# When a Central Bank Stops Explaining Itself: Donald Kohn on the Fed's Past, Present, and Future > Reflections on the August 31, 2026 Macro Musings conversation with 40-year Fed veteran Donald Kohn: why Volcker eased while inflation was still near four percent, why Greenspan refused to forecast but never refused to narrate, and why a claim that cannot be falsified is the opposite of accountable. Educational, not investment advice. Published: 2026-08-31 Locale: en Tags: Federal Reserve, monetary policy, inflation, central bank communication, reading the market ![An empty congressional hearing room where a lone older official sits at the witness table, high window light falling across the bare surface in front of him, the tiers of member seats receding into shadow](/covers/macromusings-2026-08-31-donald-kohn-on-the-fed-s-past-present-and-future-cover.png) > "He plants them as if tending a child, and then leaves them as if abandoning them — so their nature is left whole and their character fulfilled." > > — Liu Zongyuan, "The Tale of the Hunchback Gardener Guo" (Tang dynasty, c. 9th century; author's translation) The gardener in Liu Zongyuan's essay says he has no secret. He plants carefully and then walks away. The people whose trees die are not careless — they care too much. They come back daily to scratch the bark and check whether it is alive, to shake the roots and check whether they hold. The tree cannot survive that kind of attention. The parallel to a central bank is uncomfortable, because the hard question is exactly this: when to act like a parent, and when to let the market grow its own judgment. And the subtle part is that letting go and saying nothing are not the same thing at all. ## What this episode is about On the August 31, 2026 episode of Macro Musings, David Beckworth welcomed back Donald Kohn. Kohn spent forty years inside the Federal Reserve System: Kansas City in 1970, the Board staff in 1975, service under Burns, Volcker, Greenspan and Bernanke, founding director of the Division of Monetary Affairs, then Governor from 2002 to 2010 and Vice Chair at the end. He is now at Brookings, writing a memoir. The conversation runs in three movements. A retrospective — Greenspan recently died, and Kohn walks through the moments that mattered. A diagnosis of the present — inflation has now run above target for close to six years, and the soft landing never quite finished landing. And a look forward at the task forces the new chair, Kevin Warsh, has assembled, especially the ones on the balance sheet and on communication. The three movements turn out to be one question: **what entitles an unelected institution to be believed.** ## The main points **One. Volcker eased while inflation was still around four or five percent — and the reason he gave was not the real reason.** The Fed backed off in the second half of 1982. The official explanation was that the Fed was targeting money, and deregulation had broken the relationship between money, income and prices. Kohn says the actual reasons were two others: inflation expectations had started coming down — the fever had broken — and financial stability was genuinely at risk, with several large banks badly exposed to Latin American loans. His summary of Volcker's judgment: back off now, but keep the vision intact. Every speech afterward still hammered price stability. **This is the line between a tactical retreat and a thesis being retired** — a distinction many people never learn to draw, including in their own positions. **Two. Burns could see the inflation. He just did not want to pay for it.** Kohn's diagnosis of the earlier chair is blunt. Burns knew inflation was rising, but feared the unemployment cost would be too high and too persistent, and believed monetary policy was less potent than fiscal policy, cost-of-living clauses and union contracts. So he did not act. Volcker's stance was the opposite: we will incur the cost, it is a one-time cost, and it will change everything. It did — roughly two to three decades of expansion followed. **"It's a one-time cost" is something you have to believe before you can act on it,** and nothing available at the time could prove it. **Three. Greenspan's defining stretch was not the famous 1996 productivity call. It was the years of being yelled at beforehand.** After the 1991 recession came the first jobless recovery. Rates sat at what then looked like a startlingly low three percent, and pressure to ease further came from every direction. President George H. W. Bush asked publicly for lower rates — not in the style Donald Trump later used, but publicly — and afterward told David Frost, of Greenspan: "I reappointed him, and he disappointed me." It was bipartisan. Kohn remembers sitting behind Greenspan at hearings while three Democratic senators took turns shouting at him to cut. He did not cut. Kohn thinks holding that line mattered more than the elegant maneuvers that followed. In 1994 he saw inflation pressure building, tightened substantially, eased slightly in early 1995, and produced the textbook soft landing. **Four. The productivity call was hard because he was not watching the number everyone else was watching.** Conventional economists were urging him to hike: unemployment was too low, inflation had to follow. He said no, something else is going on. What he watched was unit labor costs. Unemployment was indeed low, but productivity was surging, wages were rising more slowly than productivity, so unit labor costs stayed muted — the productivity surge had a temporary disinflationary effect. Kohn used to tease him that he would take one questionable number, divide it by another questionable number, and somehow arrive at an indicator that told him exactly what was happening — "sometimes, or most of the time, not always." He still remembers a speech about Class 8 trucks in early 1995. It is the most vivid line in forty years of recollection: **this was not a man running a model. This was a man who knew the data well enough to see the economy doing something different from what everyone assumed.** By 1999, with the dot-com boom feeding consumption through the wealth effect, he started raising again. **Five. Greenspan never submitted two- or three-year forecasts, but he always supplied a narrative — and for Kohn the value of a narrative is that it can be proven wrong.** There was no dot plot then. Kohn and Mike Prell would occasionally suggest slipping the staff forecast into the published central tendency so the numbers would line up with the story. Greenspan sometimes agreed, rarely. His reason for refusing longer forecasts was simple: I do not know what happens in two or three years. But in every monetary policy testimony he laid out what was happening in the economy, why, where the risks were, how the committee saw things evolving and on what basis. Kohn's own practice as a Governor: in speeches he talked about the right-hand side of a Taylor rule — what he expected for output, employment and inflation — **and let people infer where rates were going. They are inferring. I am not saying.** Then the key sentence: the narrative has to be verifiable or falsifiable by incoming data or new analysis. That is what accountability is made of. His example is Powell — "transitory" was wrong, and Powell said so himself. But there was an analysis, a narrative, a set of stated reasons, which is precisely why it could later be checked and why he could be held to it. **An institution that never explains its thinking cannot be held accountable, because there is nothing to reconcile against reality.** **Six. Why people hate inflation so much: the raise is my achievement, the price increase is someone taking it from me.** Kohn was telling colleagues at Brookings in 2021 and 2022 that people hate inflation and this is a bad thing, when few wanted to hear it. His explanation is domestic rather than technical. I got a raise because I am a good worker and I am being rewarded; rising prices merely erode that reward. An economist looks at the same raise and says it barely kept up, there is almost no real increase in it — but the worker's experience is that it *would* have been a real increase if prices had not eaten it. Which is why short-horizon inflation expectations are so sensitive to gasoline and groceries, the things people buy constantly and cannot skip. Beckworth had sent over a long Gallup series. In the mid-to-late 1970s inflation was problem number one, at times ranking above unemployment. Post-Volcker, for decades, unemployment topped the list and inflation sat far down. Then the pandemic, the spike — **and it still has not fully come back down.** Kohn's reading of that chart: not worrying about inflation is exactly Greenspan's definition of price stability — **when households and businesses do not have to pay attention to inflation, and do not.** We have lost some of that. **Seven. Interest on reserves was argued for on grounds that have nothing to do with what it became.** Beckworth produced Kohn's June 22, 2004 testimony to the Senate Banking Committee, making the case for authority to pay interest on reserve balances. Kohn's recollection is candid: the argument was economic efficiency. Banks were being forced to make interest-free loans to the federal government, and responded with a great deal of resource-wasting effort — suppressing demand deposits, running sweep accounts. **"I don't think I was thinking at that time about the zero lower bound and QE."** The authority passed in 2006, was scheduled for 2011, got delayed by budget scoring, and was pulled forward into 2008. A 2008 internal paper laid out five options — tiered reserves, a voluntary balance target, a corridor, a floor with high balances, a wide daily band. Kohn does not recall an intense discussion of them; the financial system was frozen and the live question was how to keep control of the funds rate with a large balance sheet while buying securities in size. A tool designed for efficiency became the instrument of rate control, and then defined the operating framework for over a decade. **Eight. The discount window's stigma has two layers, and only one of them is fixable by the Fed.** The Fed was founded in large part to create a discount window so that 1907 would not recur. It did not work, because of stigma. Kohn separates the layers. The market layer: other participants see a bank borrow and conclude something is wrong — and part of that is the Fed's own doing, since it spent years discouraging borrowing. That layer is already improving. The political layer: in 2008 and 2009, borrowing from the Fed was treated as a bailout, and Dodd-Frank in 2010 required disclosure after two years. Kohn has sat in roundtables where bank treasurers were asked why they don't use the window, and the answer was — **my boss says that if he gets called before Congress to testify about it, I'm fired. Or if he gets a call from the Fed, I'm fired.** The Fed can fix the stigma it created. Convincing politicians that the window is not a bailout is much harder. Kohn was hauled before House and Senate committees over Fed lending in 2009 and 2010, so he is sensitive to it. **Nine. The composition of a task force is itself a conclusion.** Kohn admires the balance sheet task force. Karen Dynan chairs it, and the other two members took opposite positions at Jackson Hole: Stein (with Greenwood, 2016) argued a large balance sheet stabilizes the system by supplying liquidity so banks have less incentive to issue short-term runnable debt; Rajan argued more recently that the Fed supplying liquidity *encourages* banks to supply runnable liquidity to others. Kohn says honestly he does not know who is right, and suspects two things were tangled together in that period — QE, and zero rates. Silicon Valley Bank funded long-dated mortgage-backed securities with uninsured demand deposits; to Kohn that looks more like a consequence of zero rates and the expectation they would persist. The real failure, he thinks, was supervisory: nobody warned banks about interest rate risk. He gave exactly that speech at the FDIC in early 2010 — "and I was wrong, rates stayed at zero a very long time, but you still have to be careful." His reservation is about the productivity task force: three members from the AI industry, true believers. Jones is a fine economist and cautious about timing, but the group would benefit from someone more skeptical about whether and when. **"It feels like the finger was put on the scale on that one, but not on the others."** ## Going further ### "It all sounds reasonable — how do I know whether to believe it?" This is the honest problem with every earnings call and every research note. Everything said is plausible, the data is there, the logic flows. And then what? Do you believe it? Kohn's line — the narrative has to be falsifiable by incoming data or new analysis — is a usable ruler, and notice what it measures. Not whether a claim is right. Whether a claim **deserves to be tracked.** Take "transitory." What Powell offered was not just a conclusion but a chain: this is supply-side, it is a pandemic-specific dislocation, it will unwind as supply chains repair, therefore policy need not fight it. That chain turned out to be wrong. But note the consequence — **precisely because the "why" was stated, it could be reconciled against actual data months later, and it could be called wrong.** Run the ruler backward and it gets sharp. A claim that gives only a conclusion and no reasoning can never be wrong. "We remain optimistic about the long-term outlook." "Fundamentals remain solid." "This is short-term noise." No future data can overturn any of these. They sound safe because they say nothing. So before judging whether a piece of research is right, ask the earlier question: **does this claim specify what would make it wrong?** If nothing does, its value is emotional, not analytical. Point the same blade at yourself. If the reason you own that underwater position is "good company, it'll come back," you will never know when to concede, because the claim cannot be broken. Make it usable: I own it because it sits at a genuine bottleneck layer and customers cannot swap it out, so gross margin holds above a line — **and if margin breaks that line for two consecutive quarters, or a credible second supplier appears, the thesis is dead.** Now future-you has something to reconcile against. ### "Nothing bad was said — why did the market run anyway?" Warsh wants less Fed communication, and Kohn restates the two concerns fairly: that heavy Fed communication dampens the market's own ability to interpret incoming data, since everyone reads it through the Fed's frame; and that forward guidance, though not a commitment, may make the committee reluctant to change its narrative and therefore its policy. Kohn does not call these wrong — he calls them empirical questions, and cites Bill Nelson's finding that at least in 2010 through 2012, guidance did not damp market reactions to incoming data. But his counterargument is the interesting part: **markets react to new information regardless, and embedded in every reaction is their guess about how the Fed sees it.** Sit with that. Silence does not stop the guessing. It just downgrades the material people guess with — last time's behavior, headlines, the most fearful or most greedy available reading. **Silence is never neutral. Silence transfers authorship of the narrative to someone else.** The same thing happens at single-company level. The numbers are fine but the call only says "operations continue," so the market explains for itself why management won't elaborate. Good news lands without any statement of what it implies for next year's capacity allocation, so the market back-solves the meaning from price action — and usually gets it wrong. Which is why "good news, stock falls" is often not a case of fake good news. It is good news **that was never placed inside a narrative.** The issuer thinks it delivered information; it delivered a fact. What the receiver needed was where that fact sits on the map. So split the question when reading: what did this news change? And separately, did the issuer tell me what *they* think it changed? When the second answer is blank, most of the price move is filling in the blank rather than pricing the fact. ### Telling "I'm stepping back" apart from "I was wrong" When Volcker eased in late 1982, inflation was still four or five percent. In hindsight it was right: expectations had turned, the banking system had real fracture risk, and pushing on could have detonated a financial accident that reset the whole effort. In the moment, it looked nearly identical to giving up. What made the difference was that **the stated direction did not change, and every subsequent speech continued to point at it.** He retreated in pace and force, not in objective. Greenspan, taking over, did not immediately announce he would drive three or four percent down to one or two. He emphasized the price stability mandate, resisted every upward tendency, and it took most of a decade plus a productivity tailwind to actually get there. In position management this is the most commonly confused pair. Cutting exposure comes in two entirely different flavors. **Risk management**: the thesis is intact, but I cannot carry the volatility of the road ahead, or an unrelated systemic risk has appeared, so I size down to where I can sleep, and I will come back while the thesis holds. **Thesis retirement**: the thing I believed has been disproven, and this name should leave the list. Both look like "I sold some." What follows differs completely: the first preserves the watch and the re-entry condition; the second deletes the name and records which specific evidence killed it. The failure mode is disguising the second as the first — the thesis is broken but admitting it hurts, so you tell yourself you are merely stepping back, and a dead position keeps occupying a slot and your attention, waiting to get back to even. The defense is the ruler from the first section: **did you write down, in advance, what counts as thesis death?** If you did, you only have to check. If you didn't, you are left deciding in the moment, and the moment always prefers the less painful reading. ## Worth reading further - This episode, and Kohn's earlier appearance, which covers more of his career — including the after-work basketball games on the ninth floor in Kansas City and the elbow to a colleague's ribs - Kohn and Eggertsson, "The Inflation Surge of the 2020s: The Role of Monetary Policy" - Bernanke and Blanchard's decomposition, separating the early supply-driven inflation from the later demand-driven part - Greenwood and Stein's 2016 Jackson Hole paper on the financial stability case for a large balance sheet, read against Rajan's more recent contrary argument at the same conference - King and Kay, *Radical Uncertainty*, for why stating judgments and risks is more honest than producing a number - The late Robert Samuelson's book on the Great Inflation — published in 2008, timing he cheerfully described as terrible marketing ## The one thing to take away **A claim that cannot be proven wrong is not safer. It is less useful.** That single idea explains why Kohn admires Greenspan's refusal to forecast paired with his insistence on narrating, and why he still holds up Powell as a good example of accountability even though he thinks Powell was wrong. Falsifiability is not a strict standard. It is the threshold at which **a claim earns the right to be taken seriously.** One exercise for today, and it does not have to involve investing: **Take a judgment you are currently holding firmly, and write one sentence: "If this happens, I'll admit I was wrong."** It can be a view on a stock. It can equally be your assessment of a colleague, a worry about a child's habit, a read on a relationship, or your own conviction that you are "not cut out" for something. Pick the one you are most certain about — certainty and unwritten exit conditions travel together. One check while writing: **if you cannot name a single concrete, externally observable event that would change your mind, what you are holding is not a judgment. It is a position.** Positions are not wrong to have; people need them. But a position does not become more correct by being held longer, whereas a judgment becomes correctable the moment you write down what would break it. Put the sentence somewhere you will see it. You do not have to check the answer today. Time will check it for you.