Inside Room 101: A Misbehaving Yield Curve

"Doublethink means the power of holding two contradictory beliefs in one's mind simultaneously."

George Orwell, Nineteen Eighty-Four

One assumption investors stop questioning, stress-tested.

01The assumption

The Federal Reserve controls interest rates. When the Fed turns dovish, borrowing costs fall; when it turns hawkish, they rise.

And beneath it, the assumption that really matters: whatever the debate about what the Fed should do, the market's obedience to what it does is taken as given.

02Why it's believed

Because for forty years it was true — and because part of it is true by construction.

The Fed sets the overnight rate directly; that control is mechanical, and near enough absolute — the repo market's brief escape in September 2019 being the exception that proves it.

From 1982 to 2021, every easing cycle transmitted down the curve: the Fed cut, long yields fell, mortgages cheapened, the real economy got its boost.

"Don't fight the Fed" may be the most obeyed maxim in markets. A decade of quantitative easing extended the impression of control across the entire curve, and the Fed's own communication — projections, guidance, the dot plot — reinforces the idea that policy intention and the rate path are the same thing.

03The stress test — hold both beliefs up to the light

So let's separate what the Fed sets from what the market sets. In the absence of yield curve control, that is.

The Fed sets one rate: the price of money overnight. The prices critical to the economy — mortgages, corporate debt, the discount rate under equity valuations — are set further out the curve. Market prices.

A long yield is the expected path of policy plus a term premium: the risk of government largesse and inflation, boiled down to a number.

The Fed's control of the long end is not a power. It is a delegation.

Central banks possess enormous power, but much of that power derives from the collective belief that they possess enormous power. And there is the doublethink: the market holds both ideas at once — that the Fed controls interest rates, and a term premium that prices the chance it doesn't. Both, simultaneously, accepted. Orwell would have recognised the arrangement.

The long bond's independent streak is not new — we have examined it before. The question here is what happens when that independence meets an easing cycle.

In September 2024 the Fed opened an easing cycle with a half-point cut. Over the following months the 10-year yield rose by roughly a full percentage point. Mortgage rates went up after the cut, not down.

To a holder of thirty-year paper, the equation no longer added up. Fiscal deficits and rising inflation expectations drove long bond holders to demand a wider margin of safety. Negative rates increasingly look, in hindsight, like a group hallucination. Payback time. The cut lowered the rate the Fed sets and raised the rates that matter.

The 2 July jobs report was squishy — and came with downward revisions to the two before it. The Fed had turned dove in an instant, or so the market concluded: the probability of a September hike roughly halved within the session. And the 10-year yield moved two basis points.

The long end just sat there. Nothing doing.

One session proves nothing on its own — but it is precisely the experiment worth running again and again, because the shape of the response is the whole question.

04The threshold condition

The assumption fails at an identifiable point: when the term premium, rather than the expected policy path, becomes the marginal driver of long yields. Before that point, dovish news rallies the long end and the Fed's intentions transmit. Beyond it, easing steepens the curve — the front end falls on command while the long end charges more for the inflation and issuance that easing implies. The Fed keeps the steering wheel; the wheels stop being connected to it.

The test is observable and repeats on every data release. Call it the ratification test: when the market reprices the Fed dovishly, does the long end ratify the move by rallying — or does it abstain? September 2024 was an abstention that became a veto. The 2nd of July was a small abstention.

The assumption's remaining life is measured in abstentions. At some point in the accumulation, the two beliefs stop being compatible — and the market has to drop one of them.

05What to watch

The ratification test, every time. On each dovish surprise — soft data, dovish Fed communication — watch the 10- and 30-year response, not the front end. Flat or rising long yields on dovish news is the assumption failing in real time, one observation at a time.

The decomposition. Term-premium estimates (the Fed publishes them) separate long yields into expected policy and compensation demanded. A rising premium share means the marginal price-setter is the duration holder, not the policymaker.

Auction behaviour and curve shape. Demand at 10- and 30-year auctions — tails, dealer take-downs — measures the appetite for supply directly. And the curve's response to actual policy moves: a curve that steepens when policy eases is a market declining to transmit.

06One more thing

None of this says the Fed is powerless. Its control of the front end is real, and that matters. The point is narrower and more useful: the rates that price houses, corporate balance sheets and equity duration are set at the long end — and the long end answers to inflation credibility and supply, not to intention.

The 2024 episode showed what a veto looks like. The current tape is where you find out whether it was an exception or a precedent.

Room 101, Orwell wrote, holds the worst thing in the world. For a central bank, that is a bond market that has stopped listening.

In Orwell's world, the truth is whatever the Party says it is. In markets, the truth is what the market says it is — and the long end has started saying it.