Tuesday, July 23, 2013

Employment Flows

The BLS has tons of great data.  Building on yesterday's post, I was really hoping that I could find employment flow data broken out by age.  I'm coming around to the idea that so many trends that we would like to attribute politics, money supply, etc., etc. are largely demographic issues.  I was hoping to test whether the tepid JOLTS data could be a product of demographics.  Characteristics of older workers could explain the low level of quits, the high proportion of job openings to hires, etc.  But, unfortunately, I don't see it broken out that way.

The flow data is interesting, nonetheless.  Here are the historical graphs concerning flows to and from "Not in the Labor Force".  First, here is a graph of "Not in Labor Force" population.  We can see the acceleration in this category over the last 15 years, as a result of an aging workforce.  And, we can see the slight movement above trend in 2003-2005, then the slight movement below trend in 2006-2008, with a recovery back to trend in 2009 and after:



In the next graph, the blue line shows "Not in Labor Force to Employed" and the green line shows "Employed to Not in Labor Force".  I would guess that the secular growth in these categories stem from the aging labor force that I covered in the previous post.  In the 55+ age group, there are a large number of people who would not consider themselves "unemployed", but fill their days with a combination of civic involvement, odd jobs, consulting, political activities, etc.  Some of these activities would count as "employment", but for the growing number of people that meet this description, life would not fit nicely into a employed/not employed context.  We can see the long term increase of aging baby boomers here, with a temporary decline during the recession from labor market anemia, although even at the bottom of the recession, there were more flows between employment and NLF than there had been in the 90's in a booming economy with fewer older workers.  (Students could be an effect of these flows, too.)

The red line shows flows from "Not in Labor Force to Unemployed" and the purple line shows "Unemployed to Not in Labor Force".  The conventional image of the recession is that the UE>NLF flow would have increased.  But, the full picture is more complicated.  Flows in both directions have increased together.  Looking back in time, there is some cyclical behavior in these flows, but not the secular increase that we see in the employment flows.  The recent persistence of the increase in these flows is probably related to the large number of long duration unemployed.  It is interesting to note that while 11 to 12 million workers are still unemployed, there is a flow of more than 2.5 million workers between unemployed and "not in workforce" each month.  With that much circular flow, it is amazing that we don't see more noise in reported unemployment rates each month than we do.

 
 
 
To get more insight into how these flows relate to the business cycle, I have graphed the net effect of the "Not in Labor Force" flows to and from employment (blue) and to and from unemployment (red), and the net flows to "Not in Labor Force" from both employment and unemployment.  These are 12 month moving averages to eliminate noise.  There is always a net flow from employment to "Not in Labor Force" and from "Not in Labor Force" to Unemployment.  The persistence of these net effects relates to persistent aspects of the economy, such as the constant flow of new young people into the labor pool, retirements of employed people, etc.  The net NLF figure here does not account for non-employment related changes, such as population changes.  The changes in trend are what's important here.
 
First, on the total net flows to NLF (Not in Labor Force), there is probably a persistent negative flow (a flow into the workforce) over time.  Here the unsustainable trend of flows into the labor force in 2004-2005 are visible, and the unusual flows out of the labor force in 2009-2010 are visible, with the return to a neutral flow since then.
 
When we look at the separated indicators, we can see that there is a counteracting dynamic happening.  In the early phases of the last two recessions, the "Not in Labor Force" category was inflated by workers moving directly from employment to NLF, and this was countered by workers moving out of NLF and into Unemployment.  It is only later in the recession that we see relatively more workers moving from unemployment to NLF, which is then countered by relatively less workers moving from employment to NLF.  I would characterize the period in 2009 & 2010 where we saw the biggest dip in Labor Force Participation as mostly the product of a decline in workers flowing from NLF directly to Employment.  Looking back at my previous post, where we see the most anomalous movement in LFP among the youngest and oldest workers, I would speculate that this stemmed from older workers who normally tip back and forth between Employment and NLF having a harder time finding temporary employment, and younger workers who were having more difficultly getting entry level jobs in the face of stagnant hiring markets, exacerbated by the last hike in the minimum wage in late 2009. 



It seems as though there could be a leading indicator hidden in here somewhere, since early in recessionary markets we see a local minimum in flows through unemployment, a local maximum in flows through employment, a divergence of flows between NLF and Unemployment as well as between NLF and Employment, and, as shown in the last graph below, an increase in the persistence of unemployed workers.

The counteractivity of many of these indicators means that these movements can start to be visible here before they are visible in measured LFP or the unemployment rate.  Interestingly, the last chart also shows how flows in and out of NLF and the persistence of unemployment are much more volatile than the flow of workers from Employed to Unemployed.

Surely, there is a leading indicator buried in here somewhere, although the noisiness of this data might make it hard to read in real time.  These will be interesting to watch when we enter the next bear cycle.  These indicators certainly do not bear out my earlier concerns about weakness in the JOLTS data, so this is more evidence that we have a ways to go before we need to worry about a new downturn.

No comments:

Post a Comment