Core versus headline inflation: which one the market actually trades

[Fairman]

Core versus headline inflation: which one the market actually trades

Consumer price reports contain a headline number and a core figure that excludes food and energy. Sometimes they point in different directions, and the market has to choose.

In some periods, the market focuses on core because it reflects underlying trends. In others, the headline gets attention, particularly when energy prices dominate. And sometimes the focus is on a specific component, like shelter or services excluding housing.

The first reaction depends on which figure algorithms are programmed to read first. A mismatch between headline and core can produce a spike in one direction followed by a reversal as traders digest the details.

The skill is knowing which number matters in the current regime. Reading the previews and the market commentary in the days before the release usually makes it clear.

After the release, the month-over-month figure often matters more than the annual one, since it shows the latest trend. Many traders overlook that.

A useful tool is a table of the last twelve releases with the headline and core surprises and the pair's reaction. Patterns about which one dominated become visible quickly.

Which inflation component has been moving price most in recent releases?

[Fairman]

For US CPI, my impression is that core month on month has been what the market reacts to most in recent years. Headline includes food and energy, which are volatile, so a big energy move can push headline around without changing the trend. Core excludes them, so it's seen as a better guide to underlying inflation. I've seen releases where headline beat expectations but core was in line, and the dollar barely moved, and the reverse, with headline in line but a core surprise causing a big move.

[Amanhasnoname]

I used to look at the headline number first and then wonder why price went the other way. Over about a year I kept a table of CPI releases with the headline surprise, the core surprise, and the 15 minute move on EURUSD. In the eleven releases where they pointed in opposite directions, price followed the core surprise in eight, and the headline only in three.

The month over month figure mattered even more than I expected. In two cases the annual figure was in line, but the monthly core number surprised by 0.1, and price still moved 30 to 40 pips. Annual numbers carry the base effect from a year earlier, so they often hide what the latest month did.

My own routine is to write down the consensus for headline and core, monthly and annual, before the release. I then only look at core monthly first, and I do not enter until the two minute mark so that the first reversal has had time to show.