This paper (pdf) came out about two years ago, by Siwan Anderson and Debraj Ray, in the highly esteemed Review of Economic Studies, yet I haven’t seen it discussed in the blogosphere, so here goes:
Relative to developed countries and some parts of the developing world, most notably sub-Saharan Africa, there are far fewer women than men in India and China. It has been argued that as many as a 100 million women could be missing. The possibility of gender bias at birth and the mistreatment of young girls are widely regarded as key explanations. We provide a decomposition of these missing women by age and cause of death. While we do not dispute the existence of severe gender bias at young ages, our computations yield some striking new findings: (1) the vast majority of missing women in India and a significant proportion of those in China are of adult age; (2) as a proportion of the total female population, the number of missing women is largest in sub-Saharan Africa, and the absolute numbers are comparable to those for India and China; (3) almost all the missing women stem from disease-by-disease comparisons and not from the changing composition of disease, as described by the epidemiological transition. Finally, using historical data, we argue that a comparable proportion of women was missing at the start of the 20th century in the United States, just as they are in India, China, and sub-Saharan Africa today.
That is a very different interpretation from what one usually hears.