Monday, September 16, 2013

Population and Labor Force and Population


Yesterday we looked at population and prices. But Steve Waldman was not looking at population. He was looking at "civilian labor force". That's a different animal. For one thing there's an 18-year lag, or so, from when a newborn joins the population, to when he or she joins the labor force. And then, the chance that he-or-she actually does join the labor force varies over time:

Graph #1: Civilian Labor Force as a Percent of Population
A significant increase in the probability of joining the labor force began in the early 1960s, slowed just before 1980, and died out around 1990. Isn't that strange? What would cause millions and millions of people to make decisions that, when we add them up and look at the result, look like an agreement to increase our workforce to half our population. What would cause that?

For me the answer is always economic -- economic forces, or the economic concerns which give rise to economic forces, or both. Something I got from Hayek.


Oddly, FRED's "Civilian Labor Force Participation Rate" shows a pattern comparable to Graph #1, but the numbers are much different:

Graph #2: Likely a better measure of participation than is Graph #1
FRED offers 88 pages of data sets in response to a search for civilian labor force. I can't say for sure, but I'm thinking that the U.S. civilian labor force is the U.S. civilian labor force -- age, education, race, ethnicity, geography, and gender aside. I'm thinking the civilian labor force is the same in both graphs above.

Graphs #1 and #2 combined:

Graph #3: Participation Rate from Graph #2 (blue) and my calc from Graph #1 (red)
If the labor force number is the same in both ratios, then it has to be the other number that's different: It has to be the population number that's different. My ratio (the red line on Graph #3) is lower, which means my population number is higher.

If my pop number is higher, and mine is the "Total Population for the United States" then FRED must not be using total population. Maybe they're using the working-age population. That would make sense.

If -- again, if -- we're using the same labor force numbers (CLF) and different population numbers (POP and PXX say) then my graph is CLF/POP and FRED's graph is CLF/PXX. If I divide my number by their number I get

(CLF/POP) / (CLF/PXX)

Dividing by a fraction, I invert and multiply:

(CLF/POP) * (PXX/CLF)

The CLF top and bottom cancel out, and I'm left with their population (PXX) on top, and mine (POP) on the bottom. So that way I can look at their population number as a percent of Total Population:

Graph #4: FRED's Mystery Population as a Percent of Total U.S. Population
Well this result surprised me. I guess I expected something more stable. Still... what have we got?

If my assumptions are valid, the blue line shows the working-age population as a percent of total U.S. population. Should we expect an increase like that -- from the early 1960s to 1990, just as on Graph #1 -- as a result of the Baby Boom? Let's see.

According to the Wikipedia article, the Census Bureau dates the baby boom from 1946 to 1964, and an "echo" boom ("the children of the post-WWII baby boomers") from 1982 to 2000.

16 to 20 years after the 1946 start of the boom puts us at 1962-1966. That's a good match to the first leg of the increase on Graph #4, an increase that only accelerated thereafter.

16 to 20 years after the 1964 end of the boom puts is at 1980-1984. That's just at the beginning of the slowdown visible on all these graphs.

And 16 to 20 years after the 1982 start of the echo boom puts us at 1998-2002. Between those dates, another uptrend begins on Graph #4. (The vertical at year 2000 is a break in the data resulting from revised accounting. But there is a clear downtrend into the late 1990s, and a clear uptrend since 2000.)

So yes, I think so. I think Graph #4 shows the U.S. working-age population as a percent of total U.S. population.

Hmm.

1 comment:

jim said...

Hi Art,

Your graph # 4 shows the effects of the baby boomers on the work age population. FRED doesn't count 0-16 year age in their population. Before the mid 60's all the baby boomers were under 16. By 1980 all the baby boomers were over 16.

Here is a related article you might find interesting. It

http://www.kc.frb.org/publicat/econrev/pdf/12q1VanZandweghe.pdf