Promoting digital payments technology among female migrant factory workers

2020-01-27

2020-02-05

1. Use of digital payment applications for remittances
In the primary surveys, we ask whether the respondent uses digital payment applications to remit money. The variable is coded as 1 if the respondent reports using any of the following channels to remit money: mobile money, UPI-based digital payments app, net banking/bank application, and 0 otherwise.



2. Effectiveness of training sessions
During the training sessions, we collect data on whether the participant can send money to someone’s UPI id. It is coded as 1 if she can send money to someone’s UPI id during the session, and 0 otherwise.

During the training sessions, we also collect data on whether the participant can send money to someone’s bank account similar to the modus operandi to remit home through digital payment platforms. It is coded as 1 if she can send money with someone’s bank account during the session, and 0 otherwise. We will use this as a robustness check.

3. Quantification of the extent of the different barriers:
We collect the following indicators from the participants during the training sessions, which are all the necessary steps enabling participants to send and receive money:
1) Whether the participant attended the training
2) Whether the participant has her mobile phone linked to her bank account
3) Whether the participant has mobile balance to send verification SMS from her phone
4) Whether the participant has the required elements (ATM card, mobile phone and SIM card)
5) Whether the participant was able to download the app
6) Whether the participant was able to set up her UPI id

4. Use of digital payment applications for non-remittance purposes

We collect data on whether the respondent uses her phone for the following tasks:
1) to make bill payments, not for goods
2) to buy goods
3) to receive money from someone living in the city
4) to receive money from someone living outside of the city
5) to repay money she had borrowed
6) to lend money to someone

We will create an outcome variable equal to one if any of the above indicators is 1 and 0 otherwise.

The eligible population for the study is female garment workers at Shahi Exports Private Limited, Karnataka, India, who live in 19 selected Janodaya hostels (Janodaya is the NGO managing migrant hostels linked to Shahi factories). Our sample consists of 819 migrant women in these 19 hostels, all of whom use smartphones.
There are two treatment arms and one control arm. Both treatment arms include information on digital payments with facilitated sign-up, opportunities for practice, and financial incentives.

Randomization was carried out at the individual level, stratified by units and reported mobile phone registration at baseline. Within each stratum, a third of the sample was randomized into the soft intervention treatment group (n=273), a third of the sample was randomized into the intensive intervention treatment group (n=273), while the remaining third of the sample in a control group (n=273) did not receive any intervention.

The unit of analysis is the individual/respondent. The primary unit of treatment is also the individual. We chose to conduct randomization at the individual-level as we believe spillover risks are very limited, considering the barriers at stake.

The eligible population for the study is female garment workers at Shahi Exports Private Limited, Karnataka, India, who live in 19 selected Janodaya hostels (Janodaya is the NGO managing migrant hostels linked to Shahi factories) and own a smartphone. These hostels were sampled out of convenience and availability but reflect the overall population of migrant workers in Shahi garment industry in the Karnataka region. Our sample consists of 819 migrant women in these 19 hostels, all of whom use smartphones.
There are two treatment arms and one control arm. Both treatment arms include information on digital payments with facilitated sign-up, opportunities for practice, and financial incentives. In the “Classroom treatment” arm, the batch size is between 20 and 30 participants per trainer; in the “Individualized treatment” arm, one trainer works with up to 5 participants. Participants in both treatment arms also receive two reminder SMS messages.

Randomization was carried out at the individual level, stratified by units and reported mobile phone registration at baseline. Within each stratum, a third of the sample was randomized into the soft intervention treatment group (n=273), a third of the sample was randomized into the intensive intervention treatment group (n=273), while the remaining third of the sample in a control group (n=273) did not receive any intervention.

The unit of analysis is the individual/respondent. The primary unit of treatment is also the individual. We chose to conduct randomization at the individual-level as we believe spillover risks are very limited, considering the barriers at stake.

With only two treatment arms, this impact evaluation will not provide evidence on the relative effects of each intervention type (training sessions and SMS reminders). We expect the bundle (training plus SMS reminder) to have a stronger effect than smaller combinations (training sessions and SMS reminders alone). We do not expect any negative complementarities between the interventions.

Randomization took place using the statistical software package Stata. We conducted randomization using the following procedure:
1) Collect baseline data from hostels: Prior to randomization, we conducted baseline data collection in 19 hostels, saturating each hostel
2) Stratified random assignment by unit and mobile registration dummy: Within each stratum, we randomly assigned a third of the sample to control, a third of the sample to the soft intervention and a third of the sample to the intensive intervention. In case the total number of individuals in a stratum is not divisible by 3, we randomly decided whether the last individual (and second-last individual) was in either of the treatment arms or control.

We conducted randomization once, after baseline was completed.

Randomization took place at the individual level

No

819 participants (migrant women)

819

273 (equally divided)

10pp

Financial Inclusion : promoting digital payments among migrant factory workers