Thursday, May 1, 2014

A couple more notes on April Employment

Here is an update on the unemployment insurance claims numbers.  Both measures bounced up a bit this week, so the trend going through April doesn't look as strong as it did last week, but continuing claims still seems to have a promising trend.

Here is an amended version of the other graph I put together the other day, comparing the unemployment rate with the rate of unemployment insurance (using civilian labor force as the denominator).  Using the higher continuing claims level from this week, the unemployment level that would be parallel to trend would be between 6.5% and 6.6%.  (Last week it would have been near 6.4%.)  Of course, we should expect the unemployment rate to approach the lower typical level as the economy recovers, so it should do better than just moving parallel to the normal trend, though there will be noise from month to month.

I'm also trying to get an idea about the mysterious long duration unemployed group.  This is not a sophisticated model, but I'm just trying to work out a basic idea of the data here.  In mid-2011, the average duration of unemployment for people unemployed more than 26 weeks leveled out at around 80 weeks, and has remained at that level, even as the total number of long term unemployed workers has declined.  I am using a simple assumption that, starting around 2011, unemployment can be divided into two groups.  Short term unemployment began to recover to normal levels, and for these workers who were unemployed for more than 14 weeks, each quarter approximately 50% of them exited the unemployment rolls.  Additionally, there was a second group of very long term unemployed workers who were more persistently unemployed.  This group appears to have declined in a somewhat linear pattern, averaging about 80,000 workers per month.

So, by March 2014, there were about 3.7 million workers who had been unemployed for more than 26 weeks.  About 2.1 million were in the very long term unemployed group, roughly averaging 100 weeks of unemployment.  And, about 1.6 million were leaving unemployment at roughly typical rates for a normal economy, with an average unemployment duration of about 60 weeks.

Even in a typical labor economy, some workers have durations of unemployment over 26 weeks.  So, if we assume that Emergency Unemployment Insurance above 26 weeks has no effect on their subsequent duration of unemployment, we can assume that, given the availability of EUI, a typical proportion of workers would take payments from EUI.  This graph compares the reported number of EUI beneficiaries (blue) to my estimate of the very long term group of unemployed (red).  In addition, I estimate the number of EUI beneficiaries who are part of that very long term persistently unemployed group (green).  So, reading from bottom to top, the green line is the number of VLT unemployed who are receiving EUI.  From the green to the blue line is the number of unemployed workers who would be eligible for EUI in a typically functioning labor market (with 50% quarterly turnover, as described above).  And from the green to the red line is the estimated number of VLT unemployed who are not receiving EUI.

As the following graph demonstrates, the behavior of the long term unemployed has been very peculiar this cycle, coincidental with the unprecedented level of EUI benefits.  There has been a disconnect from a relationship of employment durations that is decades old.  EUI created significant frictions in long duration unemployment.  Some of this could have been from subtle effects on reservation wages, simple incentives for workers with some discretion, hysteresis among former beneficiaries, etc.

It is difficult to determine what we could expect from the current unemployed cohort.  If my graph above is accurate, then by the time that EUI was terminated, the unemployed workers who were still beneficiaries were mostly not members of the persistent very long term group of unemployed.  By 2014, the very long term unemployed lacked any direct connection to the remaining EUI program.  This would suggest that the end of the program may not have an immediate, direct impact on the very long term unemployed.  And this group is essentially the entire reason why the current unemployment rate is higher than normal recovery levels.  So, my optimism for unemployment in the first half of 2014 may have been misplaced.

The number of very long term unemployed who were not EUI beneficiaries (either because they timed out of the program or because they hadn't been eligible) has been pretty level at about 1.5 million for several years.  In effect, broadly speaking, each month, 80,000 VLT unemployed who were not on EUI would leave unemployed status, but they would be replaced by another 80,000 unemployed coming out of EUI.  But, if this was the case, and there was a churn of workers through EUI, then we should have seen the net number of VLT unemployed leaving unemployment increase as the number of EUI recipients declined.  This didn't happen.  The decline in VLT workers has remained pretty linear, about 80,000 per month.

So, it seems like we may not see a direct impact from the termination of EUI.  And, it's possible that either unemployment will continue to decline linearly or will decline less quickly if there are a large number of permanently unemployed workers in the VLT group.  However, I think there could be a complex set of ingredients here that still could lead to a decline in VLT unemployment as we leave EUI and proceed through 2014.

Here is a graph of the gross number of workers who leave LT unemployment each quarter. (This is the starting number of 15+ week unemployed minus the 26+ week unemployed who remain unemployed 3 months later.)  If my pessimistic scenarios above were true, we should have seen this flow decline as EUI wound down, if the remaining VLT unemployed were more persistently unemployed.  This month should add interesting information here.  This flow actually stabilized at around 2 million workers during the last year.  (This is for all workers, so about 1.6 million of these workers are workers in the recovered economy who have flowed into and back out of unemployment, and the remaining workers in this flow would be coming out of the VLT group.)  Last month, this flow dipped down to 1.8 million, but these are noisy data.  So, if this flow continues to decline toward 1.6 million, this could mean that the VLT group will be very persistent.  If this flow recovers back to 2 million, then this should bode very well for the labor market.  Even if EUI was only pulling in a couple hundred thousand workers into a VLT unemployment context, the removal of that program would mean that the net 400,000 workers leaving long term unemployment each quarter would affect the bottom line unemployment figure, and we would be seeing sharper decline in the net remaining number of VLT unemployed.

I think the factor that could make this true would be the inaccuracy of my assumption above that continued EUI was not affecting the long term duration of unemployment for the workers who have recently been becoming beneficiaries.  It is possible that, even though much of the labor market is back to normal, EUI was continuing to have an inflationary effect on the duration of unemployment for the workers who did end up in the program.  Even at the end of 2013, the program might have been feeding the VLT unemployment problem, and some of those EUI beneficiaries were being converted into VLT unemployed as they were exposed to the perverse incentives of the program.

Regarding the gross flow out of LT unemployment, all else equal, if that flow declines down to 1.7 million in the April report, the unemployment rate will be at 6.7 6.6%, and if it recovers to 2 million, the unemployment rate will be 6.5 6.4%.  The over/under on this month's report could indicate what we might expect as we move through the year.

No comments:

Post a Comment