Occupational Stigma and Worker Migration

Last registered on August 10, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Occupational Stigma and Worker Migration
RCT ID
AEARCTR-0019290
Initial registration date
August 02, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
August 10, 2026, 9:05 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name
Varun K
Affiliation
Harvard University
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Chinmaya Kumar
PI Affiliation
Azim Premji University
Contact Investigator

Additional Trial Information

Status
On going
Start date
2026-07-27
End date
2027-12-31
Keywords
Behavior, Labor
Additional Keywords
Migration, Social Stigma
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Many low-income workers remain in low-return local labor markets despite large spatial wage gaps, while others migrate for jobs that appear unattractive on purely pecuniary grounds. Standard accounts explain migration through wages, employment risk, liquidity constraints, and networks. We study whether occupational stigma also shapes the spatial allocation of labor. Manual work such as construction may carry a social penalty. The same work may be less socially costly when undertaken where the worker is not known or observed by their peers or reference groups. Migration may therefore be a way of escaping social stigma associated with certain jobs.

We test this mechanism in a field experiment with male manual laborers in villages near Patna, Bihar where participants evaluate the same construction job—holding the task, hours, and duration fixed—at three locations that differ in social visibility. Because each worker evaluates all three locations, the design holds constant his underlying distaste for construction work and uses within-worker differences in job choices to measure the value of performing the work away from his local social audience. This allows us to estimate the effect of occupation stigma on migration choice of workers.
External Link(s)

Registration Citation

Citation
K, Varun and Chinmaya Kumar. 2026. "Occupational Stigma and Worker Migration." AEA RCT Registry. August 10. https://doi.org/10.1257/rct.19290-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
Male manual workers in rural Bihar are presented with a series of real, incentivised job offers for construction work. The offers are described one at a time and the respondent accepts or declines each on its own terms. Offer attributes are varied experimentally across and within respondents. Choices are consequential: at the end of the module one offer is selected at random, and for a randomly chosen subset of respondents that offer is implemented as an actual referral to a contractor at the stated wage. A separate module elicits community norms regarding manual labour from households.

Intervention Start Date
2026-07-27
Intervention End Date
2027-08-31

Primary Outcomes

Primary Outcomes (end points)
Offer-level take-up: an indicator equal to one if the respondent accepts a given job offer and zero if he declines it. Each respondent contributes multiple offer-level observations.

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
1) Ranked preference over the three jobs (best-liked and second-best), elicited after the choice block.
2) Reasons volunteered for accepting or declining each offer, from pre-coded lists with an open "other" field.
3) Perceived probability that the respondent's own village would learn of the work, elicited separately for each location on a 0–10 scale.

4) Perceived difficulty, non-wage benefits, and expected pay reliability for each location.
Secondary Outcomes (explanation)
The ranking outcome is coded as two indicators: whether the respondent ranks a location other than his own village first, and whether his own village is ranked last of the three. These are informative for respondents who accept every offer, a group for whom binary take-up carries no information.

The reasons outcome is coded as one indicator per response option, plus two constructed flags: whether a respondent accepting an offer volunteered concealment or social judgment as a reason, and whether a respondent declining an offer volunteered being seen, taunts, or community talk. The option lists differ by location, so the flags are built from the location-appropriate codes.

The visibility outcome enters both as a level per location and as a within-respondent difference relative to his own village, which measures how much concealment each location buys in his own estimation.

Experimental Design

Experimental Design
Households are sampled from villages in the Patna region of Bihar. A short screening module classifies each household by whether any member performs manual or daily-wage labour. Households containing such a worker receive the job-offer module described above; households without one receive a norms module only.

Within the job-offer module, three attributes are randomised by computer prior to fieldwork: the sequence in which offers are presented, which single offer becomes binding, and whether the binding offer is implemented as a real referral. Assignments are preloaded into the electronic instrument and are not visible to the enumerator or the respondent before the relevant point in the interview.
Experimental Design Details
Not available
Randomization Method
Randomisation done in office by computer prior to data collection, using a statistical package. Assignments were generated at the household level in the sampling frame and preloaded into the SurveyCTO instrument, so that assignment is fixed before contact and cannot be altered in the field.

Randomization Unit
Individual respondent (household). All three randomised attributes — presentation sequence, which offer becomes binding, and whether the binding offer is implemented — are assigned at the household level.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
1,350 households, drawn from 25-30 villages. Randomisation is at the household level, so households are both the unit of randomisation and the unit of clustering.

Sample size: planned number of observations
1,350 households: approximately 900 containing a manual labourer, who receive the job-offer module, and approximately 450 without one, who receive the norms module only. The 900 experimental households yield up to 5,400 offer-level observations (3 locations × 2 wage levels per respondent), which are the unit of analysis for the primary outcome.
Sample size (or number of clusters) by treatment arms
Presentation order: 450 households per sequence (2 arms), among the 900 experimental households.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Harvard University
IRB Approval Date
2026-07-17
IRB Approval Number
N/A
Analysis Plan