Plenty has already been written about last week’s nonfarm payroll data. Many have pointed out the weakness underlying the relatively strong +151,000 headline number: the household survey (which feeds the unemployment rate calculation) printed a loss of 330,000 jobs (the worst of 2010 so far), the number of full time jobs continued to fall, etc. Calculated Risk wrote a great piece today on a topic that we have highlighted as very important: labor force shrinkage. We were inspired by another recent Calculated Risk post to dig into the Job Openings and Labor Turnover Survey (awesomely nicknamed JOLTS) from the Bureau of Labor Statistics.
The headline Nonfarm Payroll number is only a single number. It doesn’t tell you anything about what’s going on underneath the hood. Payrolls change for two reasons: people joining payrolls (hires) and people leaving them (separations). The JOLTS survey gives us some color on these moving parts.
First, to get a sense of the change in payrolls, we can look at the number of net hires by subtracting separations from hires. As it turns out, the JOLTS survey tracks NFP pretty tightly. (We wish this survey went back further, but it only goes back to 2000. The JOLTS data are through September while the NFP data are through October.)
You can see the dramatic job losses of 2008 and the Census-related hiring binge of early 2010 (all of the data here include government hiring/firing). The chart below looks at the behavior of hires and separations.
Hires and separations tend to move in the same direction, which at first seems strange. In a recession, you would expect hires to drop and separations to rise (fewer people hired, more people fired). In reality, what happens is that hires fall faster than separations (i.e. more people leaving payrolls than joining), driving unemployment higher.
A closer look at the two main components of separations helps us understand why this is so:
The single line item “separations” is made up of layoffs and discharges (the first thing that comes to mind) but it’s also made up of people leaving their job, which we’ll call “quits.” As it turns out, quits is usually the larger component. A few things to notice:
1. The level of layoffs tends to fluctuate less than quits.
2. Quits and layoffs seem to move in opposite directions.
3. Layoffs spiked in 2008 and overtook quits to become the largest source of separations. This has recently reversed.
4. Quits bottomed out in late 2009 and have been in an uptrend ever since: this is a good sign.
5. The level of quits is still below the previous recession’s worst levels: this is not so good, meaning that labor turnover is still at relatively sickly levels.
6. Layoffs are pretty low right now, so companies must have done most if not all of their labor force “right-sizing”.
Another interesting data series from JOLTS is job openings, which we show in the following graph. Openings, not surprisingly, tracks quits (i.e. people usually don’t quit a job without another one in hand). This has also been in an uptrend along with hires, but both are still at pretty depressed levels.
The takeaway from all this? As usual, there is more to the employment data than what the headline numbers purport to show. The JOLTS data show that the best that may be claimed about where we are in the employment cycle is that we are making our way off a very low base.