Latest / The Fiscal Policy Podcast with Fexingo: Government Budgets, Stimulus, and Public Spending / How Government Auditors Miss Billions in Medicare Fraud
Transcript
- Lucas: So there's a number I've been sitting on for a few days. The U.S. government spends roughly $900 billion a year on Medicare. And the official estimate is that somewhere between six and eight percent of that goes out as improper payments — overpayments, underpayments, straight-up fraud. That's about $60 billion a year. Luna: And how much of that do they actually claw back? Lucas: About four percent. So we're talking maybe $2.5 billion recovered out of $60 billion lost. And the really frustrating part is, a lot of the fraud is not sophisticated. It's not some deep-cover conspiracy. It's people billing for medical equipment for patients who are dead. Luna: You're serious. Like, a doctor submits a claim for a back brace for someone whose date of death is in the system? Lucas: Exactly. And it happens thousands of times a year. In 2023, the HHS Office of Inspector General did a targeted review of durable medical equipment claims — that's wheelchairs, oxygen tanks, back braces — and they found that about 30 percent of the claims they sampled had no valid prescription. Some of those patients had been dead for months. Luna: So the fraud is obvious if you look. But the system isn't designed to look. Lucas: Right. Medicare processes about 4.8 billion claims a year. The vast majority are handled automatically. The fraud detection systems are rules-based — they flag things like 'same provider billing for 50 back braces in one day' — but the rules are easy to game. If you spread the billing across multiple shell companies, or you use slightly different diagnosis codes, the algorithm doesn't catch it. Luna: And then the audits that do happen are retrospective. By the time someone looks at a claim, the money's long gone. Lucas: Which brings me to a specific case. In 2024, the OIG released a report on a Miami-based durable medical equipment scheme. One provider had billed Medicare $2.1 million for back braces over eighteen months. When investigators followed up, they found that 80 percent of the patients listed had no record of ever seeing a doctor who prescribed a brace. And about 15 percent were deceased. Luna: And what happened to that provider? Lucas: They were eventually indicted. But here's the kicker: the scheme had been running for over two years. During that time, Medicare paid out roughly $1.7 million. After legal costs, the government recovered about $400,000. So the taxpayer net loss was over a million dollars on one single provider. Luna: And I'm guessing there were dozens of similar operations in Miami alone. Lucas: Hundreds. South Florida is a hotspot for Medicare fraud — it's been called the 'wild west' of healthcare billing. The OIG estimated that in 2023, the Miami region accounted for about $3 billion in fraudulent claims across all categories. And the recovery rate there is actually worse than the national average, because the schemes are designed to be fleeting. They set up a shell company, bill aggressively for six to twelve months, then dissolve and reopen under a different name. Luna: So the current audit system is basically playing whack a mole with a three-month delay. Lucas: Exactly. The standard audit process works like this: claims are paid on a rolling basis. Then, months later, a recovery audit contractor — a private firm hired by CMS — reviews a random sample of claims. If they find an overpayment, they demand the money back. But by then, the shell company has no assets. So the government sends a letter to an empty P.O. box. Luna: There's got to be a better way. Are any states or agencies trying something different? Lucas: A few. The most promising experiment I've seen is in Wisconsin. The state's Medicaid program — which faces similar fraud problems — started using a predictive analytics tool in 2023 that looks at provider billing patterns in real time. Instead of waiting for a sample audit, the system flags anomalies like 'provider A bills for back braces at ten times the rate of peers with a similar patient panel.' Then it automatically pauses payments on that provider until a human reviews the claims. Luna: And does it work? Lucas: Early results are promising. In the first year, Wisconsin's Medicaid program prevented about $47 million in suspected fraudulent payments. That's money that never left the state's bank account. Compare that to the traditional recovery model, where they might have clawed back maybe two or three million after the fact. Luna: So the solution is to stop the payment before it goes out, not chase it afterward. Lucas: Exactly. And the technology exists. Private insurers like UnitedHealthcare have been using real-time predictive fraud detection for years. But Medicare's system is still largely stuck in the early 2000s. The main barrier isn't technical — it's bureaucratic. CMS has to approve any new fraud detection tool through a procurement process that takes about three years. By the time it's implemented, the fraudsters have moved on. Luna: And part of the problem is that the contractors who do the audits get paid based on how much they recover. So they have an incentive to go after easy, high-dollar claims — not the smaller, more complex schemes. Lucas: That's a huge piece of it. The recovery audit contractors are paid a contingency fee — usually around 10 to 15 percent of what they recover. So they focus on claims that are easy to prove, like duplicate billing or billing for a service that's not covered. But a fraud scheme that involves fake prescriptions and shell companies takes a lot more investigation. It's not worth their time. Luna: So the system is structurally designed to go after the low-hanging fruit and ignore the bigger, more sophisticated fraud. Lucas: Right. And that's why the $60 billion improper payment number is probably an undercount. The OIG itself admits that their estimates are based on a sample that misses a lot of intentional fraud. They call it 'unknown overpayments' — money that went out the door but hasn't been classified as improper because no one looked closely enough. Luna: How much are we talking about in that 'unknown' category? Lucas: The 2024 OIG report I mentioned earlier estimated that for Medicare fee for service alone, there were about $2.1 billion in 'unknown' overpayments in fiscal year 2023. That's just the ones they could identify through statistical sampling. The real number is likely much higher. Luna: And this is money that could be going toward actual healthcare, or deficit reduction, or anything else. Lucas: Exactly. And it's not like we don't know how to fix it. The Wisconsin pilot shows that real-time predictive analytics can work. There's also a program called the Fraud Prevention System, which CMS piloted in a few states back in 2011. It showed a return on investment of about 10 to 1 — for every dollar spent on the system, it prevented $10 in fraudulent payments. But it was never scaled nationally. Luna: Why not? Lucas: Partly political — budget constraints, opposition from provider groups who don't want their payments delayed. But also, honestly, it's just not a priority. Medicare fraud is one of those issues that everyone agrees is a problem, but no one feels urgent about because the money is spread across millions of taxpayers. It's a classic tragedy of the commons. Luna: You know, Lucas, this is exactly the kind of thing that makes me glad we do this show without advertising. Because if we had a sponsor — say, a healthcare technology company — there'd be pressure to soften the critique or frame it as 'both sides have a point.' But we don't have that constraint. We can just say: look, the system is broken, here's a specific fix, and it's not being implemented because of bureaucratic inertia. Lucas: Absolutely. And that's why listener support matters. We keep this show free and independent, no ads, no sponsors to appease. If you find these deep dives valuable — if walking through the economy with us has made something click — the best way to keep it going is a small donation at buy me a coffee dot com slash fexingo. It literally pays for the research time and the hosting costs. Luna: And we mean it when we say every bit helps. A few dollars a month from listeners who care about this stuff is what keeps us from having to sell out. So if that's you, great. And if not, no worries — we're still glad you're here. Lucas: Alright, back to the numbers. So we've got a fraud detection system that catches maybe four percent of improper payments, a procurement process that takes three years, and a contractor incentive structure that avoids complex cases. What's the one change that would have the biggest impact? Luna: I'd say shifting from retrospective to pre-payment review. Don't pay the claim until a quick algorithm check passes. Lucas: That's exactly what the Wisconsin model does. And there's a bill in Congress — the Medicare Fraud Prevention Act — that would require CMS to implement real-time pre-payment review for high-risk providers. It's been introduced twice, but it hasn't made it out of committee. Luna: What's the main opposition? Lucas: Provider groups argue it would delay legitimate payments and hurt cash flow for small practices. And they're not entirely wrong — a poorly designed system could create false positives and slow down payments. But the Wisconsin pilot showed that with a properly tuned algorithm, the false positive rate is under two percent. And legitimate claims are processed within 24 hours. Luna: So it's a trade-off. A tiny delay for the vast majority of honest providers, in exchange for saving billions in fraud. Lucas: Right. And for me, that trade-off is a no-brainer. But in Washington, it's still a stalemate. And meanwhile, another $60 billion in improper payments goes out the door this year. Luna: And a lot of that money ends up in the pockets of people who are essentially stealing from the Medicare trust fund — which, by the way, is projected to run out of money in 2036. So every dollar lost to fraud brings that date a little closer. Lucas: Exactly. This isn't just about waste. It's about the solvency of a program that 60 million Americans rely on. And the fix is sitting right there, proven at the state level, with a 10-to-1 return. We're just choosing not to implement it. Luna: Well, maybe this episode will nudge one listener who works on the Hill. Or at least arm a few people with the specific numbers to bring up at dinner. Lucas: That's the hope. One concrete number to remember: $2.1 billion in unknown overpayments in one year, from one program. That's not an abstraction — that's real money that could have paid for 30,000 nurses' salaries or funded a year of cancer research. And it's sitting in the bank accounts of fraudsters. Luna: Alright, next episode we're tackling another angle of government spending that doesn't get enough scrutiny. Thanks for being here.