Lag in Job Numbers Behind GDP Growth is No Worse than in Past Recoveries


At first glance, the job numbers of the last week seem to offer a mixed and confusing picture.   On the one hand, today’s headline from the Bureau of Labor Statistics certainly sounds like good news:  the unemployment rate finally dropped below 10.0% — to 9.7%.   On the other hand, today’s establishment survey of employment, which most of the time is a more reliable measure than the unemployment rate, still shows job change numbers that are negative.   Furthermore, recent numbers on claims for unemployment benefits have been discouraging.   

To reach an overall evaluation, one must take a longer-term perspective. Even the job loss reported in the establishment survey for January can be interpreted more positively:   it is less than 1/30th of the rate at which job losses were running one year ago, in January 2009.   (See the chart below, which presents 3-month averages of employment changes over the last two decades.)  The best way to sum up all the labor market numbers in recent months, both positive and negative, is they have been within measurement-error-distance of zero, roughly flat.  That may not sound great, but it is a big improvement relative to what came before.


It is inconvenient, but common, that the labor market and the rest of the economy send conflicting signals.


As a member of the NBER Business Cycle Dating Committee, I get asked whether we are ready to call officially an end to the recession.   Although the Committee looks at many indicators in reaching its decisions, the most important overall are measures of output, particularly quarterly GDP.   GDP shows growth turning from negative in the first half of 2009 to positive in the 3rd quarter of the year, and now strongly positive in the 4th quarter.  That suggests that the trough will probably turn out to have been in the middle of last year. 


Beyond output, the monthly indicators that are most important to the Committee are probably those that come from the labor market. Here is how I personally read the overall picture of the labor market that has emerged recently (refer again to the graph): 

·       The employment loss was especially bad in this recession, whether viewed in absolute terms or relative to the output decline : 8 million jobs lost, in the BLS revisions that are probably the most important news in today’s report.   (Incidentally, as Bob Gordon points out, the tendency for the labor/output ratio to fall in recent recessions is a mortal blow to Real Business Cycle theories which attribute recessions to declines in productivity.)

·        But the recovery time in jobs does not appear to be lagging behind output any more than was the case in the preceding two recessions — and perhaps less.   The worst job loss this time occurred around the time of the worst negative GDP numbers – the very beginning of 2009.  This in contrast to the preceding two recessions, when the job loss numbers waited until GDP growth actually turned positive, before hitting their worst point.   Yes, it is painful that labor market losses have not completely ended yet, a half-year after the apparent end of output losses.  Nevertheless, as I count, that still appears to be a shorter lag than in the famous “jobless recoveries” following the recessions of 1990-91 (job losses ended 11 months after the trough) or 2001 (they did not end until 21 months after the trough).


The bottom line is to reinforce the verdict of most economists: that the recession very likely ended sometime in 2009.    At this point the main thing that the NBER Committee must watch out for is the small risk that the economy could be hit in 2010 by a sudden new downturn, as the changes start to diminish;  but a real double dip recession (which might in theory count as a continuation of the same recession as 2007-09), seems increasingly unlikely.



Employment changes, monthly data, January 1990-January 2001.   Source: Bureau of Labor Statistics.