The Week the Signals Stopped Agreeing

In credit analysis there is a particular moment I was trained to respect. Two indicators that normally move together stop agreeing. The junior analyst calls it noise and smooths it. The senior analyst stops, puts down the file, and treats the disagreement as the most valuable thing in front of him.

This week produced four of them at once.

The man who refused to be a signal became the loudest signal of the week.

On Friday morning Kevin Warsh gave his first Jackson Hole address as chairman and declined, deliberately, to say what he would do. Providing forecasts to illustrate a reaction function, he argued, works better in theory than in practice, better in the lab than in the field. He went further: we should not indulge a regime in which market participants look primarily to the Fed for their next trade. On inflation he was cool — the summer's softer readings do not, in his view, show that underlying trends have meaningfully improved, and unless the Fed can be confident inflation is moving to target clearly and at sufficient speed, there is work to do.

Now set that beside the data. The July PCE report landed the same morning: 3.7% headline, 3.3% core. Ahead of the speech, fed funds futures put the odds of a September hike at roughly 38 to 40 percent. After it, around 55% of bond traders expected one. The two-year note jumped nearly eight basis points to 4.31%. Equities climbed regardless.

Read that sequence again. The inflation print — the actual new information — barely moved anything. The speech in which the chairman explained that he would not be providing a reaction function moved September odds by roughly fifteen points in a morning. A refusal to guide is still guidance. You cannot exit the forward-guidance regime by announcing your exit from it, because the announcement becomes the most closely parsed communication of the week. Warsh has diagnosed the disease correctly and then administered a dose of it.

Korea and Brazil looked at the same risk and did opposite things.

On Wednesday the Bank of Korea raised its seven-day repo rate a quarter point to 3.00% in a 6–1 vote, a second consecutive hike, and lifted its 2026 growth forecast to 3.3% from 2.6% on the back of an extraordinary semiconductor cycle. Household debt had just crossed 2,000 trillion won for the first time. Governor Hyun Song Shin's reasoning was explicitly preemptive — move early, anchor expectations fast, and you shorten the tightening and spare growth the worst of it.

Three weeks earlier Brazil's Copom voted unanimously to cut the Selic for a fourth consecutive time, to 14.00%, completing a hundred basis points of easing since March. Its own statement describes heightened uncertainty, elevated risks, and a deanchoring of longer-term inflation expectations, with the Focus survey sitting at 5.0% for 2026 against a 4.5% ceiling.

So Korea tightens preemptively to stop expectations coming unmoored, while Brazil eases in a statement that says its expectations are already coming unmoored. Australia holds at 4.35% and calls policy somewhat restrictive. Washington may well hike. And global manufacturing input prices fell almost seven percentage points in the two months to July. One disinflating world, four verdicts. This is not analytical failure. It is what happens when the common anchor is withdrawn and every committee falls back on the domestic constraint that binds hardest — chips and household leverage in Seoul, credibility and the real in Brasília.

Europe's headline disagrees with Europe's composition.

The euro area's goods surplus fell to €9.8 billion in the first half of 2026 from €82.2 billion a year earlier — a number that looks like tariff carnage and is being reported that way. Decompose it and the story inverts. Exports barely moved: €1,487.2 billion, down 0.2%. Imports rose 4.9%. The EU's energy deficit alone widened from €71.3 billion in the first quarter to €101.1 billion in the second, a swing that accounts for essentially the entire quarterly deterioration. Second-quarter exports actually rose 4.5% year on year, reversing three quarters of decline, and the EU still runs a large goods surplus with the United States, on chemicals.

Europe did not stop selling. Europe's energy bill went up. Tariffs are in the number, mostly as a base effect against the pre-tariff export surge of early 2025 — real, but nothing like the headline implies. Anyone reading that surplus collapse as a tariff story has smoothed away precisely the disagreement that mattered, and I will admit I nearly filed this piece having done exactly that.

India opened its wheat trade, and almost nobody asked why.

On 24 August the DGFT moved wheat, durum, flour, maida, semolina and atta from the prohibited list to free, effective immediately, ending a ban imposed in May 2022. The convenient reading is trade diplomacy during a tariff negotiation. The actual reading is a glut. Foodgrain production hit a record 376.6 million tonnes in 2025–26 and wheat a record 120.7 million, and the stated official rationale was that domestic prices had become depressed — farmers needed the export outlet to make the next sowing worth doing. India did not open because anyone asked. India opened because the silos are full.

Here is what I think links the four, and it is uncomfortable for anyone selling macro certainty.

Warsh is substantially right on principle. A market whose first question is always what will the Fed do is not pricing risk; it is pricing one committee's psychology. I spent fifteen years building credit-risk software in the United States and watched genuine underwriting thin out across the guidance era, because when the central bank is the answer to most questions, paying for independent analysis starts to look like a luxury. Warsh wants to hand that work back. Good.

The honest objection is about who is left able to do it. Removing the common signal does not distribute cost evenly. A primary dealer underwrites for itself. So does a global asset manager. A mid-sized exporter in Ludhiana or Coimbatore, financing receivables against a currency whose direction now depends on four central banks openly disagreeing, cannot — and the widened risk premium is paid by exactly those borrowers, who never had a research department to begin with. That is the part of price discovery that never makes the slide.

The Indian wheat trader makes the point sharper still. For four years he could not underwrite an export book at any price, because the binding variable was never harvest, freight or price — it was a notification that might arrive on a Tuesday. That is not market risk. It is policy reversibility, and no rating model I have ever built prices it honestly. This week's liberalisation is genuinely good news and it is also, quietly, provisional: what was freed by notification can be prohibited by notification.

Last week I argued that a corrupted price signal is dangerous. This week the worry is larger. An absent signal is not neutral ground — it gets filled by whatever each borrower can afford to buy, and most cannot afford much at all.

So the question I keep returning to: if the era of a single reliable macro signal really is ending, who rebuilds the capacity to underwrite from the bottom up — and who carries the cost while we wait?