Myrtle Beach, SC
Open them, and be ready for what the audit finds, because it is not what this Forum has been assuming.
Social Security has one of the lowest improper-payment rates in the federal government. Well under one percent of benefits paid. Not one percent of the improper-payment total — one percent of the program, which on a program that size is a very small number relative to nearly everything else in the budget.
That is the published figure, produced by the agency and audited, and it has been roughly stable for years.
I work in marketing and I know why the assumption ran the other way. Social Security is the biggest number in the budget, so people assume it must contain the biggest leak. Size and rate are different quantities and everybody conflates them constantly.
Medicare is a different story. The fee-for-service improper-payment rate runs in the high single digits — something over seven percent — and Medicaid is around five. On programs of that size, those rates produce tens of billions a year each.
So the ocean this file has been demanding we sail into has two very different halves. One is enormous and clean. The other is enormous and leaky, and the leaky one is the health programs.
Open both to the audit. Absolutely. But go in expecting the finding rather than a conclusion, because the finding on Social Security is going to be that it runs tighter than almost anything else in the government, and a waste campaign that promised otherwise will have to say so.
That is what auditing means. You find out.
Novi, MI
A plant that audits the side shed and ropes off the main line is protecting politics, not quality. Open the main line. What I want to do is predict what the gauges will read, because the prediction is checkable and it is not the one this Forum has been making.
Here are the published improper-payment rates on the three largest programs.
Social Security. Under one percent of benefits paid. Both retirement and disability.
Medicare fee-for-service. Somewhere above seven percent.
Medicaid. Around five percent.
Those rates are produced by the agencies under a statutory methodology and audited. They are not contested figures.
In quality terms that is a decisive difference and it should drive everything.
A one percent defect rate on a mature high-volume process is very good. Not perfect — nothing is — but a shop running at that level does not have a quality problem, it has a process operating near its practical limit. You do not send your best people there. You send them where the rate is seven.
So the sequencing conclusion follows automatically. The health programs are the target and Social Security is not, and the reason has nothing to do with which brand is politically dangerous. It has to do with where the defects are.
And the reason for the difference is instructive, because it tells you what the fix looks like.
Social Security pays a formula-determined amount to a person whose eligibility is established once and rarely changes. Age, earnings record, done. It is close to the simplest high-volume determination in government, which is why the error rate is low.
Medicare pays thousands of different procedure codes to hundreds of thousands of providers, each claim requiring a judgment about medical necessity and documentation. The dominant error category is insufficient documentation — the provider did not supply what the rule requires to establish the service was necessary. That is not fraud in most cases and it is enormously expensive.
Different defect, different fix. Social Security’s is a data-matching problem. Medicare’s is a documentation and coding problem, and it will not yield to identity matching at all, because nobody disputes the patient exists.
That distinction has been absent from this file for six months, and it is the difference between an audit that finds something and one that photographs well.
Open both. Staff the health programs. And when the Social Security audit comes back clean, publish it, because publishing a clean finding on the biggest program in the budget is the single most credibility-generating thing this whole enterprise could do.
Orange County, CA
Orange County knows the difference between a big number and a bad number, and this question conflates them. Let me put the operating figures down.
Social Security’s administrative cost is roughly half of one percent of benefits paid.
That is not a typo. The agency that writes checks to seventy million people spends about fifty cents in overhead for every hundred dollars it disburses.
I work in finance and I would be hard pressed to name a private financial institution operating at that ratio. Most fund administrators, insurers, and payment processors do not come close. The comparison is not favorable to the private sector and it is worth stating plainly by somebody who is generally on the other side of that argument.
So when this Forum has demanded that the audit go to the ocean, we should be clear-eyed that the ocean’s administrative overhead is already among the lowest of any large financial operation in the country, and its improper-payment rate is under one percent.
That does not mean leave it alone. It means the finding will be a compliment and we should be prepared to deliver one.
Now, three weeks ago a new commissioner was confirmed. And the agency is proceeding with a plan announced in February to reduce staff by roughly seven thousand positions, from about fifty-seven thousand to fifty.
Here is the analytical problem with that, stated as I would state it about any firm.
If administrative cost is half a percent and the improper-payment rate is under one percent, the efficiency case for a twelve percent headcount reduction is not obvious. There is no visible slack in the ratio. What a reduction of that size on that base does is degrade service — wait times, field office capacity, claim processing — while producing savings measured against a denominator that was already tiny.
Half a percent of overhead reduced by twelve percent is not where a trillion dollars comes from. It is not where a billion comes from.
That is the finding I would put in front of anybody serious. Audit the administration by all means. And when the audit shows the administration is already the cheapest large payment operation in America, follow the finding rather than the plan.
The health programs are a different case entirely, and that is where the analysts should be.
Bismarck, ND
Run it the other way, and then post what the numbers say.
A Democratic administration opens Social Security to a full integrity audit. Would our side object? We would not — we have been demanding exactly that for months, and we would be right.
So open them, both parties, always. That is settled and it does not need the reversal test.
What needs saying is what the audit will find, and I would rather post it now than be corrected in six months.
Social Security’s improper-payment rate is under one percent of benefits. Its administrative cost is around half a percent. Medicare fee-for-service runs above seven percent improper and Medicaid around five.
In compliance work those numbers would produce an immediate and uncontroversial examination plan. You do not allocate examiners by the size of the book. You allocate them by the error rate times the book. Two large books, one at under one percent and two at five to seven, and the answer takes about a minute.
The examiners go to the health programs.
And I would say one more thing to my own side, because we have spent six months implying that the big brands were being protected out of cowardice.
Some of that protection is cowardice. Some of it is that the numbers do not support the story we wanted to tell. Social Security is not a slush fund. It is a formula-driven payment operation running at a defect rate most private institutions would envy, and anybody who audits it honestly is going to have to say so.
Poison is not the risk. The risk is auditing it and disliking the answer, and then not publishing.
Publish it. A clean finding on the largest program in the budget buys more credibility for the next hard fight than any number of cancelled grants.
Clermont, FL
Audit the big pipes. And when you do, here is what you are going to find, and it is not what anybody expects.
Social Security’s largest integrity failure is not people stealing from it. It is the agency overpaying beneficiaries by its own error and then demanding the money back, sometimes years later, from people living on a fixed income who did nothing wrong and had no way to know.
That was a national story last year. The agency’s default was to withhold a beneficiary’s entire monthly check until the debt was cleared. It reduced that to ten percent in 2024 after the reporting became impossible to ignore. This April it raised the default back to fifty percent for retirement and survivor benefits.
So the integrity picture at Social Security is: a very low error rate overall, and where errors occur the agency’s own recovery practice is the thing causing the harm.
I consult with small businesses and this is a pattern I have seen many times. A firm discovers it has been undercharging a customer for two years, sends a bill for the whole amount, and is genuinely surprised when the customer — who paid every invoice as presented — reacts badly. The firm made the error. The customer is being asked to fund the correction out of current cash flow.
That is exactly what a clawback is, and the fifty percent default means half the monthly check.
So my answer on the question is: open the audit, and understand that the honest finding on Social Security integrity runs partly against the beneficiary and partly against the agency, and the second part is larger than most people expect.
Audit is not a benefit cut. That is the right principle and I keep it. What I would add is that an audit which finds the agency’s own recovery practice is generating hardship should say so, even though it is not the kind of finding a waste campaign was looking for.
Poison is the fear of the formula fight. Fine — do not touch the formula. But an audit that only looks in the direction it hoped to look is not an audit.
Jacksonville, FL
Ports do not exempt the largest berth from inspection because the ship is popular. Open them. What I want to do is separate the two things that get merged into one word, and then say what the inspection will actually show.
Two different subjects.
The formula. Who is eligible, at what age, for how much. Written in statute. Congress. Not an audit question and never was.
The administration. Whether the payment matched the rule, whether the identity was verified, whether the provider billed for what was delivered, whether the systems can close a file. That is an audit question and it has nothing to do with anybody’s benefit.
Keeping those words apart in every public sentence is the anti-poison. Merging them is how audits die, and both sides do the merging — critics because it protects the program, supporters because it makes an administrative finding sound like a policy victory.
Now the manifest, and it is not what this file predicted.
Social Security. Improper payments under one percent of benefits. Administrative cost around half a percent. Formula-driven determinations, eligibility established once. By the standards of any large payment operation, tight.
Medicare fee-for-service. Improper-payment rate above seven percent. Thousands of procedure codes, hundreds of thousands of providers, and a dominant error category of insufficient documentation to establish medical necessity.
Medicaid. Around five percent, with the added complication of fifty-plus state administrations running on federal match, which means the integrity problem is distributed across state systems the federal government does not operate.
So the ocean has a clean half and a leaky half, and the leaky half is the health programs, and the leaky half is also the harder one — because the fix is not identity matching. Nobody disputes the patient exists. The dispute is whether the documentation supports the claim, and that requires clinical review at a scale nobody has staffed.
That is an uncomfortable finding for a campaign that has been promising a trillion dollars, because it means the largest genuine integrity opportunity in the federal government requires hiring — medical reviewers, coders, auditors — at a moment when the policy is reduction.
I would put that on the record plainly. The place where the money actually is requires more people looking at claims, not fewer.
And the reason it has not been done is not that anybody protected the brand. It is that the payoff arrives in years and the cost arrives in the first budget, and no administration of either party has wanted to book that trade.
Open the berth. Publish the clean finding on Social Security, because publishing a favorable audit is what makes the unfavorable ones credible. And staff the health programs, which is the opposite of what the last four months have been doing.
Las Vegas, NV
You count the main cage. Nobody in this Forum disagrees, so let me talk about the thing that actually determines whether it happens.
Three weeks ago the Senate confirmed a new commissioner at Social Security, on a close vote. That matters more to this question than any argument about brands.
An audit of an agency that size is not an event. It is a two-year program requiring the agency’s own people to produce data, reconcile systems, and answer questions, and every one of those steps happens or does not happen depending on whether the person at the top wants it to. Auditors get what management decides they get. Anybody who has run a floor knows this.
So the question of whether Social Security gets a real integrity audit was largely settled on the sixth of May by fifty-three votes, and this Forum spent that week arguing about whether touching the brand was poison.
Now the numbers, because they set the assignment.
Social Security: improper payments under one percent, administrative cost about half a percent. Medicare fee-for-service: over seven percent improper. Medicaid: about five.
In my business you put the surveillance where the hold percentage is off, not where the drop is biggest. Big drop and normal hold is a good game. Modest drop and bad hold is where somebody is working you.
The health programs are where the hold is off.
So: open both, obviously, and assign the analysts by rate rather than by headline. And expect the Social Security audit to come back clean, and publish it when it does, because a shop that only publishes the findings it wanted is not running an audit. It is running a campaign with an audit-shaped logo.
Knoxville, TN
Audit the administration, leave the formula. That is the split and I want to specify the audit, because “open them up” is not a work plan.
Four questions, and they apply to any payment system regardless of size.
Is the recipient who they say they are? Identity resolution. This is Social Security’s territory and it is where the death-record reconciliation belongs.
Was the recipient eligible under the rule? Determination accuracy. Formula-driven programs do well here. Discretionary and needs-tested programs do worse.
Did the payment match the entitlement? Calculation accuracy. Social Security’s clawback problem lives here — the errors are the agency’s own and are found years later.
Was the service actually delivered as billed? Only applies where the government pays a provider rather than a beneficiary. Medicare and Medicaid. This is where the dominant error is and it is the hardest of the four, because verification requires clinical judgment on individual claims.
Run those four across the three big programs and the answer sorts itself. Social Security scores well on one, two and three, has no exposure on four, and its published error rate under one percent reflects exactly that. The health programs are exposed on four, which is why their rates run five to seven times higher.
So the design conclusion is that this is not one audit. It is two, with different methods, different staffing, and different timescales.
The identity and calculation work is data engineering. It scales, it is cheap, and it can be largely automated.
The service-delivery work is clinical claim review. It does not scale, it requires trained people looking at individual records, and the return per reviewer is high but the reviewer count has to go up.
In public works terms: one of these is a survey and the other is an inspection program, and a plan that treats them as the same thing will do the cheap one, report a number, and leave the expensive one where it has been for thirty years.
Open both brands. Split the work into its two kinds. And be honest that the half where the money is requires hiring, which is the recommendation nobody in this file has been willing to make and the only one the rates support.
