The Money

On March 17, 2026, the Federal Open Market Committee convened in Washington to decide what to do about an economy that appeared to be pulling apart. Inflation remained above target for the fifth consecutive year. A new conflict in the Middle East had driven crude oil futures up fifty percent, pushing near-term expectations higher and raising the specter of a supply shock feeding into an already elevated price level. At the same time, the labor market showed signs of fragility — job gains were low, business contacts reported delaying hiring, and several participants warned that any further decline in labor demand could push unemployment sharply higher.

The Committee held the federal funds rate at 3-1/2 to 3-3/4 percent, with one dissent. The minutes reveal a meeting consumed by a question that appeared to have no good answer: which way should the rate move when inflation argues for tightening and employment argues for easing? But beneath that question lies a more fundamental one that the minutes do not ask: where is the money coming from? What mechanism is sustaining the price level above target for the fifth consecutive year? The Federal Reserve's own documentary record — eighty-nine years of internal deliberation — provides an answer the Committee has discovered three times and never retained.

The Bind

The case for cutting is not frivolous. The lone dissenter argued that the current stance "remained restrictive and was contributing to weak labor demand and elevated downside risks to the labor market." The logic is direct: if the rate is above neutral, it is suppressing hiring. Job gains are concentrated in a few sectors. Business contacts report caution. In a low-hiring environment, the margin between balance and deterioration is thin. Many participants cautioned that "labor market conditions appeared vulnerable to adverse shocks" and that a further fall in demand "could push the unemployment rate sharply higher." A protracted conflict in the Middle East would compound this by tightening financial conditions, reducing purchasing power, and weakening growth abroad. The employment side of the mandate is exposed.

The case for holding — and for acknowledging the possibility of hikes — is equally direct. Core PCE inflation ran at 3.1 percent in January, a quarter-point higher than a year earlier. Near-term inflation expectations had risen sharply with oil prices. After five years above target, the vast majority of participants judged that the "risk of inflation running persistently above the Committee's objective had increased." Some participants highlighted the danger that "longer-term inflation expectations could become more sensitive to energy price increases" after prolonged above-target readings — the expectations-unanchoring dynamic that turned the 1970s oil shocks from transitory disruptions into a decade of entrenched inflation. Several argued explicitly that "rate increases could be appropriate if inflation were to remain at above-target levels." The price stability side of the mandate is exposed.

Both arguments are internally consistent. Both identify real risks. And they cancel each other out. A rate cut eases financial conditions for rate-sensitive sectors — housing, consumer durables, small business borrowing — and supports employment, but it signals tolerance of above-target inflation at precisely the moment when expectations are most vulnerable. A rate hike restrains inflation pressure and defends the Committee's credibility, but it tightens credit for the same sectors that are already fragile and accelerates the labor market deterioration that the majority identified as a downside risk. The Committee recognized this paralysis explicitly: the vast majority judged that "upside risks to inflation and downside risks to employment were elevated" simultaneously. Moving in either direction means accepting damage to the other side of the mandate.

The question the minutes do not ask is whether the instrument itself is adequate — whether the federal funds rate is the tool that reaches the inflation the Committee is trying to fight.

Where Money Comes From

We went into the archives.

The question put to the FOMC Insight Engine was not about interest rates or the dual mandate. It was more fundamental: does the Federal Reserve understand how money is created in the economy? The Engine searched 352,965 indexed passages spanning 1936 to 2025 — transcripts, staff memoranda, Greenbooks, Tealbooks, and public statements — and returned evidence that the answer has been known inside the institution for over sixty years.

As early as 1960, a staff memorandum to the Board described money creation not as something the central bank initiates, but as something it accommodates.

Two years later, James L. Knipe of the Board staff wrote directly to Chairman William McChesney Martin explaining the mechanism. Banks, he wrote, grow by "constantly adding to their loan and investment accounts, thereby creating new deposits." The loan creates the deposit. The deposit does not precede the loan. Reserves do not "multiply" into credit through a textbook ratio. The banking system creates money by lending, and the Federal Reserve supplies whatever reserves are needed to support that process after the fact. This was the credit channel of money creation — understood inside the institution before the academic economists who are often credited with the insight had published their landmark papers. By 1965, the staff had drawn the operational consequence: so long as the Fed prioritized interest rate stability as its operating guide, it could not simultaneously control the quantity of credit flows. The institution saw the choice — govern the price of money or govern the quantity — and chose the price. That choice, made sixty-one years before the March 2026 meeting, is still governing the Committee's deliberations today.

But the archive reveals a second channel that is, for the current moment, more consequential. In August 1961, with private credit demand weak during the recovery from recession, the Committee turned to an alternative engine for expanding the money supply.

The mechanism is precise. The government runs a deficit — it spends more than it collects in taxes. The difference is financed by issuing debt. When that debt is intermediated through the banking system — when banks buy it and the government spends the proceeds — deposits flow into the accounts of households and businesses as payroll, procurement, and transfer payments. The money supply has expanded not because the central bank pushed reserves into the system, but because the government created a financial liability and the banking system converted it into deposits.

The Committee in 1961 understood this as a practical tool: when the private sector will not create money by taking out loans, the government creates it by running deficits through banks. But there is a deeper reason the two channels are not simply additive. Banks have finite balance sheet capacity. When private credit demand is strong, banks must fund new loans — and one source of funding is selling existing securities. The Greenbook tables from 1994 document this concretely: as private lending rebounded, banks began liquidating Treasury securities to make room for business and consumer loans. The deposit created by the new loan was offset by the balance sheet adjustment required to fund it. When banks sell Treasuries to make loans, the fiscal channel's contribution to money creation shrinks even as the credit channel grows. The two channels compete for the same balance sheet space.

There is also an asymmetry between them. A private loan has a maturity — it is repaid. When a household pays down a mortgage or a firm retires a bond, the deposit and the loan extinguish simultaneously. The stock of private credit constantly decays through repayment, and the deposits it created decay with it. Government debt behaves differently. When a Treasury matures, it is refinanced — rolled into new issuance. Rollover does not create new deposits, but it preserves the debt stock, and the deposits generated by the original deficit spending that issued the bond are not destroyed by the refinancing. New deficit spending then layers additional debt — and additional deposits — on top of a stock that never shrinks through repayment. This is why the fiscal channel is the more persistent engine of money creation: private-credit-backed deposits are constantly being extinguished, while deficit-backed deposits accumulate as each year's new borrowing adds to a base that does not erode.

The offset operates even before repayment. When one firm borrows and pays a supplier, that supplier may use the proceeds to retire its own debt. The deposit created by the first loan is extinguished by the second firm's repayment. At the level of the individual transaction, the loan created a deposit. At the level of the private sector as a whole, the net deposit creation can be far smaller than gross lending — because one borrower's spending is another borrower's debt service. Government spending does not face this offset. When the Treasury pays a contractor or mails a transfer, the recipient does not use the deposit to repay the government. The money stays in the private sector. Each dollar of deficit spending is net new purchasing power in a way that each dollar of private lending is not.

The archive shows that the Fed recognized this asymmetry, even if it never quantified it. In February 1992, during a private-sector deleveraging cycle in which the Federal Reserve found itself unable to generate money growth despite providing ample reserves, Governor Wayne Angell named the structural constraint that connects the two channels.

Credit cannot grow faster than debt grows. Every deposit requires a corresponding asset on the other side of the ledger — a loan, a bond, a security. When the private sector will not supply that asset by borrowing, the government must supply it by running deficits. This is what Angell called the doctrine that central bankers never want to admit. The archive shows why they resist it: if the money supply is determined by the composition of borrowing in the economy — how much is private credit, how much is government debt — then the instrument that governs inflation is not the overnight rate. It is whatever controls the volume and composition of that borrowing.

The Forgetting

An institution that understood by 1961 that money is created through bank lending and government borrowing, and that understood by 1992 that these channels substitute for each other, would be expected to formalize that understanding into a model — one that connects the composition of debt to the trajectory of the money supply, and tells the Committee which instrument is appropriate at any given moment. The archive reveals that this model was never built. Not because the data were unavailable, but because the staff explicitly chose not to build it.

The "fractions" in question were the share of bank assets allocated to Treasury securities versus private loans — precisely the quantities that determine which channel of money creation is dominant, and whose rotation between Treasuries and loans the 1994 Greenbook would later document in real time. The staff tracked these fractions with extraordinary precision, quarter by quarter. They could see banks shifting from one asset class to the other as the private sector deleveraged in 1989 and then re-leveraged in 1994. Donald Kohn, then Director of the Division of Monetary Affairs, correctly identified the hand-off: federal borrowing was moderating while "private debt growth will be picking up." The balance sheet rotation that governs which channel dominates money creation was observed, documented, and left unmodeled.

Then the 2008 financial crisis arrived, and the same question returned. The Federal Reserve tripled the monetary base through quantitative easing. Broad money growth remained modest. Bank lending was weak. The textbook prediction — that a massive increase in reserves would produce a proportional expansion in deposits and credit — failed. And Vice Chairman Donald Kohn, the same official who had understood the mechanism fourteen years earlier, sat in the December 2008 FOMC meeting and said this:

The understanding he held as a division director did not survive his elevation to Vice Chairman. Not because his intellect deteriorated — but because the analytical framework had changed beneath him. In the 1980s and 1990s, the economics profession migrated from models that contained money — monetarist frameworks, IS-LM with an LM curve — to New Keynesian models that do not. The workhorse models the Fed relies on today, including FRB/US and its DSGE variants, have an interest rate rule and a Phillips curve. They do not have a money supply. Money, in these frameworks, is determined residually by money demand given the interest rate the central bank sets. There is no channel through which deficit spending flows through bank balance sheets into deposits that sustain the price level — because the variable that would carry that channel does not exist in the model. When the profession removed money from its analytical framework, the institutional understanding of money creation lost its vehicle. It could survive in the judgment of individuals who had operated the plumbing directly — Kohn in 1994, Angell in 1992 — but it could not be transmitted through the models to the next generation of policymakers. (As The Nominee documented, when asked why the Fed persisted with models that repeatedly failed to predict inflation, staff conceded they were "not aware of a model that fits the data as well" — a confession not of the model's strength but of the profession's inability to replace it.)

The result was visible even before the 2008 crisis. In November 1994, Michael Prell, Director of Research and Statistics — the official responsible for the Greenbook — described the staff's working practice in terms that confirmed the disappearance.

An endogenous byproduct. The institution's most senior forecaster acknowledged that money was not an input to the staff's analysis — it was an output, something that happened as a consequence of other variables the models did track. The staff understood, in practice, that money was endogenous. But because the models did not contain the channel through which money was created — through borrowing, on bank balance sheets, from two competing sources — the understanding remained operational intuition rather than analytical structure. By 2012, John C. Williams, then President of the San Francisco Fed, acknowledged that the "money multiplier has broken down" and that the failure was "not yet addressed in our textbook models of money supply." By 2015, he conceded that research incorporating these effects remained "still fairly rudimentary."

Three crises — the Volcker money-targeting failure of 1979–1982, the private-sector deleveraging of 1989–1993, and the quantitative easing disconnect of 2008–2014 — forced the Federal Reserve to confront the same structural reality. Each time, the institution documented the finding internally. Each time, it declined to formalize the insight into a model that would survive the crisis and inform future decisions. The understanding was episodic, not cumulative. It was rediscovered in each emergency and abandoned between them. (This pattern — an institution that arrives at the correct diagnosis during a crisis and then discards it when the crisis passes — echoes what The Transmission documented in the Fed's understanding of how monetary policy reaches the real economy: the channels are mapped under pressure and unmapped when the pressure lifts.)

The Ledger

The pattern the archive documents is not only historical. It is operating now. Debt held by the public has grown from roughly 79 percent of GDP before the pandemic to nearly 100 percent — an expansion driven by sustained deficits that were financed substantially through the banking system. During 2020–2021, the Federal Reserve purchased trillions in Treasuries directly; since then, those securities have been slowly returned to private balance sheets through quantitative tightening. M2 surged by more than seven trillion dollars in two years, contracted briefly in 2022 and 2023 — the first sustained M2 decline since the early 1960s — and then stabilized at a level well above its pre-pandemic trend. The trajectory is more consistent with fiscal-channel money creation than with private-credit-led expansion: the deposits appeared when the government borrowed and spent, not when the private sector leveraged up. The Fed did not initiate the expansion. It accommodated it — precisely as the 1960 staff memorandum described.

Private credit has grown, but not in the way that would offset the fiscal channel. Bank lending standards tightened sharply after the failure of Silicon Valley Bank in 2023, and borrowing costs at a federal funds rate above 3.5 percent remain elevated for rate-sensitive borrowers — households seeking mortgages, small businesses seeking working capital, commercial real estate firms seeking refinancing. Both bank holdings of Treasury securities and total loans have been rising in 2026, but that coexistence is itself significant: banks are expanding total balance sheet capacity rather than rotating from one asset class to the other. The substitution mechanism the Greenbook tables documented in 1989 and 1994 — banks selling government securities to make room for private loans, partially offsetting the fiscal channel — is not operating. Credit growth has not been strong enough to force the rotation. The two channels are running in parallel, with the fiscal channel contributing substantially to deposit creation while private credit adds to it rather than displacing it.

This creates the trap that the March minutes reveal without naming. The funds rate is simultaneously too high for private credit and insufficient for the fiscal channel. Too high for credit: the rate suppresses the private borrowing that would, through the balance sheet rotation, pull banks away from Treasury holdings and partially offset the fiscal channel's deposit creation. Insufficient for the fiscal channel: the deficit does not respond to the overnight rate, so the fiscal impulse continues to create money regardless of where the Committee sets its target range. Raising the rate further would tighten private credit even more — widening the gap between the two channels and making the fiscal channel more dominant, not less. Cutting the rate would ease private credit conditions but would also signal tolerance of above-target inflation, risking the expectations unanchoring that participants identified as a salient concern. The instrument cannot address both channels. It can only choose which one to damage.

The Word That Does Not Appear

Return now to the March 2026 minutes with this history and this present-tense evidence in hand.

The Committee discusses inflation at length. The sources are enumerated carefully: tariff pass-through into core goods prices, the surge in oil prices from the Middle East conflict, still-elevated nonhousing services inflation, the risk that near-term expectations could drift upward after years of above-target readings. Participants note that progress toward the 2 percent objective "could be slower than previously expected" and that the risk of persistent inflation "had increased." The analysis is thorough. The discussion is serious. The risks are real.

The word "deficit" does not appear in the participants' discussion.

The federal government's borrowing — the channel that the Committee itself identified in 1961 as the "most eligible medium" for expanding the money supply, the channel that Governor Angell in 1992 called the inescapable backstop when private credit weakens — is absent from the analysis of the inflation the Committee is trying to contain. The minutes discuss the effects of tariffs on prices, a relative price adjustment that shifts the composition of the price index without necessarily expanding the money supply. They discuss the effects of oil on energy costs, a supply shock that raises input prices but does not explain why the monetary system can sustain a price level above target for five consecutive years. They discuss the risk that inflation expectations could unanchor, a psychological phenomenon that is itself a consequence of persistent money creation, not a cause of it. They do not discuss the mechanism by which money is being created in the economy at a rate sufficient to keep the price level elevated.

This absence is not an oversight. The institutional architecture makes it structural. Acknowledging that the inflation is substantially sustained by fiscal deficits intermediated through the banking system would mean acknowledging that the federal funds rate — the Committee's primary instrument, the tool around which the entire meeting is organized — is insufficient for the problem. The rate still affects borrowing costs, asset prices, and financial conditions. It is not irrelevant. But a rate hike does not reduce the deficit. A rate cut does not reduce the deficit. The deficit creates deposits through the banking system regardless of where the overnight rate sits, because the Treasury will issue debt, intermediaries will purchase it, and the resulting deposits will flow into the economy as government spending. The rate can modulate the edges of this process. It cannot govern it.

But the Committee does control one instrument that enters the relevant channel: the balance sheet. The distinction is operational, and it requires precision. When the Treasury issues new debt to finance a deficit and banks buy it, the Treasury *spends* the proceeds — deposits flow into the economy as payroll, procurement, and transfer payments. That is money creation. When the Federal Reserve shrinks its balance sheet and existing Treasuries move from the Fed's holdings to the private sector, something different happens: the buyers pay with existing deposits. Those deposits are extinguished. The Treasury does not receive new spending power — it is refinancing the same debt with a different holder. The total government debt is unchanged. Only the composition of who holds it changes. But the deposit base contracts, because the private sector exchanged deposits for securities.

This is how the balance sheet constrains the fiscal channel without requiring a reduction in the deficit itself. The deficit creates deposits through new issuance. Balance sheet tightening destroys deposits through refinancing. The two operations are not contradictory — they are the opposing forces that determine, at any given moment, how much of the government's borrowing translates into money in the economy. When the Fed holds more Treasuries, more of the deficit's deposit creation survives into the broader money supply. When it holds fewer, more is reabsorbed. The balance sheet operates on the same mechanism the staff documented in the 1989–1993 Greenbook tables — bank asset composition and its relationship to deposit growth — and it is the only instrument the Committee controls that reaches this channel. The funds rate affects borrowing costs, asset prices, and financial conditions broadly, but it does not alter the fiscal-monetary transmission that sustains money creation when the deficit is the dominant source.

The balance sheet is not, however, a costless instrument. Returning Treasuries to the private sector means that banks, dealers, money market funds, and other investors must absorb them — and the exact transmission depends on who the marginal buyer is. When banks absorb the securities, the balance sheet capacity consumed is capacity that cannot fund private lending. This is the same rotation the archive described in reverse: in 1994, banks sold Treasuries to make room for loans; balance sheet tightening pushes securities back onto bank balance sheets, reducing room for loans. The credit channel, already growing too slowly to force the substitution, would face additional pressure. But the nature of that pressure differs from what a rate hike inflicts. A rate hike suppresses private credit directly — through borrowing costs — while leaving the fiscal channel untouched. Balance sheet tightening reaches the fiscal channel directly and affects credit only indirectly, through the composition of private-sector assets. The cost to credit is real but secondary; the benefit to the fiscal channel is something the rate cannot achieve at all. The choice is not between a perfect instrument and an imperfect one. It is between an instrument that addresses the dominant channel at some cost to the secondary channel, and an instrument that damages only the secondary channel while the dominant one continues unchecked.

The March minutes reveal that the Federal Reserve is currently expanding its balance sheet through reserve management purchases — acquiring Treasuries from the private sector and replacing them with reserves. On the framework the archive itself describes, this is accommodative: each purchase absorbs a Treasury that would otherwise sit on a bank balance sheet, supporting the deposit-creation process rather than constraining it. The manager noted that after April, "the monthly pace of reserve management purchases was likely to be reduced significantly." That reduction is a step in the right direction — but as long as the balance sheet is still growing, the Fed is still absorbing Treasuries that the framework says should remain in private hands. The Committee devoted the meeting to debating whether the funds rate should move up or down. It did not discuss whether the balance sheet should stop expanding and begin to contract.

The Instrument Already Available