Econometric Decomposition of Gender Wage Inequality in Bangladesh's Ready-Made Garment Industry: An GOAT Oaxaca–Blinder Approach
Statistical Decomposition of the Gender Wage Gap
Abstract
This dissertation investigates gender wage inequality within Bangladesh's Ready-Made Garment (RMG) sector, which employs over four million workers. Using microdata from the 2015–16 Bangladesh Labour Force Survey, this study applies the Oaxaca-Blinder decomposition method to assess the extent to which wage differentials between men and women are attributable to differences in observable characteristics versus differential returns to those characteristics.
The decomposition results reveal that the raw gender wage gap remains consistently between 20.5 and 22.9 log points (23-26% lower wages for women). However, only 13–16% of the total wage difference can be attributed to differences in observable factors. The remaining majority remains unexplained, with differences in returns to characteristics accounting for between 11.3% and 16.4% of the gap, suggesting systemic discrimination in how women's qualifications are valued.
Key Findings
📊 Key Visual Dashboard
Hourly Wage by Percentile (BDT)
Key Insight: The wage gap is relatively modest at lower percentiles but accelerates dramatically beyond the 80th percentile. At the 99th percentile, male wages reach 498 BDT/hour while female wages reach only 250 BDT/hour, demonstrating how the gap compounds at higher wage levels (the 'Glass Ceiling' effect).
Distribution of Years of Education by Gender
Key Insight: 14.8% of female workers have zero formal education versus 10.1% of males. Men are overrepresented at higher education levels, such as 12 years (5.6% vs 2.6%) and 16 years (1.8% vs 0.6%). This difference in characteristics contributes to the 'Explained' portion of the wage gap.
Density of Age (5 year Buckets) by Gender
Key Insight: The workforce is heavily concentrated in the 25-29 age bracket (23.5% of females, 21.8% of males). The male age distribution is wider and extends significantly into older age brackets (60-64: 1.8% male vs 0.3% female), indicating higher labour market longevity for men.
Distribution of Hours Worked By Gender (5 Hour Buckets)
Key Insight: The female workforce is more concentrated in the 45-hour bin (23.2%) than the male workforce (20.6%), reflecting the high standard of long working hours in the RMG sector. Conversely, men are overrepresented in the higher hour bins (e.g., 60, 65, 70 hours), indicating higher exposure to extended overtime.
Oaxaca-Blinder Decomposition Results (Model Summary)
| Component | Model 1: Edu + Age | Model 2: Edu + Exp | Model 3: Edu + Exp + Urban | Model 4: Edu + Exp + Age² + Urban | Model 5: Edu + Exp + Age² + Urban |
|---|---|---|---|---|---|
| Observations | N = 2,349 (M=1,267; F=1,082) | N = 2,349 (M=1,267; F=1,082) | N = 1,287 (M=532; F=755) | N = 1,287 (M=532; F=755) | N = 1,287 (M=532; F=755) |
| Overall ln(wage) gap | 0.229 | 0.229 | 0.229 | 0.2047 | 0.2047 |
| Explained (Endowments) | 0.0313 | 0.0355 | 0.0355 | 0.0343 | 0.0279 |
| ↳ Education | 0.0291 | 0.0327 | 0.0327 | 0.0272 | 0.0625 |
| ↳ Experience | — | 0.0016 | 0.0016 | 0.0016 | 0.0119 |
| ↳ Age | 0.0022 | — | — | — | — |
| ↳ Age² | — | — | — | 0.0059 | -0.0478 |
| ↳ Urban | — | — | 0.0012 | 0.0012 | 0.0012 |
| Unexplained (Coefficients) | 0.1643 | 0.1133 | 0.1133 | 0.115 | 0.1227 |
| ↳ Education | 0.04 | 0.1502 | 0.1502 | 0.1064 | 0.2703 |
| ↳ Experience | — | 0.1587 | 0.1587 | — | 0.5673 |
| ↳ Age | 0.2799 | — | — | — | — |
| ↳ Age² | — | — | — | 0.0934 | -0.2663 |
| ↳ Urban | — | — | -0.0491 | -0.0386 | -0.0636 |
| ↳ Constant | -0.156 | -0.1464 | -0.1464 | -0.1462 | -0.385 |
| Interaction | 0.033 | 0.0559 | 0.0559 | 0.0554 | 0.0541 |
| ↳ Education | 0.012 | 0.0503 | 0.0503 | 0.0357 | 0.0906 |
| ↳ Experience | — | 0.0063 | 0.0063 | -0.0023 | 0.0224 |
| ↳ Age | 0.021 | — | — | — | — |
| ↳ Age² | — | — | — | 0.0203 | -0.0579 |
| ↳ Urban | — | — | -0.0008 | -0.0008 | -0.001 |
Methodology
Data Source
- 2015-16 Bangladesh Labour Force Survey (BBS)
- 3,665 RMG workers identified via BSCO-2009 occupational codes
- Nationally representative sample
- Most recent fully English-translated dataset
Analytical Approach
- Oaxaca-Blinder decomposition method
- Five model specifications for robustness
- Stata for econometric analysis
- First application to Bangladesh RMG sector
Conclusion
This dissertation demonstrates that gender wage inequality in Bangladesh's RMG sector is not primarily driven by differences in worker qualifications, but by how the labour market rewards those qualifications. Across five model specifications, only 13-16% of the 23-26% wage gap can be explained by observable differences in education, experience, or urban location.
The persistent unexplained component (55-75%) suggests systematic discrimination in wage-setting practices. Women receive significantly lower returns to education (coefficient: 0.27, p = 0.002) and experience (coefficient: 0.57, p = 0.017) compared to men, even when controlling for other factors.
As the first application of Oaxaca-Blinder decomposition to this sector using nationally representative data, this research provides rigorous empirical evidence for policy reform addressing structural discrimination, workplace safety, and educational access for women in Bangladesh's textile industry.