Is the Jobs Data Still Reliable? Yes, at Least for Now.

When last month’s jobs report showed unexpected weakness in the labor market, President Donald Trump fired the head of the agency that produces the data and named a loyalist to run the department that produces those numbers.

That has prompted a natural question ahead of Friday’s jobs report: Can this month’s numbers be trusted?

The answer, according to economists and experts in government statistics, is yes — but with all the same caveats that always apply to the data.

Economists across the political spectrum have criticized Trump’s decision to fire the head of the Bureau of Labor Statistics, Erika McEntarfer, and his choice to replace her. The president tapped E.J. Antoni, a conservative economist with a history of distorting statistics to support his political arguments.

Antoni has not yet been confirmed by the Senate — Trump didn’t even formally nominate him for the position until Wednesday — so he isn’t yet in charge of the agency. In the meantime, it is being run by its deputy commissioner, William J. Wiatrowski, who served as acting commissioner twice before.

Erica Groshen, who led the bureau under President Barack Obama, described Wiatrowski as a “BLS lifer” who is committed to the agency’s mission of producing statistics that are free of political influence. She and others who know him said they were confident that he would speak up — or resign — rather than allow anyone to interfere improperly in the agency’s work.

In any case, Groshen and other experts said, even a commissioner with ill intentions would not be able to meddle with the data, at least not in the short run and not without anyone’s noticing. The monthly jobs report is produced on a tight schedule using a highly automated and decentralized process. Most of the data that underlies the monthly payroll figure is reported directly by companies through an electronic system that is subject to strict access limitations. The commissioner, who is the agency’s only political appointee, does not have access to the numbers until they have been made final.

“There’s not like one person in a room who can manipulate things,” said Aaron Sojourner, an economist at the W.E. Upjohn Institute for Employment Research. “There are safeguards in place.”

Sojourner did add a note of caution. He said he trusted the data “for now” but worried that Antoni, if confirmed, could undermine the agency’s reputation for objectivity. And even before then, Trump could continue to attack the agency in ways that erode its credibility with his supporters.

While the raw numbers may be free from political influence, it is still wise to take them with a grain of salt. The August employment figure that was released Friday morning is a preliminary estimate and will be revised twice as data arrives from businesses that missed the deadline for the initial report.

Those revisions can be substantial, as last month’s report demonstrated: The agency lowered its estimates for employment growth in May and June by a combined 258,000 jobs.

Other recent revisions haven’t been as large, but they have been consistently downward. Historically, that has often been a sign that the economy was losing momentum, or even entering a recession. But some economists have suggested that the recent trend could be the result of other factors, including shifting patterns in seasonal hiring.

Next week, the bureau will release a preliminary estimate of its “benchmark revisions,” an annual practice in which the monthly estimates, which are based on surveys, are reconciled with more accurate but less timely data from state unemployment insurance offices. Most forecasters expect that report to show that the surveys significantly overstated employment for the second year in a row.

Economists have grown increasingly concerned in recent years that declining survey response rates and shrinking agency budgets could be eroding the reliability of the nation’s economic data. But they say there is little evidence that aggregate measures like the monthly payroll totals and the unemployment rate have become less reliable. Indeed, revisions have become smaller over time.

“It’s an estimate — it’s not perfect,” Sojourner said. But “it’s the best information available,” he added. “It’s very valuable.”

This article originally appeared in The New York Times.