Recent revisions to U.S. payroll employment data by the Bureau of Labor Statistics (BLS) have prompted renewed questions about the reliability of economic data. In August, the BLS reported that data revisions for May and June 2025 were “larger than normal,” with total nonfarm payroll employment gains revised down by 125,000 jobs and 133,000 jobs, respectively. These changes brought job gains for each month to fewer than 20,000, while July’s figure stood at 73,000 jobs gained.
Despite these adjustments raising concerns that the labor market may be weaker than previously thought, a new analysis from Sylvain Leduc and Luiz Edgard Oliveira of the Federal Reserve Bank of San Francisco suggests that such revisions are not unusual in a historical context. The authors compared recent changes with data from the past six decades using the Philadelphia Fed’s Real-Time Data Set.
“Our updated analysis shows that these revisions remain similar to other large revisions when compared with historical data over the last 60 years. This suggests that incoming data are not generally subject to greater fluctuations—and thus may not reflect greater uncertainty—than in the past,” Leduc and Oliveira wrote.
Their research follows a March 2025 study which found that short-term revisions in payroll employment and consumer price index (CPI) inflation data over recent years were roughly consistent with those seen before the pandemic. According to their findings, “these revisions were roughly the same size over the two to three years ending in 2024 as they were in the years preceding the pandemic, on average.”
The topic has drawn wider attention recently. In June, The Economist published an article questioning whether U.S. economic data is becoming less clear. The BLS also issued a note in July about reductions in data collection related to CPI releases.
Leduc and Oliveira’s review of monthly payroll employment revisions showed that first-round adjustments for January through June 2025 were somewhat higher than those in 2023 and 2024 but still comparable to pre-pandemic averages. They noted that larger-than-usual monthly changes have occurred before, including as recently as December 2023 and January 2024.
Further analysis indicated that even though some recent monthly corrections were sizable, they fall within a range observed multiple times since November 1964—the earliest period covered by available real-time revision records. Their assessment found that there is about a seven percent chance of experiencing a revision as large or larger than June’s adjustment.
“Overall, some recent revisions to payroll employment gains data were large relative to those issued over the last year or so. Still, over the last 60 years, there have been examples of first revisions that were of similar or greater magnitude,” Leduc and Oliveira concluded.
They emphasized that current levels of volatility do not necessarily indicate increased uncertainty compared with previous decades.
The views expressed by Leduc and Oliveira do not necessarily reflect those of management at the Federal Reserve Bank of San Francisco or its Board of Governors.
