DOGE'S SAVINGS MATH FAILED THE AUDIT

01 THE CLAIM WAS BUILT FOR A PRESS RELEASE

DOGE's public savings story was easy to repeat: identify a terminated contract, grant, or lease, add up its possible ceiling, and present the total as money saved. The Government Accountability Office's review makes clear why that shortcut matters. A contract's maximum potential value is not necessarily what an agency would have paid, and canceling an item already winding down is not a new reduction.

The GAO examined $110 billion in claimed savings associated with DOGE actions from June 2025 through July 2026. Its conclusion was not that every termination was meaningless. It was that billions could not be substantiated, while DOGE took credit for actions already in progress and did not provide the underlying work needed to test the aggregate.

02 WHAT THE GAO FOUND

The central finding is an evidence problem. The watchdog could not substantiate billions in the claims it tested and said DOGE refused to show its work. That phrase is more consequential than a single arithmetic mistake: without source records, reviewers cannot determine whether a cancellation changed an agency's obligations, merely changed a line in a database, or transferred costs somewhere else.

GAO also found DOGE taking credit for terminations that were already underway. That turns timing into apparent performance. If an agency had begun canceling a lease before DOGE posted it, the termination may still be real, but attributing the entire avoided cost to a new efficiency program exaggerates the program's effect.

The audit's broader context matters too. The same initiative that advertised savings gained access to sensitive federal systems and helped hollow out agencies. A savings ledger therefore cannot be separated from questions about data controls, legal authority, continuity of services, and the cost of rebuilding capacity.

03 THE CONTRACT CEILING TRAP

Federal procurement records often show a ceiling: the maximum amount a contract could reach if all options are exercised. DOGE frequently treated that ceiling as the savings created by termination. But an option-year contract can be canceled before the option is exercised, a cost-plus agreement can settle below its maximum, and a contract can be modified without spending its full authorized amount.

This is the difference between authority to spend and spending avoided. The first is a legal or administrative limit. The second is a counterfactual: what would have left the Treasury if the government had done nothing? GAO's inability to substantiate billions means the public ledger often skipped that counterfactual step.

One public example illustrates the scale of the problem. An $8 million Immigration and Customs Enforcement contract was initially presented by DOGE as an $8 billion saving before being corrected. The correction is not a rounding error; it is a thousand-fold change in the claim's order of magnitude.

04 CUTS CAN MOVE COSTS, NOT ERASE THEM

Even a correctly recorded cancellation does not automatically equal a net gain for taxpayers. Ending a lease can impose relocation expenses. Cutting an inspection or enforcement function can reduce near-term payroll while increasing unpaid taxes, fraud losses, health risks, or delays. Eliminating technical staff can create a later bill for contractors and emergency remediation.

The GAO review is therefore best read as a floor, not a complete cost-benefit study. It challenges the reliability of the savings total; it does not by itself price every consequence of DOGE's reductions. That distinction protects the analysis from an opposite mistake: replacing an inflated benefit claim with an unsupported claim that every cut was harmful.

05 WHY THE DEFICIT DID NOT DISAPPEAR

A program can make real reductions and still fail to lower the federal deficit if spending elsewhere rises, revenue falls, or the reductions are too small relative to the budget's scale. The Hamilton Project tracker cited in the research record put federal spending at roughly $7.1 trillion in 2024 and $7.558 trillion in 2025. That movement does not prove that every DOGE action failed; it does show why a tracker of announced cuts cannot substitute for government-wide fiscal accounting.

There are also timing effects. A canceled multi-year agreement may produce no immediate cash reduction if funds were never obligated. Conversely, a workforce reduction can create severance, paid leave, litigation, and rehiring costs before any recurring payroll reduction appears. Deficit analysis needs budget authority, obligations, outlays, and receipts kept separate.

The political promise—“we found billions”—compresses all of those accounting layers into one number. GAO's work re-expands them. That is the real efficiency test: can an outside reviewer follow the money from the original obligation to the claimed saving and then into the actual fiscal result?

06 TRANSPARENCY IS AN EFFICIENCY TOOL

DOGE's public receipts were presented as proof of transparency, but a list of contract identifiers is not the same as reproducible evidence. A public ledger needs stable identifiers, the original obligation amount, the remaining balance, the action date, the responsible agency, and a clear method for preventing duplicates. It should also show corrections rather than silently removing embarrassing entries.

That standard is not bureaucratic decoration. It is how a savings claim earns trust across administrations. GAO's finding that DOGE did not provide the underlying work turns missing documentation into a fiscal issue: without an audit trail, Congress cannot distinguish a durable reduction from a press-release total, and agencies cannot learn which interventions actually worked.

07 THE LESSON AFTER DOGE

The GAO result does not settle the philosophical question of how large government should be. It settles a narrower, more practical question: the administration's advertised savings total was not reliable enough to carry the weight placed on it. The difference between a reform agenda and a branding exercise is whether its results survive independent checking.

For future efficiency drives, the sequence should be reversed. First define the baseline. Then identify the legal and operational decision that changes it. Then publish the expected and realized cash effects, including costs. Only after that should leaders add the numbers into a national total.

That may sound slower than a live leaderboard. It is. But slower arithmetic is cheaper than rebuilding a tax agency, restoring a technical workforce, or repairing public trust after a $110 billion claim dissolves under audit. The GAO's message is ultimately pro-efficiency: count only what can be shown, and make the showing durable.