A field experiment in 24 Bangladeshi garment factories randomly gave equally "marginal" men and women supervisor roles. Women started behind on ratings due to worker bias, closed the gap in ~4 months, and were retained above baseline.
~4 months
Time for performance/evaluation convergence between female and male supervisors
~80%
Share of RMG production-line workers who are women
~95%
Share of line supervisors who are men (baseline)
24
Number of factories in the study
Details
Promoter
University of Warwick, University of Oxford & University of Dhaka research team, with Better Work Bangladesh-affiliated garment factories
Period
2015-2020
Keywords
garment & apparel manufacturing, human resources, gender equality
Context
In Bangladesh's ready-made garment (RMG) sector, roughly 80% of production-line workers are women but about 95% of line supervisors are men — a gap long assumed to reflect a shortage of qualified female candidates.
Objectives
The study aimed to isolate statistical discrimination from real skill differences by testing whether 'marginal' female candidates — equally qualified to their male counterparts on baseline skills diagnostics — could succeed as line supervisors when given the chance.
Activities
Researchers from the University of Warwick, University of Oxford and University of Dhaka worked with 24 large garment factories to identify 'marginal' candidates for supervisor promotion — men and women whose baseline skills diagnostics showed no meaningful difference — and randomly assigned them to manage real production lines.
Results
At the outset, the newly promoted female supervisors received significantly lower productivity readings and worker evaluations than their male counterparts, driven by workers' negative prior beliefs about women's management ability rather than an actual skills gap; after roughly four months, performance and evaluations converged. The share of female candidates still working as line supervisors after the trial was significantly higher than the baseline share of female supervisors in the same factories, and the experience shifted male workers' expectations about promotion and tenure.
Conclusions
The published results describe averages across the participating factories rather than a single headline productivity percentage, and effects on turnover and wages beyond the trial period are not tracked in the publicly available working paper.
Implementation
Indicative cost
Low (< €50k)
Time to results
Long (> 3 years) — Field experiment ran 2015–2020 across 24 factories, with performance convergence observed after approximately four months of a supervisor's tenure.
Staffing & skills
University of Warwick, University of Oxford & University of Dhaka research team, Participating garment-factory management/HR (24 factories), Better Work Bangladesh-affiliated garment factories
Conditions for success
Rigorous baseline skills diagnostics to identify genuinely 'marginal' candidates of both genders
Random assignment of equally-qualified candidates to real production-line supervisor roles
A sustained ~4-month period for worker perceptions and performance evaluations to converge
Common failure modes
Initial worker bias led to significantly lower productivity readings and evaluations for female supervisors despite no real skill gap
Effects on turnover and wages beyond the trial period were not tracked in the published working paper
Replication kit
Reusable artefacts from this practice — as published by their sources.
Since 2004, Toronto's TRIEC Mentoring Partnership has logged 27,000+ mentoring matches for skilled immigrants; an independent, federally funded evaluation found …