Two views of the same number, and why they can point in opposite directions.
intermediate · 4 min read · 15 XP
Every calendar prints core CPI twice — once month-on-month and once year-on-year — and traders routinely quote whichever one supports the view they already hold. This lesson explains what each one actually measures, why they disagree, and which to weight when they do.
Core CPI is headline CPI with food and energy stripped out. That sounds like hiding the parts households care about most, and in a sense it is — but the reason is sound.
Food and energy prices are set largely by weather, harvests, OPEC decisions and shipping. They are volatile, they mean-revert, and no interest rate a central bank sets has any influence on them. Raising rates does not grow more wheat. Core inflation isolates the part of the price level that monetary policy can actually reach — which makes it the better guide to what the bank will do.
The one-line version
Headline tells you what households experienced. Core tells you what the central bank will react to. Currencies trade the second one.
Both are calculated from the same index. The difference is the length of the window.
That difference has three consequences.
MoM is fresher but noisier. One month of data contains real signal and real measurement noise in roughly equal measure. Desks smooth it by looking at the three-month and six-month annualised rates — take the recent monthly prints, compound them out to a yearly pace, and you get the current run rate without waiting a year for YoY to catch up.
YoY is smoother but stale. It is an average of the last twelve monthly changes. Six of those months may be describing a regime that has already ended.
YoY carries base effects. Because the comparison point is a month twelve months ago, YoY can fall purely because a large old month has dropped out of the window — with nothing whatsoever happening this month. The reverse traps traders more often: YoY can rise on a mild month simply because the month rolling out was unusually weak.
The base-effect trap
"Inflation fell to 2.8% from 3.1%" can be entirely mechanical. If last year's same month was very strong, it leaves the twelve-month window and the annual rate drops on its own. Before treating a YoY move as news, check whether the MoM prints justify it. If MoM has been steady at 0.3% for six months, a falling YoY is arithmetic, not disinflation.
Put the two side by side and you get a directional read that neither gives alone:
| Core MoM | Core YoY | What it usually means |
|---|---|---|
| Rising | Rising | Genuine reacceleration — the hawkish case, strongest currency support |
| Rising | Falling | Base effects are flattering the annual rate; momentum has turned up. Markets follow MoM |
| Falling | Rising | Annual rate still catching up to a slowdown already underway |
| Falling | Falling | Sustained disinflation — the dovish case |
The two off-diagonal rows are where the money is, because that is where the headline and the reality disagree and the first market reaction is most often wrong.
Almost every inflation-targeting central bank states its target as an annual rate — 2% year-on-year. That is what the mandate says.
But a bank does not wait a year to act. Internally it is forecasting where the annual rate will be in twelve to eighteen months, and the input to that forecast is the recent monthly run rate. Rough arithmetic: a 2% annual target is roughly 0.17% per month, so a string of 0.2% MoM prints is target-consistent and a string of 0.4% prints is not — regardless of what the annual figure currently reads.
Weaker. The MoM miss is the fresh information: current momentum has collapsed to well below target pace. The YoY beat is describing eleven months of history plus one soft month, and a favourable base effect is doing most of the work. Expect an initial spike higher on the YoY headline that fades as desks read the monthly figure — a classic algorithm-then-human sequence.
What to remember
Core CPI excludes food and energy because those prices are outside monetary policy’s reach, which makes core the better guide to central-bank behaviour. Month-on-month is the fresh signal and year-on-year the smoothed headline; because YoY carries base effects, the two can disagree, and when they do the market usually ends up following momentum rather than the annual number.