The Fed's renovation needed a price ceiling, not a prosecutor
The watchdog found mismanagement and no crime. The finding worth arguing about is buried in the recommendations: years into a multibillion-dollar build, nobody has agreed a maximum price.
The Federal Reserve is the United States central bank. Unlike most of the government, it is not given money by Congress — it covers its own costs from the interest it earns and sends the surplus to the Treasury. It is rebuilding its Washington headquarters, and the cost estimate has roughly doubled. Its internal watchdog has now reported poor management but no crime, and told the Board to agree a maximum price with its builders.
Whether a crime occurred was always the least useful question. The finding that matters is that a multibillion-dollar build is years in with no agreed price ceiling — and a central bank that funds itself has no appropriator to impose one.
The sentence nobody quoted
The most consequential line in the Inspector General’s report on the Federal Reserve’s headquarters renovation is not the one about criminal law. It is the recommendation that the Board should establish a guaranteed maximum price. Read that again with the dates attached: a project begun years ago, with an estimate that has gone from $1.3 billion in 2020 to more than $2 billion by 2024 and is now described at $2.5 billion, does not yet have an agreed ceiling above which someone other than the Fed eats the overrun.
A guaranteed maximum price is a plain idea. The builder agrees a number. If the job comes in under it, fine. If it comes in over, the builder absorbs the difference, not the client. It is not exotic; it is how a great many large commercial fit-outs are contracted, precisely because the client knows it cannot police every change order. Without one, every surprise — and an eighty-year-old marble landmark will produce surprises — travels straight through to the owner’s bill.
The question everyone asked was the least interesting one
The watchdog said it found no reasonable grounds to believe a violation of federal criminal law occurred, and identified no administrative misconduct, while outlining deficiencies in how the project was managed. That is a narrow finding and it will satisfy nobody who wanted a culprit. President Trump responded by demanding the former chair’s resignation and saying he had asked the Attorney General to study the report — which tells you how the document will be used, not what it says.
Treat the political fight as noise and the procurement finding as signal. “Mismanagement, not criminality” is the normal diagnosis for cost overruns on big buildings. It is also the diagnosis with an actual remedy, which prosecution would not supply. Nobody goes to prison for failing to cap a construction contract. Someone does, eventually, have to write the cheque.
Self-funding is the mechanism
Here is why this is a central banking story rather than a facilities story. Almost every other arm of the United States government has to go to Congress for construction money, and an appropriator with a number in hand is a crude but effective discipline. The Fed does not. It covers its own costs from the earnings on its portfolio and remits what is left to the Treasury. There is no vote, no line item, no hearing at which a ceiling is set as a condition of the money.
That insulation exists for a reason, and the reason is monetary policy. You do not want the body setting interest rates to be negotiating its operating budget with the politicians whose borrowing costs it determines. But insulation from Congress’s purse was never meant to extend to drywall and glazing. Where the external discipline is deliberately removed, the institution has to manufacture an internal equivalent: a hard ceiling, a named owner, and a recorded decision whenever the ceiling moves. The Inspector General’s recommendation is an admission that this did not exist.
What the Fed asks of everyone else
The contrast is uncomfortable. In the same week, the Fed and the Federal Deposit Insurance Corporation published feedback on the resolution plans — the “living wills” — of fifteen banking organisations with more than $250 billion in assets, documents that exist solely to force firms to write down in advance how a large, complicated undertaking would be managed if it went wrong. The Board also signed a written agreement with a small Wisconsin holding company, the standard instrument for telling a bank that its governance needs fixing.
Supervisors spend their working lives asking regulated firms a simple question: who owns this risk, what is the limit, and who has to approve breaching it. Those are exactly the questions the renovation appears to have gone years without answering about itself. A supervisor who found a bank running a multi-year, multibillion-dollar capital programme on open-ended cost-plus terms with a drifting estimate would not call it criminal either. They would call it a governance finding and require a plan.
The case against a ceiling
The strongest objection is that guaranteed maximum prices are not free. Ask a contractor to cap the cost of restoring a 1930s federal landmark, with unknown conditions behind the stonework and preservation review over its shoulder, and the contractor will price the risk it is being asked to swallow. The ceiling may sit well above the honest expected cost. Impose one now, with much of the risk already discovered and much of the work already done, and the leverage is poor: the Fed is negotiating from inside a half-finished building.
That is an argument for disclosing the contingency, not for continuing without a number. A padded ceiling with the padding visible is still better information than an estimate that moves every year for reasons the public learns about afterwards. And there is a second objection worth naming: governance reform extracted under political pressure can become a lever against independence. True — which is why the Board should do it itself, now, and own it.
Why the boring number is the best defence
The Fed’s independence is contested on substance, and it will be defended on substance. But it is attacked through the housekeeping, because the housekeeping is legible to everyone and the monetary economics is not. A published ceiling, a published contingency and a quarterly figure for cost-to-complete would retire this subject as ammunition. Refusing to produce them keeps a renovation in the same sentence as the federal funds rate for years to come.
The Board should accept the Inspector General's recommendation without waiting to be asked twice: negotiate a guaranteed maximum price, publish the ceiling and the contingency inside it, and report cost-to-complete against that ceiling in every quarterly financial statement until the project closes.
It should also do the structural thing the self-funding model requires — a standing capital-expenditure limit that cannot be raised except by a recorded vote of the Governors, published with the vote. The Fed demands documented limits and named owners from the banks it supervises. It can hold itself to the same paperwork.
What would change this view
If the Board signs a guaranteed maximum price contract and publishes it — with the ceiling, the contingency inside it and a dated completion — in time for its next annual financial statements, and then reports a quarterly cost-to-complete figure against that ceiling, my complaint is answered and the governance gap was closed by the ordinary audit process working.
If instead the final cost lands below a ceiling that was set above the current estimate, the ceiling was decoration and I was wrong to treat it as the fix. In that case the answer is a standing capital-projects limit requiring a recorded Board vote to raise, not a contract clause.
Worked from
- Fed watchdog finds mismanagement, not criminality, in $2.5B renovation
Banking Dive · 2026-10-01