Mobile Banking and Financial Inclusion

Title:

Mobile Banking and Remittances among Migrant Workers: Experimental Evidence from Bangladesh (with Jonathan Morduch, Jean Lee and Hassan Zaman).

Conference:

IGC Conference on Financial Inclusion in Bangladesh, 2016.

Synopsis:

Providing financial access to poor and excluded population is still a challenge for developing countries. Though the innovation of microcredit has demonstrated a great promise to extend the credit access to the poor, the outreach of micro-credit is by far limited and the basic banking services are still not accessible by most of the people in developing countries. For instance, Bangladesh which is well-known for its micro-credit revolution still has 76% of its population unbanked.[1] This vast unbanked population typically does not have access to basic financial necessities like savings, deposit and remitting facilities. Especially households with casual, infrequent and irregular wage earnings, such services are very important for their fluid operation of income fluctuations as well as smoothing consumptions, and risks. Typically households that depend on remittances, generally receive money through hand or informally through friends or local transport bus drivers. This process is quiet expensive, fraught with delays, and involved substantial losses due to theft. Moreover, due to this unreliability of existing remittance network, the frequency of remittance is also very limited which also creates suffering at the receiving end.

Commercial Banks, by and large, find it very expensive to have branch level operation at the remote rural areas simply because of the fact that the accumulated deposit and loan amount in these areas do not usually generate sufficient revenues to meet the expenses. Setting up ATMs in rural areas is not feasible due to lack of infrastructural support (for example lack of electricity). As a result financial inclusion of unbanked population by commercial banks is still a big challenge and an important policy objective for the developing country policy makers.

However, the remarkable rise of mobile phone users in developing countries has given a unique opportunity to extend the banking services to the unbanked population through M-Banking.[2] M-Banking simply requires an agent, mobile hand-set and a connection to perform financial transactions. This service is easy, reliable and doesn't require branch level banking (for example see the success story of M-PESA service introduced by Safaricom in Kenya). In Bangladesh at present two mobile based financial transactions facilities exist, one is operated by Dutch-Bangla Bank Limited (DBBL) which is branded as “Mobile Banking” and the other one is operated through the BRAC Bank Limited (BBL) which is branded as “bKash”[3]. Both these services have wide coverage and nation-wide agent points to do the transactions at real time using any mobile phone connection from any operator.[4] The cost of opening up an account and other transaction costs related to access for different financial services (like cash-in, cash-out, deposit, transfer of money, etc.) are very comparable between these two service providers including the accessibility and usability of the system.

Though the current the operation of Mobile-based Banking (hereafter Mobile Banking) is only limited to money transfer system, however, one could easily imagine the potential of such powerful and easily accessible service to advance in many different directions: for example it could be a doorway of financial inclusion for unbanked population, the possibility of evolving a different medium of transaction through e-cash or even the possibility of replacing the current banking practice, among others. Nevertheless, even at the current level of services offered for Bangladeshi consumers, mobile banking services could dramatically change the socio-economic environment of the typical unbanked population.

Literature:

The literature on the impact of Mobile Banking is relatively new and growing. Pioneering the study on the impact of MBB on Kenyan households is done by Morawczynski and Pickens (2009), and Morawczynski (2009) followed by Jack and Suri (2010, 2011) and Mbiti (2011). Morawczynski in her series of qualitative work using ethnographic fieldwork, observed some interesting financial behavior changes by the M-PESA users; like higher frequency of small remittances, using of cash-in service as a deposit, etc. Using similar method, Plyler et al. (2010) argued that M-Pesa has enabled small businesses to develop in rural communities. While these studies provide suggestive evidence of the impacts of MBB service, however, they are generally unable to quantify the impact. Quantitative or empirical study on this field of study so far has been done by Jack and Suri (2010, 2011) and Mbiti (2011). Empirical study using sub-location data, these series of quantitate work showed that M-Pesa improves the ability of households to smooth risks, lowers the propensity of people to use informal savings mechanisms and increases the probability of being banked. However, there has not been any study that has used household level micro data to evaluate the impact of MBB services.

Research Aim and Objective

The goal of this research is to understand the adoption of and impacts of mobile banking among urban migrant workers and their rural originating households in Bangladesh using a field experiment. The experiment will randomly vary the provision of training and incentivized offers to adopt mobile banking among migrant-household pairs. Our primary outcomes of interest include the following:

(1) Take-up and use of mobile money: We will study factors influencing demand for mobile money accounts, and measure the extent to which take-up of mobile money is influenced by provision of a basic training module on how to enroll in the service and a small monetary incentive to begin using the service. We will also collect administrative as well as survey data to track the use of mobile money after enrollment by study participants. Based on a small pilot of the training and incentive conducted in March 2014, we anticipate that takeup of mobile accounts in our study population will be very high. We plan to further explore determinants of the adoption decision, using both survey and experimental methods to better understand why adoption decisions are so responsive to the provision of small subsidies.

(2) Transfers and risk-sharing within and across households, especially in response to shocks: We will collect survey data on the incidence of shocks at the household level, and estimate the extent to which the impact of these shocks on household consumption and other outcomes are mitigated by access to mobile banking, either through increased household saving or through the faciliation of sending and receipt of transfers.

(3) Remittances by migrants: We will collect both administrative and survey data to track the impact of facilitating transfers through mobile money enrollment on remittances sent by migrant workers to their originating households, particularly through seasonal lean periods.

(4) Other outcomes related to financial access, including education and labor market outcomes: We will collect monthly work histories to see whether financial access in this form affects other outcomes, such as investments in children's education or labor market decisions. In particular, we will measure whether enrollment in mobile banking affects off-farm agricultural or non-agricultural wage labor, and whether it also affects migration decisions, as reflected in the duration of migration and the quality of the job obtained.

Study Design:

Our experiment will take place in and around Gaibandha, a district in northwest Bangladesh, and in Dhaka, the capital city of Bangladesh. Gaibandha is in Rangpur, one of the poorest regions of Bangladesh, with exposure to a seasonal famine in September through November known as the monga. Even measured outside of the monga season, Rangpur has significantly lower rates of food consumption per capita than other regions in the country (Bryan et al, forthcoming).

In order to target the issues of extreme poverty exhibited in Rangpur, the United Kingdom’s Department for International Development (DfID) included it in the set of eligible populations for its set of SHIREE projects, aimed at addressing extreme poverty. Through one of our partner organizations, a non-governmental organization known as GUK, DfID recently implemented a garment workers training program which was targeted towards the ultra-poor, even among those in Rangpur. The program, which has trained approximately 1100 workers to date, consisted of six to eight weeks of training in a fully equipped training facility at the GUK headquarters. Trainees were then assisted in finding jobs in the garment sector in Dhaka. These jobs are competitive and pay well relative to daily agricultural labor. The base salary in most salaries is 3500 Taka (approximately 47 dollars) per month. Workers are offered more generous rates for overtime work, and typically earn between 6000 and 8000 taka (80 and 107 dollars) per month in total. Most trained workers migrate to Dhaka at least initially following training, although there is some episodic remigration to Gaibandha in our sample. They thus constitute a population of likely migrants.

Our study overall will consist of six phases, conducted in cooperation with both GUK and MoMoDa Solutions, a local survey consultancy. Research assistance throughout will be provided by Niamot Ali and Rokonuzzaman Khan, two University of Dhaka master's degree graduates in Development Studies. First, we will recruit individuals from the GUK/DfID project to participate in our study. Second, we will conduct a baseline survey. Third, we will assign household/migrant pairs to either the treatment group or the control group, and will conduct the adoption experiment among individuals in the treatment group, as described in more detail below. Fourth, we will conduct an interim follow-up survey at three to four months following the initial intervention to look at impacts of adoption of mobile banking accounts. Fifth, we will conduct a longer-run endline survey at 8 to 9 months following the initial intervention. Finally, at the time of the longer-run endline survey we will conduct a survey experiment to test whether individuals have different perceptions of mobile money and cash.

The first phase of our research activities will be the tracking and consent phase. In this phase, we will work with existing staff of GUK to locate and contact individuals trained under the GUK/DfID SHIREE project and their originating households. GUK maintains a record of the locations (union and village) of all originating households of individuals trained by the SHIREE project and this list will be used to contact the originating households, in which parents and spouses of trainees reside. Parents and spouses will be asked for the contact information of trainees who have migrated to Dhaka for work. Households and individuals will be visited by our field staff, informed about the content of the study, and asked whether they would consent to participate in the study. If rates of current migration to Dhaka in our proposed sample are low, we plan to modify our sample to include current and more recent trainees of a second wave GUK garment worker training program that are very likely to migrate in the near future.

The second phase of our research activities will be the baseline survey. Once the universe of participating households and trainees has been identified, we will conduct an extensive baseline survey to track changes in outcomes from before to after our intervention. The rural baseline survey will consist of a module asking about the age and demographic information of all individuals resident in the household, including education; the work information of all members of the household; agricultural activities and investment; health; access to savings and credit; remittances; and risk and time preferences. The urban/trainee baseline survey will include similar modules, but will also include a detailed work history by month asking about migration, conditions of work and pay rates, and overall earnings.

The third phase of our experiment will be to assign individuals to treatment or control, and to conduct the mobile banking adoption experiment. One significant aspect of the adoption experiment follows the methods used in Bursztyn et al (forthcoming) to empirically estimate both the effects of information transmission through social networks and direct behavioral externalities or peer effects in social networks. Individuals may both learn about the value of a new product through social contacts, and their valuation may directly depend on the adoption decisions of their contacts. After this phase of the adoption experiment, we propose to in a surprise visit, revisit all treatment households and migrants and offer them the training and incentive to adopt. Based on our pilot, we anticipate that final rates of adoption will be very high.

The fourth and fifth phases of the experiment are an interim follow-up survey and an endline survey. These surveys will collect information very similar to or the same as the information collected in the baseline survey, covering the age and demographic information of all individuals resident in the household, including education; the work information of all members of the household; agricultural activities and investment; health; access to savings and credit; remittances; risk and time preferences; the incidence of household level shocks; and detailed work histories. Individuals will receive a small amount of compensation for their participation in each survey wave. Data from the survey will be supplemented with administrative data from bKash on the use of accounts by individuals in our survey. This will give us detailed and accurate data on savings, transfers and remittances.

The sixth and last element of our study will test whether perceptions of and decisionmaking using mobile money differ from those in cash. At the time of the endline survey, we will administer a short module on risk preferences, time preferences, and willingness-to-pay for different goods.

We will evaluate the impact of training and the subsidy on adoption of mobile banking accounts by comparing adoption rates for treatment and control households in a regression framework. Results from our adoption experiment will provide further insight into adoption decisions. We will similarly estimate effects on savings and remittances by comparing outcomes for treatment and control households. Finally, we will estimate effects on risk mitigation by first assessing whether transfers to treatment households affected by adverse shocks are greater than those to control households affected by adverse shocks, and by testing for differential impacts of shocks on consumption in the two groups.

[1] The prediction comes from a study conducted by the Boston Consulting Group (BCG) for Telenor in 2011 on the socio-economic impact of mobile financial services. For web link click here.

^[2] The basic idea of M-Banking is very simple. The purchase and transfer of "talk-time" (i.e., pre-paid cell phone credit) to others via SMS is already well established, which is long distance transfers of stored talk-times. M-Banking formalizes this by creating e-money balances that can be converted to cash through registered agents and can be accessed and transferred by SMS.

[3] With a recent limited introduction of M-Cash by Islamic Bank Limited.

[4] DBBL has 22019 listed agents whereas BBL has over 28,000 agents (based on the information available from their respective homepages.)