Each year, the BLS conducts an annual revision of job creation figures. The estimate of this revision is published at the end of the summer (this is today's figure), and the final revision is published in February.

A revision that changes everything?

The job creation figures published each month are based on surveys, whose response rates have declined significantly in recent years, which does not help to provide an accurate picture of reality.

The surveys must therefore be reconciled with "hard" data. This results in an annual revision based on the Quarterly Census of Employment and Wages (QCEW), which counts jobs covered by the unemployment insurance tax system.

What we see this year, as in 2024, is that there is a very significant revision. On a monthly basis, this represents 76,000 fewer jobs. This gives an average of 71,000 jobs created between April 2024 and March 2025. Imagine how different market reactions and conversations would have been with these figures each month.

In recent weeks, most data has shown a weakening of the labor market. This revision therefore further strengthens the likelihood of a Fed rate cut.

Yesterday, the New York Fed published its monthly survey of consumer expectations, which shows a sharp increase in the number of respondents who believe it would be more difficult to find a job if they became unemployed. The expected probability of finding a job in such a situation has fallen to 44.9%, the lowest level since June 2013.

Pressure on the BLS

Politically, however, this revision may allow Donald Trump to say that he inherited a labor market that was already weakening, since the period covers only three months of his presidency. Trump will therefore be able to say that these are "Biden's data" (Biden's figures).

He also has another obvious target: the Bureau of Labor Statistics (BLS), the agency that publishes employment figures, accused of failing in its mission to provide reliable data.

In reality, revisions are a normal process for a statistical agency. Over time, more data is collected, which allows estimates to be refined. As Neil Dutta of Renaissance Macro rightly points out, "the only country that does not revise these statistics is China." However, no one attaches much credibility to Chinese statistics.

There are two main reasons why BLS estimates are less reliable. First, because survey response rates have fallen since the pandemic. Second, because of the agency's lack of resources, with funding down 20% since 2010, according to Bloomberg's calculations.

In the immediate term, Donald Trump cannot do much more anyway. He already fired the head of the Bureau of Labor Statistics, Erika McEntarfer, in early August, following a disappointing employment report.

To replace her, he has nominated the current chief economist of the Heritage Foundation, an influential conservative think tank, E.J. Antoni, who has yet to be confirmed by the Senate.