Department of Labor Logo United States Department of Labor
Dot gov

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Trying to Understand an Unusual Employment Report

Wednesday, October 18, 2017

This column is called the Commissioner’s Corner, but I’m just keeping the seat warm until a new Commissioner is nominated by the President and confirmed by the Senate. More often than not I feel like I’m back in school, having to learn new concepts from scratch. Fortunately, I’m surrounded by incredibly knowledgeable and dedicated BLS employees who go out of their way to make sure things are done correctly and are very patient in helping me to learn new things. That certainly was the case with the September 2017 Employment Situation report, released on October 6.

The turnaround time from data collection to processing to publication of most BLS data series is very short. That’s the case with the monthly employment and unemployment information. I first saw the payroll employment information about 48 hours before we would release it, and my reaction isn’t suitable for a G-rated blog. What happened? The employment information looked so much different from the recent trend. Fortunately, some more information from that dedicated staff helped me to understand what was going on.

If you haven’t heard, it’s hurricane season. And several storms affected the United States in August and September. Evacuations, damaged businesses, and damaged homes have a lot of implications, including for the job market. Let me give you my description of what the data reveal, using (I hope) some nontechnical terms.

The monthly Employment Situation release contains information from two separate surveys—a survey of businesses (called the Current Employment Statistics program) and a survey of households (called the Current Population Survey). We get different information from each survey, but over time they typically tell a consistent story. For example, during the 2007–09 recession, the business survey showed a decline in jobs, while the household survey showed an increase in unemployment. A consistent story.

So what happened in September 2017?

The business survey asks how many workers were paid for any time during the payroll period that includes the 12th of September. An important fact to understand is that people who did not receive pay for the payroll period are not counted as employed. In September, the business survey showed that the number of jobs in “food service and drinking places” (let’s call them restaurants) declined by 105,000 from the previous month. That’s very different from the trend, which has shown consistent job gains. Workers in these jobs are typically paid on an hourly basis and don’t get paid if they don’t work. This large decline in restaurant jobs in turn affected the overall number of jobs, which declined by 33,000.

Chart showing over-the-month change in food services and drinking places employment

Editor’s note: Data for this chart are available in the table below.

In contrast, the household survey showed an increase in people who were employed and a decrease in the unemployment rate. Once again, it is important to understand definitions. In this survey, people are counted as employed if they had a job but did not work due to bad weather, whether or not they were paid. So those same restaurant workers who were not paid and therefore not counted in the business survey were counted as employed in the household survey.

And the household survey tells us more. In something we call the “bad weather” series, 1.5 million employed Americans were not at work due to bad weather during the week that included September 12. This is the highest number for that series in over 20 years. In contrast, in September 2016 there were only 24,000 people in this category. The number of “bad weather” workers was unusually high because Hurricane Irma happened to fall during the week that included September 12—the reference period for the survey. The figure was much lower for many other major weather events, including Hurricane Katrina in 2005, because those events fell outside the reference period for the survey.

Chart showing the number of people each month with a job in nonagricultural industries but were not at work because of bad weather.

Editor’s note: Data for this chart are available at https://data.bls.gov/timeseries/LNU02036012

So what I thought would be a major story turned out to be easily explained by weather events and differing definitions. And it taught me something new about the wide variety of information available from BLS.

One last note. The national information included in the monthly Employment Situation news release does not include data for the U.S. territories, such as Puerto Rico and the U.S. Virgin Islands. Data for some territories are included with the State Employment and Unemployment news release, which typically is available a couple weeks after the national data. Want to know about how recent hurricanes have affected BLS data? See our page about Hurricanes Harvey, Irma, and Maria.

Over-the-month change in food services and drinking places employment
MonthChange
Jan 201422,700
Feb 20146,500
Mar 201461,000
Apr 201436,600
May 201439,600
Jun 201425,000
Jul 201413,500
Aug 201423,300
Sep 201431,400
Oct 201424,700
Nov 201426,300
Dec 201436,400
Jan 201517,600
Feb 201546,200
Mar 201511,600
Apr 201537,100
May 201533,900
Jun 201546,500
Jul 201539,700
Aug 201526,400
Sep 201536,200
Oct 201552,900
Nov 201530,800
Dec 201538,000
Jan 201630,000
Feb 201628,100
Mar 201632,700
Apr 201616,900
May 201624,600
Jun 201621,900
Jul 201619,100
Aug 201632,300
Sep 201620,800
Oct 201611,500
Nov 201623,700
Dec 201614,700
Jan 201718,000
Feb 201720,600
Mar 201727,900
Apr 201726,100
May 201737,200
Jun 201720,500
Jul 201752,000
Aug 2017(p)8,600
Sep 2017(p)-104,700
Footnotes:

(p) = preliminary