The Real Effects of Electronic Wage Payments: A Field Experiment with Salaried Factory Workers in Bangladesh

We are working with two textile factories, a large domestic bank, one of Bangladesh’s leading mobile payments providers, and a mobile phone operator to test the effects of moving salaried garment workers from cash to electronic wage payments. We randomly assign participants to receive wage payments through a no-frills bank account, a mobile money account, or to keep receiving wages in cash. All treated workers will receive a bank account or a mobile money account, and training on how to use the account. We will then track the effects of these accounts on workers’ savings, consumption and credit use. We will also measure their use of formal and informal financial services, including vehicles for sending remittances.
In addition, we will run a follow-up experiment with a subset of the sample. All of these workers will receive a salary bonus of 1000 taka, randomly assigned to be given a week before pay day or the same day as pay day. We will examine the effect of the timing of the transfer on workers' consumption and borrowing behavior. This will provide a measure of worker liquidity constraints.

2015-02-05

2016-12-31

The primary outcomes of interest for this study are changes in savings, consumption, credit and remittance behavior among workers. In particular, we will look for consumption-smoothing and/or increased long-term savings. We expect that having access to a formal account may lead to decreased temptation spending among treated workers. Access to mobile money systems could have an effect on the ease and frequency with which workers are able to remit money. These variables will be tracked over time through several follow-up surveys relating to these topics. The high-frequency follow-up surveys will provide measures of variance for:
• Total savings and vehicles used
• Monthly consumption on durable and non-durable goods
• Remittances and within-household transfers
• Use of credit
• Large expenditures
• Meals skipped and end-of-month strategies
In addition, we will measure other traits during our baseline survey and again in our endline survey. These include:
• Attitudes towards and comfort with financial institutions
• Measures of IQ and numeracy
• Frequency of financial shocks that respondent cannot pay for
• Job satisfaction
• Mobile money systems are easily accessible in Bangladesh, but are not always used as intended. We will track whether the participants are able to use these products correctly before and after the intervention by asking questions about how they use these services.

• Treatment 1: Receive a bank account; monthly wage payments paid into the account.
• Treatment 2: Receive a bank account; monthly wage payments in cash.
• Treatment 3: Receive a mobile money account; monthly wage payments paid to the account.
o Treatment 3a: A subset of workers without a phone will also be given a phone with their account and sim card.
• Treatment 4: Receive a mobile money account and sim card; monthly wage payments in cash.
o Treatment 4a: A subset of workers without a phone will also be given a phone with their account and sim card.
• Control: Monthly wage payments in cash (status quo)
To implement the proposed intervention, we will draw from a sample of approximately 3000 urban workers located at two factories in Dhaka, Bangladesh, all of whom currently receive their monthly wages in cash. The workers will be randomly assigned to one of six treatment groups or the control. Workers in the control group will continue to receive cash wage payments. Workers in the first treatment group (T1) will receive a no-frills bank account in their name, along with training on how to use the account and will be paid using electronic wage payments into the account. Workers in the second treatment group (T2) will receive a bank account and training about how to use the account, but will continue to receive their wages in cash. Workers in the third treatment group (T3) will receive a mobile money account, along with the training necessary to use the account, and will be paid using electronic wage payments into the bank account. Workers in the fourth treatment group (T4) will receive a mobile money account, along with the necessary training, but will continue to receive wage payments in cash.
In addition, we will randomly provide mobile phones to participants in the mobile money account treatment groups (T3a and T4a) who do not already own their own mobile phones. Workers in T3 who do not have their own mobile and are not randomly assigned to receive one, will be allowed to opt-in to the mobile payment scheme.

Our study will be carried out in four phases: In the first phase, we will administer an extensive baseline survey. In addition to information on demographics, household income and consumption, the baseline will include modules on formal and informal credit, financial literacy, savings, savings goals, credit, time and risk-preferences and risk-sharing and remittance networks.
After baseline is complete, we will begin the process of opening bank and mobile money accounts for treatment groups. Second, we will administer several follow-up surveys that ask about monthly consumption, remittances, savings, and credit use to all study participants. These high-frequency consumption surveys will be repeated on an ongoing basis over the entire duration of the study and will provide measures of the variance of consumption faced by workers. Third, we will conduct an extensive end-line survey, approximately nine months after the last accounts have been opened. The end-line will follow up on characteristics measured in the baseline and recurring surveys and will pay special attention to changes in patterns of borrowing, including demand for high cost informal loans, saving, remittances and consumption that could have been affected by the intervention. We will also measure worker job satisfaction. Administrative data from the factories will allow us to measure attendance to accurately track attrition and other measures of job performance.
Finally, in the fourth stage, we conduct an additional experiment on short-term savings and consumption smoothing. All endline participants at one factory will receive a participation bonus, announced for all workers at the same time. By varying the timing of the receipt of this bonus, we can measure the extent to which workers are financially constrained at the end of the month and whether previous access to electronic wage payments affects the level of financial constraints. We randomly assign workers to one of two treatment groups, and stratify by EWP treatment:
T1: Receive bonus one week before pay day
T2: Receive bonus on pay day
A Short follow-up survey with treated workers will be conducted in the days immediately before and after pay day in which we measure the use of informal and formal credit, financial shocks, food and non-food consumption. We also conduct evaluations of worker performance with factory supervisors. Supervisors rate the workers on various standards of performance and efficiency. Each worker will be evaluated twice: mid-month, and directly before pay day. This will allow us to detect variation in performance at the end of the month caused by financial constraints.

We randomize at the individual worker unit. This is the level of randomization that maximizes statistical power, and the unit of analysis at which we observe key outcomes of interest. We obtained a list of all workers currently employed at each of our partnering factories; from this list we have excluded workers who are not directly involved in production activities (such as cleaners, nurses, truck loaders etc) and full-time administrative staff and managers. From this sample frame, we have randomized into six groups plus the control, stratifying on tenure, factory location, gender, and mobile phone ownership where applicable. We randomize using STATA.

We randomize at the individual worker unit stratifying on tenure, factory location and gender.

No

The total sample size is approximately 3000 workers.
The sample size for the consumption smoothing experiment is approximately 500 workers.

The total sample size is approximately 3000 workers at 2 factories.

T1 (bank account with electronic wage payments): 850
T2 (bank account with salary in cash): 200
T3 (mobile account with electronic wage payments): 725
T3a (mobile account with electronic wage payments and mobile phone): 125
T4 (mobile account with salary in cash): 350
T4a (mobile account with salary in cash and mobile phone): 50
Control: 700

Consumption Smoothing:
T1: (receive bonus before payday): 250 workers - stratified by EWP treatment
T2: (receive bonus on payday): 250 workers - stratified by EWP treatment