Wage Determinants in Global Apparel Value Chains: Insights from Ethiopia and Kenya

As global brands have expanded their supply chains, apparel production has increasingly shifted to low-wage countries in the Global South. For many developing economies, this shift has generated new employment opportunities, increased export earnings, and spurred the development of industrial infrastructure. However, despite rising investment and production volumes, wages in the apparel sector remain strikingly low, particularly in sub-Saharan Africa, raising important concerns about equity and sustainability.

These challenges are closely tied to the structure of the global apparel value chain. In buyer-driven chains, global lead firms exercise significant control over design, pricing, and delivery timelines. Suppliers must meet tight deadlines and low-price points to remain competitive. This pressure is especially acute in apparel assembly factories that only carry out the 'cut-make-trim' (CMT) steps in apparel production and thus are paid only for the labor time involved in assembling the garment, checking it and packing it for shipment. Many of the apparel suppliers of big brands and retailers are transnational, meaning that they have their headquarters in the home country, which is usually in Asia, and have factories in countries across the global South. Most of these factories are just CMT factories, especially in sub-Saharan African countries, although some factories in southeast Asia carry out additional functions such as sourcing textile or even producing it. To cut costs, factories that depend mostly on the labor-intensive sewing, checking and packing functions may adopt strategies that suppress wages and increase pressure on workers through more demanding labor conditions.

The research being carried out by the Creating and Capturing Value project finds that even under similar global pressures, wage outcomes in CMT factories can differ across countries, and thus national context matters. Both Ethiopia and Kenya prioritize apparel-led industrialization and attract foreign investment but diverge in key ways. Ethiopia follows a state-led model with tightly managed industrial parks, no statutory minimum wage, and weak labor enforcement. Kenya, by contrast, has a more liberalized economy with stronger labor institutions, wage-setting frameworks, and union representation. These institutional differences, alongside global cost pressures, shape the wage gaps observed between countries, as well as within a supplier country where segmented labor markets exist.

This blog draws on new empirical findings from our research examining what determines wages in the apparel export sectors of Ethiopia and Kenya. By comparing how global dynamics intersect with national labor regimes, we explain why workers performing similar jobs often receive vastly different pay across borders, and even within the same country.

How we modeled and analyzed wages

We examined the determinants of wages in Ethiopia’s and Kenya’s apparel export sectors using a linear multilevel regression model based on data from over 900 workers across 28 factories. To address the skewed distribution of earnings, wages are modeled in their natural logarithmic form. Given that workers are nested within firms and firms are nested within countries, we incorporated random intercepts at the firm and country levels to capture unobserved heterogeneity. Parameters were estimated using maximum likelihood estimation (MLE), with robust standard errors to correct for potential heteroskedasticity and model misspecification. We estimated three models: (1) separate models for Ethiopia and Kenya to identify country-specific wage patterns; (2) a pooled model without interactions to assess shared determinants across contexts; and (3) a pooled model with interaction terms to test for country-level variation in predictor effects. This multilevel approach allows us to unpack how wages are shaped by individual attributes, firm-level dynamics, and broader national contexts.

Figure 1 breaks down the sources of wage variation among apparel workers in Ethiopia and Kenya using a multilevel statistical model. Each bar represents a country-specific or pooled model. The colored segments show the proportion of total wage variance attributable to three different levels. The results from this figure explain the selection of the linear multilevel model in general and specifying both country-specific and pooled models in particular.

Figure 1: share of wage variance by worker, firm and country-level variables

Key Findings

Our analysis reveals several important patterns that help elucidate why some workers earn more than others in Ethiopia’s and Kenya’s apparel export factories. Here's what stood out:

1. Gender still matters

Men earn on average 7.6% more than women in Kenya, even when accounting for job roles and other observable factors. A comparable gap (8%) appears in the pooled model, while the male wage gap in Ethiopia is not statistically significant. Though more pronounced in Kenya, gender pay gaps in both countries reflect structural wage inequality in the apparel industry.

2. Age brings higher pay—especially in Ethiopia

Older workers earn more in both Ethiopia and Kenya, but the effect is substantially stronger in Ethiopia. The estimated age-related wage premium is approximately 1.5% higher in Ethiopia, suggesting that experience or seniority is more strongly rewarded in its apparel export sector.

3. More education, better wages

Each additional year of schooling is linked to higher wages in both countries, underscoring that education pays, even in low-wage sectors like apparel. In Ethiopia, one extra year of schooling is associated with a 1.6% wage increase, compared to 1.0% in Kenya and 1.4% in the pooled sample. While returns to education appear higher in Ethiopia, the difference is not statistically significant, pointing to broadly similar education-related wage gains across both contexts.

4. Job role matters—a lot

Line supervisors earn more than sewing machine operators across both countries. In Ethiopia, workers in finishing and quality control roles receive modest wage premiums, while in Kenya, these roles are associated with slightly lower pay. Wages for packing and cutting positions vary without a clear pattern. However, the job-role effects are not statistically significant overall, suggesting broadly comparable wage structures across different job categories in both contexts.

5. Contracts and training: a mixed bag

In Ethiopia, workers with verbal and renewable agreements earn more than those with formal and fixed contracts. That is unusual and could reflect hidden risks or informal rewards. On-the-job training boosts wages in Ethiopia, but has no clear impact in Kenya.

6. Firm-level effects: Bigger and unionized is better

Larger factories and those with collective bargaining agreements tend to pay more in both countries. However, the presence of a union or compliance certificate alone does not guarantee better pay. Notably, foreign-owned factories tend to offer lower wages across both countries, a finding that challenges common assumptions about foreign investment and wage standards.

7. Kenya pays more—much more

Kenyan apparel workers earn substantially higher wages. Even when comparing individuals with similar characteristics employed in comparable factories, Kenyan workers earn over three times as much as their Ethiopian counterparts, a wage premium of approximately 212%. Figure 2 below depicts the predicted average log wages of apparel workers in Ethiopia and Kenya after controlling for key worker-level characteristics and factory-level differences suggesting that country-level factors play a substantial role in shaping wage outcomes in the two countries.

Figure 2: predicted take-home wages by country

8. Wages vary widely between factories and countries

Wages vary markedly across factories, particularly in Ethiopia. Some firms pay significantly more than others. In addition, a substantial portion of wage variation is attributable to national context, underscoring the role of country-level differences. These patterns highlight the need for advanced statistical approaches to better understand the determinants of wage disparities.

How we explain the key findings:

The stark wage gap between Ethiopian and Kenyan apparel workers reflects deeper structural differences in labor laws and labor market institutions. The substantial wage premium observed for Kenyan workers, even after controlling for individual and factory-level characteristics, can be attributed to more protective labor legislation, the existence of sector-specific statutory minimum wages, stronger enforcement mechanisms, and more robust labor institutions. In contrast, Ethiopia’s state-led industrial strategy has prioritized low labor costs, enabled by the absence of a statutory private-sector minimum wage, weak enforcement capacity, and limited worker representation in wage-setting. These institutional divergences fundamentally shape wage outcomes across the two countries’ export-oriented apparel sectors.

The gender-based wage gaps observed align with research documenting persistent disparities in the global apparel sector. Women are concentrated in low-paid roles due to gender norms, discrimination, weak enforcement of equal pay, and the preference for 'feminized' labor in repetitive tasks. Age-related wage gains likely reflect proxies for experience and reliability, where formal promotion paths are limited. The positive link between education and wages shows even basic schooling signals skills and adaptability in an otherwise low-skill sector.

Findings on firm size and collective bargaining imply that larger factories and those with collective bargaining agreements tend to offer higher wages, likely due to greater resources and external accountability. However, the limited and uneven impact of collective bargaining agreements and compliance certifications underscores that formal systems alone are insufficient without strong worker representation and effective enforcement.