Scott Alexander of Slate Star Codex asked in his blogpost “Invisible Women” why the entry of women into the American labour force doesn’t seem to show up in the time series of GDP. Their introduction should look like a big bump, should it not ? But there’s really no mystery about any of it.
( This is very belated. I meant to post it some time ago, but I got side-tracked with foreign policy blogging in July and completely forgot about this in the draughts folder. I was going to throw it out but then I just noticed several people were still citing that blogpost as a means of implying maybe women didn’t have much of an impact on GDP. )
The comments section of Alexander’s blogpost is a veritable favela of error, confusion and misunderstanding, bloated with unnecessary hypotheses and explanations.
The shortest answer to the non-mystery is this : in 1950, both the labour force participation of American women was low, and the average hours of those women who were working were also low. By 2000 working men and women averaged about the same number of hours, but still a smaller percentage of women than of men worked in the first place.
This is not a theory, as much as something tautologically true, by simple accounting of output and hours.
Below is a plot of normalised real GDP versus total hours worked for the period 1950-90, both values set at 1950=100. The GDP data are from the BEA, and the hours worked data are extrapolated from this paper and from the Census Bureau. For simplicity’s sake, the hours data are smoothed from five observations (1950, 1960, 1970, 1980, 1990).
Real GDP a little more than quadrupled in those 40 years, whereas total hours worked in the economy went up from 175 billion hours to less than 310 billion, or less than double. The difference between the quadrupling and the near-doubling is what is called growth in labour productivity, or output per man-hour of work. Between 1950 and 1990 the percentage of total hours supplied by women went up from ~25% of the total to ~40%.
The above is true by accounting or by tautology.
The part that’s not tautological or knowable from mere accounting is the contribution of women to the productivity growth. There must be research on the topic but I haven’t looked. But you can still just eyeball the distribution of hours worked by age and sex (from the previously cited paper) and infer quite a lot :
During those 50 years, the labour force participation of old people and of young people fell, presumably in response to better retirement options and more educational opportunities, respectively. To the extent that these exits from the labour force were being matched by the entry of prime working-age women, the age-distribution effect alone would have helped boost productivity.
