How far from full labor market recovery are we? Part I

Last year around this time, I wrote a blog looking at the behavior of the unemployment rate over and after the Great Recession. I found that relative to its past historical relationship with output growth, the overall unemployment rate rose too rapidly during recession and then fell too rapidly between 2011 and 2012. I then approvingly quoted Ben Bernanke, who noted: “further significant improvements in unemployment will likely require faster economic growth than we experienced during the past year.”

In last year’s blog post, I noted that going forward the historical relationship between output growth and unemployment suggested that two straight years of 2.7 percent growth would be needed to reduce unemployment by even 0.4 percentage points. The growth rate for 2012 came in well under this at 2.2 percent. And what happened to unemployment in 2012? It fell by 0.8 percentage points.

So, another year has passed where the overall unemployment rate significantly over-performed relative to most other economic aggregates. The figure below shows the two-year change in unemployment, both actual and what is predicted from a simple regression of the two-year change on the two-year difference in growth rates of actual gross domestic product versus potential gross domestic product as measured by the CBO.1 What this captures is that the economy must grow faster than underlying trend growth (or growth in potential GDP) in order for unemployment to decline. The figure confirms that the actual unemployment rate rose more rapidly than predicted during the Great Recession, but then fell more rapidly than predicted in 2011 and 2012. By the end of 2012, the actual unemployment was a nearly 1.5 percentage points below what it would have been had the simple Okun’s relationship between output growth and unemployment continued to hold.

okun 2013

What explains this significant decline in unemployment even as output growth seems too slow to justify it? The answer, as it has been since the recovery began, is declining labor force participation. Between  2011 and 2012, the labor force participation rate fell by 0.4 percentage points, which, all else equal, can explain nearly three-quarters of the 0.8 percentage point reduction in unemployment.

An important point is that current controversies over what portion of the reduced labor force growth is cyclical versus driven by longer-run trends like demographics don’t really matter much here—the potential GDP numbers used in this exercise should account for any normal, long-term and predictable decline in labor force participation caused by demographics. So, if demography is lowering labor force participation to a greater degree in the current recovery than in previous ones, this should simply result in slower potential GDP growth and hence should make the prediction model require slower rates of actual growth in order to drive down unemployment rates.

If one wants to make the claim that the CBO is radically understating the degree of labor force drop-out that is due to demographics, I guess one can make it, but that’s the only reason to think there’s nothing puzzling about the continued downward drift of unemployment even in the face of very slow growth.

We noted a long time ago that policymakers seemed to have largely given up on presenting actual solutions to the jobs-crisis. 2013, however, is the first year that federal policymakers have succeeded in setting policy to actively make it worse, and stomping on some of the first promising trends in years. As data on GDP and jobs are released in the coming two weeks, we’ll have as series of posts examining just how far from full recovery the economy remains and what prospects for achieving it in the short-run look like.

1The regression also includes the difference between actual and potential growth rates squared and a time-trend, and is based on the 1979 to 2007 period.