The Supreme Court has twenty-three cases left to decide this month, and one of them is Trump v. Cook. The facts, briefly: last August, the President posted a letter announcing the immediate removal of Federal Reserve Governor Lisa Cook, citing uninvestigated allegations that she misstated her residency on mortgage paperwork before she joined the Board. Cook declined to leave, on the theory that the Federal Reserve Act permits her removal only “for cause” and that an unproven allegation about a 2021 mortgage is not one. A district court agreed, the D.C. Circuit left the injunction in place, the Court heard argument in January, and the decision could arrive any opinion day now.
The legal commentary has organized itself around the question of whether the President can do this. That is the right question for the lawyers. The right question for this blog is different: why would he want to? Formal theory has a surprisingly complete answer, and the answer is that he shouldn’t — because the thing being sued for is worth less than the thing being sued away. Working through why takes us past three pieces of furniture that regular readers will recognize: a commitment problem, a junk drawer, and a conservation law.
The Problem Isn’t Bad Presidents
Start with the classic result, due to Kydland and Prescott and sharpened by Barro and Gordon.1 Suppose output responds to inflation surprises,
\[ y = \bar{y} + b\,(\pi – \pi^{e}), \]
and suppose the government dislikes inflation but would like output above its natural level — it minimizes \( \frac{1}{2}\pi^{2} + \frac{\lambda}{2}(y – y^{\ast})^{2} \) with \( y^{\ast} > \bar{y} \). The trouble is timing. If the government could commit in advance to a policy, it would choose \( \pi = 0 \): expectations would settle there, output would sit at \( \bar{y} \), and nobody would pay an inflation tax. But once expectations are set, a government choosing policy afterward faces a temptation: a little surprise inflation buys a little extra output. The public knows this, anticipates it, and builds it into \( \pi^{e} \). In equilibrium the surprise never materializes — you cannot systematically surprise people who know your objective function — and the economy lands at
\[ \pi^{D} = \lambda\, b\,(y^{\ast} – \bar{y}) \;>\; 0, \qquad y = \bar{y}. \]
Same output as under commitment. Strictly more inflation. The gap between \( \pi^{D} \) and zero is pure waste, and it is generated not by bad character but by the absence of a commitment technology. The government in this model is neither foolish nor venal. It is simply free, and its freedom is the problem, because everyone else can see it.
Independence Is a Removal Technology
The standard fix is Rogoff’s: delegate monetary policy to someone who weighs inflation more heavily than you do — a “conservative central banker” with a smaller \( \lambda \) — and tie your own hands against overriding her.2 The second clause is the entire mechanism. Delegation to an agent you can costlessly replace is not delegation at all; it is ventriloquism. If the President can remove any governor whose vote he dislikes, then by the usual backward-induction argument the governor’s preferences drop out of the model entirely. She either votes the President’s temptation or is replaced by someone who will, and the public — which is solving the same game we are — sets \( \pi^{e} \) accordingly. The inflation bias reappears in full, with extra steps.
This is why “central bank independence” is a misleading phrase if you hear it as a personality trait. Independence is not a property of central bankers. It is a property of the removal architecture — of how costly it is for the political principal to undo the delegation at the moment the delegation binds. The Federal Reserve Act’s contribution to that architecture is two words long: governors serve fourteen-year terms “unless sooner removed for cause by the President.” Everything — the term premium on thirty-year debt, the inflation expectations embedded in every wage contract in the country — hangs on what “cause” means and, more importantly, on who gets to say.
“Cause” Is a Junk Drawer
Congress never defined “cause,” and readers of this series will recognize that the omission is load-bearing rather than lazy. Genuine malfeasance arrives in forms no legislature can enumerate in advance — the 1913 Congress could not have listed crypto conflicts of interest, and the 2026 Congress cannot list whatever 2040 will produce. “Cause” is the residual category that lets the statute remain adequate to a world it cannot describe: a lost key waiting on information that hasn’t arrived yet. So far, so functional.
But a junk drawer is only functional if the interested party doesn’t control what counts as belonging in it. If the President both alleges the cause and classifies it as sufficient, with no review, then the classification is constitutive: calling it cause makes it cause, in exactly the sense this blog has been worrying for months. Once “cause” means “whatever the President says it means,” the phrase “for cause” has been quietly rewritten to “at will” without anyone amending a word of the statute — which, regular readers will recall, is also how seniority behaves in sovereign debt. These protections are architectures, not clauses, and an architecture with the adversary holding the master key is a floor plan.
The model says the bias returns the moment the public understands that the protection is gone — not when a governor is actually removed, but when removal becomes a live option in the public’s model of the game. We do not have to speculate about whether markets solve for that equilibrium, because they showed their work in real time. On the August night the removal letter was posted, before any court had said anything at all, the Treasury curve steepened: short yields fell on the expectation of a more pliant, more dovish Board, while long yields rose on the inflation expectations that pliancy implies, and the dollar sold off. That is \( \pi^{e} \) moving. Nobody waited for a holding. In rational expectations models, the announcement is the policy.
The Alibi Is Worth More Than the Steering Wheel
Everything so far is the macroeconomist’s case, and it is sufficient on its own: control of the Fed buys the President no output in equilibrium and costs him an inflation premium on everything the government and everyone else borrows. But there is a second case, a purely political one, and I find myself making it a lot these days. It goes like this: a smart President should want the Fed to be independent for his own sake, because independence is a blame technology, and blame is the currency that actually matters to him.
Monetary policy is, much of the time, the business of inflicting pain on purpose. Fighting inflation means raising rates, which means mortgages get expensive, hiring slows, and some recessions are scheduled rather than suffered. Voters are retrospective: they punish the visible author of economic pain, and they are not especially careful about causal attribution. Formal theory has understood for a long time what this implies about institutional design. Fiorina called it shifting the responsibility; Weaver built a whole politics of blame avoidance on the observation that politicians care more about not being blamed than about claiming credit; Alesina and Tabellini showed that exactly the tasks where politicians will be judged on painful outcomes are the ones rationally handed to unelected technocrats; and Binder and Spindel have argued that Congress designed the Fed, in no small part, to have somebody to point at.3 “I’d love lower rates, but the Fed is independent” is not an admission of weakness. It is an asset with a positive price, and every President since the Treasury-Fed Accord has held it on his balance sheet.
Now run the counterfactual in which the administration wins Cook outright. The asset is gone — not just for this President, but for all of them, in both parties, forever. Every basis point of every mortgage becomes presidential policy. Every recession is signed. The first President to control the Fed is also the first President in ninety years who cannot say “talk to the Fed,” and he acquires that liability in exchange for a steering wheel that the Barro-Gordon result says steers nowhere: no extra output, just a higher price level and a public that has repriced his promises. He is litigating to trade a valuable option for a worthless one. The deep reason is one this blog keeps returning to: the value of a commitment device lies precisely in your inability to use the discretion it takes away. The Fed’s independence was never a constraint on the President. It was a service to him, and like many services “that have been around a long time”, its price becomes visible only upon cancellation. (This is another example of Chesterton’s fence, but we’ll return to that soon.)
Who Has Cause to Review the Cause?
The optimistic reading of the litigation is that judicial review will police the drawer: the President may allege cause, but an Article III judge decides whether the allegation belongs in the category, and the commitment survives. Maybe. But notice what that reading concedes. The commitment problem has not been solved; it has been relocated, from the Federal Reserve Act to the federal judiciary, and the judiciary’s willingness to second-guess a President’s classification is itself an equilibrium object — one the emergency docket has been estimating, in public, for over a year. The question “who has cause to review the cause?” is the Kalshi question wearing a robe. Conservation of impossibility holds, as it always does: you cannot legislate credibility into existence, you can only choose where to store the uncertainty about it.
The Court’s own attempted storage solution deserves a final word. In Trump v. Wilcox, while letting removals at other agencies proceed, the Chief Justice described the Federal Reserve as a “uniquely structured, quasi-private entity” in a “distinct historical tradition” — language widely read as pre-building an exemption for the Fed alone.4 Read through this series, that is an attempt to construct a junk drawer containing exactly one item, with the item’s name printed on the front of the drawer. And a junk drawer with one labeled item is not a junk drawer. It is a target, with its location published in the U.S. Reports. The kind of protection the Court is drafting tells every future President precisely which drawer to pry open, and stakes the credibility of the dollar on the hope that none of them ever brings a screwdriver.
With that, I leave you with this.
1 Kydland, F.E. and Prescott, E.C. (1977). “Rules Rather than Discretion: The Inconsistency of Optimal Plans.” Journal of Political Economy 85(3): 473–491; Barro, R.J. and Gordon, D.B. (1983). “A Positive Theory of Monetary Policy in a Natural Rate Model.” Journal of Political Economy 91(4): 589–610.
2 Rogoff, K. (1985). “The Optimal Degree of Commitment to an Intermediate Monetary Target.” Quarterly Journal of Economics 100(4): 1169–1189. Rogoff’s deeper point is routinely forgotten: full commitment is not optimal either, because you sometimes want discretion to respond to genuine shocks. The design problem is choosing how much discretion to destroy, which is why the removal architecture — rather than a mechanical rule — is where modern systems store the answer.
3 Fiorina, M.P. (1982). “Legislative Choice of Regulatory Forms: Legal Process or Administrative Process?” Public Choice 39(1): 33–66; Weaver, R.K. (1986). “The Politics of Blame Avoidance.” Journal of Public Policy 6(4): 371–398; Alesina, A. and Tabellini, G. (2007). “Bureaucrats or Politicians? Part I: A Single Policy Task.” American Economic Review 97(1): 169–179; Binder, S. and Spindel, M. (2017). The Myth of Independence: How Congress Governs the Federal Reserve. Princeton University Press.
4 Trump v. Wilcox, 605 U.S. ___ (2025) (per curiam). Whether a “historical tradition” can do the work of a commitment device is, of course, the question this post answers in the negative: traditions are common knowledge about behavior, and common knowledge about behavior is exactly what an off-equilibrium-path removal letter is designed to revise.