A more robust look at jobs data may hold the answer.
January 4, 2008
We've had a significant slowdown in economic growth in recent months. Continued deterioration in housing and related sectors has coupled with greater weakness in some important durable-goods industries. Amid worrying weakness in leading economic sectors, some government data were suggesting that employment had been holding up relatively well heading into late 2007. But the latest data are suggesting some chinks in the armor there as well. In Friday's national employment report, payrolls barely increased in December. Labor markets and consumer spending bear watching closely for signs of any intensifying recession.
Watching Labor Markets
Getting a good, timely read on the employment situation isn't like calling 411. Counting heads isn't as easy as it might seem. No single indicator out there is perfect. The available survey methods differ, and they can do funny things around economic "turning points" (expansion turning to recession, and vice versa).
Friday's headline number is not necessarily the truth. These reports are estimates--and subject to revision. Given past performance, it is possible that employment has already peaked.
To get a sense for the reality under the data, the best route involves looking at various data sources as well as anecdotal evidence all together, and forming a judgment based on a big-picture overview.
Some of the official government labor data worth digging into include the monthly establishment (or "payroll") survey, the monthly household survey, and weekly reports on unemployment insurance claims. Private indicators worth watching include the employment components of the national and regional purchasing manager surveys, and reports from employment services firms (more on the latter in a moment).
Comparing Payroll and Household Surveys
The U.S. Department of Labor produces monthly reports based on the payroll and household surveys, but these surveys differ significantly. The payroll survey polls companies, while the household survey polls people in their homes. The payroll survey is based on a sample of firms, with statistical methods used to adjust those results and estimate total employment throughout the country.
The household survey is also based on a sample, with the results adjusted statistically to produce a national estimate. The household survey, not the payroll survey, produces the widely reported "unemployment rate." But the household survey also produces a lesser-noticed but important estimate for total employment as well as unemployment. The household estimate for total employment is significantly more volatile on a month-to-month basis than the payroll data. However, the relative smoothness of the payroll data does not always equal precision.
The methods used to sample and statistically adjust the payroll data for trends in establishments outside the sample involve art as well as science. Some of the ambiguities involved there have led the establishment survey to produce remarkably false signals in the past. For example, heading into the 1990 recession, the initially reported state sample results were holding up well in late 1990 and early 1991, but later revised downward significantly when the "universe" data were compiled a year after the recession started.
The state-level data are aggregated by the Bureau of Labor Statistics (BLS) in the Department of Labor, and the national data take note of the possibility that out-of-sample trends may not match the sample totals. But these methods have historically been insensitive to trends at turning points. In 1992, for example, the payroll data were initially showing little if any economic recovery, even as a good one was really under way. On the other hand, the employment total in the admittedly bouncy month-to-month report from the household survey was showing a complete recovery by year-end 1992. Ultimately, the payroll data were revised more closely in line with the household number. The BLS has been working hard to improve its statistical techniques, but that experience still cautions observers not to put too much stock in any one indicator.
Where Else Can We Turn?
In 1992, reports from private employment services firms were giving off signals that things were picking up significantly, even as many market commentators and a presidential election were fixated on the payroll data and a still-high unemployment rate as evidence that we were in a "jobless recovery."
Ultimately, the payroll numbers were revised upward sharply. It made sense then, and it makes sense today, to pay attention to the surveys and personal observations of people working in the employment business. That's where Morningstar's Joel Bloomer comes in. Joel covers employment services firms such as Manpower (MAN), Kelly Services (KELYA), and Robert Half (RHI). His analysis can offer a good supplement to the standard measures.
Digging Deeper for Data
Temporary employment, which absorbs employment demand volatility, is a timely leading indicator for overall employment. For example, when a business owner starts to believe that the economy is improving, he or she will often hire a temporary employee to fill growing demand before taking on a new permanent employee, which represents a longer-term commitment to a fixed cost. As the situation improves and it becomes clear that the economy is in full upswing, companies start to hire permanent workers. The reverse is true on the downside. As the economy weakens, temporary employees are the first to be let go, eventually followed by permanent employees, assuming the situation worsens sufficiently.
So, how leading is this leading indicator? If we look at the historical temporary employment services numbers, a rarely publicized figure from the BLS' monthly Employment Situation report, we can see that temporary employment growth turned negative about eight months before the rest of employment posted its first decline in the last recession. (Temporary employment dropped 1% in December 2000, while total employment less temporary employment posted its first decline--0.1%--in August 2001.)
What Is the Data Telling Us Now?
Temporary employment growth turned negative in December 2006 and has remained negative for 12 of the last 13 months. Unfortunately, temporary employment data became available only in 1990, so it's hard to say what a long-run average lead time is, but it's safe to say this is a worrisome trend.
Data from staffing companies, the primary providers of temporary employees, lend further credence to this bearish outlook. Most staffing companies, such as those Bill mentioned earlier, have continued to see revenue growth, but a growing portion of the top-line boost has come from price increases, while in some cases volume has become a drag. As is common in many industries, price increases (bill rates, in this case) often lag the onset of a supply shortage, eventually leading to the counterintuitive situation of rising prices with falling demand (volume). If volume continues to fall, as appears likely, employers will begin to realize that there is an abundance of workers available, pressuring bill rate growth.
What Does All This Mean?
At Morningstar, our equity analysts focus primarily on bottom-up company analysis, but we'd be remiss to ignore persistent trends within economic data. As Bill said earlier, chinks in the labor market's armor are forming, and the data increasingly suggest more, and possibly larger, dents are still to come.
In an upcoming Stock Strategist article, we'll dig deeper into various attributes of temporary employment and its relationship to total employment. We'll also add a microeconomic view through a profile of our employment services coverage universe. This should let us separate good stocks from the bad, given the challenging macro environment.
Email This Page!