Last week's Catherine Rampell piece in the NYT looking at the increased amount of long-term unemployment in the most recent recession took as its main human interest an unemployed 52 year old office worker Cynthia Norton. This motivated me to look at female employment/population ratios. I have tended to be more interested in the male data because I think it's a clearer signal of the long-term threat of automation and globalization to society. For women, the primary trend in the twentieth century was the move of women from the household economy into the formal workforce, which I take to be a mixture of some complex of cultural and technological change. The obvious technological changes are the spread of labor-saving devices (washing machines, dishwashers, microwaves, etc) allowing the household economy to be more productive and thus freeing women to take work outside the home, as well as more reliable contraception, allowing women to control their fertility and thus spend a smaller fraction of their lives in child-rearing.
The graph above summarizes the long term changes.
I guess the most striking thing is that perhaps the tide has turned. Prior to 2000, recessions (such as in 1980, 1982, and 1991) tended to cause only a small pause in the increase in female employment, which then resumed rising following the recession. However, the last two recessions have caused clear outright declines in female employment in the US, and at least after the 2001 recession, the female employment population ratio didn't recover.
Thus, arguably, the same forces affecting male employment have now started to affect female employment too. The potential confound is again cultural - anecdotally, it appears to me that families in the last ten years have become more aware of the attachment needs of small children, and this may also account for some of the shift in the trend away from rising female employment. Quantitatively, it's hard to know.
Unfortunately, the BLS doesn't break out female employment by educational level, and there is not too much more I can make out of the available data.