Inflation Data: CPI and PCE

Inflation data moves currencies because it moves rate expectations, and lesson 1 established that rate expectations are most of what currencies trade on. What is less obvious is that the number in the headline is rarely the number the market reacts to.

How CPI is built

The Consumer Price Index tracks the cost of a fixed basket of goods and services bought by households. A statistics agency collects thousands of prices each month, weights them by how much people actually spend on each, and reports the change.

Two structural properties matter more than the construction details.

• The basket is fixed for a period, so it lags changes in what people actually buy. When a price rises, people substitute away from it, and the index takes time to notice.
• Year-on-year readings carry base effects. A high reading can fall sharply purely because the month being compared against was unusually high a year ago — arithmetic, not disinflation. Anyone reading only the year-on-year number will mistake one for the other.

PCE and why the Fed prefers it

Personal Consumption Expenditures is a second US inflation measure, built from what households actually spent rather than from a fixed basket. The Federal Reserve's 2% target is defined on PCE, not CPI.

The practical consequence of that last row: a country can be at target on one measure and above it on the other, and arguments about whether inflation has been beaten sometimes turn out to be arguments about which index is being quoted.

CPI still moves the market more, despite the Fed targeting PCE, for a simple reason: it is published first. By the time PCE arrives, most of its content has already been inferred from the CPI release weeks earlier, so PCE usually confirms rather than surprises.

What the other major banks target

• European Central Bank — HICP, the euro area's harmonised consumer price index, at 2%.
• Bank of England — CPI, at 2%.
• Bank of Japan — CPI, at 2%.
• Federal Reserve — PCE, at 2%.

Headline versus core

Core strips out food and energy. That sounds like removing the part people care about most, and in a sense it is — but the reason is defensible.

1. Food and energy prices are volatile and driven by things monetary policy cannot touch: weather, harvests, conflict, OPEC.
2. They mean-revert. An oil spike that fades leaves no lasting mark on the price level.
3. Core is a better forecast of future headline than current headline is, which is the whole point — central banks are setting policy for a year or two ahead.

So the market usually reacts to core, and specifically to core month-on-month, which is the freshest read on the underlying trend. A hot headline with a soft core produces a much smaller move than the reverse.

Reacting to a surprise

The move comes from the gap between the release and consensus, and it is worth being precise about how small a surprise counts as large.

Consensus for core month-on-month is typically quoted to one decimal place — 0.3%, say. A reading of 0.4% is a single decimal and routinely worth 50 to 100 pips on the major dollar pairs in the first minutes. A rounding difference is a market event, which tells you how tightly these numbers are priced.

What happens in the first minutes

• Spreads widen sharply in the seconds around the release, often several times normal.
• The first move is frequently reversed as the components are read properly.
• Stops fill far from where they were placed, which is a sizing problem rather than a broker problem — Track 2, lesson 3 covered why.
• Liquidity returns within a few minutes, and the settled direction after that is the more reliable signal.

One release also means much less than the direction of several. A single hot month is noise; three consecutive hot months change the rate path, and the currency move that follows is correspondingly larger and more durable.