# The GAO High-Risk List and Improper Payments: The Watchdog That Quantifies the Waste Federal agencies estimate roughly 3 trillion dollars in improper payments since 2003. Here is the watchdog that counts them, and why 'improper' is not 'stolen.' Author: J.A. Watte Published: July 16, 2026 Source: https://jwatte.com/blog/gao-high-risk-list-improper-payments/ --- Almost every number in this series, the ones that make a nuclear roof redesign or a rural flight look like waste, traces back to a single small agency most Americans could not name. It employs a few thousand auditors, it answers to Congress rather than to any president, and by its own accounting it returns about seventy-six dollars to the Treasury for every dollar it costs to run. It is also the source of the most misused statistic in American budget politics: the running tally, now near three trillion dollars since 2003, of federal payments that should not have gone out the door the way they did. The agency is the Government Accountability Office, and understanding what it actually measures is the difference between reading that three-trillion-dollar figure as a headline about theft and reading it as what it is, a risk estimate about paperwork. ## What it is, and who actually runs it The Government Accountability Office is the investigative arm of Congress. It describes itself as an independent, nonpartisan agency that provides Congress, executive-agency heads, and the public with "timely, fact-based, non-partisan information that can be used to improve government and save taxpayers billions of dollars," [in its own words](https://www.gao.gov/about/what-gao-does). It sits in the legislative branch, and it audits the executive branch: it examines how federal money is spent, whether programs work, and where they are exposed to waste, fraud, abuse, or mismanagement. It works either at the request of a congressional committee or as required by law. The common nickname, the "congressional watchdog," is accurate in both halves. It belongs to Congress, and it bites the agencies. The distinction matters because GAO does not run any of the programs it scores. It has no checkbook of its own beyond its operating budget, no benefits to pay, no grants to award. Its product is reports and recommendations, and its influence is entirely a matter of whether the agencies and the appropriators act on them. That is the frame to keep in mind for everything below: the dollar figures GAO publishes about improper payments are agencies' own estimates that GAO compiles and audits, and the dollar figures GAO publishes about savings are GAO's own estimates of what its work produced. Neither set is an independently certified bank balance. Both are the best structured numbers the government has, which is a different and more modest claim. ## The return on investment, and where it comes from Start with the counterintuitive part, because it is the reason the agency survives every budget cycle. For fiscal year 2024, GAO reported about 67.5 billion dollars in financial benefits attributable to its work, against an operating budget a small fraction of that. It put the ratio at about seventy-six dollars returned for every dollar invested in a single year, [in its own year-end announcement](https://www.gao.gov/press-release/gaos-work-led-67-billion-financial-benefits-behalf-congress-american-people), and it cited a longer-run figure of roughly 123 dollars per dollar averaged over six years. That same year it reported 1,232 other benefits that do not carry a dollar sign, improvements to public services, safety, and security, along with 718 products issued and 1,484 new recommendations to agencies. Two honest caveats attach to those numbers, and the second is easy to miss. First, they are GAO's own attributed estimates, not an outside audit of GAO, so the correct way to state them is "GAO reports" rather than "GAO returned." Second, the six-year, 123-to-1 figure is a routine agency metric consistent with its usual reporting, but it did not appear verbatim in the year-end announcement, so it is worth treating as the softer of the two ratios and attributing it to GAO's Performance and Accountability Report if it is doing load-bearing work. The seventy-six-to-one single-year figure and the 67.5 billion dollars are the ones stated plainly in the primary source. Even discounted heavily for self-attribution, the leverage is real, and it is why an agency that produces nothing but paper is one of the least controversial line items in the federal budget. ## The High-Risk List: the running scoreboard The mechanism that connects GAO's oversight to the specific messes in this series is a document it has published every two years since 1990: the High-Risk List. GAO started it to flag the federal program areas most vulnerable to waste, fraud, abuse, and mismanagement, or otherwise in need of transformation, [as the program's landing page records](https://www.gao.gov/high-risk-list). It is updated at the start of each new Congress, which is why the count shifts from edition to edition. The February 2025 edition, catalogued as report GAO-25-107743, lists 38 high-risk areas, [according to GAO's announcement of it](https://www.gao.gov/press-release/gao-urges-attention-2025-high-risk-list-save-billions-improve-government-efficiency-effectiveness). One start-year trap is worth defusing here, because a careless summary gets it wrong. The High-Risk List began in 1990. Government-wide tracking of improper payments, discussed below, began in 2003 under a different law. They are two separate clocks, and conflating them mislabels a thirty-five-year-old program as a twenty-two-year-old one. The 2025 edition is also a fair snapshot of how slowly these problems move. GAO added one new area, Improving the Delivery of Federal Disaster Assistance, and it flagged three areas that actually regressed since the prior update: Department of Defense Weapon Systems Acquisition, Improving IT Acquisitions and Management, and Managing Federal Real Property. The marquee entries read like a table of contents for federal risk: Medicare and Medicaid, which have been on the list for decades; DOD weapon-systems buying; the enforcement of tax laws, where the IRS projects a net tax gap of about 606 billion dollars for tax year 2022; VA health care; and the cybersecurity of the nation. Many of these are the parent categories under which the individual boondoggles in this series live. A specific stalled program is usually one line item inside a high-risk area that has resisted a clean fix for a generation. GAO does claim credit for the list working. It attributes nearly 759 billion dollars in financial benefits to high-risk work since 2003, an average around 40 billion dollars a year, and about 84 billion dollars of that since the 2023 update alone. Those are, again, GAO's own attributed estimates, and they are a different scoreboard from the improper-payments figure below. One counts benefits the watchdog says its work produced; the other counts errors the agencies estimate they made. They should never be added together or netted against each other. They are not the same kind of number. ## Improper payments: the symptom, quantified If the High-Risk List is the scoreboard, improper payments are the single number that gets quoted most and understood least. An improper payment, in the government's definition, is one that should not have been made at all or that was made in the wrong amount. Federal agencies have been required to estimate these across their programs since 2003, and GAO compiles the totals. The most recent figure is large and rising. For fiscal year 2025, agencies estimated about 186 billion dollars in improper payments across 64 programs reported by 15 agencies, [according to GAO's tally](https://www.gao.gov/products/gao-26-108694). Roughly 82 percent of that, about 153 billion dollars, was overpayments. Nineteen programs had improper-payment rates of at least 10 percent, and six exceeded 25 percent. That 186 billion was up from the prior year: in fiscal 2024, agencies estimated about 162 billion dollars across 68 programs reported by 16 agencies, [as GAO reported](https://www.gao.gov/blog/federal-government-made-estimated-162-billion-improper-payments-last-fiscal-year), which was itself down sharply from about 236 billion in fiscal 2023. The year-to-year swings are a trap of their own. The big drop from 2023 to 2024 was not systemic reform. It was mostly the winding down of pandemic programs, notably a decline of roughly 44 billion dollars in Pandemic Unemployment Assistance as that spigot closed. Read across three years, the pattern is 236 billion, then 162 billion, then 186 billion. That is not a problem being solved; it is a problem whose size tracks how much unusual money is moving through the system in a given year. The errors also concentrate in the same places. In fiscal 2024, about 121 billion dollars, roughly 75 percent of the total, came from just five programs: Medicare, Medicaid, the Earned Income Tax Credit, the Supplemental Nutrition Assistance Program, and the pandemic-era Restaurant Revitalization Fund. Strip out the pandemic entrants and the perennial leaders are Medicare, Medicaid, and the EITC, year after year. Secondary reporting of the GAO data has put Medicare alone near 54 billion dollars in fiscal 2024, roughly a third of the whole, though that per-program figure comes from outside summaries of the data rather than the headline release, so it is worth confirming against the full report before quoting a precise Medicare dollar amount. Then there is the cumulative number, the one that anchors the political fights. Executive-branch agencies' improper-payment estimates total roughly 2.8 trillion dollars since fiscal 2003 as of the fiscal 2024 reporting, rising toward about 3 trillion once fiscal 2025 is included. An earlier round-number version of this figure, about 2.7 trillion, circulated widely, including in a [House Oversight Committee request to GAO](https://oversight.house.gov/release/sessions-mfume-request-gao-examine-2-7-trillion-in-improper-payments-and-provide-anti-fraud-recommendations/) that is useful as political context but is not the data source. The current, correct framing is about 2.8 trillion dollars through fiscal 2024 and about 3 trillion through fiscal 2025. Always state the through-year, because the number moves every year, and always note the direction of the error bar: these totals are partial, because some programs are never estimated, so GAO treats the figure as a floor rather than a ceiling. ## Why it looks like a loss Lined up against the ordinary expectation that money leaves the Treasury only when it should, in the right amount, to the right person, the record is genuinely bad, and the scale of it does not shrink under scrutiny. A cumulative estimate near 3 trillion dollars over about two decades, annual totals in the 160-to-240-billion range, and a real total that GAO says is almost certainly higher because some large programs are not estimated at all. The same handful of programs top the list every year, which means the problem is concentrated, well understood, and still not fixed. The persistence is the sharpest indictment. A designation of "high risk" is supposed to be a spur to reform, yet Medicare has carried it since 1990, the tax-enforcement gap and DOD weapon-systems buying have sat unresolved for years, and three areas managed to get worse in the 2025 edition rather than better. That a program can be formally flagged as high-risk for more than thirty years, while agencies keep making nine- and ten-figure payment errors annually, is itself the evidence: oversight findings translate into reform slowly and partially, and the watchdog's bark carries no power to compel the agencies to heel. GAO can count the waste with precision. It cannot make anyone stop producing it. ## Ratio-adjusted for the mission Now change what the number is measuring, and the reading shifts, because the single most abused feature of the improper-payments statistic is the assumption baked into how it is quoted. An improper payment is not a stolen payment. GAO and the Office of Management and Budget define the category to include overpayments, underpayments, inaccurate recordkeeping, missing documentation, and eligibility-verification failures, [as GAO's improper-payments reporting spells out](https://www.gao.gov/products/gao-25-107753). Fraud is a subset, not the whole. In fiscal 2024, of the 162-billion-dollar total, about 7.9 billion was underpayments, money the government owed people and failed to send, and about 12.6 billion was classified as "unknown," neither confirmed proper nor confirmed improper. A payment that went to the right person for the right reason but arrived with a missing signature counts as improper. So does a check that was too small. Treating the full 162 billion or 186 billion as a measure of theft is not a rounding error, it is a category error, and it overstates confirmed loss by a wide and unknowable margin. The honest way to read the figure is as a measure of payment-integrity risk to be tightened, not a pile of dollars all stolen. Against that corrected reading sits the case for the watchdog itself, which is unusually strong. GAO is arguably the highest-leverage oversight investment in the federal government. It reported 67.5 billion dollars in financial benefits in fiscal 2024, about seventy-six dollars for every dollar it spent, plus 1,232 improvements to services, safety, and security that never show up as a dollar figure at all. The High-Risk List is the concrete machinery by which a good share of these problems actually gets fixed rather than merely counted: GAO credits it with nearly 759 billion dollars in savings since 2003, and in the 2025 edition ten areas showed enough progress to move in the right direction even as three slid back. The same agency whose reports supply the alarming totals is a large part of the reason any of that money is recovered. The number that makes the government look worst and the institution that makes the government work better are the same institution. There is a subtler public good here too, the one that gives this whole series its spine. The reason it is possible to write honest, sourced accounts of a nuclear roof redesign, a stalled hospital software rollout, or a rural flight subsidy is that GAO exists to produce the underlying audits in the open. A government that publishes its own failure estimates, itemizes its own high-risk programs every two years, and scores its own oversight agency's return is doing something most institutions never do voluntarily. The three-trillion-dollar improper-payments figure is embarrassing precisely because the government generated and disclosed it. The disclosure is the healthy part. ## The ledger reading The GAO story is the one entry in this series that is honestly described as a failure and honestly described as a success at the same time, because the two verdicts answer two different questions. Measured by outcomes, the record is grim: cumulative improper payments near 3 trillion dollars, annual totals stubbornly in the hundreds of billions, high-risk areas that sit unresolved for a generation, and a watchdog with no teeth to force the fixes it recommends. Measured by the value of the watchdog itself, the record is excellent: a reported seventy-six-to-one return, three-quarters of a trillion in credited high-risk savings, and a running public account of federal risk that exists nowhere else. What holds the whole arrangement together is that the second verdict is the reason the first is even knowable. The improper-payments number looks like an indictment of government, and in part it is, but it is also a product of the one part of government whose entire job is to find and publish that number. The most important discipline in reading it is the one that political discourse abandons most often: improper is not stolen, a floor is not a ceiling, and a benefit scoreboard is not an error scoreboard. Get those three distinctions right and the capstone reads cleanly. The waste is real, the mission that measures it is real, and the number everyone quotes is a measure of risk that still needs tightening, not a receipt for money that is simply gone. ## Related reading - [The Public-Money Programs Index](/blog/public-money-programs-index/): the full map of the boondoggle series, every program in it living somewhere on the High-Risk List this piece describes. - [The VA's Oracle-Cerner EHR Modernization](/blog/va-oracle-cerner-ehr-modernization/): one program inside VA health care, a long-standing high-risk area, read at the level of a single stalled contract. - [The Working Ledgers](/blog/the-working-ledgers/): the method behind this series, reading public programs through their own paperwork rather than the headline. ## Fact-check notes and sources - **What GAO is, an independent nonpartisan legislative-branch audit agency that reports to Congress and audits the executive branch:** [GAO, What GAO Does](https://www.gao.gov/about/what-gao-does). All ROI and savings figures below are GAO's own attributed estimates of benefits from its work, not independently audited totals; they are framed as "GAO reports" throughout. - **Fiscal 2024 return on investment, about 67.5 billion dollars in financial benefits, roughly seventy-six dollars per dollar in a single year, 1,232 other benefits, 718 products, and 1,484 recommendations:** [GAO year-end announcement](https://www.gao.gov/press-release/gaos-work-led-67-billion-financial-benefits-behalf-congress-american-people). The longer-run figure of about 123 dollars per dollar over six years is a routine GAO metric but did not appear verbatim in that announcement; it should be attributed to GAO's [Performance and Accountability Report for fiscal 2024](https://www.gao.gov/products/gao-25-900570) and treated as the softer of the two ratios. - **The High-Risk List began in 1990 and is updated every two years:** [GAO High-Risk List landing page](https://www.gao.gov/high-risk-list). Do not conflate this 1990 start with the 2003 start of government-wide improper-payment tracking. - **The February 2025 edition lists 38 high-risk areas (report GAO-25-107743), adds one area (federal disaster-assistance delivery), flags three regressions (DOD weapon-systems acquisition, IT acquisitions and management, managing federal real property), and reports nearly 759 billion dollars in savings since 2003 and about 84 billion since the 2023 update, alongside an IRS-projected net tax gap of about 606 billion dollars for tax year 2022:** [GAO announcement of the 2025 High-Risk List](https://www.gao.gov/press-release/gao-urges-attention-2025-high-risk-list-save-billions-improve-government-efficiency-effectiveness) and the [full 2025 High-Risk Series report](https://www.gao.gov/products/gao-25-107743). The area count is specific to this edition; the tax-gap figure is an IRS projection. - **Fiscal 2025 improper payments, about 186 billion dollars across 64 programs reported by 15 agencies, roughly 82 percent (about 153 billion) overpayments, 19 programs at rates of at least 10 percent and six above 25 percent, and cumulative estimates near 3 trillion dollars since 2003:** [GAO fiscal 2025 improper-payments report](https://www.gao.gov/products/gao-26-108694). This is a partial estimate; GAO notes the real total is likely higher because some programs are not estimated. - **Fiscal 2024 improper payments, about 162 billion dollars across 68 programs reported by 16 agencies, overpayments about 135 billion (84 percent), underpayments about 7.9 billion, "unknown" about 12.6 billion, down from about 236 billion in fiscal 2023 largely on pandemic-program wind-down (Pandemic Unemployment Assistance falling roughly 44 billion), with about 121 billion (75 percent) concentrated in five programs (Medicare, Medicaid, the Earned Income Tax Credit, SNAP, and the Restaurant Revitalization Fund), and cumulative estimates near 2.8 trillion dollars since 2003:** [GAO fiscal 2024 improper-payments summary](https://www.gao.gov/blog/federal-government-made-estimated-162-billion-improper-payments-last-fiscal-year). A per-program figure near 54 billion dollars for Medicare comes from secondary reporting of the GAO data and should be confirmed against the full report before quoting. - **The definition that improper payments are not synonymous with fraud, encompassing overpayments, underpayments, and documentation and eligibility errors:** [GAO fiscal 2024 improper-payments report (GAO-25-107753)](https://www.gao.gov/products/gao-25-107753). - **The roughly 2.7-trillion-dollar cumulative framing as political context, not a data source:** [House Committee on Oversight and Government Reform request to GAO](https://oversight.house.gov/release/sessions-mfume-request-gao-examine-2-7-trillion-in-improper-payments-and-provide-anti-fraud-recommendations/) (a congressional, secondary source). The corrected current figures are about 2.8 trillion through fiscal 2024 and about 3 trillion through fiscal 2025, per the GAO reports above. *This post is informational and journalistic, describing public programs and public records. It is not legal, financial, or policy advice. Dollar figures are drawn from Government Accountability Office reports and are, where noted, GAO's own attributed estimates rather than independently audited totals; secondary and congressional sources are labeled as such.* --- Canonical HTML: https://jwatte.com/blog/gao-high-risk-list-improper-payments/ RSS: https://jwatte.com/feed.xml JSON Feed: https://jwatte.com/feed.json Hero image: https://jwatte.com/images/gao-high-risk-list-improper-payments.webp