Each month, the Bureau of Labor Statistics (BLS) releases one of the most closely watched indicators in the U.S. economy: its estimate of how many jobs were gained or lost. These employment figures, drawn primarily from the Current Employment Statistics survey of businesses and government agencies, provide a look at payroll growth across industries. For many Americans, it is a measure of whether the economy is “working.” What often goes unnoticed is that the initial job numbers reported are not final. The BLS operates on a structured revision cycle designed to improve accuracy over time. The first estimate is based on a sample of employers and is released quickly to meet demand for timely data. In subsequent months, as more complete survey responses come in, the agency revises the figures. Sometimes the original report overestimates, other times it underestimates.
From 2010 through 2024, the BLS underestimated job growth an average of seven months per year. In 2010, 2014, and 2021, revisions were especially pronounced, with 11 months in each of those years adjusted upward. By contrast, 2023 saw only two months revised higher, signaling a period of tighter forecasting.
Key Takeaways:
📈📈😂2025 Marks a New Era of Overestimation: So far this year (December revisions are still pending), the BLS has overestimated payroll growth in 11 out of 11 months. After years where the agency often underestimated job growth, especially coming out of the Great Recession of 2008, 2025 is shaping up to be a mirror image, with initial reports consistently coming in stronger than the revised data. These revisions indicate that the economy was actually weaker, or generated far fewer jobs, than initially believed.
😐🤔😯ADP vs. BLS: ADP’s payroll numbers are another go-to source for employment information, but their data is a bit all over the place compared to the BLS’s final revised figures. Sometimes they come in higher, sometimes lower, and often the real difference only shows up after the BLS makes its revisions. ADP only looks at its private-sector clients, so it misses government jobs, self-employed workers, and businesses outside its system, etc. BLS, on the other hand, surveys a broader range of employers and uses additional data like unemployment insurance records. ADP can sometimes overestimate or underestimate compared to BLS, and the gap can shift month to month. ADP is a useful early indicator, but the BLS revisions usually provide the final word on where the labor market stands.
☯️☯️Jobs Forecasts Track the Economy or Vice Versa?: When the economy is stable and growing, BLS estimates tend to be closer to the mark. Between 2015 and 2019, revisions were usually small, reflecting steadier growth. In contrast, during difficult periods, like the 2010-2014 recovery from the Great Recession or the post-pandemic years, revisions were much larger. Coming out of 2008, the BLS often underestimated job gains, while in the post-pandemic period, it has been overestimating by a significant margin.
