Case: March 2020, and the Path to 3.75–4.00%

March 2020: the run on the market itself

The chain was the crisis one, one link further in. The shock was exogenous and arrived at the same time everywhere, so the demand for cash rose sharply and simultaneously. Funds and dealers sold the most liquid assets to raise it, which meant Treasuries — and because Treasury dealers were themselves balance-sheet constrained and the buyers of last resort were absent, the deepest market in the world stopped clearing. That is why yields rose on the day after the central bank cut to zero: the price of a bond is set by the balance sheet willing to intermediate it, and there was no balance sheet. The response was instruments that act on the market rather than on the rate: purchases of Treasury and agency securities at scale, facilities to support commercial paper and the repo market, and lending against a wider range of collateral. Each corresponds to a link in the transmission chain, exactly as in 2008 — and the speed with which they were announced is itself the signal about how serious the authorities judged the funding failure to be. The lesson is the same one, restated: when the collateral stops working, the instrument is a purchase rather than a cut. What followed was also the same: fiscal transfers at an unprecedented scale, which the fiscal lesson supplies the arithmetic for — a deficit that raised the debt ratio and an inflation episode that followed, partly from the demand support, partly from the supply constraints, and partly from the energy shock that arrived in 2022. The macro path from there to the middle of the decade is the subject of the rest of the lesson. What fell, and why it was the collateral that mattered — Equities, credit, commodities: sold to raise cash, whatever the view on the asset · Long Treasuries: sold for the same reason — the hedge failed because the motive was cash ← · Cash and the shortest bills: the only thing that rose: the thing being demanded · The response: purchases and facilities, acting on the market rather than the rate A hedge is a claim about a motive as much as about a correlation. Treasuries hedge equities when the shock is a growth shock and both are being valued; they do not hedge equities when the motive is cash and both are being sold. That distinction is the same one the 2022 case made about inflation, and it is why a single correlation number cannot describe a portfolio.

The disinflation of 2022–2026, and what it cost

Inflation peaked around 9% in mid-2022 and was down to a core rate of 2.9% by the autumn of 2026, with the policy rate at a 3.75–4.00% range after the rise of 16 September. Two facts about that path matter for the framework. The first is that the disinflation cost far less in unemployment than the historical relationship between inflation and slack would have predicted. The second is that the falling inflation was accompanied by long-run expectations that never moved much — the anchor held, which is exactly the condition the expectations lesson identified as making a disinflation cheap. Why did the anchor hold when the shock was large? The plausible account has three parts. The central bank raised rates quickly and said clearly what it was doing and why, so the framework was not in doubt even when the forecasts were wrong. The labour market was tight but wage growth never fully incorporated the energy shock, so the wage-price spiral the framework warns about did not get going. And supply constraints eventually eased, which removed the source of the shock without requiring demand to fall much — the same supply reversal that made the disinflation unusually cheap. The residue is the configuration at the end of the period: a positive real policy rate around one percent, an unemployment rate near its longer-run level, core inflation at 2.9% and still above target, and a yield curve whose front end has come down from the peak. That is not a normalisation to the pre-pandemic world; it is a higher-rate equilibrium with an intact anchor, and the questions it leaves are the ones this subject has been building the tools to answer: what the neutral rate is now, how much restrictive policy is being delivered through conditions rather than the rate, and what the term premium is charging for the supply of debt. • Expectations held through the largest inflation shock in four decades, which is why the cost was low. • Wage growth never fully incorporated the energy shock, so no spiral formed. • Supply eased, removing the shock without requiring much demand destruction. • The residue is a positive real rate, an intact anchor and inflation still slightly above target. The two great disinflations make an instructive pair: the early 1980s, where the anchor had to be rebuilt with a deep recession, and the mid-2020s, where it never broke. The difference is the state of credibility at the start, which is why the expectations series is worth more than any single inflation print.

What the pair of episodes says about the framework

The two cases bracket the range of what a funding shock and an inflation shock can do. March 2020 shows the fastest possible route from a shock to a policy response, with the transmission channel operating through the collateral market and the instrument being a purchase rather than a rate. The disinflation shows the slowest: three years from peak to something near target, with the cost decided by whether expectations moved rather than by how large the initial shock was. Both are consistent with the same framework, which is the point of studying them together — the channel is stable, the instrument changes with the diagnosis, and the outcome depends on the state of the anchor. For a portfolio process, three conclusions follow. Diversification has to be assessed against the shock that arrives, not against an average: in March 2020 only cash worked, and in 2022 only real assets worked, and a portfolio built for one of those states was wrong in the other. The instrument the authorities choose is a real-time diagnosis that costs nothing to read. And the state of inflation expectations is the variable that decides how expensive the next few years will be, which makes it a monthly reading rather than an academic interest. The last conclusion is the one to carry forward. The framework built across this subject — growth and inflation, the reaction function, expectations, transmission and the credit channel — is not a forecasting machine. It is a way of describing what a policy action is aimed at, what it can and cannot reach, and which state of the world would make the answer wrong. That is the deliverable, and the capstone asks for it in writing. • Assess diversification against the shock that arrives, not against an average correlation. • The instrument a central bank chooses is a free diagnosis of the problem it sees. • Inflation expectations are the variable that prices the cost of the next episode. • The framework describes mechanisms and wrong-states, not forecasts. The two episodes also explain why the same policy rate can be described as restrictive by one participant and neutral by another: the answer depends on the neutral rate, which is unobservable, and on the conditions being delivered through spreads and the currency rather than through the rate alone.

The basis trade and the balance-sheet constraint behind the dash for cash

The previous read established that the Treasury market stopped clearing in March 2020 because the balance sheets willing to intermediate it were absent. Why they were absent is the most useful piece of plumbing in the whole episode, and it has two halves. The first half is **dealers**, who are the market’s intermediaries: they buy from sellers and hold the paper until a buyer appears, funding the inventory in the repo market. After the post-2008 reforms, a dealer’s capacity to do that is constrained by capital and leverage rules that charge for balance-sheet usage, which means the cost of warehousing a billion dollars of Treasuries rises exactly when volatility rises and the paper is most plentiful. Intermediation is therefore pro-cyclical by construction: the capacity to absorb selling shrinks at the moment when selling is largest. The second half is the **hedge funds** on the other side of a very large, very quiet position — the cash-futures basis trade. The basis trade works like this. A Treasury futures contract and the cheapest-to-deliver bond underlying it are the same risk, but they trade at slightly different prices, with futures usually the richer. A fund sells the futures and buys the bond, financed in the repo market, and collects a very small, very reliable difference on a very large notional — a trade that might earn a few basis points of return on a position levered fifty times. It is not a market-direction bet; it is a liquidity-and-funding bet, and it depends on two assumptions holding at once: that the repo funding stays available at a stable rate, and that no one forces the position to be unwound early. In March 2020 both assumptions broke together. Repo funding for the specific collateral tightened sharply as everyone wanted cash, so the cost of carrying the bond rose just as the futures-basis widened against the fund, and the fund had to sell the bond into a market whose dealers were already constrained. The forced selling of the most liquid asset in the world is what turned the dash for cash into a Treasury-market failure, and it is the same mechanism — leverage plus a funding shock plus a constrained intermediary — that the subject has now met in the 1998 case, in 2008, and again in the 2022 gilt episode. Two consequences follow for reading the present. The first is that **reserve balances and repo rates are a first-order part of the macro picture**, not plumbing to be skipped: when the quantity of reserves in the system falls far enough that the money market stops being perfectly elastic, the policy rate becomes sensitive to ordinary swings in the Treasury’s cash balance and in settlement flows, and the central bank has to intervene with facilities at the margin. That is why the spread between the effective overnight rate and the administered rate is monitored, and why a central bank announcement about the pace of balance-sheet runoff can move funding markets more than a data surprise. The second is that the official response to the 2020 episode was designed around this diagnosis rather than around the level of the policy rate: purchases at scale put a balance sheet back on the other side of the market, the facilities made repo funding available against a wider collateral set, and the standing facility that now operates as a permanent backstop exists because a market that depends on private balance-sheet capacity will periodically lack it. The transferable lesson is the one at the top of this lesson, in a plumbing key: when the question is who will hold the risk, the answer depends on who has the capacity, and capacity is a regulated quantity that moves the wrong way in a crisis. The links in the chain, in order — Dealer balance-sheet capacity: Capital and leverage rules make intermediation pro-cyclical — capacity falls in stress · The basis trade: Short futures, long the bond, repo-funded; a funding bet levered many times over ← · The funding break: Repo for the collateral tightens, the basis moves against the fund, the position unwinds ← · The forced sale: Selling the world’s most liquid asset into the least willing market The linkage to the policy lesson is direct: the instrument that works when the problem is the market’s balance-sheet capacity is a purchase or a facility rather than a rate cut, and the announcement of that instrument is the clearest available signal that the authorities have diagnosed a funding failure rather than a demand problem.

Sources

• The March 2020 dash for cash and the Federal Reserve’s facilitiesFederal Reserve documentation and published facility terms, March 2020; contemporaneous market evidence
• The 2022–2026 disinflation and the anchoring of expectationsFOMC statements and projections through September 2026; the 16 September 2026 decision
• The sacrifice ratio and the cost of disinflation under differing credibilityMR7 on expectations; standard Phillips-curve literature