Closed Means Nothing If the Work Was Never Done
Here is the sentence that should be carved above the entrance to every regional VA office in America: the claim status “closed” is defined as “the claim is complete with no further action possible.” That is the government’s own definition, straight from VBA Manual 21-4.
Now consider that the Office of Inspector General just spent thirteen months proving that VBA staff closed thousands of these supposedly complete, no-further-action-possible claims while the actual required action, the one affecting a veteran’s monthly paycheck, had never happened at all.
Somewhere a bureaucrat defined “closed” and somewhere else an entirely different bureaucrat closed a case that was, by any honest reading, wide open. This is the kind of contradiction that would get you fired from a lemonade stand. At the VA, it just gets you a “concur” stamp and a target completion date sometime next year.
A Third of Cases, A Thousand Ways to Fail a Veteran
The report in question, released last week and quietly filed under a review number nobody will remember, examines how the Veterans Benefits Administration handles “adverse actions,” the bureaucratic euphemism for cutting a disabled veteran’s monthly check. Under 38 C.F.R. Section 3.103, before VA can reduce or terminate a veteran’s disability compensation, it has to notify them, give them sixty five days to respond with evidence, and offer a hearing.
This is not some newfangled progressive innovation. It is basic due process, the kind of thing that has been settled law in this country since well before most of the claims processors involved were born.
The OIG sampled 229 cases out of an estimated pool of 27,100 and found errors in roughly a third of them, projecting that about 9,300 cases involved at least one mistake (see the report’s Appendix C for the full statistical gymnastics, margin of error and all). Break down the error types and you get a portrait of an agency that treats due process as optional homework.
About 4,000 cases involved insufficient notice to the veteran, meaning VA moved to cut a check before telling the person whose money it was. Nearly 2,900 involved uncorrected “codesheets,” the internal ledgers that track a veteran’s disability history, left inaccurate after the fact. Another 2,900 had botched effective dates, sometimes off by a month in either direction (see Example 2 in the report, where a claims processor miscounted sixty days and cost a veteran an extra month of benefits they were never entitled to, an overpayment of roughly $300 that will eventually have to be clawed back from someone who did nothing wrong).
Then there is Example 3, which deserves its own museum exhibit. A veteran’s lung cancer rating was proposed for reduction from 100 percent to 30 percent. The due process letter went out. The veteran did not respond. So far, bureaucratically unremarkable. Then a different claims processor, working an entirely separate benefit for the same veteran, canceled the “end product” tracking the reduction, which under VA’s own system definitions automatically closed the case as complete.
The reduction that was supposed to happen never did. VA overpaid the veteran about $43,000 over sixteen months, at a running rate of $2,700 a month, and nobody noticed until the OIG went looking. This is not fraud. Nobody here is a villain twirling a mustache. It is simply a records system so brittle that one employee’s unrelated paperwork can silently erase another employee’s entire adjudication, and an agency with no automated way to notice when that happens.
If this sounds familiar, it should. Federal courts have spent decades wrestling with agencies that treat procedural notice requirements as a formality to be checked off rather than a substantive right to be honored. The Supreme Court’s own due process jurisprudence in the benefits context traces back to Mathews v. Eldridge, which set the very balancing test the VA is nominally supposed to be applying here. And this is far from the VA’s first rodeo with the OIG on exactly this theme.
Earlier reviews have flagged similar breakdowns in how VBA tracks pending claims through its “end product” workload system, a structure originally built to measure employee productivity rather than to safeguard veterans’ rights, which may explain why it is so good at counting closed cases and so bad at noticing which ones should not have been closed.
They Knew in 2024. They Made a Video.
What makes this report more than a dry compliance exercise is the timeline buried in its own pages. In March 2024, over two years before this report’s publication, a senior quality reviewer at VBA’s Compensation Service flagged on an internal call that an “alarming number” of adverse action claims were being closed without final decisions. The recommended fix was to remind new staff to finish their paperwork.
Nobody followed up.
The OIG then found the exact same problem persisting through September 2025, in nine of twenty sampled cases, a 45 percent failure rate that is arguably worse than the original estimate. This is bureaucracy’s favorite trick: identify the problem, hold a call about the problem, and mistake the call for the fix.
Meanwhile, VBA officials proudly told the OIG team about a new training initiative called “Minute Videos,” rolled out in December 2025 with more planned for March 2026. The OIG actually watched them. The December batch had nothing to do with adverse actions. The March batch addressed the topic but, in the OIG’s own words, did not address the deficiencies identified in this review or the actions recommended to mitigate them.
Somewhere in a windowless conference room, someone approved a training video series that fails to train anyone on the thing it needed to train them on, and that someone almost certainly got a positive performance review for shipping content on schedule.
VA’s own fix for the letter generating problem, meanwhile, has a “tentative delivery timeline” stretching from a 2023 start date to early 2027, a four year odyssey to standardize a form letter.
For context, the Government Accountability Office has spent years documenting how federal agencies routinely blow through their own modernization timelines, sometimes by a decade or more, for systems far more complex than a rating decision letter.
The Bare Minimum, Arriving Four Years Late
The dollar figures here, at least $16.9 million in improper payments and nearly $1 million a month in ongoing overpayments and underpayments, are almost beside the point, though Congress will fixate on them because dollar figures are legible in a way that human suffering is not.
The real story is that a veteran who lost a leg or a lung in service to this country can have their monthly check quietly reduced, restored, reduced again, or simply left in bureaucratic limbo because two different employees in two different systems never talked to each other, and no software exists to flag the contradiction.
VA concurred with every finding. VA agreed to every recommendation. VA has, as of this writing, resolved 184 of the 191 specific errors the OIG identified, which sounds impressive until you remember those 191 are drawn from a sample that represents thousands more the agency has not yet found.
The recommendations themselves (standardized letters, an extra layer of signoff, an automated tracking report) are not radical. They are the bare minimum any competent organization would have built a decade ago. That the largest disability benefits system in the country is only now getting around to it, under OIG pressure, with completion dates stretching into 2027, tells you less about incompetence than about priorities.
Building a system that reliably protects a veteran’s paycheck has simply never been urgent enough. It took an audit to make it urgent for exactly as long as this news cycle lasts.
Bureaucracy Times 