Saturday, September 7, 2013

Quibbling about Labor Force Participation

Tyler Cowen links to this piece at the Washington Post.

It claims that there are 3 reasons for lower Labor Force Participation:

1) Demographics
2) The bad economy
3) Disability Insurance

On #3, I think it correlates highly with demographics, with some cyclical behavior, as can be seen in the chart from the article, so I think that this is really a secondary issue.  I'm not going to dig into the details here.

My quibble is on #2.  We have had a business cycle, and so, clearly, we should see some cyclical effects.  But, there is nothing unusual about this cycle, once we adjust for demographics.

The article makes two common errors:

1) Because the trend for LFP has been in secular decline since the late 1990's, the very strong labor market of the 2000's was obscured by the downward trend.  The article cites this graph.
epop graph-thumb-615x395-82792
I haven't reviewed the referenced studies, but note that both of them make the LFP of 2006 appear to be below trend.  That is very unlikely.  So, both of these forecasts are starting out too high.  But, if we look at the slopes of these forecasts, they both predict LFP falling at .25% to .3% per year for the next 10 to 20 years.  That's basically what we are seeing.  At the bottom of the last recession, in 2004, LFP was at 66%.  Coming up on 2014, we are nearing 63%.  That's a 3% decline in 10 years, which is similar to the rate of change these forecasts predict.

2) Using LFP for 25-54 year olds does not account for all demographic issues.  The article uses this graph, and says, "obviously, retirement can't explain this":
First, I will just point out that it should be visually obvious, looking at LFP (the blue line) in this graph, that there are long term, non-cyclical factors pulling down the trend, and that we have just experienced a very typical cycle.  There is a little bump from 2004 to 2009, and, correcting for trend, it's basically the same bump you see in the late 90's, the late 80's, and the late 70's.

But, I'll take a moment to disaggregate this.  Here are men's LFP's by age:

There are several things to note here:
1) There is a significant fall-off in LFP as workers age to 45-54.  There are a large number of 45-54 year olds right now.  So, some of the demographic effect is still captured when using 25-54 year olds.

2) There might be increasing cyclicality among 25-34 year olds, which could be a long term trend or could be related to this specific cycle.  But, this age group's LFP has been recovering to trend strongly.

3) There is a very long term trend of LFP decline among all age groups.  Before 2000, this factor was obscured in the aggregated measures by the rising LFP of women.  But, even if there wasn't an issue with aging, we would still need to expect more than a 1% decadal decline in LFP.

4)  On the whole, the LFP's are not exceptionally low compared to trend.

Here are women's LFPs:
For each age group, behavior is similar, and, after a rise to the mid 70s, all age groups have appeared to take on the long term decadal decline rate of more than 1%.

Interestingly, there is no drop-off for women going into the 45-54 age group.  Here is a close-up of the past 14 years.  As with the men, there is a strong downward linear trend, within each age group.  There is a typical cyclical fluctuation of less than 1% above & below trend.  There is nothing especially unusual about the current cycle.

An unwillingness to internalize normal long term downward trends is causing undue hand wringing about cyclical LFP's.

1 comment:

  1. My thoughts here:
    This is good stuff for cohort analysis. If I were more of a labor economist, I'd put more thought into it.