Jobless Claims Statistics - How Useful Are They?

Updated February 2015

Every Thursday morning at 09:30 Eastern Time, the investment world gets a vague notion of what is happening in the American economy through the Initial Unemployment Claims data released by the Department of Labor.  A recent paper by David Wiczer, an economist with the Federal Reserve Bank of St. Louis provides us with some guidelines regarding the usefulness of the Initial Unemployment Claims data.

Let's start with a graph from FRED showing the history of initial claims back to 1967:

During the last recession, initial claims hit a peak of 665,000 at the end of March 2009 and have since fallen to their current level of 297,000 in late November 2014.  The peak in March 2009 was the second highest level seen since 1967; in October 1982, the nearly 50 year peak of 695,000 was hit, slightly above the Great Recession peak.  The current level of 297,000 is touted as being one of the lowest levesl of new claims going back to February 2006, a rather significant significant achievement, at least on the surface.

The author of the paper notes that while the steady decline in the number of initial unemployment insurance claims reflects an improving economy, it also reflects two other issues:

1.) A long-term trend of fewer job separations as shown on this graph:

2.) A trend of fewer hires which still have not reached their pre-Great Recession levels as shown on this graph:

Initial unemployment claims are used as a proxy for the number of workers who are newly separated from their jobs.  If we look at this graph, we can see that the number of new claims as a percentage of the total labor force is at its pre-Great Recession level which, on the surface, would suggest that the economy has healed:

However, this apparent "healing" could be for two reasons:

1.) A decline in the number of new claims.

2.) The fact that the labor force has grown very little since the Great Recession.

The last graph shows us that the level of new claims is lower than at any time during the 1980s and most of the 1990s.  There are long-term trends that affect the rate of new claims that have nothing to do with an improved labor market since the 1980s.  For one, the drop in the level of new claims is a partly a result of changes in legislation that have impacted the number of workers that are eligible to collect unemployment insurance.  This is quite clearly shown on this graph that shows the declining fraction of unemployed workers that are covered by UI in dark green:

  

The percentage of the workforce that is covered by UI has dropped from over 4 percent at the end of the Great Recession to 2 percent in mid-2013.  This means that fewer newly unemployed workers can apply for UI benefits and, as a result, they do not appear in the weekly statistical Department of Labor roundup of new claims.

Secondly, as I noted above, the rate of separations and hires have changed during the post-Great Recession period as you can see on this graph:

Job separations are a poor predictor of changes to the unemployment rate.  When unemployment increases, it may increase because of an increase in the number of workers separating from their jobs (voluntarily or involuntarily) or because currently unemployed workers are finding new jobs at a lower rate.  Work by other economists shows that the employment exit/separation rate has steadily declined even during recessional spikes in the unemployment rate.  A study by Robert Shimer shows that ninety-five percent of the increase in the unemployment rate during the 1991 and 2001 recessions was a consequence of a reduction in the job finding probability rather than being related to increases in the rate of separations.  Since the end of the Great Recession, as we can see on the previous graph, a low level of separations also corresponds to a low level of hires, suggesting that the idea that a low level of separations means a healthy job market is far from the truth in the post-Great Recession economy.  During  and since the Great Recession, the rate of separations fell along with the rate of hires meaning that there were fewer people to initiate unemployment insurance claims with no effect on unemployment.

As we can see, using the weekly jobless claims data is not a particularly reliable indicator of the employment situation in the United States.  While it is tempting to use the weekly data release to give us a sense of how the economy has improved since 2008 - 2009, the conclusions of the author suggest that factors other than an improving job market may well be responsible for the apparent improvement in the number of new jobless claims.