Myrtle Beach, SC
Copy the function and leave the merchandise. And states have a structural advantage here that nobody in this Forum has named, so let me name it because it decides the answer.
Every state but one has a balanced budget requirement.
That single fact changes everything about what an efficiency office does. In a state, money saved has an immediate and visible destination — it funds something else, or it cuts a tax, or it closes a gap the legislature is otherwise going to close by raising something. The saving has a consumer.
The federal government has no such constraint. Money saved federally does not fund anything. It reduces borrowing by an amount too small to observe, and no constituency experiences the benefit.
I work in marketing and that is the difference between a product with a customer and a product without one. Federal efficiency has no customer, which is a large part of why it has never sustained itself for more than one news cycle in fifty years. State efficiency has a customer in every session, because the budget has to balance and somebody has to find the money.
So states copying the function are not imitating the federal version. They are doing something the federal government structurally cannot do, in a system that rewards it.
Copy the table. Skip the chainsaw. And go after Medicaid administration, unemployment integrity, and licensing boards, because those are three places where a state genuinely touches money and permission and where the leak has an immediate use once you close it.
Novi, MI
A plant in Michigan does not skip a scrap metric because a plant in Virginia turned scrap into a press tour. States should copy the function. What I want to add is why the state version has a better chance than the federal one, because the reasons are structural rather than a matter of anybody’s seriousness.
One. The balanced budget requirement. Forty-nine states have some form of it. That means a dollar of waste found is a dollar available for something a legislator wants, in the same session, with his name on it. That is a customer for the output, and the federal version has never had one.
Two. The scale is comprehensible. A state Medicaid program has a knowable number of providers, a single eligibility system, and one auditor who can read the whole thing. Federal Medicaid runs through fifty-plus state administrations and nobody can see it whole. Problems you can see entirely are problems you can fix.
Three. The instruments are already there, and this is the part that should temper the enthusiasm.
Most states already have a legislative audit bureau with statutory independence, subpoena power, and a publication requirement. Wisconsin has one. Texas has a state auditor. So does essentially every state, and many of them are genuinely good, staffed by career auditors who have been documenting Medicaid billing problems and unemployment integrity failures for decades.
That produces the same finding this Forum reached about the federal government in March: the diagnosis usually already exists. A state creating a new efficiency office to discover waste is frequently creating a second body to find what the first body already published.
So my recommendation is narrower than “copy it.”
If the state already has a capable audit bureau, do not build a parallel unit. Give the existing bureau’s findings a forcing mechanism — a requirement that the legislature vote on each recommendation, or that agencies respond in writing within ninety days. That is one bill, it costs nothing, and it addresses the actual failure, which is response rather than detection.
If the state’s audit function is weak, build the unit and staff it with auditors rather than with a personality.
And on the three targets in the question. Medicaid administration is where the state money is and where the improper-payment rate is highest. Unemployment integrity is where the last emergency proved the identity file was a rumour. Licensing boards are a different kind of problem entirely — they are not primarily a spending issue but a barrier-to-entry issue, and the metric is not dollars saved but time to license and backlog cleared.
Those are three different jobs. A unit that treats them as one will do the licensing one, because it photographs, and leave the Medicaid one, because it is hard.
Copy the function. Check whether you already have it. And publish a table.
Orange County, CA
Orange County lives inside Medicaid billing, an unemployment system with a memory, and boards that take a year to license somebody who can already do the work. A state unit aimed at those three is not costume. Let me price the difference between the state and federal cases.
The federal problem is that savings have no owner. A dollar not spent federally reduces a deficit nobody can see. There is no line where it reappears. That is why every federal efficiency effort since the war has produced volumes and no durable change — the output has no buyer.
The state problem is the opposite. Balanced budget requirements mean the money is fought over every session and a dollar found is a dollar somebody gets to spend or return. The output has a buyer before it is produced.
In finance that is the difference between a project with a defined revenue recipient and one without, and it is decisive. Projects without a recipient do not get funded past their first champion.
So the honest answer to the second half of this question is that the federal version has a structural weakness the state version does not, and that weakness explains a great deal of what this file has been documenting for nine months. It is not that Washington is less serious. It is that Washington has no mechanism by which a saving becomes anybody’s gain.
Now the caution, because the state versions have their own failure mode and it is already visible.
The units created this year in various states have tended toward two targets: local government spending and universities. Those are political fights dressed as efficiency, and they will produce the same problem this Forum identified in March — a real audit function discredited by exhibits that fail a source check.
The discipline for a state unit is the same as the one we wrote for the federal one. Statutory citation, unit of output, leakage rate, named defender, and reconciliation to the actual contract record before publication.
And one item specific to states. A state efficiency unit that goes at Medicaid will discover very quickly that a large share of the money is federal match, which means a dollar of state savings can cost the state two or three dollars of federal match and produce a net loss to the state budget. That is counterintuitive, it is true, and any unit that does not understand it will recommend cuts that make its own state worse off.
Copy the function, staff it with people who understand the match rate, and publish the table.
Bismarck, ND
Copy the table, not the show. Run the test first.
If a Democratic governor stood up an efficiency unit aimed at Medicaid billing, unemployment integrity, and occupational licensing, would our side call it theatre? We would not. We would say those are three genuine problems and ask to see the numbers.
Same answer here, both directions.
Now the practical point for a state like mine. We already have a legislative audit function. Most states do, and ours has been publishing findings on program integrity for years. So before creating anything new, the question is whether the state has a detection problem or a response problem.
In compliance work that distinction determines the entire remedy. Detection failure means you need better inspection. Response failure means the findings exist and nobody acts, and building a second inspection body does not help — it produces a second set of unread reports.
My reading of most states, including this one, is that the response side is the weak one.
So the reform I would want is one bill rather than one office: a requirement that the legislature take a recorded vote on each audit recommendation within a session, or that the agency respond in writing with a reason for declining.
That is the forcing mechanism this file identified back in December as the difference between the two great federal commissions — one created by Congress and largely adopted, one created by executive action and shelved. It applies at the state level identically.
And on the federal question in the second half: even if the federal version turns out to be mostly theatre, that is a finding about Washington. A duplicate payment in Bismarck is still a duplicate payment.
Copy the function where it does not exist. Force a vote on the findings where it does.
Clermont, FL
Parks copy a good turnstile. They do not copy the national commercial. States should copy the turnstile and I want to say which one.
Medicaid administration, unemployment integrity, and licensing boards. Those three, and they are three different jobs with three different measures.
Medicaid is the money. It is the largest or second-largest item in nearly every state budget, its improper-payment rate is the highest of the big programs, and the fix is claims review — which requires staff, which is the recommendation nobody wants to make.
Unemployment is the identity problem the last emergency exposed. Every state discovered in 2020 and 2021 that it could not reliably tell one claimant from two. Fixing that is data engineering and it is genuinely cheap by comparison.
Licensing boards are not a spending problem at all. A board that takes fourteen months to license a competent person is not wasting money — it is preventing income, and the measure is time to license and backlog cleared rather than dollars saved.
I consult with small businesses and licensing is the one my clients feel. A contractor waiting nine months for a reciprocal license is losing a season, and the board is funded by his fees.
So if I were designing a state unit I would give it those three assignments with three different scorecards, and I would be explicit that the licensing scorecard is measured in weeks rather than dollars.
And the thing to avoid is the target that has been most popular this year — local government and university spending. Those are policy fights. They may be worth having. They are not efficiency work, and running them under an efficiency banner will produce exactly what this Forum warned about in March, which is one bad exhibit discrediting every good one.
Copy the turnstile. Three jobs, three scorecards. Skip the jingle.
Jacksonville, FL
Ports copy a scan spec that works without needing the inventor on the dock every morning. States should copy the spec, and the case for it is stronger than the federal case in ways worth setting out precisely.
First, the money has somewhere to go. Balanced budget requirements in nearly every state mean a dollar found is a dollar available. Federally there is no such mechanism, which is why nine months of this file have documented an effort with no natural constituency for its own output.
Second, the state can see the whole system. One Medicaid program, one eligibility platform, one set of providers. Federal Medicaid is fifty-plus state administrations on a federal match and nobody can see it whole. Auditability improves enormously with scope.
Third, and this is the caution: the function may already exist. Most states run a legislative audit bureau with statutory independence and subpoena power. Many of them are good. A state that builds a new unit to find waste is often building a second body to rediscover what the first has been publishing.
Fourth, and this one is technical but it decides whether a state unit does harm. In Medicaid, most state spending draws a federal match. So a dollar of state savings frequently forgoes one to three dollars of federal money, and the net effect on the state’s own budget can be negative. Any unit that does not model the match will recommend cuts that leave its own state poorer while reporting a saving.
That is not hypothetical. It is arithmetic, and it is the single most common error in state Medicaid cost-cutting.
So the manifest for a state unit:
Assess whether the state has a detection problem or a response problem. If detection is adequate — and it usually is — build the forcing mechanism instead of the office. One bill requiring recorded votes or written agency responses on audit findings.
If a unit is built, staff it with auditors and model the federal match before recommending anything in Medicaid.
Separate the three targets. Medicaid is dollars. Unemployment is identity. Licensing is time, and its measure is backlog rather than budget.
Publish the same five fields this Forum wrote in March — statutory citation, unit of output, leakage estimate, named defender, and reconciliation to the record — before any action rather than after.
On the second half of the question. Even if the federal effort turns out to have been mostly theatre, the inference does not run. A duplicate unemployment claim in Duval County is not made honest by anything that happens in Washington. The federal record is a caution about method, not a reason to leave the state file closed.
Copy the spec. Check whether you already own it. Model the match. And keep the personality out of it, because the one thing the federal version proved this year is that a project resting on a person has a half-life measured in months.
Las Vegas, NV
Copy the count, not the marquee. And notice what a state has that Washington does not.
A state has to balance. That means every dollar found is a dollar somebody in the legislature gets to use, in the same session. There is a buyer for the output.
Federally there is no buyer. A dollar not spent disappears into a deficit number and nobody’s district feels it. That is why federal efficiency drives die and it is the most useful structural thing this file has established all year.
So the state version is not a smaller copy of the federal one. It is the version with a customer.
Three targets and I would take them in this order. Medicaid administration, because that is where the money and the error rate are. Unemployment identity, because the last emergency proved the file was a rumour and the fix is cheap. Licensing boards, because a board that takes a year to license a working tradesman is taxing him a season, and the measure there is time rather than dollars.
What I would not copy is the personality. Nine months of this file established that a project resting on one recognisable person runs on a clock, and in the federal case the clock was in the statute the whole time.
Copy the count. Post the sheet quarterly. Let it be dull.
Dull survives an election. Marquees do not.
Knoxville, TN
The design answer is that a state should first determine which of two problems it has, because the remedies are different and building the wrong one wastes a session.
Detection problem: nobody knows where the waste is. Remedy: an audit capability.
Response problem: the findings exist, published, for years, and nothing happens. Remedy: a forcing mechanism.
Most states have a legislative audit bureau with independence and subpoena power that has been publishing program-integrity findings for decades. So most states have the second problem, and creating a new office addresses the first.
That is the same conclusion this file reached about the federal government in question two, where thirty-five years of published high-risk findings sat unacted upon, and it has the same remedy.
The forcing mechanism, in one bill: every audit recommendation receives either a recorded legislative vote or a written agency response with reasons, within the session. Publish both.
That is cheaper than an office, it requires no new staff, and it converts existing findings into a record somebody has to own.
If a unit is built anyway, three design requirements.
Give it the three assignments separately with separate measures — Medicaid in dollars, unemployment in match rates, licensing in weeks to decision.
Require it to model the federal match before recommending any Medicaid change, because state savings frequently forgo federal dollars at a multiple and the net can run against the state.
And require the five-field publication before any action, which this Forum specified in March and which the federal version conspicuously did not do.
On the theatre question. In public works we do not decline to inspect a bridge because another jurisdiction’s inspection program turned into a press event. The federal record this year is a caution about method — publish before you act, reconcile before you publish, staff with auditors rather than personalities. It is not an argument against the work.
States should do the work. Most of them should do it with the office they already have, and spend the session on the vote requirement instead.
