The federal government spent $9.5 billion on paid leave for workers in 2025, and a large share of that spending was tied to a buyout program rolled out under President Trump’s Department of Government Efficiency, according to a new audit. The figure represents a sixfold increase from 2023.
The audit comes from the Government Accountability Office, which reviewed how the government pays workers who are on leave while still employed. The GAO found that the program, which let workers resign or stop working while still collecting pay, was the main driver of the spending. But the audit also found that the government does not know exactly how much the program cost or whether it delivered the promised savings.
Buyout Program and Its Costs
The Deferred Resignation Program, or DRP, was launched by the federal Office of Personnel Management in January 2025. Under the program, workers who opted in could resign or stop working but still receive pay through Sept. 30, 2025. The program was part of the Department of Government Efficiency’s push to shrink the federal workforce.
The Trump administration projected that 200,000 federal employees would opt into the program. The GAO estimated that 144,312 employees accepted the buyout offer based on administrative leave data. The program was offered as a voluntary option, meaning workers chose to participate rather than be forced out.
The paid administrative leave associated with the DRP peaked in July 2025, when 2.5 million of 3 million leave workdays were due to the program. That peak reflects the period when the largest number of workers were collecting leave pay while off the payroll.
The Spending Numbers
The $9.5 billion in paid leave costs in 2025 dwarfs earlier years. The combined administrative leave salary costs for 2023 and 2024 were $3.2 billion, according to the audit. The jump from those combined years to 2025 shows how quickly the program expanded.
The DRP accounted for $6.7 billion of the total paid leave costs in 2025. The remaining portion came from other administrative leave categories, though the audit does not break down that figure further.
Use of paid administrative leave increased by 435% from 2023 to 2025, according to the audit. That growth tracks directly with the rollout of the buyout program, which created a large pool of workers who remained on the books but were not actively working.
Savings Argument
OPM Director Scott Kupor defended the spending in a statement, arguing that the $9.5 billion was a one-time cost for a significant reduction in the federal workforce. He framed the program as a bargain for taxpayers.
Kupor said the report failed to highlight the difference between “a one-time expense ($9.5 billion) to reduce the size of the federal government by 270,000 employees and the $40 billion per year savings in taxpayer dollars that this reduction provides.” He added that “That 400% return on investment is a massive benefit to the taxpayer.”
The administration’s position rests on two claims: the savings are real, and the savings outweigh the costs. The audit does not address the savings figure directly, but it does raise questions about how the government knows what it saved.
Verification Problem
The GAO found significant limitations in the data it received from agencies. According to the audit, OPM “does not know the actual costs of the paid administrative leave used for workforce reduction efforts” because DRP-associated leave was lumped in with general paid administrative leave numbers. The agency cannot separate the DRP costs from other leave costs, which means the full financial picture is obscured.
The audit also found that the government lacks a tracking mechanism to understand whether government-wide cost-saving goals are being met. Without that data, federal leaders may lack information to determine whether the program achieved its intended outcomes.
GAO recommended that OPM address the transparency of its paid administrative leave data and create a new category for leave used to support workforce reduction efforts. The recommendation suggests that the current system is not adequate to measure the program’s performance.
DOGE Background
The DRP was part of the broader push by the Department of Government Efficiency, known as DOGE, to shrink the federal government. The department’s efforts included cutting federal contracts, leases and grants, which DOGE tracked through its “Wall of Receipts” webpage.
The DOGE-era buyout program was designed to reduce the federal workforce by encouraging voluntary departures. Workers who took the buyout resigned or stopped working but continued to receive pay until Sept. 30, 2025, when the program ended.
The program’s design meant that workers stayed on the payroll for months after ceasing their duties. That arrangement drove the administrative leave costs, since workers were counted as employed even though they were not working.
Inflated Savings Claim
DOGE claimed that $215 billion of federal funding had been cut through its efforts. A separate GAO report from last month challenged that figure, finding that about $110 billion of the $215 billion was inflated.
That report said DOGE was not transparent about how it calculated its savings and could not provide sufficient information to verify 96% of its reported methods. The audit also noted that the long-term savings goals of the leave program could not be determined.
The discrepancy between the DOGE claim and the GAO estimate is substantial. DOGE’s published figure is more than double the GAO’s estimate, which raises questions about how the department arrived at its numbers.
| Program | Amount | Role |
|---|---|---|
| DRP paid leave costs, 2025 | $9.5 billion | Main driver of spending |
| Combined 2023–2024 administrative leave costs | $3.2 billion | Earlier comparison point |
| DOGE claimed savings | $215 billion | Total savings figure |
| GAO verified savings | About $110 billion | Verified portion of DOGE claim |
The key facts from the audit:
- $9.5 billion in paid leave costs in 2025, up sixfold from 2023
- $6.7 billion of that total was DRP-related
- 144,312 employees estimated to have accepted the buyout offer
- Paid administrative leave peaked at 2.5 million of 3 million leave workdays in July 2025
- OPM director Scott Kupor called the program a “massive benefit to the taxpayer”
- GAO could not verify the actual costs of DRP-associated leave
- GAO recommended creating a new category for workforce reduction leave
What the Audit Actually Says
The GAO’s report does not settle the question of whether the program was a good deal. It documents the spending and flags the gaps in verification. The audit does not address the savings figure directly, but it does say the government has not proven them.
The audit’s core finding is that the government does not know how much the program cost. The lumping of DRP leave with general leave makes the actual costs impossible to calculate. That uncertainty matters because the administration’s entire argument rests on the savings figure.
The $40 billion annual savings claim is a central part of the program’s justification. The audit’s admission that the costs cannot be verified puts that claim in question. A one-time expense is easy to defend, but the savings are supposed to justify the expense. If the savings cannot be verified, the defense loses its foundation.
The GAO’s recommendation for a new category of leave is modest, but it could change how the government tracks these programs in the future. The audit’s findings suggest that the current system produces data that is useful for budgeting but not useful for measuring performance.
Who Is Protecting Whom
The audit raises a basic question about accountability. The government spent nearly $10 billion on leave for workers who were not working, and it does not know how much that spending actually cost. That is a strange position for a government to be in.
The administration’s response, through Kupor, emphasizes the savings. The audit’s response, through its findings, emphasizes the unknowns. The two positions sit side by side, and the public has to judge which one carries more weight.
The audit’s findings do not settle the question of whether the program was a good deal. They document the spending and flag the gaps in verification. The audit does not address the savings figure directly, but it does say the government has not proven them.
The GAO’s recommendation for better data is a reasonable request. The government should know how much it spends on programs like this, and it should know whether those programs deliver what they promise. The current system falls short of that standard.
Where the paper stands
The paper backs the citizens and small business owners whose tax dollars fund these payments, and is against the government office that wrote its own authority to spend so freely. The spending grew from $3.2 billion in 2023 and 2024 to $9.5 billion in 2025, driven by a buyout program that let workers resign and still collect pay. The government cannot say what the program cost or whether it saved money, and the audit recommends a new category for leave tied to workforce reduction so the government can track the spending.
The program was offered voluntarily, meaning workers chose to take it. But the administration’s claim of a $40 billion annual savings figure rests on data the government itself admits it cannot verify. The paper is against offices that gather power and write their own authority without clear limits, and the audit shows that is exactly what happened here.
Readers should note the gap between the $40 billion savings claim and the audit’s finding that the government does not know the actual costs of the leave. The audit’s recommendation for a new leave category is a step toward transparency, but the current system is already in place and the spending has already happened. The paper will continue to watch how the government responds to the audit’s findings, and whether it fixes the verification problem before the next round of spending begins.
See the video the story is built around at Cbsnews.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.