The U.S. added 911,000 fewer jobs over the 12 months that ended in March, the BLS said, trimming by a bit more than half the 1.79 million jobs the official data now show. If that holds, it would bring the average pace of seasonally adjusted employment gains from 147,000 jobs a month over the period to a bit over 70,000.
The leisure and hospitality sector, which includes businesses such as restaurants and hotels, likely had 176,000, or 1.1%, fewer jobs in March than the official data show, the BLS reported. Other sectors registering significantly fewer jobs included retailing, professional and business services, wholesale trade and manufacturing. In percentage terms, the information sector registered the biggest hit.
By total numbers of jobs lost, Tuesday’s report marks the largest preliminary revision on record going back to 2000. When measured by percent of total jobs lost, the revision represents the largest since 2009, in the wake of the financial crisis.
Tuesday’s report is just the preliminary part of an annual process in which the BLS updates the job figures from its monthly employer survey using more comprehensive data from state unemployment tax records. The official revision will come in February, and economists suspect it might not be quite as negative.
But the report, even though it is for a period that ended nearly a half year ago, could amplify worries that the job market, which has shown signs of significant weakening since March, started from a lower baseline than previously known. And it comes at a time when the BLS has come under fire from President Trump over the accuracy of its data.
“The benchmark revisions make clear the economy President Trump inherited was even weaker than we thought,” the White House said in a statement. The White House also said the revisions showed “this is exactly why we need new leadership to restore trust and confidence in the BLS’s data on behalf of the financial markets, businesses, policymakers, and families that rely on this data to make major decisions.”
A smaller increase in job growth for the period ended March 2025 had been widely anticipated by private-sector economists and officials at the Federal Reserve. A footnote in a speech by Fed Chair Jerome Powell last month said he expected the level of employment “will be revised down materially” in Tuesday’s report.
The latest revisions cover the months preceding and including the September-December 2024 period, when the Fed lowered interest rates to guard against the risk of a sharper slowdown in hiring.
nvestors in interest-rate futures markets widely expect the Fed to cut interest rates by a quarter-percentage-point at their policy meeting next week. Those expectations were little changed after the release of Tuesday’s payroll revisions, according to CME Group.
For its monthly jobs figures the BLS relies on a survey of about 121,000 employers. The BLS on Tuesday said that its research suggested the overestimation of employment was likely the result of two factors. First, businesses within its survey reported higher employment in its survey than they did in their quarterly tax reports. Second, businesses that responded to its survey had stronger employment than those that had been selected for the survey but didn’t respond.
One of the survey’s shortcomings is that brand new businesses aren’t included in it, while the BLS also can’t quickly determine whether an employer that stops responding to its survey has gone out of business. The BLS uses a statistical model, based on historical data, to estimate these employer “births” and “deaths.” But even though this birth-death model mitigates the problem, the job figures in monthly employment reports can still drift away from what is happening on the ground.
The annual revision process is how the BLS corrects for this drift. To do this it relies on data from state unemployment-tax records that nearly all employers are required to fill out, and which covers about 95% of U.S. employment. The data comes with a substantial lag and is also subject to revision itself, because some tax filers are late.
In its preliminary revision report in August of last year, the BLS estimated the U.S. created about 818,000 fewer jobs than previously estimated for the year ended March 2024. After that report came out, President Trump, then running for election, said that former President Joe Biden and former Vice President Kamala Harris had been cooking the numbers, and that the only reason the data had come to light was because it was leaked.
When the BLS made its official revision in February, the change was actually smaller: about 598,000 fewer jobs, when unadjusted for seasonal swings.
That was still counted as a large downward revision, however. Likewise, even if it isn’t quite as dire as what the BLS reported Tuesday, the official revision will probably still be big.
Economists at UBS suspect part of the problem is that the pandemic’s aftermath created distortions that the BLS’s birth-death model is having trouble keeping up with. When the economy was rebounding strongly, a lot of new businesses got formed—indeed, the model’s not adequately accounting for this contributed to an annual revision for the 12-months ended March 2023 that raised the U.S. employment count by an additional 506,000 jobs.
But the number of business births and deaths subsequently normalized more quickly than the model could keep up with, the UBS economists argue, leading to job growth getting overstated.
The annual revisions also tend to be large at the start of recessions, and the birth-death model cannot adequately capture the drop in new business formations and increase in businesses shutting down.
By JUSTIN LAHART/Wall Street Journal
