Warsh is right on Fed reform, but wrong on AI-led rate cuts

Kevin Warsh’s argument for remaking the Federal Reserve deserves a serious hearing. The Cato Institute lays the case out plainly: a swollen remit, an expansive balance sheet and too much faith in discretionary steering have given the central bank room to err. But the same essay drives its sharpest point into Warsh himself. He wants lower rates because artificial intelligence will soon crush inflation, he says. Markets should hear that as a risky macro bet — not a durable policy framework.

Warsh is not some think-tank hobbyist. Fortune recently profiled him as a plausible contender for Fed chair, which means his ideas are being read as a live template for the post-Jerome Powell era rather than a seminar-room provocation. A reform agenda built around tighter accountability, a smaller balance sheet and less theatrical forward guidance is one thing. An easing case built on a productivity surge that has not yet shown up decisively in the inflation data is something else entirely.

His package’s sturdier half sits at the center. In a Hoover Institution interview, the former Fed governor said “AI will be a significant disinflationary force” and told the audience he believed the US was at the front end of a productivity boom. The reform themes around that claim are easier to defend than the monetary conclusion. Its balance sheet sits above $7 trillion, and a Mercatus Center brief argues that shrinking that footprint, clarifying the institution’s scope and making its choices easier to audit would leave fewer channels for mission creep. Even investors who disagree on the speed of quantitative tightening can follow the logic: a central bank that does less, explains less theatrically and leans less on constant narrative management creates fewer surprises of the wrong kind.

Years of crisis-era balance-sheet activism and heavy verbal signaling have encouraged traders to treat the Fed as the dominant author of financial conditions, not just a rate setter. In a panic, that may be unavoidable. As a standing operating philosophy, it is far less defensible. Warsh’s strongest case is that the institution should reclaim discipline — narrow its job description, reduce the reflex to fine-tune every wobble in asset prices, and let the price of money transmit through a clearer rule set.

Where the inflation case slips

The package starts to wobble on cyclical policy. Warsh’s claim, delivered again in that Hoover interview, is that AI will act as a “significant disinflationary force.” Cato pushes back — and not without reason. Faster productivity can eventually lower unit costs. It can also raise expected growth, support investment, lift real incomes and, crucially for central bankers, push the neutral interest rate higher rather than lower. The lesson in that world is not that the Fed should cut early. Policy may need to stay firmer because the economy can run hotter without immediately stalling.

US consumer-price inflation sits at 3.8 per cent, still well above the Fed’s 2 per cent target. A chair who eases on the back of a hoped-for supply shock is effectively asking markets to front-run the evidence. The burden of proof runs the other way. Productivity booms show up in the data after firms reorganize, invest, retrain and scale. Monetary policy works with lags, but it cannot be run on venture-capital time.

Cato cites MIT economist Daron Acemoglu’s estimate that AI could lift total factor productivity by only 0.66 per cent over a decade — modest once you spread it across years. Goldman Sachs research, cited in the same piece, projects a 1.5 percentage-point annual lift. Those are not rounding errors around a common base case. They describe entirely different macro worlds. A central bank can study both. It cannot responsibly cut rates as though the optimistic one has already arrived.

Even if the technology ultimately lowers costs in parts of the economy, the first-round effects can be inflationary in finance and labour markets. Companies spend on chips, software, data centres and hiring before efficiency gains spread broadly. Asset prices rerate around the prospect of higher margins long before measured productivity settles into national accounts. That sequencing is exactly why a future supply-side dividend should make policymakers more cautious about declaring victory too early, not less.

What reform would mean for markets

For investors, the practical takeaway is to separate Warsh’s governance critique from his rate prescription. On governance, he says the Fed should be smaller in footprint, clearer in mandate and less reliant on constant rhetorical shepherding. That could matter for everything from term premia to rate volatility — a less interventionist Fed would probably tolerate more market price discovery and offer fewer comfort blankets when asset prices wobble. On rates, the argument is that an AI boom is close enough, and disinflationary enough, to justify easier money sooner. That claim is the weaker one by a wide margin.

It is also the claim most likely to unsettle markets if it migrates from the speech circuit to the policy table. Chair Jerome Powell and Vice Chair Philip Jefferson have spent the past cycle arguing, in different registers, that the Fed must stay anchored to realized inflation and labour-market evidence rather than to optimistic narratives about what supply might soon deliver. Warsh’s institutional reforms could fit inside that discipline. His inflation thesis does not. Once traders conclude a chair is leaning on a forecast that flatters the case for cuts, every data print carries a second question: is the Fed reacting to the economy, or to a story it wants the economy to become?

Cato’s critique matters for a narrower reason. It gives markets a cleaner way to price the Warsh package. Take the reform impulse seriously. A Fed with a narrower remit, a leaner balance sheet and less appetite for performative guidance would look meaningfully different from the institution investors have traded for much of the past decade. But treat the AI disinflation argument as a hazard light. A productivity boom may yet arrive. It is not a sound basis for easing while inflation remains at 3.8 per cent and the evidence is still this thin.

Warsh’s most convincing line is that the Fed needs tighter rules and fewer moving parts. His least convincing is that AI makes that easier by solving inflation on schedule. The first argument is about institutional design. The second is a wager. Markets should know the difference before they start pricing them as the same trade.