Digitizing G2P Payments in Bangladesh

The primary intervention, initiated by the government of Bangladesh, rolls out digital payments to a random selection of households receiving social services benefits. The control group consists of households that continue receiving payments through the pre-existing cash payment method. Additional cross-treatments test the relative impact of a financial literacy campaign, differing frequencies of payments (as determined by the government), and an advance schedule for payments.

2019-04-01

2022-04-01

1. Impact of digital payments on payments infrastructure: We will examine whether guaranteed transactions extend the payments infrastructure by collecting individual-level data on access to digitized payment (confirmed receipt of funds), whether the account is active (at least one transaction in survey month), the number of transactions, and the quantity of funds in the account, as well as agent-level data on the number of accounts opened monthly, the number of clients the agent sees, and the daily volume of transactions processed.

2. Impact of digital payments on broader DFS usage: We will examine whether digital G2P payments lead to broader DFS usage by collecting individual-level data on the magnitude of formal vs informal savings and credit, recipient trust in banks and mobile banking, and the number of family and
 community members engaged in DFS, as well as agent-level data on the number of non-G2P clients serviced, the number of cash-out transactions, and the demand for credit, insurance, or new bank accounts.

3. Impact of digital payments on financial inclusion for women: We will examine whether women specifically benefit from digital G2P transfers using individual-level data on bank account access, financial literacy, number of transactions, likelihood of suffering from a consumption shock (in health, marriage, education, agriculture), bargaining power in the household, and occupational choice, each disaggregated by gender; in addition to agent-level data on the gender composition of clients and the likelihood of interacting with a female client.

4. Optimal payment design: We aim to optimize the design of Bangladeshi social support payments by examining the varied timing of payments and whether clients receive reminders or information and collecting individual-level data on the regularity of financial activity as well as persistence in activity.

5. Impact of digital payments on corruption: We will measure the impact of digization on corruption, leakage, and targeting by collecting individual-level data on whether the transfer was received and the size of transfer, how quickly the transfer was obtained (transaction time and costs), and the variance in transaction costs, as well as agent-level data on the amount received from the government (using administrative data) and amount received by intended vs unintended clients (to measure leakage and targeting).

The study will involve a baseline survey and an endline survey 12 months later, permitting four rounds of digitized payments to be processed in the intervening period. The study will take place across 16 districts selected by the Bangladeshi government, with 4,000 DSS-eligible households across 122 subdistricts (approximately 33 households per subdistrict) and 2,250 agents across 122 subdistricts (approximately 18 agents per subdistrict).

The randomized experimental arms will be as follows:
Control (61 subdistricts including 2,000 households): No intervention; households continue to receive payments in cash directly from banks at the subdistrict center, according to the status quo schedule for the benefits they receive.
Treatment – Digitization program (61 subdistricts including 2,000 households): Households will have bank accounts opened and will be able to access payments through either local union digital centers (UDCs) or mobile money.
Cross-treatment I – Financial literacy campaign (1,000 households within Control and 1,000 households within Digitization Treatment): Households will receive a 30-minute training session on mobile banking. The one-on-one training will take place in the household, directly between the surveyor and the participant. The surveyor will teach the participant how to use the mobile money platform (BKash), then run three practice transfers of funds between the surveyor’s mobile phone and the participant’s.
Cross-treatment II – Frequency of payments (1,000 households within Digitization Treatment): Households will receive the same yearly total amount in regularly scheduled monthly transfers.
Cross-treatment III – Payment schedule provision (1,000 households within Control and 1,000 households within Digitization Treatment): Households will receive a brief monthly text message providing the schedule of future payments for the remainder of the year (e.g., “Your next payments will be delivered on the following dates: MM/DD/YYYY, MM/DD/YYYY, and MM/DD/YYYY.”).

Baseline and endline surveys will be administered at two levels:
1. Household-level survey: The survey will collect information on household demographics, mobile phone usage, banking usage, attitudes towards banking systems, status and experience with benefits transfers, and social agency and socioeconomic status.
2. Agent-level survey: The survey will collect data for a random sample of financial service access points (an “agent” can be a mobile banking agent, a bank branch employee, a banking correspondent, a post office worker, an MFI or a co-op employee, or any individual granted permission to transfer G2P payments).

randomization done in office by a computer

Randomization of the Digitization Treatment will take place at the subdistrict level. Randomization into cross-treatments will take place at the household level.

Yes

122 subdistricts

4,000 households and 2,250 agents

Control: 2,000 households
Digitization Treatment: 2,000 households
Cross-treatment I - Financial Literacy: 2,000 households (1,000 within Control and 1,000 within Digitization Treatment)
Cross-treatment II - Frequency of Payments: 1,000 households within Digitization Treatment
Cross-treatment III - Payment Schedule Provision: 2,000 households (1,000 within Control and 1,000 within Digitization Treatment)