This month’s employment report shows unexpected and conflicting results and presents a perfect opportunity to explain how it is produced.
The report is comprised of two surveys, the “establishment” survey and the “household” survey, both performed by the Bureau of Labor Statistics (BLS). The establishment survey takes a sample of about 150,000 businesses (establishments), counts the number of people on the payrolls, and the average hourly wages (among many other things). The household survey takes a sample of about 60,000 households to calculate the unemployment rate, the size of the labor force, and the participation rate (among many others). Since the establishment survey is so much larger and covers about one-third of all payrolls, it is generally regarded as giving a more thorough picture of the labor market each month. Note that since both surveys use a sample, the results are presented as estimates, and are therefore subject to error.
The surveys can give different impressions in any given month, like now, because there are several differences in the way the surveys are performed. Just knowing that there are differences is probably enough for most readers. But if you are interested, here are the details according to the BLS:
- The household survey is limited to workers 16 years of age and older. The establishment survey is not limited by age.
- The household survey includes people on unpaid leave among the employed. The establishment survey does not.
- The household survey includes agricultural workers, self-employed workers whose businesses are unincorporated, unpaid family workers, and private household workers among the employed. These groups are excluded from the establishment survey.
- The household survey has no duplication of individuals, because individuals are counted only once, even if they hold more than one job. In the establishment survey, employees working at more than one job and thus appearing on more than one payroll are counted separately for each appearance.
(If you really want to get into it, the report produces a mind-numbing amount of detail including the tables at the bottom of this page:
This month the establishment survey showed that the US economy added only +20k jobs in February, far short of the +180k expected.
- One part of the explanation for the weakness is that it was a payback from a very strong month in January and a strong month in December. A second part of the explanation is that since the error around the survey (mentioned above) is a massive +/- 100k, we occasionally get results like this which are way different than expected.
- And a third part of the explanation is that the economy is slowing. Job losses and weakness were widespread across industries. Notable job losses occurred in construction with -30k jobs, due in part to bad weather, the biggest loss in over eight years. The services sector, which accounts for 85%-90% of the economy, gained only +53k jobs vs a 12 month average of +166k
- However hourly wages rose 0.4% m/m, pushing the y/y rate up from 3.1% to 3.4%, the highest in almost 10 years. And because of weak inflation, real wages have grown sharply from +0.5% y/y six months ago to +1.9 y/y today, the highest in 31 months. How can that be if the rest of the survey shows weakness? First, there is still plenty of demand for labor - employers have a record high 7.3 million jobs open while there are only 6.2 million people looking for jobs, so of course there is wage pressure. But a lament we have been hearing for a long time is that employers can’t find people with the right skills for the jobs. So there are plenty of jobs to be had at good wages, but it’s possible that this month weak job growth in part reflected a lack of skilled labor. A second explanation for the difference between weak job growth and strong wage growth is again, the job growth estimate has a lot of uncertainty around it. In other words, from month to month, wage growth is much, much more stable than job growth, as you can see in the charts.
The “household” survey showed that the unemployment rate dropped from 4.0% to 3.8%. The broader U6 measure of unemployment which includes part time workers who want a full time job, and discouraged workers, fell from 8.1% to 7.3%, the largest drop on record. In both cases the BLS attributes the decline in part to the end of the government shutdown.
So overall the employment report is not quite as contradictory as it seems. The household survey suggests things are on track, there was just a blip for two months from the government shutdown. The establishment survey in part suggests that the weak jobs number is due to payback, sampling error and a weakening economy, while decent wage growth is a product of supply and demand for labor, and its inherent stability month-to-month. Overall, there is no need to panic, especially over one report. A rule of thumb is that it takes three months to make a trend. It’s possible the labor market is weakening, but it is still strong.
But about that weakening…. Keep in mind that other recent indicators have also been less than stellar, including personal consumption, income, retail sales, manufacturing, housing, and the divergence in the consumer confidence survey. Estimates for Q1 GDP growth are a little upsetting – most of them are less than 1%. The economy is expected to do better in Q2 and Q3, but for now, it is slowing, and the Fed is certainly on hold.