Longer quote from EJ Antoni in Townhall:

First and foremost, BLS must become radically transparent. For most data users, the agency is a black box. Only those who devote a tremendous amount of time to the subject understand the practical implications of the minutiae of how the agency collects and processes data. Although much of its methodology is already available, it’s often difficult to find on the website and isn’t in layman’s terms.

…

The Handbook of Methods should also be expanded significantly with explanations readily understandable to people without doctorates in statistics or economics. People are much more likely to trust BLS products if they understand where the numbers come from.

BLS releases have FAQs. For instance, the employment situation has this set of FAQs. In other words, unless one is illiterate, or too lazy to search for the acronym “FAQ”, you’re pretty hopeless anyway in using any index. Moreover, Dr. Antoni is a newly appointed “strategic adviser” for Truflation. Has he read through the documentation for Truflation’s price index? Talk about incomplete and opaque (you have to read a bit to find out there are no adjustments for quality of goods in the Truflation price index, for instance).

Antoni continues:

Serious methodological updates are also needed, especially for the Birth-Death model, which has comically overstated job growth for years, exacerbating the differences between initial and final monthly payroll estimates. Seasonal adjustments have been problematic too, as they are likely adjusted too frequently, causing changes in the trend of nonfarm payrolls to initially be misattributed to seasonality. Similarly, all indirect measurements for price data ought to be reevaluated to avoid the recent health insurance price debacle.

 

Antoni is a young’un so he can be excused for not realizing that the birth-death model, whatever its deficiencies at a given instant, can understate as well as overstate net job growth. Indeed, the creation of the research series adjusting the household employment series to the NFP concept can be understood as exactly the BLS response to the understatement critique. Here’s the discussion of the alternative series, and here is the data. As of today, here are official NFP, household series adjusted (as well as implied preliminary benchmark, which probably understates what the final benchmark will reveal):

Figure 1: CES nonfarm payroll (NFP) employment (blue), implied preliminary benchmark NFP (tan), civilian employment series adjusted to NFP concept (green). Source: For NFP, FRED series PAYEMS; for civilian series, BLS, for implied benchmark NFP, author’s calculations.

Another remark:

BLS should also hold accountable any employees who have prematurely shared market-moving data prior to public release. The agency must ensure such leaks do not occur again, possibly via audit trail tracking of who has access to sensitive data and when.

and

Trust was squandered by years of inaccurate data, leaks, and failures to adapt to a changing economy [emphasis added – MDC]

This is pretty rich, given that we know Mr. Trump did exactly that, in 2018:

President Trump broke years of presidential protocol on Friday morning with a tweet that signaled a strong jobs report was on its way from the Labor Department, an hour before the report was released.

While the White House brushed off any notion that Mr. Trump had crossed a line, legal experts said the tweet raised possible insider trading concerns and economists said it was a blatant misuse of presidential power.

No need to mention the July 2025 incident regarding a premature comment by Mr. Trump on GDP (a release from BEA).

Anyway, more on Dr. Antoni’s plans for the statistical agencies are laid out in Project 2025 (of which he was one of the authors).

Personally, I don’t think we should be taking any sort of advice on statistics from a guy who in his dissertation thinks a parameter is mysteriously small, when it’s actually pretty big, and he’s forgotten that he’s measured his regression left hand side percentage point variable in decimal form, and his right hand side in percentage points… (check it out, it’s in his dissertation available online)