Philadelphia Fed Confirms BLS Is BS

2nd Quarter Job Numbers Were Pure Fiction

The “official” narrative during the second quarter of the year was that there was good job growth, a reflection of a labor market that was (and allegedly still is) strong.

The April Employment Situation Summary claimed 428,000 jobs were added that month.

The May Employment Situation Summary claimed 390,000 jobs were created during the month.

The June Employment Situation Summary claimed 372,000 jobs were created during the month.

All told, the Bureau of Labor Statistics claimed 1,190,000 jobs were created during those three months.

Last week, the Philadelphia Federal Reserve shredded the narrative, stating that the actual number of jobs created during those three months was a whopping 10,500.

Last week, the Philadelphia Fed released its Early Benchmark Revisions of State Payroll Employment for the second quarter. The Philly Fed releases quarterly benchmark revisions to augment the BLS data, which gets benchmarked just once per year.

The large revisions occur primarily because the preliminary state estimates are based on a small sample of firms, while subsequent benchmark revisions incorporate other BLS data based on a full count from nearly all firms. Moreover, the BLS issues its benchmark revisions for state employment estimates just once a year. However, the full count of data is issued quarterly, which offers an opportunity for researchers to create their own early benchmarks on a timelier basis.

The Philly Fed’s early benchmarks are based on the BLS Quarterly Census of Employment and Wages, which contains more comprehensive payroll and employment information than the monthly Employment Situation estimates (which are based on sampling). The benchmarking revisions performed by the Philly Fed takes that QCEW data—which uses the BLS data with its most recent revisions and updates—to update the BLS annual benchmarks for employment.

The BLS data itself gets revised in between the monthly releases, as late data that would normally go into the monthly release is received. After the April Employment Situation Summary was released, total non-farm employment data was revised, resulting in the total jobs created figure for April being revised downward from 428,000 to 368,000 on a seasonally adjusted basis.

May’s jobs created number was revised downward to 386,000.

June’s jobs created number was revised downward to 293,000.

The BLS’ own revised figures for total non-farm employment, therefore, indicate that total job creation for the second quarter was 1,047,000 jobs rather than 1,190,000, a decrease of 143,000 jobs.

Yet if the Philly Fed’s early benchmarks are at all accurate, these downward revisions were not nearly enough.

In the aggregate, 10,500 net new jobs were added during the period rather than the 1,121,500 jobs estimated by the sum of the states; the U.S. CES estimated net growth of 1,047,000 jobs for the period.

Payroll jobs in the nation remained essentially flat from March through June 2022 after adjusting for QCEW data:

• Less than the 3.0 percent growth indicated by the sum of the states
• Less than the 2.8 percent growth indicated by the U.S. CES estimates

According to the Philly Fed, there was effectively no job growth at all during the second quarter of the year.

Readers may recall my skepticism about the credibility of the BLS Employment Situation Reports on several occasions, including May and June of this year.

BLS Jobs Numbers Look Good, BUT....

The BLS Report Is In

The June report was particularly suspect, as the internals of the data simply did not support the robust job growth that was being claimed.

One reality is unavoidable. Despite the proclamation at the top of the labor report that payrolls grew by 372,000 in June, the BLS data shows that total employment, non-agricultural wage and salary employment, non-agricultural self-employment, and part-time employment all declined in June. The “increase” of 372,000 payroll jobs is ultimately just Lou Costello Labor Math. When the labor force is declining there can be no increase in jobs, period.

Even the May Job Opening and Labor Turnover Summary report had significant problems with credibility.

Signs Of More Lou Costello Labor Math?

Despite what the Employment Situation data indicates on job growth, the JOLTS data indicates that a significant number of workers separated during the 2020 recession have still simply not come back into the labor force. The labor market is currently “tight” not just because the demand for workers has risen since the pandemic, but also because the supply of workers remains stubbornly smaller than it was.

The “Great Resignation” is continuing even into the current recession. And Lou Costello math continues to be the dominant discipline at the BLS.

The Philadelphia Fed has essentially confirmed my suspicions about the jobs numbers reported by the BLS being little more than Lou Costello Labor Math.

And the BLS Household Survey portion of the Employment Situation Summary reports largely confirms the Philadelphia Fed’s early benchmarks (the jobs numbers that are claimed each month come from the Establishment Survey).

During the second quarter of the year, according to the Household Survey the number of employed persons dropped by 353,000 in April, rose by 321,000 in May, and dropped by 315,000 in June.

The Household Survey data for the second quarter makes the jobs claims of the Establishment Survey mathematically impossible. One cannot have more than one million new jobs and have the number of employed people decrease by 347,000 in the same calendar quarter. One or both of the employment metrics, the Household Survey and the Employment Survey, is simply wrong.

The Household Survey data indeed has its own question marks, such as the June data showing a decline in the number of employed people, the number of people not in the labor force, and the number of unemployed people all in the same month. Given that in theory a person has to be in one of those three demographics, how all three can decline at the same time is a bit of a mystery, as a decrease in employment should produce an increase in either unemployment or people not in the labor force.

However, the Household Survey does show overall employment growth during the first quarter, followed by an employment contraction during the second quarter.

The Philly Fed’s first quarter early benchmark revisions supports this data, as it showed that the BLS actually understated job growth during the first quarter.

In the aggregate, 1,695,800 net new jobs were added during the period rather than the 1,614,800 jobs estimated by the sum of the states; the U.S. CES estimated net growth of 1,616,000 jobs for the period.

Payroll jobs in the nation grew 4.6 percent from December 2021 through March 2022 after adjusting for QCEW data:

• More than the 4.4 percent growth indicated by both the sum of the states and by U.S. CES estimates

Going by the Philly Fed’s early benchmarks, there was a small surge of employment growth during the first quarter which tapered off during the second.

The projection by the Philly Fed of problematic job growth

While none of these employment metrics perfectly align with each other, and while all of these employment metrics can be challenged as to accuracy and methodology of computation, the Philly Fed’s early benchmarks are far closer to the Household Survey than to the Establishment Survey, which suggests that the Household Survey is far more reliable as a gauge of employment growth than the Establishment Survey.

At a minimum the confluence of the Philly Fed’s early benchmarks with the Household Survey add considerable weight to the assessment of the job growth reported in the Establishment Survey as frequently little more than Lou Costello Labor Math. Whether the data is simply flawed or if the data is fudged, time and again the Establishment Survey simply does not add up.

Contrary to what the White House wants us to believe, the US economy is not producing lots of jobs, and has not been producing lots of jobs. On the contrary, the employment outlook remains problematic and uncertain.

Employment during a recession tends to be that way.