drop in short-duration unemployment from the December Household Survey doesn't seem to show up in the JOLTS data at all. Here are updated graphs:
Below is the Beveridge Curve - the relationship between job openings and the number of unemployed workers. After a decades-long shift to the left, the relationship began to shift to the right in 2008.
I have attributed this shift, partly, to the sizable minimum wage hikes in 2008 and 2009 and to extended unemployment insurance. EUI was discontinued at the end of 2013, and the minimum wage level, compared to average wages, is trending back to 2008 levels. So, especially because of EUI, I expect to see a shift back toward the 2008 curve position over the next 6 to 12 months. It looks like we have already started to see this reshifting in December and January.
Here is a short paper from the Boston Fed, which disaggregates the Beveridge Curve by age and by reason for unemployment. It's findings are very interesting. There is a bifurcation of results. Many of the subpopulations show essentially no shift in the curve. As of January 2013, the additional unemployment, relative to job openings, was mostly the result of 2 groups, split roughly 50-50:
1) Job leavers, new entrants, and re-entrants, between 20 and 34 years old.
2) Job losers, 45 years and older.
The second group is the group where we would expect to see a shift, if extended unemployment insurance were the culprit. In some of my past posts, I have noted how the inflated long term unemployment levels that might be attributed to EUI seemed to be heaviest among the older age groups. The relationships shown in the paper suggest that EUI may be even more skewed toward older workers than my basic model estimated.
This also pretty closely corroborates my estimates for the excess unemployment attributable to EUI. I had attributed excess unemployment topping out at 1.2% in mid-2011 and falling to 0.9% by January 2013. The Boston Fed paper shows a shift in the Beveridge Curve attributable to job losers topping out at 1.9% in mid-2011 and falling to 1.1% by January 2013.
The puzzle for me is, why is the other half of the explanation for the shifted Beveridge Curve concentrated among 20-34 year old job leavers, new entrants, and re-entrants? There could be a number of reasons, none of which seem like good news. The most hopeful cause would be the minimum wage, because at least that would be an easily identified, self-inflicted wound. But, the shift does not show up in the 16-19 year old data. Now, some funny things appear to happen in labor force participation in these minimum wage episodes, so that doesn't count MW out as a culprit, but it makes it harder to blame MW with confidence.
The gap from the old Beveridge Curve in January 2013 was about 2%. About 1% of this was from 45+ year old job losers and about 1% was from the 20-34 year olds. The gap had maxed out at nearly 3% in late 2011, and nearly 2% of the gap was due to the job losers. But, the gap coming from the younger workers had been persistently around 1% since mid-2011.
To update the data to December 2013, the closest data I could find to the authors' were age-based unemployment levels, without the distinction for reason for unemployment. But, I can use this data to get an idea of what has happened since January. What I find is very good news. Here are the updated Beveridge Curves for 20-35 Year Olds and for 45+ Year Olds.
The good news is that both groups are making strong moves back to the earlier Beveridge Curve trendlines. In my data, in January 2013, the 20-34 Year Olds accounted for 0.8% and 45+ Year Olds accounted for 1.0% of the excessive unemployment rate. By December 2013, the excess rates had decreased to 0.5% and 0.8%, respectively. Employment seems to be recovering in both groups.
In the most recent months, my independent estimate of the excess unemployment due to EUI, from comparing short duration and long duration unemployment, is 0.9% in December 2013 and 0.8% in January 2014.
Unemployment has been declining, parallel to the previous Beveridge Curve for the past two years, but I think we are going to see a reversion back toward the old curve this year.