = 0 Indicate the employment status of respondent, Encoded as (Employed, Entrepreneur, Merchant, Farmer) = 1, (Unemployed, Student) = 0

Source: Throughout, F CFA (Franc of the African Financial Community) refers to the local currency. The exchange rate during the survey period was about 500 F CFA = $1 US.

17

5. Results This section reports our main results. Table 4 presents estimates of the impact of mobile money on individuals’ choice of deposit instruments. We analyze the likelihood of mobile money user compared to non-user of mobile money to make deposit using formal and/or informal financial mechanisms. Specifically, we test whether mobile money increases the likelihood of individuals to make deposit using formal institutions such as banks and credit unions than informal deposit mechanisms compared to non-users of mobile money. Across the columns, we find the coefficients of the variable of interest MM user statistically significant in columns 4 and 5. But in columns 1 to 3 this coefficient is not significantly different from zero. According to the results, the use of mobile money increases the likelihood of individuals to make deposit in mobile money account but it decreases their likelihood to participate in informal deposit mechanisms respectively by (175.3%)10 and (0.30%). However, we find no difference in the likelihood of mobile money user compared to mobile money non-user to make deposit in formal institutions such as banks and credit unions. Our results are consistent with the findings of Mbiti and Weil (2011) and Morawczynski and Pickens (2009) that mobile money acts as a substitute of informal deposit mechanisms. Looking at the effect of socioeconomic characteristics on individuals’ choice of deposit instruments, we distinguished formal financial institutions including banks, credit unions and mobile money from informal saving mechanisms as well as ROSCA. Regarding deposit using formal institutions (columns 1 to 4), we find the coefficient of married positive and significant (columns 1, 3 and 4) implying that a household of more than one individual have more likelihood to make deposit using formal instruments especially credit unions and mobile money accounts than single individual. This result may reflect the fact that households need better insurance and security of their money according to household expenditure (such as healthcare, education) leading them to make deposit in formal institutions. The reported results also show the coefficient of irregular income negative and significant in column 2 but positive and significant in column 3 and 4. Thus, irregular income individuals have less likelihood to make deposit in banks but have more likelihood to make deposit in credit union and mobile money account than regular income individuals. In fact people with irregular income are less likely to integrate informal saving groups. This may be due to the unpredictable income and the high cost of bank services that dissuade individuals to use them. 10

The coefficients reported in all our tables are the log odds of the use of mobile money on the choice of savings instruments. To obtain the odds ratio, we simply compute the exponential of log odds.

18

Table 4. Choice of deposit instruments and mobile money adoption

MM user Age Age squared Married Rural Male Occupation Irregular income Person in charge Education Income Income squared Constant

Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients)

Likelihood ratio test (H0: nullity of coefficients)

Deposit in formal institutions (1) -0.165 (0.283) -0.038 (0.154) 0.001 (0.002) 0.901** (0.374) 0.268 (0.303) -0.054 (0.274) 0.157 (0.461) -0.276 (0.307) 0.299 (0.263) 0.840*** (0.190) 2.921*** (0.704) -0.341*** (0.112) -6.504*** (2.430)

Deposit in bank account (2) 0.374 (0.298) 0.115 (0.159) -0.001 (0.002) 0.293 (0.396) -0.450 (0.318) -0.097 (0.280) -0.205 (0.508) -0.960*** (0.302) 0.338 (0.274) 1.141*** (0.233) 0.415 (1.201) 0.144 (0.225) -8.226*** (2.747)

379 0.1975

379 0.3359

63.97*** 128.93***

Full sample Deposit in credit union account (3) -0.111 (0.236) -0.094 (0.129) 0.001 (0.002) 0.587** (0.284) -0.149 (0.238) -0.180 (0.227) -0.032 (0.407) 0.756*** (0.250) 0.305 (0.215) 0.123 (0.145) 2.101** (1.018) -0.310* (0.183) -2.076 (2.243)

Deposit in mobile money account (4) 5.167*** (0.610) -0.034 (0.218) -0.000 (0.003) 1.270*** (0.425) 0.246 (0.389) 0.661* (0.367) -1.224** (0.593) 0.731* (0.382) 0.132 (0.356) 0.475* (0.252) 1.790 (1.123) -0.273 (0.190) -6.997** (3.249)

Deposit in savings groups or at home (5) -1.268* (0.679) -0.294 (0.826) 0.003 (0.013) 0.188 (1.197) -2.548* (1.497) 0.176 (0.948) 2.989 (2.116) -0.989 (1.065) 0.689 (0.574) -0.440 (0.581) -1.871 (3.115) 0.178 (0.399) 13.813 (15.034)

379 0.0488

378 0.5893

169 0.2759

77.95***

23.79**

114.07***

51.31***

200.71***

57.56***

331.92***

16.66*

% correct prediction (y=1) 68.28% 66.45% 58.85% 94.16% 93.21% % correct prediction (y=0) 76.58% 86.61% 62.03% 84.38% 57.14% Note: Dependent variables: deposit using formal financial institutions, deposit using bank account, deposit using credit union account, deposit using mobile money account and deposit using informal mechanisms are all dummies. Deposit using formal financial institution equals 1 if respondents make deposit using bank account and/or credit union account, and 0 otherwise. Deposit using bank account equals to 1 if respondents make deposit using bank account, and 0 otherwise. Deposit using credit union account equals to 1 if respondents make deposit using credit union account, and 0 otherwise. Deposit using mobile money account equals to 1 if respondents make deposit using mobile money account, and 0 otherwise. Deposit using informal mechanisms equals to 1 if respondents make deposit using informal mechanisms, and 0 otherwise. The variable of interest, MM user is also a dummy that equal to 1 if respondents use mobile money, and 0 otherwise. The coefficients reported in the table are the log odds of the use of mobile money on the choice of savings instruments. To obtain the odds ratio, we simply compute the exponential of log odds. Robust standard errors in brackets. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

19

The variable income is entered in nonlinear form (income and income squared) to highlight how an increase in the income affect individuals choice of saving instruments. Our results show the coefficient of income positive and significant in columns 1 and 3. Then, individuals with high income have more likelihood to make deposit using formal institutions, while their have less likelihood to participate in informal deposit mechanisms than low income individuals. Moreover, greater increase of income decreases the likelihood to use formal deposit instruments indicating that an increase in the income after a certain threshold leads individuals to diversify their deposits in other instruments to lessen the risk related to each type of deposit instruments. Some explanations of these results may stem from the high costs of opening formal account and fees of account maintaining that discourage poor individuals to access formal financial services (Allan, Massu and Svarer, 2013; Beck, Demirgüç-Kunt and Honohan, 2009; Kendall, 2010). The level of education shows difference in the choice of deposit instruments. We show the coefficient positive and significant in columns 1, 2 and 3, indicating that greater level of education increases the likelihood of individuals to make deposit using formal institutions especially bank and mobile money accounts. In column 4, we find that the coefficient of male positive and significant implying that male have more likelihood to make deposit in the mobile money account than female. The results also report the coefficient of occupation negative and significant indicating that unemployed and student have more likelihood to make deposit in the mobile money account than individuals with paid activity. Turning now to informal deposit mechanisms (column 5), we find that rural has negative and significant impact on participating in informal deposit mechanisms. This result indicates that individuals located in rural areas have less likelihood to participate in informal deposit mechanisms than those located in urban area. This result confirms the findings of Carpenter and Jensen (2002) who argue that urban areas facilitate the formation of informal deposit mechanism such as ROSCA, and that rural individuals are disadvantaged in access to even informal finance methods. One of the most important questions about mobile money is to understand to what extent it can improve formal financial access. Although mobile money reduces the likelihood to make deposit through informal mechanisms, there is no evidence that it may bring out informal finance participants toward formal finance.

20

To highlight this issue, we investigate whether the use of mobile money increases the likelihood of individuals who participate in informal deposit mechanisms to make deposit in formal financial institutions such as banks and credit unions. We use the following specification as almost all respondents report participating in informal financial mechanisms:

PROBy i  1  ( 1   2 MMuseri   3 MMuseri  PIM i   4 PIM i  X i ) where



(2)

is the cumulative distribution function of logistic distribution.

In the model (2), yi stands for our three dependent variables that characterize individuals’ choice of deposit instruments to capture the effective deposit instruments used. It is a dummy variable that alternatively stands for: deposit using formal instrument, deposit using bank account and deposit using credit union account. MMuseri represents the use of mobile money. PIM i is a binary variable that stands for participating in informal financial mechanisms. It is a dummy variable that equals to one when the individual participates in informal financial mechanisms (savings groups, saving at home and/or with a neighbor), and zero otherwise. We add the interaction between mobile money use and the participation in informal deposit mechanisms. X i is the same vector of control variables used in equation (1) that we also interact with the participation in informal deposit instrument. The coefficient of interest is given by the total effect of being mobile money user and participating in informal deposit mechanisms (  2   3 ). The coefficient  2 give the effect of being mobile money user and not participating in informal deposit mechanisms. Table 5 presents the results. We show the coefficient of variable of interest while not significant in column 2 and 6, this coefficient is positive and significant in column 4. Thus, mobile money increases the likelihood of individuals participating in informal deposit mechanisms to make deposit using bank account by (4.32%). These results suggest therefore that mobile money may improve formal financial access and acts as a channel through which individuals can reduce their use of informal financial services. Moreover, the results also show the mobile money user positive and significant in table 1 and 3. Reflecting that the use of mobile money increases the likelihood of individuals who do not participating in informal mechanisms to make deposit in formal institution especially in bank account. While participating in informal mechanisms lessen the effect of mobile money on the likelihood of individuals to access and/or use formal financial services, but on the whole mobile money facilitates the transition from informal mechanisms toward formal financial services. 21

Table 5. Choice of deposit instruments and mobile money adoption: individuals participating in informal financial mechanisms

MM user MM user x Participate in informal mechanisms Participate in informal mechanisms x Age Age squared Married Rural Male Occupation Irregular income Person in charge Education Income Income squared Constant

Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients)

Individuals participating in informal financial mechanisms Deposit in credit union Deposit in formal institutions Deposit in bank account account Total effect Total effect Total effect (1) (2) (3) (4) (5) (6) 15.060*** 21.865*** -0.401 (1.490) (5.388) (1.654) -15.332***

-0.272

-20.402***

1.463***

0.201

-0.2

(1.552)

(0..421)

(5.298)

(0.546)

(1.704)

(0.399)

-0.447* (0.240) 0.007* (0.004) 0.534 (0.532) 0.098 (0.545) -0.316 (0.454) -0.617 (0.841) 1.025** (0.472) 0.539 (0.404) 1.303*** (0.302) 1.252 (2.037) -0.035 (0.388) 0.823 (1.448)

0.477 (0.330) -0.007 (0.005) -0.039 (0.821) -2.183** (0.928) -0.068 (0.642) 0.064 (0.855) -0.897 (0.578) 0.472 (0.543) 2.014*** (0.515) 4.746** (2.155) -0.660* (0.343) -21.354*** (5.338)

-0.454** (0.189) 0.008** (0.003) 0.010 (0.471) -0.285 (0.506) -0.781* (0.427) -1.006 (0.757) 1.432*** (0.456) 0.519 (0.384) 0.822*** (0.254) 2.537* (1.539) -0.346 (0.259) 0.912 (1.483)

178 0.2591 1582.05***

178 0.4823 39.36***

178 0.1721 40.01***

324.47***

424.93***

355.65***

% correct prediction (y=1) 64.04% 75.86% 80.20% % correct prediction (y=0) 78.13% 90.00% 55.84% Note: Dependent variables: deposit using formal financial institutions, deposit using bank account, deposit using credit union account are all dummies. Deposit using formal financial institution equals 1 if respondents make deposit using bank account and/or credit union account, and 0 otherwise. Deposit using bank account equals to 1 if respondents make deposit using bank account, and 0 otherwise. Deposit using credit union account equals to 1 if respondents make deposit using credit union account, and 0 otherwise. The variable of interest, MM user is also a dummy that equal to 1 if respondents use mobile money, and 0 otherwise. The coefficients reported in the table are the log odds of the use of mobile money on the choice of savings instruments. To obtain the odds ratio, we simply compute the exponential of log odds. .Robust standard errors in brackets. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

22

According to the remaining explanatory variables, the results show the coefficient of rural negative and significant only in column 3. Indicating that individuals located in rural area and participating in informal deposit mechanisms have less likelihood to make deposit in banks than those located in urban area. We also find the coefficient of male negative and statistically significant only in column 5 implying that male participating in informal deposit mechanisms have less likelihood to make deposit in credit unions than female. In columns 1 and 5, the coefficient of variable age is negative and significant but the coefficient of variable age squared is negative and significant. Then, older individuals have more likelihood to make deposit in formal institutions especially credit unions than least aged individuals. The results for variables irregular income, education and income are similar to that find in Table 4. Our results show the coefficient of irregular income positive and significant (columns 1 and 5). Then, irregular income individuals participating in informal deposit mechanisms have more likelihood to make deposit in formal institutions especially in credit unions than those with regular income. Across columns, the coefficient of education is positive and significant reflecting that increasing education level, increases the likelihood of individuals participating in informal deposit mechanisms to make deposit in formal institutions such as banks and credit unions. Similarly, the coefficients of income is positive and significant in columns 3 and 5, while the coefficient of income squared is negative and significant only in column 3. Thus increasing the level of income leads to an increase in the likelihood of individuals who use informal mechanisms to make deposit in formal institutions such as banks and credit unions. While greater increase in the income reduces the likelihood to make deposit in formal institutions. Overall, our results show that mobile money acts as a substitute of informal deposit mechanisms, while we find no evidence concerning the relation between the use of mobile money and formal deposit instruments. However, we find that mobile money increases the likelihood of individuals participating in informal mechanisms to make deposit in banks. Therefore, mobile money may contribute to improve formal financial access by bringing individuals using informal finance toward formal finance. Nonetheless, one may question the effects of mobile money on the barriers to formal financial access such as the low and unpredictable income of poor people, the remoteness or lack of formal financial infrastructures (rural/urban), gender discrimination (female/male) and the lack of financial literacy (less educated/high educated).

23

With regard to this question, we test whether mobile money can bringing down barriers to formal financial access by looking at its effects on formal deposit instruments by distinguishing low and high income individuals, irregular and regular income, rural and urban, female and male, and less and high educated individuals. We use the following specification in equation (3) to restrict our sample on sub-samples of low and irregular income, rural, female and less educated individuals. PROB  yi  1   1   2 MMuseri   3 Subsamplei   4 MMuseri  Subsamplei   5 Subsamplei  X i   6 X i 

(3) where  is the cumulative distribution function of logistic distribution.

In Table 6 we report estimates of the effect of mobile money on the choice of deposit instruments for individuals with different level and type of income. The coefficients of interest are  2 and the total effect given by  2 +  4 . Considering the level of income, we show the coefficient of low income individuals negative and significant in columns 2, while the coefficient of high income individuals is positive and significant in columns 1 and 3. These results indicate that for low income individuals mobile money decreases their likelihood to save in formal institutions by 0.46%, then mobile money acts as a substitute of formal deposit instruments. While mobile money increases the likelihood of high income individuals to make deposit in formal institutions by 2.96%. Considering the type of income, we find the coefficient of individuals with irregular income significant in column 4. Implying that the use of mobile money increases the likelihood of irregular income individuals to make deposit in bank account by 3.04%. Our findings suggest that mobile money appears to be a substitute of formal financial institutions for low income individuals, while for irregular income individuals mobile money acts as a complement of bank account. The results also indicate that for high income individuals mobile money acts as a complement of formal financial instruments. We turn now to the remaining set of individuals demographic and socio-economics characteristics that appear to be barriers to the access of formal financial deposit instruments. In Table 7, we present the effect of mobile money on individuals’ choices of deposit instruments on the basis of their location, gender and level of education. Considering individuals assumed to have access to formal institutions (urban, male and high educated), the results show that the use of mobile money increases the likelihood of urban to make deposit in bank account by 2.15%. These results imply that individuals located in urban area use mobile money in complement to bank account.

24

Table 6. Choice of deposit instruments and mobile money adoption: Low, irregular vs. High, regular income. Low vs. High income

MM user Low income MM user x Low income Controls Low income x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients)

% correct prediction (y=1) % correct prediction (y=0)

Deposit in formal institutions Total effect (1) (2) 1.084** (0.514) 11.017** (5.521) -1.866*** -0.782** (0.633) (0.368) YES YES

Deposit in bank account (3) 0.665 (0.409) -2.053 (5.770) -0.642 (0.618) YES YES

Total effect (4)

0.023 (0.464)

Deposit in credit union (5) 0.234 (0.354) 8.509* (4.989) -0.810 (0.509) YES YES

379 0.220 68.34***

379 0.326 96.51***

379 0.111 46.53***

139.01***

195.74***

90.09***

82.46% 66.67%

81.29% 75.89%

61.98% 72.73%

Total effect (6)

-0.576 (0.366)

Irregular vs. Regular income

MM user Irregular income MM user x Irregular income Controls Irregular income x Controls

Deposit in formal institutions Total effect (1) (2) -0.057 (0.465) -7.209 (6.559) -0.073 -0.130 (0.608) (0.393) YES YES

Deposit in bank account (3) -0.228 (0.433) 9.075 (6.822) 1.340* (0.699) YES YES

Total effect (4)

1.112** (0.548)

Deposit in credit union (5) 0.157 (0.333) -21.111*** (6.990) -0.354 (0.498) YES YES

Total effect (6)

-0.198 (0.371)

Observations 379 379 379 Pseudo R2 0.253 0.390 0.121 Wald χ2 (H0: nullity of coefficients) 71.07*** 108.54*** 45.18*** Likelihood ratio test χ2 (H0: nullity of 154.38*** 228.23*** 95.33*** coefficients) % correct prediction (y=1) 71.27% 70.97% 67.19% % correct prediction (y=0) 80.18% 87.50% 67.38% Note: Dependent variables: deposit using formal financial institutions, deposit using bank account, deposit using credit union account are all dummies. Deposit using formal financial institution equals 1 if respondents make deposit using bank account and/or credit union account, and 0 otherwise. Deposit using bank account equals to 1 if respondents make deposit using bank account, and 0 otherwise. Deposit using credit union account equals to 1 if respondents make deposit using credit union account, and 0 otherwise. With the interactions, the total effect is given by the sum of the coefficient of the interaction term plus the coefficient of the use of mobile money (MM user), and the p-value obtain through the Wald test is reported below. Low income individuals are those with less than 50 000 F CFA (around $100US) per month. Irregular income individuals are those who specify having irregular income by answering the following question: “Do you have regular or irregular income?” The responses are encoded as irregular income = 1, and regular income = 0. The coefficients reported in the table are the log odds of the use of mobile money on the choice of savings instruments. To obtain the odds ratio, we simply compute the exponential of log odds. Controls included: age, age squared, married, rural, male, occupation, irregular incomes, person in charge, education level, incomes level and incomes squared. According to the subsamples we remove respectively controls incomes level and incomes squared, and irregular incomes. Robust standard errors in brackets. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

25

By contrast, we find that the use of mobile money decreases the likelihood of high educated to make deposit in credit union account. Then, mobile money appears to be a complement of formal financial instrument especially bank account for urban, but a substitute of credit union account for high educated individuals. Regarding the excluded group, while mobile money have no effect on the likelihood of rural to make deposit in formal financial institutions, it increases the likelihood of female to make deposit using bank account by (3.58%). Similarly, the use of mobile money increases the likelihood of less educated individuals to make deposit in bank and credit union by respectively (2.78% and 2.07%). Thus, these findings suggest that mobile money acts as a complement of formal financial institutions such as banks for female, bank and credit union for less educated individuals. Overall, our results show that mobile money may help individuals overcome barriers to formal deposits instruments especially for formally excluded individuals by bringing them from informal finance to formal finance. Mobile money appears to be a substitute of formal deposit instruments for low income individuals, but acts as a complement of formal deposit instruments for high income individuals, urban and more interesting for female, irregular income and less educated individuals. Moreover, by reducing the use of informal deposit mechanisms, mobile money appears as a stepping stone toward formal financial institutions for poor people especially female, irregular income and less educated individuals that have less access to formal financial services.

26

Table 7. Choice of formal deposit instruments and mobile money adoption: Low vs. High access to formal finance

MM user Rural MM user x Rural Controls Rural x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients) % correct prediction (y=1) % correct prediction (y=0)

MM user Female MM user x Female Controls Female x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients) % correct prediction (y=1) % correct prediction (y=0)

Deposit in formal institutions Total effect (1) (2) -0.337 (0.401) -5.907 (6.652) 0.280 -0.056 (0.584) (0.424) YES YES 379 0.212 70.70*** 135.42*** 67.54% 79.28%

Rural vs. Urban Deposit in bank account Total effect (3) (4) 0.764* (0.402) 7.381 (6.883) -0.864 -0.100 (0.612) (0.461) YES YES 379 0.386 98.20*** 226.54*** 70.97% 87.05%

Deposit in credit union Total effect (5) (6) -0.260 (0.344) -7.703 (5.261) 0.268 0.008 (0.490) (0.348) YES YES 379 0.089 32.62* 78.69*** 67.19% 61.50%

Deposit in formal institutions Total effect (1) (2) -0.085 (0.478) 4.882 (5.023) -0.124 -0.210 (0.610) (0.379) YES YES 379 0.210 76.44*** 134.57*** 68.66% 77.48%

Female vs. Male Deposit in bank account Total effect (3) (4) -0.564 (0.427) 15.911*** (5.842) 1.840*** 1.276** (0.704) (0.559) YES YES 379 0.390 101.94*** 228.64*** 74.84% 84.82%

Deposit in credit union Total effect (5) (6) -0.233 (0.343) -6.709 (4.762) 0.258 0.025 (0.497) (0.359) YES YES 379 0.086 40.86** 77.04*** 65.63% 63.64%

Less vs. High educated Deposit in formal institutions Deposit in bank account Deposit in credit union Total effect Total effect Total effect (1) (2) (3) (4) (5) (6) MM user -0.612 0.574 -0.650** (0.440) (0.365) (0.323) Less educated 9.476* 11.630* 6.083 (5.331) (6.943) (5.155) MM user x Less educated 1.206** 0.594 0.448 1.021* 1.378*** 0.727** (0.585) (0.385) (0.674) (0.567) (0.492) (0.371) YES YES YES Controls Less educated x Controls YES YES YES Observations 382 382 382 Pseudo R2 0.224 0.339 0.104 Wald χ2 (H0: nullity of coefficients) 223.01*** 270.09*** 193.02*** Likelihood ratio test χ2 (H0: nullity of coefficients) 139.58*** 199.31*** 83.01*** % correct prediction (y=1) 69.00% 64.97% 69.59% % correct prediction (y=0) 80.18% 85.78% 63.30% Note: Dependent variables: deposit using formal financial institutions, deposit using bank account, deposit using credit union account are all dummies. Deposit using formal financial institution equals 1 if respondents make deposit using bank account and/or credit union account, and 0 otherwise. Deposit using bank account equals to 1 if respondents make deposit using bank account, and 0 otherwise. Deposit using credit union account equals to 1 if respondents make deposit using credit union account, and 0 otherwise. With the interactions, the total effect is given by the sum of the coefficient of the interaction term plus the coefficient of the use of mobile money (MM user), and the p-value obtain through the Wald test is reported below. Less educated individuals are those with primary education level or less (about six years of schooling at best). To obtain the odds ratio, we simply compute the exponential of log odds. Controls included: age, age squared, married, rural, male, occupation, irregular incomes, person in charge, education level, incomes level and incomes squared. According to the subsamples we remove respectively controls rural, male and education level. Robust standard errors in brackets. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

27

6. Discussion of the potential mechanisms Overall, we find that mobile money contributes to financial inclusion by reducing the use of informal deposit mechanisms but increasing the likelihood to make deposit using formal financial instruments. In this section, we address the mechanisms underlying the use of mobile money as a complement or a substitute of informal and formal deposit instruments. The prominent way by which mobile money improves the likelihood of individuals to make deposit using mobile money account is the advantage that it provides such as safety, competitive costs, ease of use, proximity, and costs and possibility of transfers. But other mechanisms could be at work especially the attributes related to others deposit instruments that may lead individuals to consider mobile money as a complement or substitute of informal and formal deposit mechanisms. We identify four attributes that are mainly cited as factors that lead individuals choice of deposit instruments between informal and formal methods (Allan, Massu, and Svarer 2013; Kendall 2010; Kendall et al. 2011). The first is the risk relying on both formal and informal finance. Informal deposit mechanisms (ROSCAs) are find to be quite risky and often incur losses as member can disband without warning, and proceed outside of the scope of the formally regulated and supervised financial services system. While, formal deposit instruments tend to be less risky than informal deposit instruments, they are not immune from risk, failure and fraud although well-regulated and supervised financial institutions (Kendall, 2010). Mobile money have the advantage to be secure as it involves banks (in partnership with mobile operators) that are formally regulated and supervised. Moreover, all financial transactions performed through mobile money account are traceable. In this situation, mobile money as well as formal deposit instruments can be preferred as a deposit account according to the risk associated with informal financial mechanisms. The second is costs of financial services that play a critical role in the choice of deposits instruments. Many formal institutions charge higher fees of transactions and account maintenance to their customer and the minimum balances require lead individuals to prefer informal deposit option that have the advantage to be at lower costs. However, informal deposit mechanisms as well as ROSCA, often charge high costs when member face emergencies and need to move ahead in the queue to access their deposit or savings. Thus, as mobile money offers competitive costs comparing to both informal and formal finance, this may motivate

28

individuals to use mobile money as a saving account. The third is associated to the level of liquidity that informal and formal finance provided. For both deposit instruments, liquidity refers to the accessibility or the rapidity with which individuals can access their deposits. In developing countries, accessing deposits made in a formal financial institution is not realistic according the long distance people have to travel to reach an agency and the time they would spend in long queue. Similarly, when participating in a rotating savings groups, members cannot access their deposits when needed according to preset order for the pot assignment. Mobile money account appear be appropriate to make deposits according the large network of mobile money agent throughout the country that insure the conversion of electronic money into cash (and vice versa) and allow people to access their deposits when the need arises. The fourth rely on the quality of financial services that is the reliability and privacy associate with informal and formal financial instruments. Informal deposit mechanisms (savings groups) are often unreliable as they may disband when needed most and some savings groups may break down when the local economy experience a downturn that may simultaneously default members (Kendall 2010). Thus, the reliability of savings groups entails the perception of each participant of the probability that these types of risk may occur. Privacy is a desirable quality for financial instrument that helps individuals resist the temptation to spend out deposits or savings to assist family or friends (Collins et al. 2009). However, savings groups based on social relationship used social pressure to motivate member to make regular deposits and often lead members to share problem with others. Moreover, when receiving the pot individuals may face many losses by lending the money to family and friends which often do not paid back. By contrast, formal deposit account as well as mobile money account are always reliable, safe and personal (as the account is password protected), and allow individuals to determine their own frequency to make deposits. In this way, it seems that quality may lead individuals to prefer mobile money to informal deposit mechanisms while it may not affect the choice between mobile money and formal financial instruments.

29

To provide evidence on the mechanisms describe above that is the attributes of formal and informal financial mechanisms that may motivate mobile money users to make deposit in the mobile money account, we consider the following specification: PROBDMM i  1  (1  2 Attributesi  3 Attributesi  MMuseri  4 MMuseri  X i )

(4)

where  is the cumulative distribution function of logistic distribution.

In equation (4), DMM i , deposit made using mobile money account, is the dependent variable. It is a binary variable that indicate the response of individual i to the following question “During the past 12 months, did you deposit your money in your mobile money account?” It is encoded as one if the individual i (mobile money user) response is YES, and encoded zero if the response is NO. X i represent the same set of control variables in equations above (age, marital situation, location, gender, occupation or employment status, person in charge, education level, level and type of income).

Attributesi is the independent variable of interest and consist of risk, cost, illiquidity and low quality related to formal, and informal financial mechanisms and that may influence the decision to make deposit in the mobile money account. In fact, we compare individuals perception of risk, cost, illiquidity and low quality associated with formal and informal financial mechanisms to those related to the mobile money. Thus, we compute our set of attributes as follow: Attributesi 

Attributesi Formal / Informal MM _ Attributesi

We distinguish attributes related to formal financial methods to those related to informal financial mechanisms. For instance, to obtain the attribute risk related to formal financial institutions compared to those of mobile money, we divide individual i perception of risk related to formal financial methods by her/his perception of risk related to mobile money account. All the variables are measured on 5-point Likert scale ranging from one (low) to five (high). Summary statistics are reported in Table A.2. in Appendix. Table 8 presents results of the impact of formal and informal finance attributes on the likelihood of individuals to make deposit in the mobile money account given by the coefficient of interest that is the total of effect of

 2  3 .

Considering factors related to formal financial

institutions, our results show the coefficients of interest associated with cost, illiquidity and low 30

Table 8. Factors affecting the choice of making deposit using mobile money account. Full sample Deposit using mobile money account Attributes of Formal/Mobile money

MM user x Attributes of informal/mobile money Attributes of informal/mobile money

Attributes of Informal/Mobile money

Risk

Cost

Illiquidity

Low quality

Risk

Cost

Illiquidity

Low quality

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

21.867***

12.818***

9.843**

-3.968

0.694

1.178

1.459**

238.482***

(8.217)

(3.821)

(4.045)

(3.126)

(0.427)

(1.213)

(0.666)

(52.746)

-21.615***

-8.684**

-8.235*

6.462** (Omitted)

(Omitted)

(Omitted)

-236.947***

(8.110)

(4.096)

(4.315)

(2.967)

Controls included

NO

NO

NO

YES

NO

NO

NO

YES

MM user x Controls included

YES

YES

YES

YES

YES

YES

YES

YES

Total effect

(52.744)

0.252

4.134***

1.608**

2.494**

0.694

1.178

1.459**

1.534**

(0.784)

(1.327)

(0.714)

(0.984)

(0.427)

(1.213)

(0.666)

(0.633)

Observations

146

146

146

341

73

72

72

347

Pseudo R2 Wald χ2 (H0:

0.134

0.322

0.210

0.6606

0.383

0.373

0.385

0.6870

256.27***

270.30***

256.83***

27.25***

26.78***

18.86*

426.02***

451.12***

436.18***

487.73***

487.29***

488.32***

nullity of coefficients)

Likelihood ratio test χ2 (H0: nullity

384.38***

394.18***

of coefficients)

% correct 95.87% 92.56% 94.21% 72.03% 88.46% 88.24% 86.27% 69.66% prediction (y=1) % correct 20.00% 60.00% 40.00% 94.44% 76.19% 76.19% 71.43% 94.55% prediction (y=0) Note: Dependent variables: deposit using mobile money account is a dummy that takes the value 1 if respondents make deposit using mobile money account, and 0 otherwise. Robust standard errors in brackets. In columns 4 and 8, we add the variable MM user used in the interaction terms in the controls variables. We use a logistic model specify in equation (3). Controls included: age, age squared, married, rural, male, occupation, irregular incomes, person in charge, education level, incomes level and incomes squared. Table A.2. in the Appendix gives definitions and summary statistics of the attributes of formal and informal financial mechanisms that consist of risk, cost, illiquidity and low quality. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

31

quality positive and significant (columns 2 to 4). Implying that individuals who perceive formal financial methods costly than mobile money, the use of mobile money increases their likelihood to make deposit in the mobile money account. Similarly, individuals who perceive formal financial methods illiquid and lower quality than mobile money, the use of mobile money increases their likelihood to make deposit in the mobile money account. Our results suggest that individuals may prefer to use mobile money account for deposit than formal financial institutions due to the relative low fees of transactions and the convenience that mobile money provides. Regarding attributes related to informal financial mechanisms, we find the coefficients of interest associated with illiquidity and low quality positive and significant. Reflecting that individuals who perceive informal financial methods illiquid and of lower quality than mobile money, the use of mobile money increases their likelihood to make deposit in the mobile money account. These findings support the fact that making deposits using informal mechanisms such as savings groups makes difficult the access to money when a need arises. Similarly, saving at home makes difficult to resist the temptation to spend out deposits on unnecessary items or to excessively assist family and friends. Thus, individuals may prefer mobile money account that is personal, allows avoiding unneeded expenditures and facilitate access to deposit when the need arises.

7. Endogeneity issue In this section we discuss the potential endogeneity problem. In order to confirm that the observed differences in the choice of deposit instrument between mobile money users and nonusers and among subsamples (individuals participating in informal mechanisms, low/high income, irregular/regular income, rural/urban, female/male, less/high educated) are genuinely due to the use of mobile money, we replicate the estimations reported in section 5 and 6 using standard instrumental variables methods. In fact, we have assumed that the effect of the use of mobile money on individuals’ choice of deposit instruments to be independent given the control variables included in the regressions. Therefore, the estimated coefficients are valid only if the use of mobile money is not correlated with the error term conditional on the control variables. Furthermore, making deposit in any of 32

financial instruments may influence the decision of individuals to use mobile money as cheaper and convenient deposit instrument. Also, banked individuals may use mobile money to easily make deposit in their bank account. To address this potential endogeneity issue resulting from simultaneous effects, we resort to instrumental variable estimation of choice of deposit instruments. Following Jack and Suri (2014) who consider the distance to the closest agent, we use the same excluded instrument that we measure through the following question “What distance did you travel to reach a mobile money agent?” The response are encoded on 5-point Likert scale, 1 (less than 1 km), 2 (1 to 2 km), 3 (2 to 5 km), 4 (5 to 10 km) and 5 (10 km and more). We also use the capacity to use cell phone to perform monetary transactions as a second excluded instrument. This variable is measured using the following question “How do you rate your ability to use a cell phone to perform monetary transaction?” And the responses are encoded using 5-point Likert scale ranging from 1 (very low) to 5 (very high). The correlation matrix reveals a relatively low correlation between these two instruments (   0.37 ). Table 9 presents results of the reduced form for predicting the use of mobile money. As expected, the coefficient the distance to the nearest agent is negative and significant. Implying that the further the mobile money agents, the harder it may be for individuals to access and use mobile money services. So, individuals have more likelihood to use mobile money if the distance from the nearest retail agent is relatively shorter. Similarly, we find the coefficient of capacity to use cell phone to perform monetary transactions positive and significant. This finding suggests that more able individuals can use cell phone to perform monetary transactions, more likely they may adopt/use mobile money services. We report Chi-square Wald test for the weakness of the instruments and Chi-square Sargan test for the exogeneity of instruments (and each corresponding p-value). For the Wald test, the statistic is 23.45 and significant at 1% allowing us to confirm that our instruments are not weak. The Sargan test do not reject the null hypothesis for the exogeneity of instruments with statistic of 0.6 and not statistically significant.

33

Table 9. Reduced form analysis of the use of mobile money

Distance to the nearest agent Capacity to use a phone to perform monetary transactions Age Age squared Married Rural Male Occupation Irregular income Person in charge Education Income Income squared Constant Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients) % correct prediction (y=1) % correct prediction (y=0) Wald χ2 (H0: Weakness of instruments) p-value Sargan test χ2 (H0: Exogeneity of instruments) p-value

Full sample Adoption/Usage of mobile money -5.922*** (1.229) 1.988*** (0.525) -0.601 (0.589) 0.008 (0.009) 4.214*** (1.330) 2.361*** (0.810) 1.265* (0.750) 1.194 (1.319) -0.430 (0.968) -0.132 (0.818) 1.440** (0.633) -2.244 (4.622) 0.130 (0.899) 18.967 (12.205) 380 0.915 58.09*** 424.88*** 97.91% 98.94% 23.45 0.000 0.06 0.999

Notes: Dependent variables: the use of mobile money. The use of mobile money is a dummy variable that equals to 1 if respondents use mobile money, and 0 otherwise. The excluded instruments are distance to the nearest agent and the capacity to perform monetary transactions using cell phone. Robust standard errors in brackets. We use a logistic model specify in equation (4). *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

34

Table 10. IV Results. Choice of deposit instruments and mobile money adoption / individuals participating in informal financial mechanisms. Full sample

Pr (MM user) Controls included Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients)

% correct prediction (y=1) % correct prediction (y=0)

Pr (MM user) Pr (MM user) x Participating in informal mechanisms

Participating in informal mechanisms x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of

Deposit in formal institutions

Deposit in bank account

Deposit in credit union account

(1) -0.158 (0.296)

(2) 0.431 (0.317)

(3) -0.155 (0.250)

Deposit in mobile money account (4) 4.871*** (0.521)

YES

YES

YES

YES

YES

377 0.199 63.61***

377 0.343 77.12***

377 0.049 23.50**

376 0.552 135.27***

167 0.296 56.82***

131.94***

205.97***

60.22***

313.98***

17.96

68.91% 76.36%

66.88% 87.00%

60.73% 62.37%

91.50% 82.96%

91.88% 57.14%

Deposit using informal mechanisms (5) -1.785*** (0.582)

Individuals participating in informal financial mechanisms Deposit in formal Deposit in credit union Deposit in bank account institutions account Total Total Total effect effect effect (1) (2) (3) (4) (5) (6) 40.655*** 22.525*** -0.440 (0.464) (5.595) (1.654) -40.968

-0.313

21.197***

1.328**

0.213

-0.228

(0.000)

(0.464)

(5.530)

(0.559)

(1.711)

(0.436)

YES

YES

176 0.2670

176 0.490 38.35***

YES

176 0.177 39.81*** 359.05** 328.47*** 428.30*** coefficients) * % correct prediction (y=1) 86.73% 77.19% 80.00% % correct prediction (y=0) 55.56% 91.60% 56.58% Notes: Pr (MM user) is the independent variable of interest that is the predicted value of mobile money use that we obtain from the reduced form estimation in Table 9. Robust standard errors in brackets. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

In Table 10, 11 and 12, we replicate our estimations in section 5 (results) using the predicted value of mobile money use. The findings are consistent with our previous results.

35

Table 11. IV Results. Choice of deposit instruments and mobile money adoption: Low, irregular vs. High, regular income.

Pr (MM user) Low income Pr (MM user) x Low income Controls Low income x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients)

% correct prediction (y=1) % correct prediction (y=0)

Pr (MM user) Irregular income Pr (MM user) x Irregular income Controls Irregular income x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients)

Deposit in formal institutions Total effect (1) (2) 1.005* (0.526) 11.302** (5.519) -1.734*** -0.729 (0.654) (0.389) YES YES 377 0.217 67.06***

Low vs. High income Deposit in bank account Total effect (3) (4) 0.756* (0.422) -0.786 (5.957) -0.690 0.066 (0.656) (0.502) YES YES 377 0.332 95.02***

140.46***

200.74***

91.58***

68.54% 78.18%

70.13% 85.65%

62.83% 71.51%

Deposit in formal institutions Total effect (1) (2) 0.055 (0.476) -7.509 (6.419) -0.257 (0.637) YES YES 377 0.255 71.46*** 157.60***

Irregular vs. Regular income Deposit in bank account Total effect (3) (4) -0.098 (0.450) 7.824

-0.202 (0.424)

(6.897) 1.228 (0.752) YES YES 377 0.394 110.31*** 232.00***

1.130* (0.603)

Deposit in credit union

(5) 0.123 (0.362) 8.804* (5.003) -0.662 (0.534) YES YES 377 0.109 45.55***

Total effect (6)

-0.539 (0.393)

Deposit in credit union

(5) 0.136 (0.349) 21.185*** (6.835) -0.414 (0.530) YES YES 377 0.121 45.28***

Total effect (6)

-0.278 (0.398)

97.79***

% correct prediction (y=1) 71.54% 72.08% 67.54% % correct prediction (y=0) 79.09% 87.89% 67.74% Notes: Pr (MM user) is the independent variable of interest that is the predicted value of mobile money use that we obtain from the reduced form estimation in Table 9. Robust standard errors in brackets. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

36

Table 12. IV Results. Choice of deposit instruments and mobile money adoption: Low vs. High access to formal finance.

Pr (MM user) Rural Pr (MM user) x Rural Controls Rural x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients) % correct prediction (y=1) % correct prediction (y=0)

Pr (MM user) Female Pr (MM user) x Female Controls Female x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients) % correct prediction (y=1) % correct prediction (y=0)

Pr (MM user) Less educated Pr (MM user) x Less educated Controls Less educated x Controls Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients) % correct prediction (y=1) % correct prediction (y=0)

Deposit in formal institutions Total effect (1) (2) -0.264 (0.419) -5.864 (6.628) 0.186 -0.078 (0.608) (0.441) YES YES 377 0.212 69.56*** 138.12*** 68.16% 79.09%

Rural vs. Urban Deposit in bank account Total effect (3) (4) 0.833* (0.427) 8.041 (7.003) -0.968 -0.135 (0.644) (0.482) YES YES 377 0.394 95.71*** 232.10*** 72.73% 88.34%

Deposit in credit union Total effect (5) (6) -0.280 (0.370) -7.560 (5.240) 0.263 -0.017 (0.519) (0.365) YES YES 377 0.089 32.36* 81.15*** 71.20% 61.29%

Deposit in formal institutions Total effect (1) (2) -0.105 (0.480) 4.871 (5.006) -0.061 -0.166 (0.626) (0.402) YES YES 377 0.210 76.32*** 137.13*** 69.66% 78.18%

Female vs. Male Deposit in bank account Total effect (3) (4) -0.332 (0.439) 14.721** (5.768) 1.517** 1.185** (0.713) (0.562) YES YES 377 0.394 100.50*** 232.18*** 74.68% 85.65%

Deposit in credit union Total effect (5) (6) -0.318 (0.358) -6.766 (4.756) 0.345 0.027 (0.523) (0.381) YES YES 377 0.087 40.68** 80.27*** 64.92% 61.83%

Deposit in formal institutions Total effect (1) (2) -0.555 (0.460) 8.973* (5.418) 1.153* 0.597 (0.610) (0.401) YES YES 377 0.218 207.40*** 140.90*** 68.16% 80.00%

Less vs. High educated Deposit in bank account Total effect (3) (4) 0.522 (0.381) 12.824* (7.559) 0.643 1.166** (0.702) (0.590) YES YES 377 0.347 303.04*** 208.07*** 66.88% 85.65%

Deposit in credit union Total effect (5) (6) -0.639* (0.344) 7.101 (5.189) 1.269** 0.631* (0.515) (0.384) YES YES 377 0.100 218.76*** 87.01*** 70.16% 61.83%

Notes: Pr (MM user) is the independent variable of interest that is the predicted value of mobile money use that we obtain from the reduced form estimation in Table 9. Robust standard errors in brackets. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

37

Table 13. IV Results. Factors affecting the choice of making deposit using mobile money account.

Risk (1) Pr (MM user) x Attributes of informal/mobile money

Full sample Deposit using mobile money account Attributes of Formal/Mobile money Attributes of Informal/Mobile money Cost Illiquidity Low quality Risk Cost Illiquidity Low quality (2) (3) (4) (5) (6) (7) (8)

2.203

3.592

-1.696

-2.070

12.554**

-8.316

-8.977

1.249

(4.890)

(3.662)

(4.157)

(2.180)

(4.919)

(6.364)

(7.001)

(2.250)

-1.724

0.640

3.276

4.540***

-11.560**

9.480

10.154

0.619

(4.933)

(3.609)

(4.372)

(1.756)

(4.768)

(6.321)

(7.181)

(1.838)

Controls included

NO

NO

NO

YES

NO

NO

NO

YES

Pr (MM user) x Controls included

YES

YES

YES

YES

YES

YES

YES

YES

Total effect

0.478 (0.660)

4.233*** (1.367)

1.580** (0.708)

2.470** (1.049)

0.994** .4843047

1.165 (1.297)

1.177* (0.695)

1.868** (0.781)

Observations Pseudo R2 Wald χ2 (H0:

145 0.117

145 0.303

145 0.190

340 0.6188

86 0.474

85 0.439

85 0.445

345 0.6015

nullity of coefficients)

18.44

25.01**

27.96***

140.11***

35.25***

38.84***

31.75***

155.57***

424.10***

448.83***

433.83***

365.63***

492.70***

489.79***

490.31***

354.96***

Attributes of informal/mobile money

Likelihood ratio test χ2 (H0: nullity of coefficients)

% correct 95.00% 92.50% 94.17% 91.55% 86.36% 92.31% 92.31% 88.89% prediction (y=1) % correct 16.00% 56.00% 32.00% 87.37% 80.00% 70.00% 70.00% 87.56% prediction (y=0) Note: Pr (MM user) is the independent variable of interest that is the predicted value of mobile money use that we obtain from the reduced form estimation in Table 9. Robust standard errors in brackets. In columns 4 and 8, we add the variable MM user used in the interaction terms in the controls variables. We use a logistic model specify in equation (3). Controls included: age, age squared, married, rural, male, occupation, irregular incomes, person in charge, education level, incomes level and incomes squared. Table A.2. in the Appendix gives definitions and summary statistics of the attributes of formal and informal financial mechanisms that consist of risk, cost, illiquidity and low quality. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

Table 13 shows the estimations for the attributes related to formal and informal financial mechanisms compared to those related to mobile money that may induce mobile money users to make deposit in the mobile money account. The results are similar to our previous findings that cost, illiquidity and low quality associated with formal financial methods compared to those related to mobile money, are attributes that may lead mobile money users to make deposit in the mobile money account. Considering informal financial mechanisms, we also find the coefficient of risk related to informal financial mechanisms compared to those of mobile money positive and significant. Thus, the risk associated with informal financial methods compared to those of mobile money, is a factor that may also lead mobile money users to make deposit in the mobile money account.

38

8. Conclusion This paper examines the effect of mobile money as a substitute or a complement of informal and formal financial instruments, and its potential to enhance financial access for disadvantaged individuals. In developing countries, the predominance of informal finance associated with the underdeveloped formal financial system raises questions about the effect the growing technology of mobile money may have on the improvement of financial access. The paper addresses this issue. We use an original dataset obtained from a survey we conducted in Burkina Faso in May 2014, and find that while the use of mobile money has no impact on deposit using formal financial instruments, it decreases the likelihood to use informal deposit mechanisms. Our findings also show that mobile money increases the likelihood of individuals who participate in informal mechanisms to make deposit using formal financial instrumentsbank and credit union accounts. In further investigations, we show that among disadvantaged groups, mobile money increases the likelihood of female, individuals with irregular income and those who are less educated to make deposit in bank and credit union accounts. The mechanisms underlying these results reveal that cost, illiquidity and low quality associated with formal financial methods compared to those related to mobile money are attributes that may lead individuals to use mobile money account to make deposit. Similarly, we find that perception of risk, illiquidity and low quality associated with informal deposit mechanisms compared to those of mobile money increase the likelihood of individuals to make deposit using mobile money account. Inadequate access to financial services is widespread in Burkina Faso. Mounting evidence suggest that various socioeconomic constraints depress savings/deposits even among those with access (Allan, Massu, and Svarer 2013; Kendall 2010; Kendall et al. 2011). In settings where the technology of mobile money exists, bridging the gap in individuals’ access to formal financial services is not overstating. However, the banking system regulations need to be adjusted to take into account the new scalable technology-enabled business models. In fact, in Burkina Faso, mobile money is operated in partnership between banks and mobile network operators that render difficult the supervision of mobile money services. Thus, it is critical for the Central Bank (BCEAO) and the “Autorité de régulation des communications électroniques” to build strong and adequate regulation framework and supervision to support the wide range of services, especially

39

deposits/savings services that can be provided through mobile money. Although mobile money services are mobiquity (mobility and ubiquity), mobile money providers have inclination to concentrate their services in locations where formal activities are already available especially in urban areas. Government and policymakers may act through specific strategies to motivate mobile money providers to reach remote areas to ensure access to basic formal financial services throughout the country. Expanding mobile money agent networks by facilitating retail stores to start mobile money business especially in rural areas may help reduce the gap in formal financial access between urban and rural areas. Policies that focus on and motivate female, less educated and informal savings groups’ access to and usage of mobile money services should also be encouraged. More specifically, promoting the creation of an informal savings groups mobile money account linked to individuals’ mobile money account and that allows transfers between both accounts may reduce the need of cash exchanges and favor electronic money. Thus, mobile money may in turn bring out individuals from informal financial methods toward formal financial institutions by increasing the likelihood of individuals to access/use bank and credit union accounts. However, our results should be interpreted with caution given the lack of data on the amount allocated to each financial instrument. Further work will be needed to refine the analysis with detailed data.

40

References Allan, Alice, Maude Massu, and Christine Svarer. 2013. Banking on Change: Breaking the Barriers to Financial Inclusion. Allen, Franklin et al. 2014. “The African Financial Development and Financial Inclusion Gaps.” Journal of African Economies 23(5): 614–42. Association Professionnelle des Systèmes Financiers Décentralisés du Burkina Faso, AP/SFDBF. 2014. Rapport D’analyse Des Performances Financières de 2010-2012 Des Membres de l'AP/SFD-BF. Banque Centrale des Etats de l'Afrique de l'Ouest. 2014. Commission Bancaire Rapport Annuel. Beck, Thorsten, Asli Demirguc-Kunt, and M. S. Martinez Peria. 2008. “Banking Services for Everyone? Barriers to Bank Access and Use around the World.” The World Bank Economic Review 22(3): 397–430. Beck, Thorsten, Asli Demirgüç-Kunt, and Patrick Honohan. 2009. “Access to Financial Services: Measurement, Impact, and Policies.” World Bank Research Observer 24: 119–45. Carpenter, Seth B., and Robert T. Jensen. 2002. “Household Participation in Formal and Informal Savings Mechanisms: Evidence from Pakistan.” Review of Development Economics 6(3): 314–28. Collins, Daryl, Jonathan Morduch, Stuart Rutherford, and Orlanda Ruthven. 2009. Princeton and Oxford: Princeton University Press Portfolios of the Poor: How the World’s Poor Live on $2 a Day. Princeton and Oxford: Princeton University Press. Conroy, John. 2005. “APEC and Financial Exclusion: Missed Opportunities for Collective Action?” Asia-Pacific Development Journal 12(1): 28–29. Demirguc-kunt, Asli, and Leora Klapper. 2012. “Measuring Financial Inclusion. The Global Findex Database.’'.” Policy research working paper, World Bank 6025(April): 1–61. Demirgüç-Kunt, Asli, and Leora Klapper. 2012. “Financial Inclusion in Africa: An Overview.” World Bank Policy Research (June). Demirgüç-Kunt, Asli, Leora Klapper, and Dorothe Singer. 2013. “Financial Inclusion and Legal 41

Discrimination against Women: Evidence from Developing Countries.” World Bank Policy Working Paper 6416 (April). Demirgüç-kunt, Asli, and Ross Levine. 2008. “Finance and Economic Opportunity.” Policy research working paper, World Bank Working Research Paper (January): 1–32. Demombynes, Gabriel, and Aaron Thegeya. 2012. “Kenya’s Mobile Revolution and the Promise of Mobile Savings.” World Bank Policy Research Working Paper 5988 (March). Dermish, Ahmed, Christoph Kneiding, Paul Leishman, and Ignacio Mas. 2012. “Branchless and Mobile Banking Solutions for the Poor : A Survey of the Literature.” Innovations: Technology, Governance, Globalization 6(4): 81–98. Dupas, Pascaline, and Jonathan Robinson. 2013. “Savings Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya.” American Economic Journal: Applied Economics 5(1): 163–92. Gash, Megan, and Bobbi Gray. 2015. “Understanding Household Resilience of the Poor in Burkina Faso Interim Report.” Consultative Group to Assist the Poor Working Paper. Honohan, Patrick, and Thorsten Beck. 2007. Finance Making Finance Work for Africa. World Bank. International Finance Corporation. 2013. Access to Finance Sub-Saharan Africa. Jack, William, and Tavneet Suri. 2014. “Risk Sharing and Transactions Costs: Evidence from Kenya’s Mobile Money Revolution.” American Economic Review 104: 183–223. Karlan, Dean, Aishwarya Lakshmi Ratan, and Jonathan Zinman. 2014. “Savings by and for the Poor: A Research Review and Agenda.” Review of Income and Wealth 60(1): 36–78. Kendall, Jake. 2010. “A Penny Saved: How Do Savings Accounts Help the Poor?” Financial Access Initiative, Focus Note (2005): 1–22. Kendall, Jake, Bill Maurer, Phillip Machoka, and Clara Veniard. 2011. “An Emerging Platform: From Money Transfer System to Mobile Money Ecosystem.” Innovations: Technology, Governance, Globalization 6(4): 49–64. Kendall, Jake, Robert Schiff, and Emmanuel Smadja. 2014. “Sub-Saharan Africa : A Major 42

Potential Revenue Opportunity for Digital Payments.” McKinsey Quarterly (February): 3–8. Klein, Michael, and Colin Mayer. 2011. “Mobile Banking and Financial Inclusion: The Regulatory Lessons.” World Bank Policy Research Working Paper … (May). De Koker, Louis, and Nicola Jentzsch. 2013. “Financial Inclusion and Financial Integrity: Aligned Incentives?” World Development 44: 267–80. Ky, Serge, Clovis Rugemintwari, and Alain Sauviat. 2015. Does Mobile Money Affect Saving Behavior ? Evidence from a Developing Country. Mas, Ignacio. 2009. “The Economics of Branchless Banking.” Innovations: Technology, Governance, Globalization 4(2): 57–75. ———. 2010. “Savings for the Poor.” World Economics 11(4): 1–12. ———. 2012. The New Microfinance Handbook: A Financial Market System Perspective Beyond Products: Building Integrated Customer Experiences (2012). ed. Ledgerwood Joanna. The World Bank (2013). Mas, Ignacio, and David Porteous. 2015. “Pathways to Smarter Digital Financial Inclusion.” Forthcoming in the Journal of Financial Transformation (42): 1–28. Mbiti, Isaac, and David N Weil. 2011. “Mobile Banking: The Impact of M-PESA in Kenya.” National Bureau of Economic Research. Mohan, R. 2006. “Economic Growth, Financial Deepening, and Financial Inclusion.” Reserve Bank of India Bulletin (September): 1305–20. Morawczynski, Olga. 2009. “Saving Through the Mobile Phone - The Case of M-PESA.” MicroBanking Bulletin (19): 7–14. Morawczynski, Olga, and Mark Pickens. 2009. “Poor People Using Mobile Financial Services: Observations on Customer Usage and Impact from M-PESA.” Consultative Group to Assist the Poor (August). Nair, Ajai, and Renate Kloeppinger-Todd. 2007. “Reaching Rural Areas with Financial Services: Lessons from Financial Cooperatives in Brazil, Burkina Faso, Kenya, and Sri Lanka.” Agriculture and Rural Development, World Bank. 43

Pande, Rohini et al. 2012. Does Poor People’s Access to Formal Banking Services Raise Their Incomes ? EPPI-Centre, Social Science Research Unit, Institute of Education University of London. Porteous, David. 2006. “The Enabling Environment for Mobile Banking in Africa.” Bankable Frontiers Associates Boston USA (April): 1–57. Radcliffe, Dan, and Rodger Voorhies. 2012. Consultative Group to Assist the Poor A Digital Pathway to Financial Inclusion. Ramada-Sarasola, M. 2012. “Can Mobile Money Systems Have a Measurable Impact on Local Development?” Robinson, Marguerite, and Graham Wright. 2001. MicroSave Working Paper Mobilising Savings. Sarma, Mandira, and Jesim Pais. 2011. “Financial Inclusion and Development.” Journal of International Development 23: 613–28. Shem, Alfred Ouma, Roseline Misati, and Lucas Njoroge. 2012. “Factors Driving Usage of Financial Services from Different Financial Access Strands in Kenya.” Savings and Development 36(1): 71–89. Thieba, Daniel. 2013. Les Coopératives D ’ Épargne et de Crédit Au Burkina : Étude de Cas Du Réseau Des Caisses Populaires Du Burkina Faso. Capra International Inc., Canada. Triki, Thouraya, and Issa Faye. 2013. “Financial Inclusion in Africa: An Overview.” African Development Bank.

44

Appendix Table A.1. Correlation matrix. Full sample. Deposit using formal institution s Deposit using formal institutions Deposit using banks Deposit using credit unions Deposit using mobile money Deposit using informal MM user

Depos it using banks

Deposit using credit unions

Deposit using mobile money

Depos it using inform al

MM user

Age

Age square d

Married

Rural

Male

Occup ation

Irregular income

Person in charge

Educat ion

Inco me

Income squared

1 0.546

1

0.661

0.023

1

0.088

0.140

0.052

1

-0.042

-0.107

0.109

-0.038

1

0.054

0.137

0.005

0.785

-0.111

1

Age

0.201

0.227

0.013

-0.018

-0.130

-0.027

1

Age squared

0.195

0.223

0.008

-0.036

-0.130

-0.045

0.990

1

Married

0.196

0.162

0.084

0.154

-0.07

0.131

0.607

0.579

1

Rural

-0.006

-0.139

0.005

0.127

-0.176

0.146

0.251

0.245

0.226

1

Male

0.085

0.125

-0.04

0.066

0.028

0.007

0.267

0.254

0.107

0.032

1

Occupation

0.082

-0.014

0.070

-0.107

0.06

-0.078

0.430

0.384

0.348

0.159

0.115

1

Irregular income

-0.097

-0.294

0.145

0.062

-0.047

0.043

0.100

0.08

0.106

0.302

-0.096

0.197

1

Person in charge

0.055

0.066

0.076

0.089

0.057

0.107

-0.003

-0.009

0.024

-0.03

-0.008

0.047

0.015

1

Education

0.275

0.449

0.017

0.185

-0.054

0.162

-0.148

-0.139

-0.169

-0.227

0.017

-0.391

-0.358

-0.013

1

Income

0.359

0.417

0.089

0.086

-0.03

0.071

0.441

0.426

0.358

-0.052

0.271

0.342

-0.131

0.055

0.127

Income squared

0.32

0.407

0.056

0.074

-0.025

0.07

0.427

0.417

0.349

-0.078

0.255

0.297

-0.145

0.063

0.135

1 0.97 7

45

1

Table A.2. Summary statistics and variables description (Mechanisms). Variable

Definition

Factors related to informal financial mechanisms (Initial measure) Indicate individuals perception of risk associated with informal mechanisms, Risk of informal encoded as (Very lower) = 1, (Lower) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Indicate individuals perception of cost associated with informal mechanisms, Cost of informal encoded as (Very lower) = 1, (Lower) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Illiquidity of Indicate individuals perception of illiquidity associated with informal mechanisms, informal encoded as (Very low) = 1, (Low) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Low quality of Indicate individuals perception of quality associated with informal mechanisms, informal encoded as (Very low) = 1, (Low) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Factors related to formal financial mechanisms (Initial measure) Indicate individuals perception of risk associated with formal mechanisms, encoded Risk of formal as (Very lower) = 1, (Lower) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Indicate individuals perception of cost associated with formal mechanisms, encoded Cost of formal as (Very lower) = 1, (Lower) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Illiquidity of Indicate individuals perception of illiquidity associated with formal mechanisms, formal encoded as (Very low) = 1, (Low) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Low quality of Indicate individuals perception of quality associated with formal mechanisms, formal encoded as (Very low) = 1, (Low) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Factors related to Mobile money (Initial measure) Risk of mobile Indicate individuals perception of risk associated with mobile money, encoded as money (Very lower) = 1, (Lower) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Cost of mobile Indicate individuals perception of cost associated with mobile money, encoded as money (Very lower) = 1, (Lower) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Illiquidity of Indicate individuals perception of illiquidity associated with mobile money, mobile money encoded as (Very low) = 1, (Low) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Low quality of Indicate individuals perception of quality associated with mobile money, encoded as mobile money (Very low) = 1, (Low) = 2, (Medium) = 3, (High) = 4, (Very high) = 5 Factors related to informal financial mechanisms compared to mobile money (Computed) Risk of Indicate the perception of risk associated with informal mechanisms compared to informal/mobile mobile money, encoded as (low) = 0.33, (High) = 5 money Cost of Indicate the perception of costs associated with informal mechanisms compared to informal/mobile mobile money, encoded as (Low) = 0.33, (High) = 5 money Illiquidity of Indicate the perception of illiquidity associated with informal mechanisms informal/mobile compared to mobile money, encoded as (Low) = 0.67, (High) = 5 money Low quality of Indicate the perception of quality associated with informal mechanisms compared to informal/mobile mobile money, encoded as (Low) = 0.4, (High) = 5 money Factors related to formal financial mechanisms compared to mobile money (Computed) Risk of Indicate the perception of risk associated with formal mechanisms compared to formal/mobile mobile money, encoded as (low) = 0.33, (High) = 3 money Cost of Indicate the perception of costs associated with informal mechanisms compared to formal/mobile mobile money, encoded as (Low) = 0.5, (High) = 4 money Illiquidity of Indicate the perception of illiquidity associated with informal mechanisms formal/mobile compared to mobile money, encoded as (Low) = 0.5, (High) = 3 money Low quality of Indicate the perception of quality associated with informal mechanisms compared formal/mobile to mobile money, encoded as (Low) = 0.2, (High) = 5 money

Obs.

Mean

Std. Dev.

Min

Max

228

3.39

.93

1

5

229

2.29

.81

1

5

230

3.13

.97

1

5

380

3.01

.65

1

5

283

2.36

0.66

1

4

282

2.94

0.46

1

4

287

2.90

0.54

1

5

376

2.59

0.68

1

5

203

2.35

0.64

1

4

204

2.29

0.71

1

4

203

1.93

0.66

1

5

383

2.55

0.65

1

5

95

1.76

0.79

0.33

5

94

1.11

0.51

0.33

5

94

1.79

0.74

0.67

5

369

1.26

0.50

0.4

5

150

1.03

0.46

0.33

3

150

1.46

0.57

0.5

4

151

1.70

0.66

0.5

3

361

1.05

0.37

0.2

5

46

Table A.3. Correlation matrix (Attributes of formal savings instruments)

Deposit using MM Risk

Deposit using MM 1

Risk

Cost

Illiquidity

Quality

Age

Age squared

Married

Rural

Male

Occupation

Irregular income

Person in charge

Education

Income

-0.014

1

Cost

0.171

0.137

Illiquidity

0.067

0.082

0.345

1

Quality

-0.046

-0.366

-0.277

-0.322

1

Age

-0.018

-0.237

-0.046

0.063

0.205

1

Age squared

-0.036

-0.213

-0.039

0.07

0.193

0.990

1

Married

0.154

-0.234

0.06

0.033

0.131

0.607

0.579

1

Rural

0.127

-0.134

0.145

0.254

-0.098

0.251

0.245

0.226

1

Male

0.066

-0.270

0.039

0.1

-0.007

0.267

0.254

0.107

0.032

1

Occupation

-0.107

-0.15

-0.1

0.013

0.128

0.430

0.384

0.348

0.159

0.115

1

Irregular income

0.062

0.008

0.019

0.082

-0.072

0.100

0.08

0.106

0.302

-0.096

0.197

Person in charge

0.089

0.1

0.030

0.043

-0.052

-0.003

-0.009

0.024

-0.03

-0.008

0.047

0.015

1

Education

0.185

0.016

0.068

0.015

0.093

-0.148

-0.139

-0.169

-0.227

0.017

-0.391

-0.358

-0.013

1

Income

0.086

-0.243

-0.120

-0.164

0.221

0.441

0.426

0.358

-0.052

0.271

0.342

-0.131

0.055

0.127

1

Income squared

0.074

-0.236

-0.132

-0.174

0.206

0.427

0.417

0.349

-0.078

0.255

0.297

-0.145

0.063

0.135

0.977

Income squared

1

1

47

1

Table A.4. Correlation matrix (Attributes of informal savings mechanisms)

Deposit using MM Risk Cost Illiquidity Quality Age Age squared Married Rural Male Occupation Irregular income Person in charge Education Income Income squared

Deposit using MM 1 0.056 0.065 0.224 -0.229 -0.018 -0.036 0.154 0.127 0.066 -0.107 0.062 0.089 0.185 0.086 0.074

Risk

Cost

Illiquidity

Quality

Age

1 0.047 0.356 -0.188 -0.2 -0.193 -0.196 -0.048 -0.147 -0.2 -0.124 0.026 0.255 -0.116 -0.133

1 0.215 -0.062 0.008 -0.012 0.007 0.308 -0.023 0.039 0.050 -0.010 -0.096 -0.066 -0.048

1 -0.23 -0.121 -0.105 -0.002 -0.097 -0.085 -0.166 -0.19 0.092 0.278 0.052 0.041

1 -0.002 0.006 -0.033 -0.308 -0.114 0.089 -0.009 0.013 -0.151 0.044 0.042

1 0.990 0.607 0.251 0.267 0.430 0.100 -0.003 -0.148 0.441 0.427

Age squared

1 0.579 0.245 0.254 0.384 0.08 -0.009 -0.139 0.426 0.417

Married

Rural

Male

Occupation

1 0.226 0.107 0.349 0.106 0.024 -0.169 0.358 0.349

1 0.032 0.159 0.302 -0.03 -0.227 -0.052 -0.078

1 0.115 -0.096 -0.008 0.017 0.271 0.255

1 0.197 0.047 -0.391 0.342 0.297

Irregular income

Person in charge

1 0.015 -0.358 -0.131 -0.145

1 -0.013 0.055 0.063

Education

Income

1 0.127 0.135

1 0.977

Income squared

1

48

Table A.5. Correlation matrix (Reduced form for predicting the use of mobile money).

MM user

MM user Distance to the nearest agent Capacity to perform monetary transactions via cell phone Age Age squared Married

Distance to the nearest agent

Capacity to perform monetary transactions via cell phone

Age

Age squared

Married

Rural

Male

Occupation

Irregular income

Person in charge

Education

Income

1 -0.888

1

0.464

-0.368

1

-0.027

0.043

-0.041

1

-0.045

0.060

-0.055

0.990

1

0.131

-0.070

0.128

0.607

0.579

1

0.146

-0.088

-0.039

0.251

0.245

0.226

1

Male

0.007

0.009

0.035

0.267

0.254

0.107

0.032

1

Occupation

-0.078

0.069

-0.154

0.430

0.384

0.348

0.159

0.115

1

0.043

-0.030

0.015

0.100

0.080

0.106

0.302

-0.096

0.197

1

0.107

-0.097

-0.007

-0.003

-0.009

0.024

-0.030

-0.008

0.047

0.015

1

Education

0.162

-0.126

0.366

-0.148

-0.139

-0.169

-0.227

0.017

-0.391

-0.358

-0.013

1

Income

0.071

-0.043

0.133

0.441

0.426

0.358

-0.052

0.271

0.342

-0.131

0.055

0.127

1

Income squared

0.070

-0.047

0.1161

0.427

0.417

0.349

-0.078

0.255

0.297

-0.145

0.063

0.135

0.977

Rural

Irregular income Person in charge

Income squared

49

1

Table A.6. Deposit in mobile money account and the use of mobile money for subsamples. (IV Results are also reported).b

MM user

Participating in informal mechanisms (1) 27.449***

Total effect Controls included Subsample variable x Controls included Observations Pseudo R2 Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients) % correct prediction (y=1) % correct prediction (y=0)

Total effect

(5) 5.748***

(6) 6.642***

(1.093)

(1.206)

(0.821)

(1.266)

-21.521***

25.976** (10.564) -1.325

12.763* (6.638) -15.578***

-2.715 (8.942) -1.264

2.801 (8.723) -0.521

-9.375 (6.554) -2.231

(7.566)

(1.314)

(1.332)

(1.467)

(1.262)

(1.453)

5.928*** (1.026)

4.990*** (0.838)

4.678*** (0.752)

5.051*** (0.836)

5.227*** (0.959)

4.411*** (0.713)

NO YES

YES YES

YES YES

YES YES

YES YES

YES YES

178 0.6415 1745.57*** 459.76*** 88.52% 89.74%

378 0.627 130.03*** 351.11*** 90.91% 87.05%

378 0.641

378 0.634 106.22*** 354.72*** 92.21% 87.95%

378 0.630 133.92*** 352.84*** 93.51% 86.61%

381 0.616 327.80*** 344.55*** 94.16% 86.34%

Female vs. Male

Less vs. High educated

(6) 6.044*** (0.860) -7.858 (6.552)

358.46*** 92.21% 86.61%

Deposit in mobile money account Low vs. Irregular vs. Rural vs. High Regular Urban income income

(1)

(2)

(3)

(4)

(5)

27.653***

5.278***

6.933***

6.509***

5.348***

(8.329)

(0.648)

(0.814)

(1.250)

(0.740)

-21.645*** (7.755)

21.253** (8.839) -0.200 (1.088)

0.290 (6.747) -2.388** (1.137)

-3.842 (9.170) -1.894 (1.433)

-0.319 (7.725) -0.158 (1.161)

6.008*** (1.061)

5.077*** (0.874)

4.545*** (0.794)

4.614*** (0.701)

5.190*** (0.894)

Subsample Pr (MM user) x Subsample variable

Less vs. High educated

(1.012)

Participating in informal mechanisms Pr (MM user)

Female vs. Male

(8.140) Subsample MM user x Subsample variable

Subsamples Deposit in mobile money account Low vs. Irregular vs. Rural vs. High Regular Urban income income (2) (3) (4) 6.315*** 20.255*** 6.314***

-1.972* (1.090) 4.072*** (0.669)

Controls included Subsample variable x Controls included Observations Pseudo R2

NO YES YES YES YES YES YES YES YES YES YES YES 176 376 376 376 376 376 0.642 0.587 0.610 0.604 0.602 0.578 1793.43*** 158.96*** 160.83*** 119.91*** 137.16*** 336.32*** Wald χ2 (H0: nullity of coefficients) Likelihood ratio test χ2 (H0: nullity of coefficients) 460.88*** 331.72*** 343.79*** 340.54*** 339.32*** 327.52*** % correct prediction (y=1) 88.33% 90.20% 90.85% 90.85% 92.16% 88.24% % correct prediction (y=0) 90.52% 87.44% 87.00% 87.44% 86.10% 87.00% Notes: Dependent variable: deposit in mobile money account is a dummy that takes the value 1 if respondents make deposit using mobile money account, and 0 otherwise. Pr (MM user) is the independent variable of interest that is the predicted value of mobile money use that we obtain from the reduced form estimation in Table 9. Robust standard errors in brackets. Subsamples are defined as follow: we compare individuals participating in informal mechanisms to those who don’t; Low to High income individuals; Irregular to Regular income; Individuals located in Rural to those in Urban area; Female to Male; and Less to High educated individuals. Controls included: age, age squared, married, rural, male, occupation, irregular incomes, person in charge, education level, incomes level and incomes squared. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

50

Table A.7. Deposit using informal mechanisms and the use of mobile money for subsamples. (IV Results are also reported). Subsamples Deposit using informal mechanisms Irregular vs. Rural vs. Regular Urban income (Predicted value of MM user) (7) (8) -6.273*** -7.193***

Low vs. High income

Irregular vs. Regular income

Rural vs. Urban

Female vs. Male

Less vs. High educated

Low vs. High income

(1) -5.158***

(2) -4.641**

(3) -7.197***

(4) -5.585***

(5) 10.854***

(6) -7.373***

(0.981)

(1.976)

(1.383)

(1.147)

(1.244)

(1.640)

(0.892)

-66.102*** (12.359)

25.738 (0.000)

292.922 (0.000)

-151.465 (0.000)

-101.934 (0.000)

-103.575*** (17.007)

-1.593

-10.061***

5.836***

4.817***

-40.957***

(2.326)

(3.083)

(1.428)

(1.351)

-6.750*** (2.110)

-14.702*** (1.212)

-1.361** (0.575)

Controls included

YES

YES

Subsample variable x Controls included

YES

MM user

Subsample MM user x Subsample variable

Total effect

Observations Pseudo R2 Wald χ2 (H0: nullity of

Female vs. Male

Less vs. High educated

(9) -5.728***

(10) 11.204

(1.475)

(1.150)

(21.378)

27.304 (0.000)

260.802 (0.000)

-135.783 (0.000)

1,218.140 (0.000)

-25.216**

-57.164*

4.356***

4.404***

-747.777

(3.333)

(12.162)

(30.216)

(1.657)

(1.084)

(0.000)

0.768 (0.906)

-30.103*** (3.042)

-32.589*** (12.028)

-63.436** (30.199)

-2.838** (1.185)

-1.324 (1.027)

-736.573*** (21.378)

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

143 0.705

169 0.690

169 0.649

169 0.520

169 0.794

141 0.789

167 0.730

167 0.674

167 0.531

167 1.000

42.38***

40.79***

38.41***

30.86***

46.84***

47.11***

43.19***

39.90***

31.60**

58.86***

coefficients)

Likelihood ratio test χ2 (H0: nullity of coefficients)

% correct prediction 99.26% 97.53% 97.53% 97.53% 99.38% 99.25% 98.75% 100.00% 100.00% 100.00% (y=1) % correct prediction 57.14% 71.43% 57.14% 57.14% 57.14% 71.43% 57.14% 57.14% 42.86% 100.00% (y=0) Notes: Notes: Dependent variable: deposit using informal mechanisms is a dummy that takes the value 1 if respondents make deposit using informal financial mechanisms, and 0 otherwise. Pr (MM user) is the independent variable of interest that is the predicted value of mobile money use that we obtain from the reduced form estimation in Table 9. Robust standard errors in brackets. Subsamples are defined as follow: we compare individuals with Low income to those with High income individuals; Irregular to Regular income; Individuals located in Rural to those in Urban areas; Female to Male; and Less to High educated individuals. Controls included: age, age squared, married, rural, male, occupation, irregular incomes, person in charge, education level, incomes level and incomes squared. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

51

Table A.8. Data sample characteristics. (Deposit instruments)

Full sample Gender Female Male Marital situation Married Single Person in charge Age < 30 >= 30 Education level Less than secondary education level At least secondary education level Occupation / employment status Paid activity Unpaid activity (include student) Income level and type Income ranging from 10 000 to 50 000 FCFA Income more than 50 000 FCFA Irregular income Regular income

Individuals that report Deposit Deposit using using credit mobile unions money

Deposit using informal mechanisms

Full sample

Mobile Money users

/

50,37

39,75%

49,14%

40,25%

41,98%

49,38% 50,62%

49,02% 50,98%

41,61% 58,39%

51,26% 48,74%

45,40% 54,60%

64,71% 35,29%

48,40% 50,86% 52,10%

54,90% 45,10% 52,45%

58,39% 41,61% 66,46%

52,76% 46,73% 60,30%

57,67% 42,33% 55,83%

39,41% 60,00% 44,12%

50,62% 49,14%

48,53% 51,47%

38,51% 60,87%

48,24% 51,26%

46,63% 53,37%

61,76% 37,06%

41,73% 57,53%

36,27% 63,73%

19,88% 78,88%

41,71% 57,29%

34,97% 65,03%

43,53% 56,47%

80,99% 15,56%

77,45% 18,14%

82,61% 16,15%

85,93% 13,07%

76,07% 19,63%

80,00% 17,06%

48,64% 51,36% 47,90% 51,60%

43,63% 56,37% 50,00% 49,51%

26,71% 73,29% 29,81% 69,57%

42,21% 57,79% 55,28% 44,72%

40,49% 59,51% 51,53% 47,85%

61,18% 38,82% 44,71% 54,71%

Deposit using banks

Usage of mobile technology Mobile phone user 99,26% 99,02% 99,38% 100% 99,39% MM user 50,37% / 58,39% 50,25% 97,55% Usage of deposit instruments Formal 88,89% 95,10% Bank 39,75% 46,08% 41,21% 48,47% Credit Union 49,14% 49,02% 50,93% 52,76% MM 40,25% 77,94% 49,07% 43,22% Informal 41,98% 35,78% 32,92% 47,74% 34,97% Source: Authors’ analysis of the survey data collected in May 2014 in Burkina Faso. Throughout, F CFA (Franc of the African Financial Community) refers to the local currency. The exchange rate during the survey period was about 500 F CFA = $1 US.

52

98,82% 42,94%

31,18% 55,88% 33,53%