Intra-household Decision-Making and Women’s Labor Market Choices

This project examines how husbands influence women’s employment choices. First, using a discrete choice experiment, the study elicits wives’ and husbands’ preferences over workplace characteristics relevant to the wife’s employment (wage, gender composition of co-workers, transport, skill development, and children allowed in the workplace). Each spouse privately evaluates eight hypothetical job offers for the wife; each compared to an outside option of “no job”. This stage measures individual preferences and the extent to which husbands’ and wives’ preferences for the wife’s job align or diverge. Second, couples are brought together to make joint decisions over the same job offers, allowing the study to analyze whose preferences dominate when the couple decides together. Third, the study tests whether low-cost information interventions can increase husbands’ support for their wives’ employment.

2025-11-11

2026-01-17

This study has three main families of primary outcomes, corresponding to the three core research questions:
1. Preferences over job characteristics (Stage 1 – individual choices)
2. Bargaining outcomes in joint decisions (Stage 3 – couple choices)
3. Workshop approval and participation (Stage 4 – policy intervention)

1. Preferences over Job Characteristics (Stage 1 – Individual Choices)
In Stage 1, each spouse separately evaluates eight hypothetical job offers for the wife, each compared to the outside option of “no job.”
For each respondent i and job offer j, I observe:
• Wife’s individual choice: binary indicator equal to 1 if the wife accepts the job offer and 0 if she rejects it.
• Husband’s individual choice (for his wife): binary indicator equal to 1 if the husband wants his wife to accept the job offer and 0 if he prefers that she rejects it.

These binary indicators are the main dependent variables in the Stage 1 analysis. Using standard discrete-choice models, the probability of accepting a job is estimated as a function of the experimentally varied job attributes.

Primary outcomes from Stage 1:
1. Spouse-specific preferences over job attributes
o For wives and husbands separately, I estimate how the different job attributes affect the probability that the wife accepts a job.
2. Preference gaps between wives and husbands
o I will compare husbands’ and wives’ preferences over job attributes and over the alternative-specific constant, which captures the average preference for the “no job” option relative to the offered jobs (holding all job attributes constant), and test whether they differ significantly.

2. Bargaining Outcomes in Joint Decisions (Stage 3 – Couple Choices)
In Stage 3, after completing the individual tasks, the couple is reunited to evaluate the same set of job offers together. For each job, they discuss briefly and make a joint decision about whether the wife would accept or reject the job.
For each couple i and job j, I observe:
• Wife’s private choice from Stage 1 (accept = 1, reject = 0)
• Husband’s private choice from Stage 1 (accept = 1, reject = 0)
• Joint choice from Stage 3 (accept = 1, reject = 0)

Primary outcomes from Stage 3:
o Alignment between joint and individual choices: the extent to which the joint decision matches each spouse’s private choice when they initially disagree. This outcome indicates whose preferences are more likely to prevail during joint decision-making
o Relative bargaining weight on the wife’s versus the husband preferences: a bargaining-weight parameter of how much influence the wife’s preferences have in joint decisions compared to the husband’s.

3. Workshop Approval and Participation (Stage 4 – Policy interventions)
In Stage 4, following the randomized policy treatments targeted at husbands, the study measures husbands’ support for the wife’s participation in an employment-related workshop.
Primary outcome:
• Husband’s approval of workshop participation: binary indicator equal to 1 if the husband states that he wants his wife to be invited to the workshop on women’s employment and local career opportunities, and 0 otherwise.

See above

As secondary outcomes, I will collect:
• Social norm threshold for workshop approval (Stage 4 – policy interventions):
To measure conformity to social norms, husbands will report how many men in their neighborhood they believe would allow their wives to attend the workshop on women’s employment, and how many of these men would need to say “yes” before they themselves would allow their wife to go. A lower threshold indicates weaker perceived social constraints. I will test whether the information treatments reduce this threshold.
• Wife’s workshop participation (Stage 4 – policy interventions):
Measured as actual attendance at the workshop, conditional on the husband’s approval. This captures realized behavior among the subset of wives whose husbands consented to the invitation.

Heterogeneity Analyses
I will conduct heterogeneity analyses to examine how individual preferences and bargaining outcomes vary with key baseline characteristics.
• Heterogeneity in individual preferences (Stage 1):
I will relate estimated preferences over job attributes, and the implied probabilities of job acceptance, to characteristics such as demographic factors (e.g. age, education, number and age of children) and gender norms and attitudes toward women’s work. Analyses will be conducted for wives and husbands.
• Heterogeneity in bargaining outcomes (Stage 3):
I will examine whether bargaining outcomes vary with characteristics such as preferences for marital harmony, proxies for the wife’s bargaining power (e.g., education, age gap), type of marriage and household (arranged vs. love marriage; joint vs. nuclear household), gender norms, and job characteristics (e.g., childcare availability, gender composition, wage level).
Additional Robustness and Measurement Checks
I will conduct a number of robustness checks such as I will control potential biases related to enumerator influence, social desirability, comprehension, and respondents’ confidence about their choices.

See above

The study aims to document how spouses evaluate different workplace characteristics for the wife’s employment, how joint decisions are made, and which policy interventions can increase husbands’ support for women’s work.

Not available

Randomization is conducted by computer. After completing the individual and joint choice tasks, couples are randomly assigned (simple randomization) to one of three arms:
1. Control (no information),
2. Legal Protections video,
3. Household Welfare Framing video.
Other randomization procedures (within the DCE):
• The DCE consists of 3 blocks; block assignment is randomized at the couple level.

The unit of randomization is the couple. After completing individual and joint choice tasks, each couple is randomized into one of three groups:
1. Control (no information),
2. Legal protections video,
3. Household welfare framing video.
Although the script is delivered to the husband, the couple is treated as the unit of assignment since subsequent outcomes are measured at the couple level.
Within the DCE, block assignment is also randomized at the couple level.

No

Couples will be randomly allocated to one of the three policy treatments.

Given the budget constraints and required logistics, I expect to reach about 1,000 married couples, which translates into 2,000 individuals. I will sample 1,000 households across 6 Nagar Panchayats, selecting 7 wards per Nagar Panchayats (42 wards in total), targeting ~24 households per ward.

I plan to survey around 1,000 married couples in semi-urban areas in the district of Lucknow. Randomization is at the couple level for the information treatment, with couples assigned evenly across three arms:
• Control (no information): around 334 couples
• Treatment 1 (Legal Protections): around 333 couples
• Treatment 2 (Household welfare framing): around 333 couples
Total: 1,000 couples

Sample size adequacy for the DCE (Stage 1) Using Orme’s rule-of-thumb, N≥(500×c)/t×a with c=5 attribute parameters, t=8 job choices, and a=2 alternatives, the minimum required sample is about 156 respondents. The planned sample of 2,000 individuals (each with 8 tasks) far exceeds this, indicating ample precision to estimate preferences over the main job attributes. Stage 1 – Preference differences (acceptance probabilities) For each spouse, I define an acceptance share as the fraction of the 8 offers accepted. At the couple level, I consider the gap: wife’s acceptance share minus husbands. Under conservative assumptions (baseline acceptance probability of 50% and no cross-spouse covariance), the standard deviation of the couple-level gap is 0.25. With 1,000 couples, a two-sided 5% test with 80% power can detect a difference in the mean gap of roughly 0.022 (≈ 2.2 percentage points). Stage 3 – Joint alignment (whose preference the joint decision follows) In disagreement cases (where spouses’ private choices differ), I define an indicator equal to 1 if the joint decision follows the wife and 0 if it follows the husband. Under equal bargaining power, the alignment rate is 50%. If couples disagree on about 25% of 8,000 potential couple–job observations, I observe roughly 2,000 disagreement cases. Treating this as a one-sample proportion with baseline 0.5, power calculations indicate that, at 5% significance and 80% power, the MDE is about 0.031 (≈ 3.1 percentage points), i.e. detecting a shift from 0.50 to about 0.531. Stage 4 – Information treatments and workshop outcomes Couples are randomized into three arms (control, legal information, household welfare framing), with ~333 couples per arm. The main binary outcomes is husband’s approval of inviting his wife to the workshop. 1. Husband approval of invitation o Baseline approval rate assumed at 50% (SD ≈ 0.50) under conservative assignment. o With 333 couples in control and 333 in a treatment arm, a two-sample proportion test (5% significance, 80% power) can detect a treatment–control difference of about 0.10–0.11 (≈ 10–11 percentage points). Overall, the planned sample provides strong power for detecting modest differences in wives’ and husbands’ job acceptance patterns, in joint bargaining outcomes, and in the effects of information treatments on husbands’ approval and wives’ workshop attendance.