The calm before inflation's second act

It was meant to look like the all-clear. In July American consumer prices were 3.4% higher than a year earlier, down from a peak of 4.2% in May, and the core measure, which strips out food and energy, eased to 2.5%. A few weeks later Kevin Warsh, the new chairman of the Federal Reserve, stood in Jackson Hole and said the central bank's "predominant focus right now should be on prices" — and traders promptly began calling September's policy decision a coin-flip, with ten-year Treasury yields climbing to their highest level since January 2025. A central bank edging toward tightening while the numbers cool is a signal that the calm carries a cost, and that cost will be paid in the price of money, not in next month's statistic.

The most reassuring part of the story is arithmetic, and it has little to do with the Fed. The spring's surge was an energy event, set off when war with Iran constrained shipping through the Strait of Hormuz and sent fuel prices soaring. Prices that spike in one year fall out of the year-ago numbers if they then stop rising. Gasoline fell 9.7% in June and 2.9% in July, mechanically pulling the headline rate down even as energy remained 14.7% costlier than a year earlier. Strip energy out and the picture darkens. The Fed's preferred gauge, the personal-consumption-expenditures (PCE) price index, was up 3.7% over the year and rising at an annualised 4.1% over the previous six months, by Warsh's tally; core PCE stood at 3.3% in July, its highest since 2023 and, outside the pandemic era, since the early 1990s.

Households have noticed the difference between the headline and their bills. Short-run inflation expectations in the University of Michigan's survey jumped from 3.4% in February to 4.6% in June. Warsh concedes that the summer's CPI and PCE readings were "better than expected" but "do not tell me that underlying trends have meaningfully improved." That sentence is a warning in bureaucrat's clothing: the chairman sees a flattering headline and a stubborn core.

He also sees institutions, not one of which answers to him. The forces assembling the next wave are not those behind the last. The surge of 2021-22 was largely a demand shock — stimulus-fuelled spending against constrained supply — and higher rates could lean on it directly. The present one is cost-push, and each of its three legs is partly immune to interest rates.

Tariffs are the clearest. Importers absorbed much of the 2025 duties rather than pass costs on, delaying the effect; that deferral is now expiring. The Minneapolis and St. Louis Feds put the tariff contribution to core inflation at 0.2 to 0.4 percentage points as of July, and the pattern is spreading — clothing-and-footwear prices, up 0.3% in December, rose 3.5% in July. New Section 301 levies followed in late July, and firms tell the New York Fed they plan further price rises. The effective tariff rate did fall below 7% in May after the Supreme Court pruned the 2025 regime: a retreat, not a repeal. Even with duties stripped out, the Minneapolis Fed reckons core inflation would still run about a point above target.

War is the second leg. Brent crude still trades near $91 a barrel. If the Gulf conflict widens, the arithmetic turns ugly: one scenario, computed by Lutz Kilian and colleagues, sees a 20% shortfall in supply pushing oil to $167 and adding 1.8 percentage points to 2026 headline inflation. The energy market has already produced one false peace this year; it is not the place to assume the danger has passed.

Then there is artificial intelligence, the leg with no war and no tariff. Prices of video and information-processing equipment are up 12.2% over the year, against an average annual decline of 6.5% in 2015-19, as the AI build-out bids for scarce chips and data-centre space; the Minneapolis Fed estimates that one category alone adds 0.4 points to core inflation, an effect at least as large as tariffs. Business investment is growing at its fastest pace since 2021, more than half of it AI-related by Warsh's own tally. Raising rates to cool that boom would tax the economy's liveliest investment without lowering the price of a chip.

This is the predicament the new chairman inherits, and why the market listens so closely. Higher rates work on inflation by making borrowing, hiring and spending hurt — a mechanism that, applied to a supply shock, produces a slowdown before it produces cheaper prices. Yet the alternative carries the worst memory in central banking. Since 1948 American inflation has re-accelerated a dozen times, but only once has it run to three successive higher waves: in the 1970s, when the Fed loosened while prices were still elevated. Warsh, who faults his own institution for 65 months of sustained elevated inflation and who describes himself as committed to "a discipline, not a decision," warned that market-based measures of expectations look strong "until they don't." He is less worried about the next report than about the one after it. In July the Federal Open Market Committee voted 9 to 3 to hold rates at 3.5-3.75%, three members already pressing for higher rates.

For the investor, the useful distinction is between the number everyone sees and the structure nobody quotes. The July print was the least informative datum of the summer; the pass-through still arriving, the energy tail, the price of compute and the Fed's willingness to break a boom to kill an expectation are doing the real work. Short rates near 3.5-3.75% against 3.4% inflation leave cash with barely any real return, so a renewed surge would tax cash savers quietly. The judgment would turn on three readings: the path of oil and the Strait of Hormuz, whether the tariff pass-through fades as it seemed to do in 2025, and whether wage growth begins chasing prices — the one leg not yet present, and the only one that turns a cost shock into a wage-price spiral.

Inflation's second act is not inevitable. What is true is that the tools for taming it are not in the hands that usually wield them. Monetary policy can always create a recession; it cannot produce cheap oil, repeal a tariff or talk a data-centre boom into submission. That is why the chairman is signalling resolve at the moment the data look most benign, and why "be cautious" is best read not as a forecast but as an instruction to watch the Fed's choice among its own errors. The calm of July was arithmetic. The risk is what the arithmetic was hiding.