Four Reasons Job Growth Numbers Don’t Match

One of the ways to track how the U.S. economy—in particular, the labor market—is performing is to keep an eye on employment growth. You could check several sources and data series if you wanted to know how many jobs were added to the U.S. economy each month, for example.

Via the St. Louis Fed’s online economic database FRED, the chart below shows monthly employment gains from January 2021 to January 2022 using four different indicators. Three of the data series are from the U.S. Bureau of Labor Statistics, or BLS, and one is from the ADP Research Institute.

But wait, what gives? Why are the numbers for employment gains different across the indicators every month?

St. Louis Fed Senior Economist Charles Gascon outlined four reasons for the differences, other variations to bear in mind, and how to get a sense of the overall trend.

Reason No. 1: Where the Data Come From

“All of these numbers come from surveys or samples of the population, so they are not as precise as they appear at first glance,” Gascon said.

And, he said, they are revised, meaning they are changed as new information comes in.

One example: A Jan. 24, 2022, FRED Blog post suggested by Senior Economic Education Specialist Diego Mendez-Carbajo plotted revisions to “All Employees: Total Nonfarm Payrolls” data for December 2020 to November 2021 from the BLS. (See the ALFRED chart below.)

The FRED Blog highlights interesting information in FRED, and the chart was plotted using ALFRED, which shows archival data—that is, data as of a certain release date or vintage. This particular chart shows vintages over 12 months (from Feb. 5, 2021, through Jan. 7, 2022).

Gascon and Paul Morris, a former St. Louis Fed research associate, suggested a 2017 FRED Blog post on revisions to employment numbers for the St. Louis Fed’s District.

That post pointed out that revisions happen “because counting new jobs is a difficult process that relies on samples and advanced statistical techniques.”

Which brings us to Gascon’s second reason for the differences between the employment gains numbers.

Reason No. 2: The Challenges of Estimating

Gascon laid out some statistics:

• Total U.S. employment is about 150 million.
• About 10 million people move into or out of employment each month.
• Job growth is the net change in employment, and the difference between those moving into and out of employment is “typically a change of about 0.1%,” Gascon said.

“So, if you think about getting an estimate of the net change of 150 million each month, this is a difficult task,” he said.

Reason No. 3: The Type of Data Used

“Total Nonfarm Private Payroll Employment” numbers are published monthly by the ADP Research Institute in collaboration with Moody’s Analytics, according to institute information about the report. The institute is connected to payroll and human resource services company Automatic Data Processing Inc.

ADP’s payroll database, which includes about a fifth of U.S. private payroll employment, allows the institute to estimate employment change before the BLS employment report is released, according to an article on the ADP report’s methodology (PDF).

But the early information might not be fully accurate.

“ADP uses data from their clients to produce an early estimate of what they expect the private-sector employment growth to be, but it can be off,” Gascon said.

Reason No. 4: Different Groups Are Measured

The BLS conducts a survey of businesses and one of households, Gascon said.

“The household and businesses surveys may differ in employment growth because they sample slightly different groups due to the methods,” Gascon said.

For example, here’s what the BLS has to say about some of the differences between the Current Employment Statistics survey, (or payroll survey), which goes to businesses, and the Current Population Survey, (or household survey).

“Employment estimates from the payroll survey are a count of jobs, while the household survey provides an estimate of the number of employed people,” the BLS said in a comparison of the surveys. “If a person changes jobs and is on the payrolls of two employers during the same reference period, both jobs would be counted in the payroll survey estimates. The household survey, on the other hand, will simply reflect one employed person in its measure.”

The surveys also go out to different numbers of respondents. So the levels they show are different. But that doesn’t mean each data series always tells a different story.

If you convert the series in FRED to tally the growth rate, “you’ll see all series move in a similar direction,” Gascon said.