PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • The solution we propose is designed for Africa’s poorest countries where a confluence of conditions holds. First, in a primarily agrarian economy, household income is scarce and seasonal [2]. Nevertheless, rural households, while cash-constrained, want to send their children to school and prioritize paying school fees; to do so, parents often rely on relatives and friends [3]. This community of funders is often non-local, and relatives either have to travel in person to pay school fees or risk sending money via a third-party. Observational studies drawing from detailed survey data have shown that these remittances—typically from relatives who have emigrated or relocated to urban areas for employment in non-agricultural sectors of the economy—are critically important for livelihoods more broadly in rural sub-Saharan Africa, both due to overall poverty levels and vulnerability [4–6], and because the formal banking and financial sectors are underdeveloped. [7]. In the absence of formal financial services, remittances have been shown to improve access to public services [8], provide insurance against income shocks [9], and even stimulate agricultural productivity [10]. And yet, remittances are not working as well as they should. Three obstacles limit the effectiveness of remittances as an informal insurance scheme in this context –transaction costs, control, and cultural norms. The first has been well-documented: the cost of sending remittances can be prohibitively high. The World Bank estimates that, globally, sending remittances costs an average 7.32% of the amount sent, and this share increases to 9.42% for sub-Saharan Africa [11]. Second, even within the same household, individuals may hold different preferences [12]: this matters because remittance-senders and remittance-recipients may diverge in how they would like the money spent, and absent a control mechanism, senders may remit less than they otherwise would. Finally, migrants who have to travel back home to remit may face added pressure to share the wealth [13, 14]; as a result, they may prefer not to send money home at all. Remittances thus offer a potential solution, but they also do not entirely fulfill their development promise. A more recent wave of experimental (randomized control trial) studies on remittances have shed light on the possibility for mobile money, with its reduced transaction costs and technological flexibility, to address some of these inefficiencies. An field experiment using an encouragement design in rural Mozambique showed that access to mobile money promoted increased urban-rural remittance flows over meaningful time scales (3 years at publication). Importantly, these remittance flows translated into reduced vulnerability to hunger shocks for receiving households, along with the ability to pay some school-related expenses (books and uniforms) even in the face of shocks. [15]. A similar experimental study in Bangladesh also resulted in increased remittance flows to rural households, reduced lean-season vulnerability, increased savings and schoolings, and decreased child labor. [16] Two other experiments that did not directly test the potential for mobile money based remittances to address the control and cultural norms issues, nevertheless offer suggestive findings in favor of such systems: A study in El Salvador showed that when remittance-senders are given the option to send their remittances via joint bank accounts (in other words, when they are given greater control over how the money is used), they send more money home [17]. Finally, a mobile money based savings experiment with farmers in Mozambique found that interventions with groups of farmers in the same network reduced pressure to “share the wealth” and resulted in greater savings by all farmers involved. [18] These two studies suggesting that mobile money accounts could potentially be configured to alleviate the control issues and cultural norm pressures that presently impede optimal remittance flows. However promising for remittances, the impact of mobile money in most of sub-Saharan Africa has been limited [19, 20] by the fact that uptake has been slow [21] in those countries that have most to gain from its widespread adoption. This is certainly the case in our study country, Benin. At the national level, education rates remain low, despite recent policies to address the issue (Fig 1(a)). In October 2006, Benin implemented part of the promise of its 1990 Constitution tasking the State with ensuring free public education and making primary school compulsory (Articles 12 and 13). On October 14, the Council of Ministers decreed free public schooling for all, but this was implemented only at the primary level. As of 2010, responding to a dire gender disparity in secondary education, the government has gradually exempted female students from paying school fees for the first cycle of secondary education as well. Figure data removed from full text. Figure identifier and caption: 10.1371/journal.pone.0198240.g001 Educational attainment and mobile phone usage/ownership in Benin, 2005-2017. (a) Educational Attainment by Gender. Educational attainment has risen over the past decade in Benin, but at different rates for men and women, particularly for secondary school. (b) Mobile phone usage (Afrobarometer rounds 3 & 4) and ownership (Afrobarometer rounds 6 & 7) in Benin. Access to mobile technology has risen rapidly over the past decade. Location of Benin shown in inset (red). Responses from Afrobarometer rounds 3-7 (2005-2017) [1]. The country’s efforts to expand public education have paid off to some extent. Indeed, the proportion of Beninois who have completed at least primary education rose from 29% in 2005 to 43% in 2017. But gains have been slower in secondary education, where school fees come into effect. Six percent had completed at least secondary education in 2005; this rose to only 13% in 2017 [1]. Disconcertingly, the disparity in educational attainment rates between men and women has widened over time (Fig 1(a)). Rural families surveyed for an unrelated study nevertheless express a strong desire to send children to school, but report deep financial constraints. [22] Family members and friends in Benin who want to pay for school fees for children living elsewhere have limited options for sending remittances—in-person travel, expensive or inconvenient standard bank transfers, taxi or bus driver. These options suffer from the inefficiencies described above, and are costly for the families and the schools: families incur significant risks and transaction costs, and schools often take months to collect their fees. At the same time, as it has across sub-Saharan Africa, mobile phone usage and ownership have risen rapidly in Benin (Fig 1(b)). In 2008 (Afrobarometer Round 4), just over half of the surveyed respondents nationally reported ever using a mobile phone; by 2017, over 91% of respondents reported that they or someone else in their household owned a mobile phone. Our solution thus leverages two facts about rural households in Benin to make remittances work better: rural households want to send their children to school, and they own cell phones. A mobile money school-fee payment system thus promises to solve the above-challenges: rural households already have access to mobile phones; remittance-senders have control over the use of their remittance; and mobile money transactions are private. Our product is a bundle of three technologies, developed between May and October 2015 and refined from October 2015 to October 2016, meant to address key needs in remittance networks and rural education, and to catalyze mobile money uptake and use more broadly. These components, their rationale, and details of their development are described below: A mobile money platform: The individual-to-institution capability was developed in conjunction with MTN, our mobile provider partner, to interface with their existing mobile money platform, called ‘MoMo’. It is a USSD menu-based interface, designed by our team, for use on feature phones. This is a critical piece of our intervention, as we are operating in a region where feature phones—but not smartphones—are prevalent. Donors access the school fee payment menu, select a destination school, input a unique student ID code (that we generate for students), and select the amount and purpose of the transfer (e.g., school fee payment, donation to school, etc.). On the receiving end, the school has an account in the name of the principal and accountant; the accountant can cash out the account as needed, up to twice per week, to transfer the money into its official institutional bank account. Simple transaction-tracking software: Most record-keeping in rural schools is done by hand, but school administrators want to know more about who is paying, and when, to better budget during the school year and try to raise funds for students in need. In addition, local parent associations—who work directly with their local school administrators to craft the school annual budget—lament the lack of standardized and timely information on school finances. We access the transaction stream from the school fee payment platform via a SOAP-based API provided by MTN at a fixed time interval (once per minute), parse it into fields, and send confirmation receipts by SMS to the school accountant. In addition, transaction information is entered automatically into a web-based student payment database. The accountant can also manually add in-person payments to the student payment database. At present this transaction-tracking platform is hosted on Amazon Cloud Servers and accessed via modem over the MTN cellular network in Benin. The student payment database was designed for easy data entry by school administrators (normally the accountant) and multi-function use. Data can be sorted and viewed by student, by date, by class. Stored fields include the student ID, class, year, a contact phone number if the student has one, as well as any information from payment transactions (e.g., donor phone number). The initial design of the web-based system included bells and whistles, such as dynamic searches, which slowed down the download speeds of the web pages. Hence we scaled back on some of the features to make the pages lightweight. In the final version, each page was less than a few kilobytes in size and allowed for fast access via the MTN modems. SMS-based messaging platform: At present, any communication between school administrators and students/parents/donors (those who pay school fees on behalf of students) is done by word of mouth through the school’s parent association, and is limited in scope and efficiency. The messaging platform uses MTN and Moov (the other main network in the pilot region) modems to distribute mass SMS messages to subsets of the greater school community. This could be an announcement by the administration about a meeting, school closure, or exam date to the entire student body, or a distinct and targeted fundraising appeal to donors who have used the mobile money based payment system. The messaging platform is integrated with the student and payment database, and designed for easy use by administrators with low technological proficiency. The pilot implemented the innovation in Kalalé, Benin, a commune of approximately 170,000 people, 80 to 90% of whom depend on agriculture for their livelihoods, and where the median resident survives on less than $1.90 a day [22]. We chose Benin in part because our team already had extensive fieldwork experience, a network of relations, and a history of successful interventions previously implemented in collaboration with a local community-based development organization there. In June 2014, our team traveled to Benin to introduce the idea to the pilot school and to discuss our innovation with stakeholders at all levels—the school administration, students, the parent-teacher organization, the local community-based development organization, commune administrators, members of the diaspora within Benin (who pay school fees for dozens of students), the two main cell phone companies, and the Ministries of Secondary Education and of Telecommunication. Our proposal was met with significant enthusiasm and enjoyed buy-in from local stakeholders to the highest levels of government. Similar success was achieved with the mobile money division of the country’s largest cell phone company. The public secondary school in Kalalé serves a total 1,870 students, and is similar in size to regional secondary schools around the country. At the time we implemented our pilot, school fees were covered only for girls from grades 6 to 9. All students had to pay their annual school fees of CFA12,000 (approximately $25) from grades 10 to 12. A key goal of this pilot study was to catalyze mobile money use. Accordingly, we intentionally chose a very remote and low-resource setting for pilot testing of this project. Mobile money use was predictably very low, and this meant that implementation by necessity required much more than simply advertising the existence of a new product and waiting to see if it would be used. CEPRODE (the Centre de Promotion de la Démocratie et du Développement), a non-profit non-governmental association, led the implementation in the field, which consisted of the following steps. Stakeholder meetings: Our team first conducted a set of meetings with key stakeholders in June 2014 to assess feasibility and support for the project. We followed up with all key stakeholders at the start of the pilot (August 2015), including local government officials, the Ministry of Education, MTN, school administrators, and the local parent association. All reaffirmed support, with the full meeting of the parent association providing key feedback on the desirability for this capacity along the dimensions discussed above.Student survey: At the start of the 2015 school year, as administrators were organizing students into classes and gathering information from them, we interviewed each student over the age of 13 in the Kalalé school to find out where they came from (many originate from outside the main village), how they usually payed their school fees, difficulties they had encountered, etc… Most important, we gathered contact information for their donors—those who contribute to their education costs. Overall we gathered data for 629 students (this represented all consenting students grades 9 through 12), with contact information from 662 donors around the country. A majority of donors were parents of students, though there were also aunts/uncles, friends, and former teachers/mentors. We obtained contact information of donors from the students themselves.Donor outreach: The CEPRODE team attempted to contact all donors, subject to IRB restrictions which limited our total number of attempts to 3; the team successfully reached and interviewed 241 donors. Donors were informed about the new platform available for Kalalé students, told how they could enroll in and access the platform on their own, and invited to registration events if they wanted help enrolling. We also administered a small survey of donors.Embedded donor survey experiment: One objective of this pilot study is to improve our understanding of the kinds of messages that effectively induce mobile money uptake and donations to the school. To do so, we tested two possible interventions. First, grounded in the social science literature on network and peer effects [23], we tested whether telling subjects that they were part of a larger community of mobile-money donors to the school would induce greater adoption of the technology (Peer effect). Second, following other impact evaluations testing the determinants of mobile-money uptake [15], we tested the effectiveness of a matching gift incentive (Altruism). Finally, we combined both messages to test whether a more forceful intervention highlighting both peer effects and matching incentives might induce technology adoption and donations (Peer effect + Altruism). Donors were randomly assigned to either the control condition, which received no particular additional message, the Peer effect condition, the Altruism condition, or the combined Peer effect+Altruism condition: Peer effect: “We want you to know that we are reaching out to an entire community of school-fee payers like yourself. There are 2,000 of you, which means that if you all choose to use this system, we will successfully have created a community of mobile money users in Kalalé.”Altruism: “In addition, to help you in this process, we are offering to donate money to the CEG-Kalalé if you enroll and pay your student’s school fee in full by the payment deadline. You can choose for this donation to go either to the general school fund, the school library, the campus enclosure, or the computer center. Which would you prefer?”Peer effect + Altruism: “We want you to know that we are reaching out to an entire community of school-fee payers like yourself. There are 2,000 of you, which means that if you all choose to use this system, we will successfully have created a community of mobile money users in Kalalé. In addition, to help you in this process, we are offering to donate money to the CEG-Kalalé if you enroll and pay your student’s school fee in full by the payment deadline. You can choose for this donation to go either to the general school fund, the school library, the campus enclosure, or the computer center. Which would you prefer?” Registration events: We hosted a series of registration events between October and December 2015, in and around the pilot region, to help the donors we had contacted (and others who were interested) register for MTN mobile money and learn, in person, how to make a payment with the system. We purchased the SIM cards for those who did not have them, and also offered a credit incentive for coming. 236 of the 241 contacted donors came to the registration events, along with approximately 125 others who had heard by word of mouth, but had not been a part of our survey sampling frame.Monitoring of transactions: With the system in place, we then monitored transactions remotely, and our CEPRODE team worked with the school accountant to make sure he could enter the data he wanted, access information, cash out the school’s account correctly, etc.Testing of messaging platform for communication and fundraising: Due to challenges with the MTN platform—which migrated in December 2015—we only tested the messaging platform twice: once to advertise the fall fee payment deadline and once to announce the new access code for the migrated MTN menu.Human subjects approval for all steps listed above was granted by the UC San Diego Institutional Review Board, #150857S, prior to implementation. A Pre-Analysis Plan for the project was also registered with EGAP (http://egap.org/registration/1426).
rdf:type