The following article, 911,000 Jobs That Never Were, was first published on The Black Sphere.

I can see Joe Biden now, asking himself as he feels his pockets: “Now where did I put those jobs?”

No surprise for Americans to learn that the Biden administration lied about job creation. Turns out, nearly 911,000 jobs that were counted between April 2024 and March 2025 weren’t real. These “jobs” remind me of the 81 million votes that Biden supposedly got. But I digress.

The Bureau of Labor Statistics (BLS) corrected the jobs numbers after reviewing more comprehensive data, post-firing of the previous director of the BLS during Biden’s final stretch in office. And what of the legitimacy of the data used by consumers, media, and policymakers, and the claims now being made by the Trump White House about what they “inherited.” Yes, Trump really inherited a mess.

The Big Reveal and What Changed

• As mentioned, the preliminary benchmark revision dropped job-growth numbers by 911,000 for the year ending in March 2025.

• In addition, monthly job-gains reported earlier (~147,000/month) are now halved, to about 70-71,000/month, once corrected.

• Major sectors hit hardest: leisure & hospitality (down ~176,000 jobs), professional & business services (-158,000), retail (-126,000), manufacturing, government jobs, etc.

The BLS does this kind of correction every year via its benchmark process: comparing survey-based monthly estimates to more complete administrative records like state unemployment insurance tax filings and business birth/death data. However, this year’s correction is the largest preliminary downward revision since these benchmarks began being tracked regularly.

The Historical Arc: Are Such Errors New?

Far from novel. The BLS has had to revise its job estimates downward (and occasionally upward) before. Two relevant examples:

• The year ending March 2024 saw an earlier major revision: initially the estimates were revised downward by ~ 818,000, though that final correction was somewhat smaller (~598,000) when all data were in.

• The “birth-death” model (how many businesses open vs. close) is perennially sticky; it’s by its nature an estimate, necessarily imperfect. In periods of rapid economic change (post-COVID, inflation shocks, supply chain jolts), its margin of error tends to grow.

So yes, humans, computers, and economists screw up sometimes—but this is not a trivial correction. Over-reporting nearly a million jobs over a year is a lot of air in the tires.

The Political Spin Machine Goes to Work

Enter the hypocrisy, take a seat—it’s showtime.

• Trump’s camp is celebrating, of course. The White House issued a statement saying the revision “shows the economy President Trump inherited was even weaker than we thought” and accusing the Biden years of having “job growth…overstated by roughly 1.5 million workers” when combining past corrections.

• They also say this justifies replacing the BLS head (Erika McEntarfer was fired) and installing E.J. Antoni as commissioner to “restore trust.”

• Predictably, the matter of “illegal immigrants, poor data, government handouts, and flood of federal spending” shows up in the rhetoric. But while those might sound like sharp commentary, they’re not proven reasons for the overstatement. The BLS is less about narrative, more about method. The BLS itself says this benchmark revision is a regular procedure.

• Also, note the irony: critics call for reform, better data, more transparency—but political pressure on statistical bodies often undermines those very qualities. When the story becomes “the bureaucracy is biased,” data integrity becomes a tool in the political fight rather than a neutral fact.

What This Means (Beyond the Soundbites)

So, is this just fodder for tweets and campaign ads, or does it affect real life? Let’s unpack:

1. Trust in economic data is shakier than ever. If job numbers can be off by almost a million, imagine inflation, trade, GDP, or debt figures under pressure. Public skepticism is going to rise. If people believe numbers are inflated or manipulated, they’ll discount official reports wholesale—which creates its own danger.

2. Federal Reserve policy decisions often rely on signals from employment. Overestimating job growth can delay interest rate cuts, or misread inflation risks. Indeed, analysts say that weakening shown by this revision gives more ammunition for rate cuts.

3. Legacy and political narrative. Biden’s defenders will now need to explain away these inflated numbers; Trump’s team will hold them up as proof of deception or incompetence. But nuts to both: Polishing or vilifying numbers after the fact doesn’t always change the lived reality of paychecks, inflation, or cost of living.

4. Agency morale and independence are at stake. Firing a statistical commissioner after weak data reports, or amid controversy, can risk signaling to staff: toe the party line, or you’re out. That’s not how robust, impartial statistics work.

5. Expect more revisions. The preliminary number for this year is 911,000 fewer. Final benchmark (due early next year, February 2026) could adjust it further.

Where the Hypocrisy Dances In

• Critics who decried the “rigged numbers” under Democrats now push the same data as vindication. Fair enough—if you believe the data were inflated. But where was the scrutiny last year when numbers pointing up were being cheered? Selective indignation is still hypocrisy.

• Calling out “illegal immigrants” as the cause, when in fact the BLS revision process depends largely on administrative data like tax records, unemployment insurance filings, and birth/death business data. That’s not immigration, that’s bookkeeping. If anything, overly blaming one factor like immigration is distraction from method errors. (Also: the effect of unauthorized immigrants is largely irrelevant in QCEW data, which counts employers’ reports.)

• Demanding immediate policy action (rate cuts, leadership overhaul) based on preliminary data is as reckless as celebrating early with inflated data. If patience is a virtue, it’s one both sides should practice.

Context & Historical Perspective

To understand how this fits into broader U.S. economic history, some folks to consider:

• Great Depression & New Deal data reform: One of the outcomes of the chaos of the 1930s was a demand for better statistical agencies and more rigorous labor data (unemployment, job creation, etc.). The idea then was simple: governments without good data are governments flying blind. The BLS was formed in 1884, but its role, methods, and visibility expanded drastically during the 20th century as economies got more complex.

• Pre- and post-2008 financial crisis: Many thought job numbers in recession recovery periods were overestimated because of flaws in models that try to estimate new businesses. The “birth-death” model has been criticized for giving overly optimistic estimates during good times, and underestimating closures during slumps. The shock this time is magnitude and how quickly corrections mount up.

• COVID-19 disruptions: The pandemic threw business birth/death estimates off, supply chain problems, labor force participation shifts, remote work, and “new business formations” that are harder to track quickly. All of those stress test every statistical method used by agencies like BLS. So, this isn’t just politics—it’s methodology under pressure.

What Should Come Next (If We Care About Truth Over Spin)

• Transparency in methodology: Let the public and independent economists see how the BLS builds its “birth‐death” model, how it handles nonresponses, delays, business openings/closings, immigration adjustments, etc.

• Protect the independence of statistical agencies: If BLS leadership can be fired because numbers look bad, expect people inside to self-censor or pull punches.

• Timely releases of administrative data: The lag in administrative data is one reason we need revisions. If state tax records etc. could flow more quickly, the preliminary numbers would be more accurate.

• Civic and media literacy: Journalists, policymakers, and the public should treat monthly job reports with caution—and understand that preliminary = subject to change. Treat them as “good faith estimates,” not gospel.

• Balanced political discourse: Sure, point out failures. But also avoid converting statistical corrections into blanket accusations of criminality or conspiracy unless there’s evidence. One side’s disaster should not be the other side’s victory lap.

So what’s the big irony here?

The people who cried “lies!” about job reports under one administration are now holding up those same inflated numbers as evidence that the previous administration was lying. But inflation in job reports, like inflation in prices, often catches everyone unawares until it’s too late.

We’re left with several uncomfortable truths:

• Our job-creation narrative can shift dramatically overnight—not because of scandal, but because of better counts.

• Political claims grounded on preliminary or inflated numbers are brittle.

• The economy’s state—real, lived, measured—is more modest (or more fragile) than many hoped or said.

I see a system where data, power, and narrative collide—and the human costs (workers, families, trust) rarely get the limelight. The “911,000 jobs that weren’t” might just be the biggest job number punchline of the season.

Continue reading 911,000 Jobs That Never Were …