Predicted probability of Mexican women and Mexican men participating in the labour force
depending on the share of agricultural, industrial and service jobs in the municipality where they live. 

Claudia Goldin showed that female labour participation rates (FLPRs) tend to peak in low-income countries where agriculture predominates, decline in middle-income countries with a growing industrial sector, and rise again in high-income countries with a service-oriented economy. The main hypothesis to explain the decline of FLPRs in middle-income countries is that this pattern is driven by the expansion of industrial activities in some regions of these countries, as jobs in this sector are typically performed by men. In the first chapter of my doctoral thesis I evaluate that specific hypothesis of the U-shaped feminization theory using subnational data. The analysis uses microdata from Mexico, the Latin American country with the highest share of industrial jobs and one of the lowest FLPRs in the region. 

The empirical strategy relies on probit regressions to estimate women’s likelihood of participating in the workforce depending on the share of jobs in agriculture, industry, and services in the municipality where they reside. The analysis is based on a repeated cross-sectional dataset derived from household surveys conducted between 2005 and 2019. 

Surprisingly, the results are not in line with the hypothesis. The figure presented above shows that women’s likelihood of participating in the workforce increases as the share of jobs in the industrial sector is higher, while it declines if they live in municipalities with a higher percentage of agricultural jobs. The data suggests that lack of labour demand is one of the main reasons why women living in agricultural regions of Mexico are not working.

Comparative results between men and women also reveal interesting patterns. In the case of men, their likelihood of being in the workforce does not vary significantly depending on the sectoral distribution of employment in their place of residence. Their probability of participating in the labour force consistently exceeds 90%, regardless of whether they live in agricultural, industrial, or service-oriented regions of the country. In contrast, the likelihood of women’s participation changes substantially depending on the share of jobs in agriculture, industry, and services in their place of residence. Their probability of participating in the workforce typically hovers around 30% if they live in municipalities where agriculture predominates, but can increase to 60% if they reside in regions with a service-oriented economy. 

The results were obtained after considering both paid and unpaid work in farms and businesses, and they hold after controlling for individual, household, and municipal characteristics. Hence, one of the main contributions of this research to the literature is its demonstration that the sectoral distribution of employment significantly influences women’s likelihood of participating in the workforce, while it does not play a decisive role for men.