Skip to content
Hugin
An empty office chair beside a bare wooden desk in a dim room, one shaft of cold window light falling across the floor.

Hugin News

Federal agencies spent $9.5 billion on salary for people who were not working. GAO says $6.7 billion of it was the deferred resignation program — and no one is tracking the rest of it separately.

4 min read

Original editorial artwork generated for Hugin.

GAO's September 15 report finds that federal use of paid administrative leave rose 435 percent between 2023 and 2025, and that agencies in its review spent an estimated $9.5 billion in salary costs on it in 2025 — six times the 2023 figure. About $6.7 billion of that is attributed to the deferred resignation program, under which OPM let employees who agreed to resign or retire stop working immediately and be paid through September 30, 2025. The finding underneath the money is an accounting one: workforce-reduction leave is booked in the same bucket as ordinary administrative leave, so, in GAO's words, OPM 'does not know the actual costs' of it. GAO also found agencies reported 144 percent more paid administrative leave in pay periods containing a public holiday — holidays being logged as administrative leave — and OPM does not plan to fix the published historical data.

gaofederal-workforceopmpublic-spendingadministrative-leave
1source receipts1source hosts4 minread timelinkedprimary source

Paid administrative leave is an ordinary tool. An agency puts someone on it during an investigation, or a building closure, or a transition. GAO's phrasing is neutral and exact: "an excused absence without loss of pay or charge to leave. It is a cost to taxpayers as employees receive full pay without performing job duties."

In 2025 it became something much larger than a tool.

The numbers

GAO's review of payroll data found that federal agencies' use of paid administrative leave increased by 435 percent from 2023 to 2025. GAO estimates federal agencies in our review spent $9.5 billion in salary costs on paid administrative leave in 2025, a sixfold increase from 2023.

The cause is named in the report's title and in its text: "In 2025, the Office of Personnel Management (OPM) directed agencies to use paid administrative leave to support the deferred resignation program, which generally allowed federal employees to be placed on leave until they resigned or retired by September 30, 2025."

GAO's estimate of that share: "Using program assumptions and internal time and attendance data provided by payroll service providers, GAO calculated about $6.7 billion of that amount was associated with deferred resignation program."

So roughly $6.7 billion of $9.5 billion is the program. The remaining $2.8 billion is everything else administrative leave is used for — and, as the next section shows, some of it is probably not administrative leave at all.

The bucket problem

The sharpest sentence in the report is not about the size of the number. It is about where the number lives.

OPM does not know the actual costs of the paid administrative leave used for workforce reduction efforts, including the deferred resignation program.

Why not: "the paid administrative leave used for workforce reduction efforts is reported with other types of general paid administrative leave." There is no separate code for it.

And the consequence, in GAO's words:

To calculate long-term savings, OPM needs to know short-term costs of paid administrative leave used for these efforts. However, OPM cannot easily and accurately do this... Without a mechanism to track paid administrative leave for workforce reduction efforts, federal leaders may not have the data needed to understand whether government-wide cost saving goals are being met.

GAO notes the stated purpose the program was meant to serve: "One of the administration's stated principles for current workforce reduction efforts is to trim the budget topline by reducing full-time equivalent positions."

The saving and the cost are real, and they are booked in different places. That is not an allegation about anyone's motives. It is a chart of accounts that cannot produce the subtraction.

The data is also noisy, and GAO says which way

GAO found a specific, checkable defect: "GAO found that agencies reported 144 percent more paid administrative leave used in pay periods with a public holiday in 2023 through early 2025."

The interpretation is immediate: "Holidays should not be reported as paid administrative leave, and OPM has issued guidance to help address this data issue." A federal holiday is not an excused absence; it is a holiday. Some agencies were coding it as one anyway.

GAO says this "could overstate the actual amount used" — so its own $9.5 billion carries a stated direction of error, which is worth more than a number with no error attached at all.

OPM has issued guidance going forward. What it will not do is go back: "OPM does not plan to retroactively fix such historical errors in data released to the public." Which means the public dataset for 2023 through early 2025 contains known errors that will stay in it.

What GAO asked for

Two recommendations, both still open as of publication:

  1. That OPM publicly disclose the data reliability issues that remain unaddressed, so that "users make informed decisions about how to use these data."
  2. That OPM create a new category of paid administrative leave for workforce reduction efforts, so the cost can be separated from ordinary administrative leave.

Neither asks anyone to change a policy about who may be put on leave. Both ask the government to be able to count what it already did.

Source links