Labor Data Watch
December 2025
The BLS headline above (+50K) is what was reported on release day. Subsequent revisions changed it to -17K — a 67K swing. The headline was not just wrong — it was directionally wrong. The 24-month charts below use the revised figure.
What the Data Shows
BLS reported +50K jobs in its preliminary release — a number that would later be revised to -17K after the annual benchmark. ADP counted +41K independently. Both sources agreed: 2025 was the weakest job growth year since 2003, averaging just 49K per month versus 168K in 2024.
Average hourly earnings rose 3.8% year-over-year per BLS, while ADP tracked job-stayer wage growth at 4.4%. With CPI inflation at 2.9%, workers are gaining purchasing power by both measures — though the margin above inflation has been narrowing since mid-2024.
The headline U-3 rate edged down to 4.4% from November’s four-year high of 4.5%. But the broader U-6 — which counts discouraged workers and the involuntarily part-time — sat at 8.4%. That 4.0-point gap hides millions of people whose labor market experience is far worse than the headline suggests.
BLS revised October down by 68,000 jobs (from -105K to -173K) and November down by 8,000 (from +64K to +56K). Combined: 76,000 fewer jobs than headlined. The majority of recent months have been revised downward. Weeks later, the annual benchmark would cut 898,000 more jobs from all of 2024-2025.
ADP closed December at $257, down 21% from its May 2025 peak of $326. Paychex closed at $112, down 29% from its peak of $158. The companies that physically count America’s paychecks have been in sustained decline — institutional investors with billions on the line are pricing in employment weakness.
Jobs Added Each Month
24-month comparison: how many jobs did each source say were added?
The 24-month trend shows clear deceleration. Early 2024 averaged 150K+ jobs per month. By late 2025, BLS showed outright job losses in five of the last twelve months (post-revision), while ADP remained positive but at much lower levels. The October 2025 divergence is the most extreme in the dataset: BLS at -140K versus ADP at +103K — a 243K gap driven by the government shutdown’s impact on data collection.
Wages vs. Inflation
The question that matters most: are workers actually getting ahead?
Both wage measures have stayed above CPI inflation for the full 24-month window, meaning workers have gained real purchasing power. But the trend is downward — BLS earnings growth peaked at 4.4% in January 2024, declining to 3.8% by December 2025. ADP shows the same fade, from 5.2% to 4.4%. The margin above inflation is narrowing as CPI firmed in late 2025.
Unemployment Rate
The official number vs. the real number — including people who gave up looking and those stuck in part-time work.
The U-3/U-6 gap widened sharply in late 2025 — from a steady 3.4-3.5 points through 2024 to 4.0+ points by December 2025. November’s 4.2-point gap was the widest since the pandemic recovery. This widening signals growing structural underemployment: more people involuntarily part-time, more discouraged workers leaving the labor force entirely.
The Revisions
What they quietly changed about the months you already forgot.
Net revision: -76K jobs October’s 68K downward revision is the largest single-month revision in this 24-month window. The original October figure of -105K — already negative — was revised to -173K, meaning the job market during the government shutdown was significantly worse than initially reported. Combined with November’s -8K, 76,000 jobs were quietly erased from prior headlines.
The Market Already Knows
ADP and Paychex together process payroll for over 40% of American workers. Their stock prices reflect what institutional investors — with billions on the line — actually believe about the labor market.
When the companies that physically count the paychecks see their stock prices fall 21-29% over seven months while the government reports the weakest job growth since 2003, the divergence is not academic. ADP and Paychex together process payroll for over 40% of American workers. Their sustained decline since May 2025 tells a story the headline jobs numbers took months of revisions to catch up to.
Discussion
Methodology
Labor Data Watch compares government employment estimates from the Bureau of Labor Statistics (BLS) Employment Situation report against private-sector payroll data from the ADP National Employment Report, produced in partnership with the Stanford Digital Economy Lab. We track every metric from initial release through all subsequent revisions.
Divergence grades are calculated using z-scores based on the trailing 60-month absolute difference between BLS and ADP private payroll estimates. Grades range from A (strong agreement) to F (extreme divergence requiring investigation).
Every number cited includes a source citation. No editorial judgment is applied to the data — we present both sources side by side and let the divergence speak for itself. Leadership attribution records who was in charge when each data point was released, creating a chain of accountability.
Note: The BLS numbers in the 24-month charts reflect the most current revised figures, including the February 2026 annual benchmark revision (-898,000 jobs). The headline stat cards show the preliminary numbers as originally reported, because those are the numbers that drove the public narrative.