Much has been written about the increase in the female labor supply, attributing it to increases in productivity in home production, women's emancipation, a drop in fertility, availability of daycare, the increasing importance of brains over brawn and the narrowing of the gender wage gap. But what can explain the current disparities in female labor market participation across regions of the United States. Obviously the urban or rural nature of the region will matter. But there are larger disparities even across the large US metropolitan areas. For example, the participation rate for high school educated women is 52% in New York City while it is 78% in Minneapolis.
Dan Black, Natalia Kolesnikova and Lowell Taylor conjecture and verify that this has to do with commuting times. The evidence is overwhelming. Not only do cities with longer commuting times have lower female participation, this participation increased more slowly in agglomerations where commuting times increased faster.
These results highlight once more that the United States should reevaluate very seriously its transportation policy. Increasing commuting times not only reduce welfare because people have less time for leisure or work, but it turns out it basically prevents women from working. It appears that men compensate this by working more when their wife is not working, but this does not strike me as Pareto improving.
Thursday, January 29, 2009
Subscribe to:
Post Comments (Atom)
2 comments:
You mention the reasons the literature brought forward for the increase in the female labor force participation. Yet none of them seem to appear in the regressions of this paper. If, for example, daycare costs are higher in NYC than Minneapolis, and you do not take that into account, would this not bias the results they obtain?
This paper has a rather elaborate model (for labor economics, at least) that you fail to mention. But given this model, the empirical work is rather underwhelming. I think the empirics basically only capture correlations, but certainly not causation. Too much is left out compared to the model.
Post a Comment