Sunday, August 10, 2014

Yes, women's labor force participation matters

A bunch of social-conservative types on Twitter keep telling me that women's labor force participation doesn't matter for GDP growth. As evidence they keep linking me to this post by Scott Alexander. Here's Scott's case:
[W]hy did the entry of women into the workforce produce so little effect on GDP? Here’s the graph of US women’s workforce participation: 
And here’s the graph of US GDP. 
In about 50 years – 1935 to 1985 – we went from 20% of women in the workforce to 60% of women in the workforce. Assuming a bit under 100% of men in the workforce, that’s an increase of almost 50% over the expected trend if number of women in the workforce had stayed constant. 
I’ve already admitted I find the fixedness of the GDP trend bizarre. But how on Earth do you unexpectedly raise the number of people in the workforce by 50% and still stick to exactly the same GDP trend? 
The idea here is that the GDP growth trend looks so smooth that it has to be a law of nature. But that's certainly not the case for other countries. For example, here's a graph for Japan, on a log-log scale:

If Japan's growth rate were constant, that line would be straight. Instead, it's concave, meaning that growth slowed substantially over time. So we know that countries' economic growth rates are not fixed laws of nature - they can be affected by events.

So what were some of the events that affected U.S. GDP growth over the last century? Here's total labor force participation in the U.S. from 1948 to 2014:

As you can see, there was a big increase between around 1968 and around 1989. A lot of that was women entering the workforce (and men not leaving).

Now what else affects GDP? Population, but U.S. population growth was pretty constant. How about productivity? David Beckworth has a good graph:

We see that from the late 1960s through the mid-1900s, productivity stagnated noticeably. So the gains from increased labor force participation was one of the things that counteracted the productivity slowdown at that time.

In other words, when Scott asks:
Are we imagining here that if women hadn’t entered the workforce, GDP would have suddenly deviated downward from the trend, and totally by coincidence women rushed in to save the day?
The answer is: Yes!

(Now, one might wonder if TFP and LFPR are related. When more people work, does productivity go down? Interesting question. Economists usually assume the two are unrelated. And there are definitely lots of episodes in which the two move in the same direction instead of canceling out - Japan in the 1990s is one of these, since female labor force participation and TFP both abruptly flattened out in that decade. But some people have argued that higher employment slows down TFP in some situations. Suffice it to say that this is a minority viewpoint, but can't be ruled out.)

Actually, there have been a number of economists who have looked at a bunch of factors and tried to isolate the effect of female labor force participation on growth. Here is a summary of much of the relevant research. The conclusion is that yes, female labor force participation does contribute substantially to per capita GDP. The puzzle Scott presents us with is actually no puzzle.

Scott also asks this question:
And then when we ran out of interested women to add to the work force, again by coincidence the GDP stabilized back to its trend line?
The answer is: No. Growth has been slower since women stopped joining the labor force. The trend is your friend til the bend at the end! Scott's GDP graph is zoomed-out, so it's hard to see the "bend at the end", but it's there. There was an acceleration of productivity growth in the late 90s and early 2000s, which - again - partially canceled out the end of women's entry into the labor force. But since then, growth has been slower; there has been no return to the growth rates of the 1960s or even the 1990s.

Now, a couple of caveats. Caveat 1 is that "home production" is not counted in GDP, but matters for living standards. Caveat 2 is that women's labor force participation can affect fertility (by decreasing it or by increasing it), which may affect long-run TFP growth, and will certainly affect total GDP by changing the population size.

But the idea that women's entry into the labor force has been unimportant for GDP is not correct. My advice to social conservatives looking to give economic justifications for traditional gender roles: Stop linking to this post.


  1. Understanding the discontinuity in the slope of the TFP chart around 1973 is perhaps the greatest public policy challenge of our age.

    We know that chemistry and physics made huge leaps in the 19th century that took over a hundred years to fully implement. I think we know that the great depression held back GDP even as technology continued to advance. I think we know that the US government and private industry (Bell, IBM, Xerox, GE, Boeing) poured enormous sums into R&D during the 1950s and 1960s. It was probably inevitable that GDP growth would slow down once the technological backlog had been caught up and the low hanging fruit picked. The most recent world changing invention was the microprocessor - 45 years ago. The internet is important but: it arose from government research in the 1960s and appeared in the mid 1980s - thirty years ago (according to Wikipedia).

    If we want to increase GDP growth we probably need to increase science funding while being realistic and recognizing that we are mining ever thinner "ore".

    1. To show just how much technical advance has slowed down look at the SR71 - the fastest airplane that ever flew. It took sixty years to go from the Wright Brothers to the SR71. The SR71 first flew in 1964. Fifty years ago. It is doubtful we will see its records broken in the next ten years.

    2. Anonymous5:11 PM

      That really isn't a very good example of the overall picture.

    3. The focus on technology is overdone. There was also trade and currency policy. Germany and Japan came online. Demographics. The 80s saw the end of the baby boom entering the work force as the lump passed through the snake. The Fed tried to compensate but there was still "stagflation." Again, overblown. Then politics and policy turned to the right in the 80s.

    4. @Peter Ultimately the underlying driver of improvements in TFP In America is technology. The technical advances between 1850 and 1950 are responsible for the rise in the western standard of living. The factors you refer to may have some transient influence. The chart of TFP shows a marked discontinuity in the slope in 1973. Something happened around that time. The question is whether that something can be identified and reversed.

    5. @anonymous I used the SR71 because it is such a cool looking airplane and because it represents such a sudden stop in progress FIFTY years ago. Overall progress has continued but the effective pace has slowed down. The F18s that are bombing Iraq are a thirty year old design. The microprocessor, the defining invention of our time, was invented in 1970. Where is the invention that will have the impact on life 10, 20, 30 or 40 years from now that the microprocessor has already had?

    6. Anonymous11:03 AM

      There is no need to trouble yourself about the discontinuity starting in 1973 . . . US Peak Oil. The unprecendented human expansion, economic or otherwise, has been an oil driven phenomenon. There are those who believe that alternative technologies are coming to save the day. I am not one of them.

    7. "We know that chemistry and physics made huge leaps in the 19th century that took over a hundred years to fully implement. "

      How do we know this?

    8. How do we know this?

      By 1865 we had Perkins (chemistry), Maxwell(electro-magnetism) and Bessemer(cheap steel). They each opened a whole new world of possibilities and directly consequential inventions. In the case of Bessemer, cheap steel made possible internal combustion engines which made possible airplanes. Electromagnetism made possible electrical lighting and power, not to mention television and radio. Chemistry made possible synthetic fertilizers and the green revolution of the 1960s. It took until the late 1960s before we had caught up with the opportunities that flowed directly from the state of scientific knowledge in 1865.

  2. Anonymous3:46 PM

    Looks demographically related to the Boomers. Once the drain ends by the 2020's and the Echo's stabilize things, you will understand.

  3. I think they are linked, more labor means lower labor costs and less incentive to invest in technology, only when women stopped increasing, China was there to take their place but that doesn't count as gdp. When we run out of new sources of labor and labor costs rise, it will then become worthwhile to develop and adopt new technology. Not there yet, but we may be approaching that.

  4. As you can see on your graph, labor force participation went from only 59% to 63%. All these women entered the labor force, but it's not like that made it so a much higher percentage of people were working. This was probably countered a lot by an aging population and much more retirees, among other things.

  5. "Caveat 1 is that "home production" is not counted in GDP, but matters for living standards."

    Right, i.e., the increase in GDP from women working probably overstates the improvement in living standards, because a homemaker doing work isn't counted as contributing the same amount as how much someone would pay a housekeeper to do the same.

    But obviously it increases GDP, and almost certainly there is still some net improvement, or else they wouldn't work. (Well, there is the case that people might be letting their home standards deteriorate because they need money to spend on other necessities.)

  6. Hmmm. Women getting into the workforce increases GDP if their new shiny salaries that they can spend on things aren't cancelled out by a decrease of the men's salaries, no?

    I am a feminist so, as such, I am happy women got the choice of being financially independent. But, as a worker, I am less convinced. We always mention technology and globalisation as factors to explain the stagnating salaries for the bottom 50% of the population. I suspect LFPR also played a role...

  7. The material feminists had long argued that women's work needed to be moved from the home into the marketplace and valued accordingly. They were in favor of child care, cleaning services, and food prepared in public kitchens. They were generally considered insane radicals. Nowadays, a huge percentage of all children do spend their time in day care, not at home or running around the neighborhood, more and more people do hire cleaning services or are doing a lot less cleaning around the house, and more and more meals are prepared in restaurants, supermarkets, and factories rather than in home kitchens. All of these tasks, formerly performed by women for "free", were added to the GDP as this kind of production was moved out of the home and into the economy.

    This type of work, since it was performed by women, is generally undervalued, just as anything a woman does is undervalued in the marketplace. What is the dollar value of a baby or teaching someone to use the toilet? Priceless in the sense of being beyond valuation or perhaps priceless in the sense of having zero value? If nothing else, these were all low value service sector jobs and hard to optimize using the usual industrial techniques. A high capital day care center is only slightly more efficient than a low capital one. A high capital industrial kitchen tends to produce low cost food, so its very effectiveness reduces its measured efficiency. The economy now has a whole new and vital sector which is highly resistant to productivity improvements.

    Despite this, we would have expected the GDP to rise as this sector appeared on economic radar. Instead we had a series of satanic child abuse cases, all bogus, but all highly reflective of our anxiety about the transition of child rearing from the home into the marketplace. I think people underestimate how much has changed. Childhood has been totally reinvented in response to parental fears and the need to monetize every moment of a child's life as a sequence of "activities". Our eating habits are dramatically different as well and not completely to our benefit.

    The ratchet point came when Volkert crushed inflation by raising interest rates in 1979. The price of a house went up from 600 labor hours to 800 labor hours for a median worker and remained that high. Suddenly, buying a house required two incomes, and a lot of women were glad to provide one of them. If you have read Vonnegut's 'The Foster Portfolio', you would understand why there was no going back. If anyone had the fantasy of doing so, Reagan's crushing of the air traffic controller's union in 1981 crushed this fantasy as well. Volkert and Reagan are not usually considered feminist icons, but they should be reconsidered.

    The material feminists had won the war, though they still have a lot of battles to fight. The limited impact on productivity, as it is measured economically, was not surprising. I think any old school material feminist could have predicted it.

    P.S, Yeah, this is a bit of a polemic, but the economic concept of value is highly filtered by societal biases and expectations. I learned about material feminists back in the 1960s when they were still considered nutcases, not prophets. Over the years I've watched as their program, developed in the 18th and 19th centuries was slowly implemented by others for their own purposes in a strange transformational synergy. It was really a major societal phase change, but no one is seems to be trying to understand it and its ramifications.