Why today's jobs numbers don't matter, in two charts
As the economic recovery drags on, people understandably greet each month’s new jobs report with a lot of interest—and sometimes histrionics—especially when the news is good. Today’s numbers are now out, with the BLS reporting just 88,000 new jobs in March. Within minutes of the release, Twitter was filled with analysis of the "weak" increase.
Through all the excitement, it’s worth remembering that these are merely the BLS’s preliminary estimates of the gain, or loss, in jobs. Each month’s release actually generates several new numbers, including a preliminary estimate for the previous month, a revision to the estimate for the month before that, and a final revision of the estimate for three months ago.
In other words, each month’s preliminary estimate—the one that gets all the attention—can change. And it often changes a lot.
To take some extreme examples, in June 2011, the preliminary estimate for job creation during May 2011 came in too low—by 191,000 jobs. In September 2008, at the depth of the recession, the preliminary estimate was too rosy by 300,000 jobs. The magnitude of each of these differences is enough to make news in terms of changes in total jobs. And those were just the adjustments to the estimates, not even the estimates themselves.
Worse yet, it’s not hard to find examples where the revisions actually reverse job gains to job losses, or vice versa. In July 2000, the preliminary estimate predicted a loss of 108,000 jobs. The final estimate actually showed a gain of 166,000 jobs. The reported loss no doubt generated nail-biting worry in the headlines, despite actual growth in the job market.
As the economy tumbled into recession in 2007, the preliminary estimates of job destruction underestimated the drop in jobs. Since the economy began its slow recovery in mid-2009, preliminary estimates underestimated the growth in jobs. In other words, one of our most-watched indicators of the health of the labor market consistently understated how quickly conditions were deteriorating during the recession and have understated the strength of the recovery.
Even though these estimates are carefully labeled as “preliminary,” they can influence policy choices such as how and whether to stimulate the economy or whether to extend unemployment insurance benefits. When the preliminary estimates consistently over- or under-state for months on end, they compromise the policy debate.
This is not to say that BLS does a bad job; they do a great job with a very complex task. But it takes time to be certain how myriad factors affecting our macroeconomy change each month. So don’t put too much stock in today's 88,000—it’s going to change.