Breadwinning Gender Norms: A Field Experiment in Urban India

We vary wages on hypothetical jobs we offer to be in the vicinity of their husband’s wage.

2023-10-04

2024-03-31

Women’s take-up decision of hypothetical jobs we offer in a phone survey. We measure their interest at different points in time, including before and after they have had time to speak with their husbands.

We want to study how women's job preferences/ labour supply changes as we offer them jobs at different wages. This will allow us to understand how women’s take-up decisions for different jobs change in relation to their husband’s income. We will also study their changes in job preferences before and after bargaining with their husbands.

We also measure information on gender attitudes and beliefs and whether or not the husband’s income varies month to month.

We elicit the participant's gender attitudes and beliefs through a follow-up survey.

The study will target participants who are married and have recently graduated from our partner organization’s vocational training program. The partner organization will provide us with the list of trainees.

We will contact the trainees and ask them if they are interested in completing a short survey. If they are interested, we will conduct the first phone survey over the phone.

First Phone Survey
The participants will be answering some demographic questions at the beginning of the survey. Women who are no longer married or estranged from their husbands and women whose husbands aren’t working or decline to provide their income will be screened out and the survey will end for them.

Following the demographics section, we will ask the wife to put her husband on speaker phone and collect the husband’s demographics in the presence of the wife. We will elicit the husband’s income in front of the wife. Subsequently, we will continue collecting additional demographic information from the wife. We will then elicit the wife’s interest in applying for new job opportunities. For our main analysis, we will drop participants who initially reject all jobs. We will also drop participants for whom half of husband’s wage is more than the highest real job wage, or four times the husband’s wage is less than the lowest real job wage (at a given center). If they are interested we will explain to them the placement opportunities and the process we will be using to implement their choices. We will also explain to them the compensation they will receive for participating in the survey.
We will go through the placement opportunities one by one. We will offer up to 30 hypothetical jobs and 10 real jobs.

We randomize wages on hypothetical job offers to be between 0.5 and 1.5 times the husband’s income, which we use to test whether women avoid jobs that pay slightly above their husbands.

As secondary analysis, we will randomize:
Treatment- whether we WhatsApp the job details to husbands and ask them to discuss them
Control- whether we Whatsapp the jobs to the women and ask them to discuss the job opportunities with their husband and family

This will allow us to test whether decisions change after involving the husband. This is because it is important to try to capture not just initial decisions but what happens after the family is involved (as they will be when people actually have to attend jobs). We will exclude couples where the husband declines to provide their phone number.

Before ending the survey, we will confirm with the wife whether she wants to make any changes with regard to her interest in job opportunities and record the changes (if any). We do this before sending any of the jobs to their husband.

We conclude the survey by fixing a time for a follow-up call 2-3 days later.

Second Phone Survey
After 2-3 days we will recontact subjects at the agreed upon time. We will re-elicit the trainees’ job preferences as we go back through each job they expressed interest in and ask if they are still interested in applying for the same.

Third Phone Survey
The research team will then analyze job preferences from the phone surveys. We will randomly select one job opportunity for each trainee. If the trainee has applied for that job, they are allocated the job. If multiple trainees are allocated to the same job and are interested in that position, we will randomly allocate one to the job.

Surveyors will then contact trainees to ask them a few follow-up questions to elicit gender norms which include questions on gender attitudes and a question on whether problems would arise in their marriage if they (the trainees) earned more than their husband.

We conclude the phone survey by notifying them of the placement outcome, connecting them with the placement officers if they have been allocated a job and sending them their compensation for participating in the survey.

Analysis
We will test for a discontinuity or differential change in elasticity of the wife’s labor supply in the vicinity of the husband’s income.

We will analyze heterogeneity by gender attitudes and beliefs, including breadwinning norms attitudes, second order beliefs of peer attitudes, and purdah/ghunghat/pallu practices. These characteristics may be related to the size of any treatment effects. We will also analyze heterogeneity by whether or not the husband’s income varies month to month, which may affect the relevance of comparing the offered income to the husband’s income.

Randomization done by Qualtrics

Treatments are randomized at the individual level

Yes

400 - depending on our ability to recruit participants. Expected participants 300.

12,000

No treatment arms in the main specification. Randomizing wages above and below the husbands income - 6000 observations above and 6000 below.

Calculating power for a regression discontinuity at the husband’s income, with power of 0.8 and significance-level of 0.05, and 300 participants and 30 observations per person we have a minimum detectable effect of 6.9 percentage points.

Analysis Plan