Please use this identifier to cite or link to this item:
Gaddis, Isis
Klasen, Stephan
Year of Publication: 
Series/Report no.: 
Courant Research Centre: Poverty, Equity and Growth - Discussion Papers 71
A large literature claims that female labor force participation (FLFP) follows a U-shaped trend over the course of economic development. This feminization U hypothesis is motivated by secular patterns of structural change in combination with education and fertility dynamics. We show that empirical support for the hypothesis is rather feeble and hinges on the data used for the assessment. The PWT 7.0 revision of international GDP estimates paints a completely different picture of the relationship between aggregate GDP and FLFP than the previous PWT 6.3, with the U coming out much stronger under PWT 7.0 than under PWT 6.3. The feminization U also tends to vanish if we use dynamic instead of static panel data methods. Moreover, differences in levels of FLFP across the world related to historical contingencies are much more important determinants of women’s employment opportunities than the muted U patterns found in some specifications. Given the large margins of error in international GDP estimates at purchasing power parities (PPP) and the sensitivity of the U-relationship we propose an alternative way to explore the effect of structural change on FLFP. We use data on sector-specific growth, which do not require PPP comparisons and allow for a direct test of the effect of structural change on women’s economic activity. Our results suggest that agriculture, mining, manufacturing and services generate different dynamics for FLFP, but the effects are small in magnitude. We conclude that the feminization U hypothesis, especially its declining portion, has little relevance for most developing countries today.
Female Labor Force Participation
Economic Development
Structural Change
Purchasing Power Parties
Document Type: 
Working Paper

Files in This Item:

Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.