BLS to add new category to unemployment measurements

The Great Recession has changed the nature of joblessness so much that the Bureau of Labor Statistics will add a new category to its surveys in order to provide data on long-term unemployment.  Thanks to an “unprecedented rise” in long-term joblessness, the survey will now count people unemployed for more that 260 weeks, rather than just the 99 weeks which had been the upper limit on its survey forms:

The change is a sign that bureau officials “are afraid that a cap of two years may be ‘understating the true average duration’ — but they won’t know by how much until they raise the upper limit,” says Linda Barrington, an economist who directs the Institute for Compensation Studies at Cornell University‘s School of Industrial and Labor Relations.

Likening recessionary unemployment spikes in recent decades to a storm at sea, she says, “The waves are getting higher, and we want to understand the intricacies of how they’re made up.”

The change involves the form used for the bureau’s Current Population Survey, based on interviews with thousands of the unemployed. Currently, no matter how much longer than two years someone has been out of work, the form allows interviewers to check off only “99 weeks or over.” Starting next month, jobless stints of “260 weeks and over” can be selected on the response form.

“The BLS doesn’t make such changes lightly,” Barrington says. Stacey Standish, a bureau assistant press officer, says the two-year limit has been used for 33 years.

A two-year limit hampers economists’ ability to compare this recession’s effect on the job market with another severe one in the early 1980s, Barrington says.

That may still be problematic.  The BLS didn’t tally a 260-week-plus category back then either, so the comparison today is actually more consistent than it will be with a newly-added category.  In order to compare across time periods with this new category, the BLS will have to rely on secondary sources to determine how many people were unemployed five years or longer in earlier recessions, as they do now.

Adding the category now will build benchmarks for future comparisons, though, and that will be important if we go through a cycle like this again.  It will also inform on the efficacy of policies in the present aimed at curtailing the long-term joblessness that threatens to create a permanent underclass, as well as give more data on the nature of long-term joblessness.

But still, one has to wonder whether this change will still get the BLS what it seeks.  The problem is that one cannot accurately determine the average length of unemployment when capping the response at 99 weeks.  Extending the cap to 260 weeks really doesn’t solve that problem, either.  In fact, talking about average length of joblessness when forcing people into broad categories is really a data type mismatch that makes claims like “the average length of unemployment has increased from 29.4 weeks in November 2009 to 34.5 weeks last month” entirely suspect.  The BLS should be collecting specific lengths of joblessness in order to determine averages and trends, not check boxes for categories.  Why not just ask each respondent for the specific number of weeks he or she has been unemployed, or simply ask for the last date of employment, and let the computers determine the actual averages?

Update: I should have noted that this change will have no effect on the traditional unemployment rate reporting.

• WaPo: Say, there seems to be a lot fewer workers these days
• Video: “It’s a permanent unemployed level”
• Jobs report: Only 88,000 jobs added; Update: “This is a punch to the gut”

Blowback