Median Unemployment Duration by Age Cohort

Okay, can BLS make it any more difficult to get at this data? grumble grumble rippensnatz... I ahd to end up going to some ftp directory and extracting data to flat files then import to MS Excel - bug or feature?

The data shows the other side of the UE story, that while the UE rate among the youngest cohort is the highest, the duration of that UE is the lowest. In addition to the 16 to 19 cohort, I emphasize the 44 to 55 cohort as that represents peak earning years. But among all cohorts we see increasing median duration of UE with increasing age.

So there appears to be another strong predictor of UE duration, a person's age (the other being the duration itself, the longer unemployed the likelier to remain there).


And here is what the absolute numbers look like for 27+ weeks UE duration by age cohort...

Different This Time - More on Labor Force Declines

As the population grows each year, how have they moved into the economy? Are things different this time?

Starting with BLS annual data for Civilian noninstitutional population 16 years and over, Civlian labor force 16 years and over and Not in labor force 16 years and over I took the year over changes for each series and constructed plots:


Looking at the changes in the over 16 population and labor force, we see only three years in which the year over year change in the labor force was negative - 1951, 2009 and 2010 - and the 1951 event is characterized by the only year over year decline in the civilian population. In fact, the year over year change for all three series was negative in 1951. This is not the case in 2009 and 20010, where there is ~1% population growth, year over year declines in the labor force, and year over year increases in not in the labor force (seen in following graph).


Here we see the population changes and the not in labor force changes, and there are three years that stand out for the increase in not in the labor force being greater than the increase in population (excepting 1951 op cit) - 1976, 2009 and 2010 - with by far the largest outlier being the 1976 data point.


In summary, it is clearly different this time, as only 2009 and 2010 year over year changes show consecutive declines in the size of the labor force and consecutive year over year increases in the not in labor force greater than the increases in population.

Continuing Declines in Labor Force Participation Rate

The plot above shows the 4 week moving average of the seasonally adjusted weekly Initial Claims numbers, and that same curve shifted forward 99 weeks, showing that the peak number of '99ers' is now rolling off the back end of their extended benefits. There appears to be a relationship with the decline in the labor force participation rate also shown above, which looks a tad strange as it is a monthly series and ICSA is a weekly series.

We can see that the labor force particpation rate began to decline around six months to a year after the ICSA began to ramp up.



UEMPMEAN appears to have a strong relationship with the declines in the labor force participation rate, and the average duration is likely to continue to rise as it appears the probability of remaining unemployed has become in part of function of unemployment duration... the longer you are unemployed, the likelier you are to remain that way.

Both of these suggest to me that we will continue to see declines in the labor force pariticipation rate... which might well drive the U-3 headline unemployment rate but for all the wrong reasons.