Three Inflations, One Hammer
Headline inflation is a single number. In early 2026, it lands within touching distance of central bank targets across much of the rich world. But the number is a sum, not a thing. A 2.4% U.S. CPI in February 2026 isn't the same animal as a 2.4% reading two years earlier, or as the 2.0% the European Central Bank is approaching. Same headline, three different stories underneath. The interesting question is which story.
The three drivers
Disaggregate the headline and three components emerge that respond to almost entirely different forces.
Goods inflation tracks tradables: refrigerators, cars, t-shirts. It's driven by supply chains, trade policy, and the global production network. Through January 2026, U.S. core goods inflation ran at 1.9% year-over-year, against a pre-pandemic average of negative 0.6% (Minneapolis Fed). Tariffs are the obvious suspect, but a March 2026 note from the same bank argues the cross-category pattern doesn't match what tariffs alone would produce. Whatever the cause, goods inflation is a trade-and-supply phenomenon.
Services inflation tracks non-tradables: a haircut, a meal out, a doctor's visit. The dominant cost input is wages. A March 2026 FEDS Note from the Federal Reserve Board finds that non-housing services have the most widespread price increases across the United States, the euro area, Canada, and the United Kingdom — and that broad-based wage growth, particularly in services-heavy sectors, is the candidate driver. Services prices are sticky because wage contracts are sticky.
Import inflation tracks the cost of bringing goods across borders: FX moves, commodity shocks, shipping. In the euro area, the March 2026 ECB Economic Bulletin reports import prices for non-food consumer goods falling 3.6% year-over-year, with manufactured food import prices having swung from +10.6% a year earlier to −3.6% — a reflection, the bulletin notes, of past euro appreciation and Chinese export redirection.
One hammer
The central bank has, essentially, one tool: the short-term policy rate. The rate affects each component, but on different timescales and through different channels.
On imports, the rate operates almost instantly through exchange rates: raise rates, the currency strengthens, imports get cheaper. A monetary surprise can move FX in minutes.
On goods, the rate acts through demand: higher rates dampen consumption, weaken pricing power, eventually pull prices lower. The lag is months to a year, and the channel gets muddied when tariff or supply shocks are dominant.
On services, the rate acts through the labor market: higher rates slow hiring, eventually cool wage growth, eventually cool service prices. The chain is long — typically a year and a half — and works through the unemployment rate, which means services disinflation is paid for in jobs.
One tool. Three transmission speeds. Targeting the average distorts the components.
A note on the underlying geometry
There is a deeper sense in which the headline number under-determines the underlying state. Write headline inflation as the weighted average
where the weights are CPI basket shares. One linear constraint on three unknowns. The same 2% headline is consistent with an entire two-dimensional family of (goods, services, imports) compositions.
This is the same logic that drives the conformal bootstrap in theoretical physics — a program described in Quanta Magazine in 2017 — where accounting-like consistency conditions, on their own, define a feasible region rather than a unique theory. Economists know the equivalent intuition under different names: Charles Manski's program of partial identification (Tamer's 2010 Annual Review of Economics survey is the standard entry point), and Wynne Godley's sectoral-balance identities. The headline doesn't tell you where on the plane you are. Only the components do.
Same headline, opposite movements
The 2026 data make this concrete. U.S. core goods inflation is rising (Minneapolis Fed: +1.9%). Euro-area consumer goods import prices are falling sharply (ECB: −3.6%). On services, both regions show the same stickiness; on goods, the regions point in opposite directions. J.P. Morgan Global Research forecasts 2026 core CPI of 3.2% in the U.S. against 1.9% in the euro area — a one-and-a-third point gap built almost entirely out of the goods component.
Same headline category. Opposite movements. A monetary policy stance calibrated to the average misses the diagnosis.
Who pays
The three inflations also fall on different people.
Goods inflation is regressive. Lower-income households spend a larger share of their budget on goods — food, clothing, durables — than richer ones do. When goods prices rise, they feel it first. Tariffs are, quietly, a consumption tax with a steep income gradient.
Services inflation tilts the other way. Wealthier and older households spend more on services — healthcare, education, professional fees, leisure — so sticky service prices chip at real wealth and at pensions. Wage earners sit on both sides of the wage-services loop and are partially insulated. Retirees on fixed incomes are not.
Import inflation shows up first in energy. A single commodity shock can dominate the headline, particularly in import-dependent economies. The euro area in 2026 — with a fresh Middle East conflict pushing energy prices — is living that example.
Same 2% headline. Three very different incidence patterns. A central bank that delivers 2% by squeezing services taxes one group; one that lets goods inflation run taxes another; one that imports its disinflation from a strong currency does something else again. The number on the front page is identical. The country, in any meaningful sense, is not.
Closing thought
A headline target is therefore not just policy-ambiguous: it is distributionally ambiguous. The number is useful in the way that a national average temperature is useful for telling you whether the planet is warming. It does not tell you what coat to wear today, and it does not tell you who is cold.
Data sources: Tariffs can't explain rising goods inflation (Minneapolis Fed) | Is the Inflation Process Different After the Pandemic? (Federal Reserve Board) | Economic Bulletin Issue 2, 2026 (ECB) | Global Inflation Forecast 2026 (J.P. Morgan) | Using the Bootstrap (Quanta Magazine) | Partial Identification in Econometrics (Tamer 2010, Annual Review of Economics)