'Phony' BLS Statistics: The Assault on Government Data

The monthly jobs report is built from two independent surveys, not one. The household survey and the establishment survey are collected separately, use different methodologies, and are checked against each other every month. ^[1][6]

Conducted monthly for BLS by the Census Bureau, the Current Population Survey samples about 60,000 eligible households — roughly 110,000 individuals each month — to measure the unemployment rate, labor force participation, and demographic breakdowns. ^[7]

This survey collects payroll data directly from approximately 622,000 worksites, representing about 119,000 businesses and government agencies. It measures total nonfarm employment, hours, and earnings. ^[8]

On August 21, 2024 — three months before that year's presidential election, not after it — the Bureau of Labor Statistics announced a preliminary benchmark revision showing that initial employment estimates for April 2023 to March 2024 had been too high by about 818,000 jobs (a figure later finalized at a smaller 589,000). Critics, including President Trump, have cited it as "proof" of manipulation. ^[5]

Benchmark revisions happen every year, using Quarterly Census of Employment and Wages (QCEW) data — comprehensive records drawn from state unemployment insurance tax filings — to true up the sample-based monthly estimates. A second, larger preliminary revision of 911,000 jobs followed in September 2025 for the April 2024–March 2025 period. Federal Reserve Bank of Cleveland economists who examined the full 1980–2025 revision history called the recent stretch "elevated" relative to the historical average but found no "structural break" in the data — meaning the size of the corrections, while unusually large, still falls within the pattern of routine annual revisions rather than pointing to manipulation. ^[13]

The Bureau of Labor Statistics operates as an independent statistical agency within the Department of Labor, stating that it conducts its work "with independence to ensure that our data and analyses are objective and free of partisan influence." ^[3] Its commissioner serves a fixed four-year term subject to Senate confirmation; McEntarfer herself had been confirmed by an 86-8 bipartisan vote in January 2024, part of why former commissioners from both the Trump and Obama administrations called her August 2025 firing "groundless." ^[5]

No credible investigation has found evidence that BLS data is manipulated for political purposes. The Federal Reserve incorporates BLS employment data directly into its policy deliberations — the January 2025 Federal Open Market Committee minutes noted that an upcoming benchmark revision "could provide more clarity regarding labor market conditions." ^[12]

An October 2025 review by Bloomberg Economics, published through the Peterson Institute for International Economics, dug into decades of BLS revisions "with an eye to detecting any evidence of malfeasance" and concluded "none has been identified." ^[9]

A separate analysis of monthly payroll revisions since 2000 found error rates that "don't differ much by administration," undercutting claims that BLS numbers tilt toward whichever party controls the White House. ^[15]

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) is the total number of unemployed people as a share of the civilian labor force. U-6, the broadest measure, adds discouraged workers, others marginally attached to the labor force, and people working part-time for economic reasons. Both measures are published monthly and have been calculated on a consistent basis for decades. ^[4]