There is a peculiar ritual that plays out on the first Friday of every month in America. A constellation of analysts, economists, and financial pundits — housed in glass towers from Manhattan to Menlo Park, armed with regression models and decades of credentialed expertise — issues its collective prophecy about the state of the American labor market. The Bureau of Labor Statistics then releases the actual number. And increasingly, the prophecy bears only a passing resemblance to reality.
On Friday, the ritual produced one of its more embarrassing spectacles in recent memory — though in a direction that Wall Street, for once, might not mind being wrong about. Nonfarm payrolls rose a seasonally adjusted 178,000 in March, a reversal from the 133,000 decline in February, and far better than the Dow Jones consensus estimate of 59,000.
That is not a minor rounding error. That is a miss of nearly 120,000 jobs — roughly triple what the experts predicted. The stock market, closed for Good Friday, could not even react. The number simply landed in the silence, waiting for Monday like an unopened verdict.
The question worth asking is not simply why the forecasters missed, but what the persistent, structural inability to accurately read the American labor market tells us about the models, the institutions, and the ideological assumptions baked into both.
The Anatomy of the Miss
To understand why Wall Street got March so spectacularly wrong, one must first understand why February looked so terrible — because the March number is, in large part, a correction of misread February signals rather than a genuine acceleration in hiring.
February’s loss of 92,000 jobs came in part due to a since-resolved strike at Kaiser Permanente that sidelined some 31,000 workers in California and Hawaii. Add to that a bout of harsh winter weather that suppressed construction and outdoor labor, and a data collection methodology that the Bureau of Labor Statistics itself has been refining amid what one economist described as greater month-to-month volatility from model adjustments. Health care companies added 76,400 jobs in March, boosted by the return of those 31,000 Kaiser Permanente employees following the end of the strike.
In other words, a significant portion of March’s apparent surge was simply February’s artificially depressed number bouncing back. The consensus of 59,000 failed to fully account for the mechanical reversal a strike resolution would produce — a forecasting failure that is less excusable than it might appear, since the strike’s end was public knowledge weeks before the report.
But the forecasters’ troubles run deeper than one missed strike. Volatility in monthly payroll employment figures has been trending upward, with the six-month trailing standard deviation increasing to 102,000 in February from a recent minimum of 45,000 in July 2025. When a single standard deviation of error in your forecasting tool is larger than the entire number you are trying to predict, the honest response is epistemic humility, not refined decimal-point estimates. Wall Street offered the decimal points anyway.
The Broken Tools of Prediction
Part of the problem is that the instruments Wall Street has long relied upon as leading indicators have become unreliable in the altered landscape of the post-pandemic, post-immigration-surge American economy.
The ADP National Employment Report — released two days before the BLS number and long treated by traders as a preview — showed a stronger private sector in both January and February than the official BLS data confirmed, a divergence that has eroded confidence in ADP as a predictive tool. This ought to surprise no one who reads the fine print: ADP explicitly states that its National Employment Report is an independent measure of private-sector employment and is not intended to forecast the Bureau of Labor Statistics monthly jobs report, a disclaimer the financial press routinely ignores in its pre-report coverage.
Beyond the tool problem, there is the model problem. The standard frameworks Wall Street uses to project payrolls were calibrated in a labor market defined by mass immigration, relatively stable demographics, and consistent Federal Reserve policy. None of those conditions hold today.
For most of the past decade, the rule of thumb was that the economy needed to add somewhere between 100,000 and 150,000 jobs per month to keep the unemployment rate from rising. That benchmark has now dramatically downshifted in response to the demographic and immigration realities of 2026. New research from the Dallas Federal Reserve suggests the so-called break-even rate — the number of jobs needed each month just to keep unemployment stable — has fallen to near zero, and may even be negative. Net unauthorized immigration turned negative in February 2025 and averaged around -55,000 per month in the second half of last year, with deportations and voluntary departures exceeding new arrivals by a wide margin.
Wall Street’s models were not updated for this reality. They were still solving for a labor market that no longer exists, inputting fresh data into old equations and wondering why the outputs were wrong.
The Deeper Structural Story the Headline Conceals
Even as the media prepares its triumphant “jobs surge” narratives, the honest analyst must resist the temptation to mistake a corrected anomaly for genuine vitality. The March number looks dramatic precisely because February looked so bad — and February looked so bad in part because the prior months had already been quietly revised downward. March’s gains followed a revised loss of 133,000 jobs in February — a steeper decline than the previously reported 92,000. January’s figure was revised upward by 34,000, to 160,000. Combined, January and February employment came in 7,000 lower than prior estimates.
This is the BLS revision problem, and it is chronic. The bureau has spent the past year-plus issuing initial numbers that are later revised — often significantly. Last September, the BLS estimated that benchmark revisions for the year prior to March 2025 would show 911,000 fewer jobs than previously reported. Goldman Sachs estimated the final count would land between 750,000 and 900,000. When a government statistical agency is systematically overstating job creation by nearly a million positions over a year, and Wall Street is building its consensus forecasts on those overstated numbers, the entire exercise becomes a kind of compounding error — fiction layered upon fiction until the benchmark revision forces a reckoning.
Beyond the revision problem is the concentration problem. Without the health care sector, over the past year there would have been a net loss of more than half a million jobs. March was no different: health care companies added 76,400 jobs, with new jobs heavily concentrated in health care and social assistance, which accounted for more than half the jobs created last month. As Indeed Hiring Lab’s economists noted in their post-report analysis, job gains in the healthcare and social assistance sector again did much of the heavy lifting, continuing a pattern of concentrated growth that has propped up the headline numbers for well over a year now. A labor market running on a single cylinder — particularly one driven largely by low-wage home health aide positions serving an aging population — is not a strong labor market wearing a flattering mask. It is a structurally fragile economy in demographic transition.
Of the 60,620 planned cuts announced in March, AI was cited as the reason for 15,341 of them, according to Challenger, Gray & Christmas. White-collar payrolls, meanwhile, have contracted for more than two years straight. The entry-level professional job market — the traditional gateway for college graduates into the productive economy — is quietly disappearing beneath a wave of automation and corporate consolidation.
What the Expert Class Keeps Missing
There is something almost philosophically interesting about the recurring failure of elite forecasters to accurately model the American economy. These are not unintelligent people. They command the finest academic credentials, access to proprietary data sets, and years of institutional experience. And yet, month after month, the numbers make them look like they are guessing.
Part of the answer is structural: economic forecasting is genuinely hard, and the current environment — an ongoing war with Iran driving energy prices sharply higher, with the average price of gasoline topping $4 a gallon for the first time since 2022, alongside an immigration crackdown restructuring the labor supply, alongside AI disrupting white-collar hiring at a pace that no historical model was designed to capture — is extraordinarily complex. No forecaster should be mocked for missing in a moment of genuine uncertainty.
But part of the answer is ideological. Much of the financial establishment spent the past several years insisting that the American labor market was structurally robust, that the pandemic gains were durable, that rising wages were a sign of health rather than a stagflationary warning signal. Wages rising while jobs disappear is not healthy — it is the textbook definition of a stagflationary signal. Companies are paying more to retain the workers they have while refusing to hire new ones. The expert class was slow to acknowledge this because acknowledging it would have required conceding that the policy environment of the past several years — the spending, the deficits, the open immigration that masked structural demographic decline — had produced a brittle rather than resilient economy.
The Trump administration’s immigration enforcement has been blamed, not without some basis, for tightening the labor supply. But the supply tightening, as the Dallas Fed’s research makes clear, actually reduces the number of jobs needed to maintain full employment. White House National Economic Council Director Kevin Hassett cited rising productivity stemming from artificial intelligence improvements as another factor restraining businesses’ need to hire. These are not signs of a failing economy — they are signs of an economy in transition, one whose dynamics the old models were never designed to capture.
Reading the True Ledger
The prudent investor, the prudent policymaker, and the prudent citizen reads past the headline number to the underlying ledger — and that ledger, even after March’s welcome surprise, is not a document that warrants celebration.
The number of long-term unemployed — those out of work for 27 weeks or more — held at 1.8 million in March but has risen by 322,000 over the past year. The labor force participation rate remained at 61.9%, stubbornly below pre-pandemic levels. The unemployment rate dipped partly because the labor force — those working or looking for work — dropped by 396,000 in March, meaning fewer people were competing for jobs. A falling unemployment rate achieved by workers leaving the labor force entirely is not a recovery; it is a statistical convenience.
With inflation well above the Fed’s target and energy prices surging as the Iran war continues, markets expect little movement from the central bank this year — with futures pointing to a 77.5% probability the Fed will stay on hold through the end of the year. The Federal Reserve remains trapped: a labor market that needs support but an inflation picture that forbids relief.
Wall Street missed the March jobs number by a gargantuan margin because it was trying to fit a profoundly transformed economy into frameworks built for a different era. The immigration flows have changed. The demographic composition has changed. The nature of job creation — narrowing to healthcare, hollowing out white-collar sectors, increasingly pressured by AI — has changed. The geopolitical environment has changed. What has not changed is the confident issuance of precise forecasts by institutions that have no particular reason for precision, and a financial media ecosystem that greets each monthly miss with brief embarrassment before resetting to treat next month’s consensus as authoritative.
March’s 178,000 was good news, to the extent that good news exists in a structurally stagnant labor market. But it is the kind of good news that a prudent person receives with gratitude and without illusion — aware that one strong month, driven largely by a strike reversal and a demographic bottleneck, does not change the fundamental challenge of building an economy where the next generation of Americans can find not just work, but the kind of meaningful, well-compensated work that builds families, communities, and a civilization worth having. That challenge will not be solved by forecasting models. And it will not be visible in next month’s headline number.
