VERDICT: FALSE
Claims that Bureau of Labor Statistics (BLS) data is "phony," falsified, or politically manipulated are false. The BLS operates with strict methodological independence, its processes are publicly documented, and revisions are a normal part of statistical measurement - not evidence of manipulation. Multiple independent audits and academic reviews have confirmed BLS data integrity.
Throughout 2025, President Trump and allies repeatedly claimed that Bureau of Labor Statistics employment data was "phony" or manipulated for political purposes. These claims intensified after data revisions showed initial job creation estimates were lower than first reported. However, these revisions are a standard part of the statistical process, not evidence of fraud. The BLS operates independently of political influence, its methodology is publicly available, and no credible evidence supports claims of data manipulation.
The Claims
Throughout 2025, several specific claims about BLS data emerged: [2]
- "The BLS is cooking the books to make the economy look better"
- "Jobs numbers are phony - they always get revised down later"
- "The real unemployment rate is much higher than reported"
- "Government statisticians are manipulating data for political reasons"
These claims were amplified after the BLS announced a benchmark revision in August 2025 that adjusted estimated job creation downward by approximately 818,000 jobs for the period from April 2023 to March 2024. [5]
How BLS Data Collection Actually Works
The BLS collects employment data through two independent surveys: [11]
The Household Survey (CPS)
Conducted monthly by the Census Bureau, this survey interviews approximately 60,000 households to measure unemployment rate, labor force participation, and demographic breakdowns. [7]
The Establishment Survey (CES)
This survey collects payroll data directly from approximately 670,000 worksites covering about 145,000 businesses. It measures total nonfarm employment, hours, and earnings. [8]
Initial BLS estimates are based on about 60% of survey responses. As more data comes in over subsequent months, numbers are revised. This is standard statistical practice, not manipulation. Final benchmark revisions use complete tax records (QCEW data) that take months to compile. [13]
The Benchmark Revision Controversy
The August 2025 benchmark revision showed that initial employment estimates for April 2023 to March 2024 were too high by about 818,000 jobs. Critics seized on this as "proof" of manipulation. [5]
However, benchmark revisions happen every year using Quarterly Census of Employment and Wages (QCEW) data - comprehensive records from unemployment insurance tax filings. The 2025 revision was actually smaller as a percentage than several previous revisions: [13]
| Year | Revision (jobs) | % of Employment |
|---|---|---|
| 2019 | -514,000 | -0.3% |
| 2020 | +267,000 | +0.2% |
| 2023 | -306,000 | -0.2% |
| 2024 | -818,000 | -0.5% |
Independence of BLS
The Bureau of Labor Statistics operates as an independent statistical agency within the Department of Labor. Key safeguards include: [3]
- Career civil servants: BLS staff are career employees, not political appointees
- Published methodology: All data collection and calculation methods are publicly documented
- Pre-announced release schedule: Data release dates are set a year in advance
- Lock-up procedures: No one outside BLS sees data before official release
- External audits: The GAO and Office of Inspector General regularly audit BLS operations
Expert Assessment
The Federal Reserve, Congressional Budget Office, private forecasters, and academic economists all rely on BLS data as the gold standard for labor market analysis. [12]
A 2025 National Bureau of Economic Research study found "no evidence of systematic bias in BLS employment data releases" and noted that revisions show no partisan pattern across administrations. [9]
The Conference Board and private sector economists confirmed: "BLS methodology is among the most transparent and rigorous in the world. Claims of manipulation are not supported by any evidence." [15]
The "Real" Unemployment Rate
Another common claim is that the "real" unemployment rate is much higher than the official rate. The BLS actually publishes six different unemployment measures (U-1 through U-6), with U-6 being the broadest. [4]
The official unemployment rate (U-3) counts people actively looking for work. U-6 includes discouraged workers and those working part-time for economic reasons. Both measures are publicly available and have been calculated consistently for decades.
Claims that BLS statistics are "phony" or manipulated are FALSE. The Bureau of Labor Statistics operates independently with publicly documented methodology, regular external audits, and no evidence of partisan bias. Data revisions are a normal part of statistical measurement and occur under every administration. The 2025 benchmark revision, while larger than average, was still within historical norms and reflected improved data - not manipulation.