Is May Employment Data Good or Bad? (It Depends)

Key Takeaways

I’ve been asked by several clients to explain the apparent disconnect in the latest employment data. The May payroll (or “establishment") survey showed the number of new jobs came in well above expectations—339,000 versus estimates of around 200,000.

At the same time, the household report showed a significant loss of jobs—310,000. The unemployment rate, which keys off the household number, moved up to 3.7%, which was higher than the expected 3.5%.

So which number is right?

Two Employment Surveys, Two Sets of Data

Both sets of data are collected and reported by the Bureau of Labor Statistics (BLS), but there are significant differences in how they’re calculated.

Employment estimates from the payroll survey are a count of jobs from the payroll records of over 100,000 employers. In contrast, the household number comes from roughly 60,000 households surveyed by the Census Bureau and provides an estimate of the number of employed people.

This means that the household survey includes a broader swath of employment groups (e.g., agricultural workers, self-employed, etc.) than the payroll survey. Also, importantly, the household survey only counts each person one time, whereas the payroll survey will count multiple jobs even though it may be the same person who works them.

Which Is More Accurate, the Household Survey or the Payroll Survey?

Some people may assume that the household survey is more accurate. But I hesitate to claim that one statistic paints a more accurate picture than the other. Because they measure slightly different things, they may both be considered “accurate” despite pointing in different directions.

For example, the current discrepancy is easily explained by what’s going on in the economy. The labor market remains tight. People are being hired by corporations, so the establishment survey shows an increase in jobs.

But in many cases, these same folks are simply coming from the ranks of the self-employed, and therefore, show as job losses in the household survey. That would also be true if the newly created job were a second or third job for some workers.

Although the two metrics point in opposite directions, they can both be correct based on what they’re measuring. I would advise viewing both measures to figure out what’s really going on, rather than only relying on one of the measures to the exclusion of the other. So, in answer to the question of which employment statistic is right? I think the right answer is that both are in fact correct.

What Does the Employment Data Mean for the Economy (and Potential Recession)?

Looking at the payroll numbers, I don’t believe that the strong labor market will be enough to stave off a recession.

First, the majority the May job gains were in the hospitality and leisure sectors, which had been decimated during the pandemic. Many of these “new” jobs may simply be replacing self-employed household numbers. That might explain, at least in part, the big difference in the two labor market stats. Indeed, leisure accounted for 75% of the job gains. That’s the lowest-paying sector of the job market.

Further, if you survey all the economic indicators across the economy, you find virtually everything loaded on the bearish/recessionary side of the scale, with the sole exception of the supposedly strong labor market, which is backed only by conflicting data. I sincerely doubt that the economy will magically shift in favor of this one indicator, especially since employment is a lagging indicator.