10%管理人占据八成市场，证券私募二八分化加剧

Question

The May and June new jobs reports from BLS (Bureau of Labor Statistics) were incorrect by a factor of ten. How does this happen, and what confidence should be placed in the July report?

• May report revised down from 144,000 to 19,000,
• June report revised down from 147,000 to 14,000
  (BLS)

Or as one media source headlined: "May and June data slashed 88%".

Answer 1

From Bloomberg:

The downward revision to the prior two months was largely a result of seasonal adjustment for state and local government education, BLS said in earlier comments to Bloomberg. Those sectors substantially boosted June employment only to be largely revised away a month later.

But economists say the revisions also point to a more concerning, underlying issue of low response rates.

BLS surveys firms in the payrolls survey over the course of three months, gaining a more complete picture as more businesses respond. But a smaller share of firms are responding to the first poll. Initial collection rates have repeatedly slid below 60% in recent months — down from the roughly 70% or more that was the norm before the pandemic.

In addition to the rolling revisions to payrolls that BLS does, there’s also a larger annual revision that comes out each February to benchmark the figures to a more accurate, but less timely data source. BLS puts out a preliminary estimate of what that revision will be a few months in advance, and last year [2024], that projection was the largest since 2009.

From the older piece linked therein, dating from almost a year ago:

The number of workers on payrolls will likely be revised down by 818,000 for the 12 months through March — or around 68,000 less each month — according to the Bureau of Labor Statistics’ preliminary benchmark revision. It was the largest downward revision since 2009.

Before the report, the BLS’s initial payrolls figures indicated employers added 2.9 million total jobs in the period, or an average of 242,000 per month. Now the monthly pace is more likely to be around 174,000 assuming the change is distributed proportionately, still a healthy rate of hiring but a moderation from the post-pandemic surge.

[...] For most of the recent years, initial monthly payroll data have been stronger than the QCEW figures. Some economists attribute that in part to the so-called birth-death model — an adjustment the BLS makes to the data to account for the net number of businesses opening and closing, but that might be off in the post-pandemic world.

Others have argued there’s another reason behind that discrepancy: immigration. Because the QCEW report is based on unemployment insurance records — which undocumented immigrants can’t apply to — the data are likely to have stripped out up thousands of unauthorized workers that were included in the initial payroll estimates.

The BBC adds (about this recent event) that:

Some speculated that [the adjustments] could reflect a hit to small businesses, which are typically slower to respond to surveys and are especially vulnerable to tariffs.

TLDR: it's complicated. The data sources themselves are full of potential bias and uncertainty, given how they are collected.

And if you believe some former commissioner, it doesn't help that the agency's budget has been under pressure for years.

Erica Groshen, a former BLS commissioner and current senior economics advisor at the Cornell University School of Industrial and Labor Relations, said Trump’s most recent cuts add to decades of the BLS struggling to publish accurate and detailed data despite budget cuts and political pressure. The agency’s budget is down by about 20% in real terms since 2009, she said.

Interestingly, perhaps, the White House has just put out a presser titled:

BLS Has Lengthy History of Inaccuracies, Incompetence

I guess it's going the way of USAID or something. Or it might not. Skimming the lengthy 'indictment', it seems entirely focused on finding fault with current commissioner, who Trump has just fired. There's no attempt I can see to put the errors in more historical perspective, in the White House presser.

Also intersting, perhaps, unlike the SocialTruhts posted by Trump personally that charged the commissioner with inflating (just) the Biden-era numbers

Trump explained in a Truth Social post that he was “just informed” that the nation’s employment reports were “being produced by a Biden Appointee” and charged that McEntarfer “faked the Jobs Numbers before the Election to try and boost Kamala’s chances of Victory.”

the more official WH document says the bias of this BLS was more uniformly in that direction: "consistently published overly optimistic jobs numbers — only for those numbers to be quietly revised later", which it proceeds to exemplify with similar happenings under both Biden and Trump. However, the WH piece charges that the errors

enabled the Federal Reserve to continue its disastrous policy of keeping interest rates high.

N.B.

Trump doubled-down in a separate social media post, arguing that the July BLS report was “RIGGED in order to make the Republicans, and ME, look bad.”

But I'm not sure if Trump meant that the numbers were "rigged" up or down in order for the latter to happen, given that the WH presser says that the overreporting hurt (via the Fed interest rate).

A couple of months ago, Trump was praising BLS data:

In May, the White House said that April's jobs report "proved" that Trump was "revitalizing" the economy. In June, Trump posted, "GREAT JOBS NUMBERS" on Truth Social.

So, I guess the WH wants high numbers reported to the press, but low ones reported to the Fed... Or at least the Fed to act as if the latter was the case.

(And yeah, Trump is hardly the first president to claim/suspect that the BLS is against him personally, rather than suffering from some structural/methodological issues that are fairly hard to correct, when it come to labor data collection.)

---

Footnote:

From BLS' own pages it's rather hard to get any idea how much that prelim data differs from their revised ones... You get a lot of large tables with very low percentage errors, like the typical one claimed to be 0.1%, which may not have much bearing on the common man's idea of measurement error, in this context (as in the OP/AP's headline, for instance).

The 'trick' is basically (exemplified by) this:

Compared with the sample-based, seasonally adjusted published estimate for March 2024, total nonfarm employment had a revision of ?589,000 or ?0.4 percent.

So basically, being off by half a million only counts as a 0.4% error (in how the BLS judges itself). However, if one admits this (total/absolute employment, not change thereof) as the basis of reporting (errors), a 73K change is basically almost totally insignificant, statistically speaking (0.05% or so), so hardly worth reporting at all! Yet it's a headline in the BLS blog. What is extremely reliable [statistically], in terms of how the BLS measures [their] errors, is to say that ~150 million Americans are still employed, but that's not what the BLS newsfeed headline says!

N.B.: The BLS actually does give the raw numbers before and after adjustments, on the right page, but they don't compute any sort of percentage error for the change-in-change (e.g. between prelim and final data). So, that helps them avoid saying they were off by 88% in prelim data compared to later/adjusted one (as in the OP's headline).

Slightly interesting, perhaps, the BLS disses data from State agencies as too noisy to report as a sum:

the BLS routinely warns that because of statistical limitations, it “does not compile a ‘sum-of-states’ employment series and cautions users that such a series is subject to a relatively large and volatile error structure.”

But apparently that level of caution doesn't extended to reporting preliminary data from their own surveys, when e.g. less than 60% of the firms have responded by the end of the month.

Answer 2

Question

How could the BLS jobs created in May and June 2025 been so far off?

Because President Trump laid off the statisticians and economists who generate these reports and now the reports aren’t based on economic data but guesses.

He’s also created a new class of government worker who can be fired for not assisting the President with his policy agenda. A real problem for agencies like BLS which are supposed to inform policy not implement it. (Close paraphrase from an interview from a previous BLS commissioner from May, warning of exactly this, given be low in last quote).

• DOGE layoffs may have compromised the accuracy of government data

The Trump administration’s drastic cutbacks of government spending and the federal workforce have economists, researchers and statisticians sounding the alarm that the reliability and accuracy of economic data could become a casualty to those efforts. … (a statistical technique that, when boiled down to very rough terms, essentially means more educated guesses).

• How abnormal are the revisions in this month’s jobs report?

The BLS also included revisions to the preliminary jobs numbers reported earlier, stating: “Revisions for May and June were larger than normal. The change in total nonfarm payroll employment for May was revised down by 125,000, from +144,000 to +19,000, and the change for June was revised down by 133,000, from +147,000 to +14,000.”

While it is typical for the BLS to issue revisions, the size of these revisions raised some eyebrows, especially among those who favor the Trump administration’s economic policies.

• Mass resignations at labor department threaten workers in US and overseas, warn staff – as more cuts loom

“The department has gotten 20% smaller, before any formal reductions in force are announced. A lot of people headed for the exits because so many different components of the Department of Labor have been threatened by reduction in forces [Rifs],” said an employee at the Bureau of Labor Statistics (BLS), a key government data agency, who requested to remain anonymous for fear of retaliation. “God only knows how much smaller it will be when the Rifs are announced.”

• The Bureau of Labor Statistics Is Not the Problem

Losing highly reliable, accurate and transparent data on the health of the world's largest economy would be perilous.

• Economists Are Seriously Alarmed About Official Data Under Trump

• Trump fires lead official (BLS) on economic data as tariffs cause markets to drop

• The Erosion of Trust in U.S. Economic Data: A Growing Concern for Policymakers and Markets

• Don't Trust Government Numbers

• We cannot trust gov’t economic numbers any longer

• Data credibility fears fueled after Trump orders firing of labor official

• Trump responds to his failure to create jobs by firing the U.S. labor statistics chief

Confronted with evidence of his own failures, the president could've changed direction. He instead found it easier to fire his labor statistics chief.

• Economists don’t know what’s going on

Blame crumbling statistical offices

• A former BLS commissioner explains how data could be less reliable under Trump

MAY 2, 2025

Marshall-Genzer: So what is the Trump administration proposing, then?

Groshen: To create a new classification of civil servants where those civil servants could be fired by the president for not helping the president to fulfill the president's policy agenda … who is supposed to be affected is anybody who is in a policy-influencing position. And the whole purpose of federal statistics is to inform policy. So you could imagine a new spurt of inflation or a drop in inflation, and the administration wants to play it up or downplay it, or a spurt of employment growth or a drop in the number of jobs or rising unemployment rate, and the administration wants to take full advantage of it for political purposes. And then, I mean, I'm not saying that they would, but this opens the door to that right, in a way that the door was not opened before.