(Bloomberg) -- The US created 818,000 fewer jobs in the year through March than initially estimated — the largest downward revision in 15 years, the Bureau of Labor Statistics said on Aug. 21. While those adjustments usually fly under the radar, the recent figures have caused a commotion — both on Wall Street and in Washington DC. Former President Donald Trump, the Republican nominee, has claimed that the early estimates were fabricated and the Biden administration “defrauded the people of our country.” But while the adjustment was a big one, the revisions are part of a decades-long, albeit confusing, exercise aimed at achieving a more accurate picture of the labor market.
What caused the commotion?
It’s rare that a preliminary benchmark revision to payrolls gets attention outside the economics community but interest in the report gained immense traction on Wall Street and in politics. First, there were forecasts for a steep decline. The actual figure reaffirmed bets US central bankers will begin lowering interest rates soon.
With the November presidential election drawing closer, Trump leaned on the figures to cast doubt on the integrity of the Biden administration and its economic record.
Adding to the confusion, the agency responsible for the report failed to publish the figures until well after the 10 a.m. release time. That prompted economists from at least three Wall Street banks to call the BLS for the number. A scramble ensued, with other firms and media outlets, including Bloomberg News, trying to obtain the figures, too.
How do payroll revisions work?
The monthly payrolls data that the BLS publishes every month comes from a establishment survey of some 119,000 employers representing nearly 630,000 workplaces throughout the US — which accounts for about a third of all US jobs.
This sample of employers is benchmarked once a year primarily to information from the Quarterly Census of Employment and Wages program that collects data from states’ unemployment insurance tax records. The figures that were just released are the agency’s preliminary benchmark estimate — the final numbers for that period are due in February.
While the QCEW represents more than 95% of US employers, the drawbacks are that it is a less timely data source and subject to revisions itself.
What explains the discrepancy?
In addition to the smaller sample size, the BLS employment data also relies on a so-called “birth-death model” — an adjustment the agency makes to the data to account for the net number of businesses opening and closing. Some economists argue that adjustment has been off due to pandemic-related distortions. According to Anna Wong of Bloomberg Economics, the main reason for the inaccuracy of nonfarm payrolls are outdated assumptions in the agency’s model. “We estimate the birth-death model contributed half of the downward revisions,” she wrote.
Is there more to it?
Because the QCEW report is based on unemployment insurance records — which undocumented immigrants can’t apply to — the data are likely to have excluded some unauthorized workers that were included in the initial payroll estimates. However, the biggest revisions came from white-collar sectors that are less likely to employ undocumented immigrants, which suggests the impact from immigration was smaller than some economists had anticipated.
Was the size of the revision unusual?
While Wednesday’s data marked the largest revision since 2009, previous prints have also been noteworthy. In 2019, the figures were revised down by more than half a million payrolls.
There have also been sizeable upward revisions in recent years. The BLS undercounted payrolls by nearly half a million in 2022 when the economy was recovering from the pandemic. And in 2006, the upward revision was over 800,000.
Other government surveys are also routinely revised, and the numbers often change as more responses are collected. Response rates have been declining for years — especially since the pandemic — which can negatively affect the quality of the data and make them prone to larger revisions.
How do the revisions change our understanding of the labor market?
The revisions suggest the labor market started moderating much sooner than originally thought, while reaffirming it remains relatively healthy. Assuming the changes are distributing proportionately throughout the year, the figures indicate the US added some 174,000 jobs every month — compared to 242,000 before. That’s still above the pre-pandemic average.
What does that mean for the Federal Reserve?
The Fed was already on track to start lowering interest rates in September before the revisions came out. Now that inflation has largely subsided, the figures are likely to give policymakers further assurance that it’s prudent to start easing policy before the labor market cools too much. “We see little impact on Fed policy,” wrote Aditya Bhave at Bank of America Corp. “It won’t be spooked by news that the labor market was ‘solid’ rather than ‘strong’ last year.”
What about Trump’s fraud accusations?
It’s not the first time GOP members have accused a Democratic administration of cooking the books when it comes to the labor market. Under President Barack Obama, they argued the government was favoring an unemployment metric that made the economy look better. Democrats traded similar accusations when Republicans in Congress rewrote the way the economic effects of legislation are measured.
But there’s no evidence the BLS figures are fraudulent. If anything, the discrepancy between the early estimates and the revision comes from the agency’s methodology not being able to capture the unique post-pandemic dynamics as well.
--With assistance from Mark Niquette and Molly Smith.
©2024 Bloomberg L.P.
