There is a conversation I have had four or five times this year, in Ciudad de México and in Helsinki, and it goes the same way each time. A company has debt maturing inside twenty-four months. The refinancing assumption in the model is built off an expected path for the central bank policy rate. And the policy rate is not the variable that is going to price that refinancing.

September made the point unusually cleanly, so it is worth walking through.

What the Fed did, and what it declined to do

On 16 September 2026 the FOMC raised the federal funds target range by 25 basis points to 3.75–4.00%, unanimously, 12–0. It is the first increase since 2023. The median dot in the accompanying projections sits at 4.1% for the end of 2026, which implies one more move of the same size this year.

Warsh described the action as removing a dose of accommodation, said that inflation is too high and has been for too long, and framed the hike as supporting a timelier return to the Committee’s 2% goal — a goal defined in terms of PCE inflation.

What he did not do is tell anyone what comes next. He confirmed he took no part in the Summary of Economic Projections, describing the dot plot as the views of his colleagues. That is consistent with what he had already said at Jackson Hole on 28 August 2026, where he declined to restore forward guidance on the grounds that in normal times it “risks creating ambiguity in the name of clarity”, and warned against what he called the hall-of-mirrors problem: “we should not indulge a regime in which market participants are looking primarily to the Fed for their next trade.”

So the position is: the rate is moving, the direction is up, and the path is not being published. For a treasurer, that is not a communications preference. It is a modeling instruction.

The measurement problem underneath it

At Jackson Hole, Warsh said of the better summer inflation readings that “they do not tell me that underlying trends have meaningfully improved”, and assessed policy settings bluntly: “I would be hard pressed to describe broad financial conditions as restrictive.” He pointed to credit spreads near historical lows, strong issuance volumes and easier lending standards.

The arithmetic supports him, and it also shows why the answer depends on which gauge you pick.

Core PCE ran at 3.3% year on year in the July 2026 reading. Core CPI came in at 2.76% for August 2026, with headline CPI at 3.71%. Against a policy midpoint of 3.875%, the implied real policy rate is roughly +0.6% measured against core PCE and roughly +1.1% measured against core CPI.

Those are different vintages of data and the comparison is rough, but the spread between the two gauges is the point rather than the artifact. Same central bank, same month, two defensible numbers half a point apart — and the 2% objective is defined against PCE, which is the higher of the two.

The long end did not wait

Here is what actually happened to the cost of long money while the policy debate ran.

Instrument	Level	Note
US 30-year Treasury	5.29% (17 Sep 2026)	10-year at 4.94% the same day
German 10-year Bund	3.57% (15 Sep 2026)	Highest since June 2009
Japanese 10-year JGB	Above 3.00% (15 Sep 2026)	First time since 1996
French 10-year OAT	Above the Italian BTP	Crossover first occurred September 2025

This is not a European story, or a Japanese story, or a US fiscal story. It is the same move happening in four sovereign markets with four different central banks, four different fiscal positions and four different political situations. When that happens, the explanation is usually not local. It is term premium — the compensation investors demand for holding duration — repricing globally.

The French and Italian crossover is the detail I would not skip. A French government bond yielding more than an Italian one is a hierarchy inversion that would have read as a typo for most of the last twenty years. It happened in September 2025 and it has not reversed. Sovereign risk hierarchies are being rewritten, not just re-rated.

Why this is a supply story, not a distress story

It matters which one it is, because they call for opposite responses.

If corporate credit were cracking, you would see it in credit spreads, and you do not. Warsh’s own description — credit spreads near historical lows, strong issuance, easing lending standards — is the opposite of funding stress. Growth is not cracking either: the ISM manufacturing index read 54.6 in August 2026, an eighth consecutive month of expansion.

What has been happening alongside is an energy shock. Crude prices rose sharply through late August and September 2026 — the Mexican export blend, to take the series I watch most closely, closed at USD 94.3 per barrel on 8 September and USD 104.78 on 11 September. Sticky inflation with firm growth and rising energy costs is not stagflation. It is a set of conditions in which long-dated nominal debt is simply worth less, and investors are repricing it accordingly.

What this means if you sign off on capital allocation

Three consequences, in the order they usually bite.

Your hurdle rate is probably anchored to the wrong instrument.If the discount rate in your investment cases was derived from a policy-rate path plus a spread, it is tracking a variable the Fed Chair has explicitly said he will not publish, and ignoring the one that has actually moved. Rebuild it off the relevant sovereign long end plus your actual observed credit spread.

Refinancing at the long end is now a materially different transaction than refinancing at the short end.That is a real choice with a real price, and in most models it is not a choice at all — it is a single blended assumption. Splitting it is usually a morning’s work and it frequently changes which projects clear.

The currency of your debt is now a bigger decision than it was.The bunds, JGBs and Treasuries moved together but not by the same amount, and a Mexican company with euro-denominated obligations or a Nordic company funding a Mexican subsidiary is carrying a basis between two curves that have both moved. That is a quantifiable exposure, and it is quantifiable before it becomes a loss.

The question I would ask your finance team this quarter

Not “what do we think rates will do” — nobody knows, and the central bank has said out loud that it is not going to help you guess.

Ask instead: which curve are we actually pricing against, and when did we last stress it? If the answer involves a policy-rate forecast, the model is measuring the wrong thing. If the answer is the long end but the last stress test predates 2026, the scenario range is too narrow — the moves above are larger than most stress cases written two years ago contemplated.

Capital allocation decisions taken on a stale cost of capital do not fail loudly. They fail as a portfolio of projects that all looked marginally positive and collectively did not earn their funding. That shows up three years later, in the return on capital employed, long after the assumption that caused it has been forgotten.

On the related question of what actually determines where large, long-horizon capital lands, I have written about the Finnish case in Why did Google choose Finland for €13bn of AI capacity?. On the trade framework side, USMCA wasn’t renewed — and why Europe now matters more covers the other variable that moved this year.

Where I fit

I am KHT-qualified in Finland — exam passed in 2005, membership currently inactive after more than twenty years abroad — with Big 4 training at Ernst & Young in Helsinki, San Jose, London and Paris, and nearly two decades running operations in Mexico across construction, mining and retail. I work as a fractional CFO and advisor to mid-sized companies, and rebuilding a cost of capital that nobody has questioned in three years is one of the more common first engagements.

If you have debt maturing inside twenty-four months and the refinancing assumption has not been revisited this year, that is a conversation worth having before the maturity, not at it. The first one is free.

[email protected]

← Back to Insights