PCE: The Fed's Preferred Inflation Gauge

If beef becomes expensive and households buy chicken instead, PCE adjusts its weights to reflect that. CPI, with a fixed basket, continues to price the beef.

Substitution is real consumer behaviour, so PCE is closer to describing what people actually spend, and it produces a lower reading as a result. The counter-argument is that substituting to a cheaper good is a reduction in living standards that the index treats as costless, which is a genuine criticism rather than a technicality.

CPI counts only what households pay directly. PCE counts the full cost of healthcare including the portion paid by employers and by government programmes.

Since that is the large majority of US healthcare spending, PCE has a much larger healthcare weight and CPI has a much larger shelter weight. The two indices are therefore weighted toward different parts of the economy and respond to different pressures.

CPI is released first and generates the larger immediate market reaction. PCE arrives a couple of weeks later and is largely predictable from CPI and the producer price index by then, so it moves markets less.

Policy, however, is set against PCE, which makes it the one that matters for the decision even though it is the one that matters less for the day's trading.

The two percent target is defined on the personal consumption expenditures price index, not on the consumer price index, and the choice was made deliberately in 2000. Three properties drove it.

1. 1The weights updatePCE weights are recalculated continuously from the national accounts, so the basket reflects what people are buying now rather than what they bought during a past survey period.
2. 2The scope is widerIt counts spending made on households' behalf, principally employer and government healthcare payments, which is a large part of consumption that households never see a bill for.
3. 3History is revisedPCE is revised as better source data arrives. That is a disadvantage for anyone trading the print and an advantage for a committee trying to understand what happened.

The consequence of the first property is that PCE typically runs a few tenths below CPI over long periods. It is not a different inflation; it is the same phenomenon measured with a basket that updates as substitution happens, and the gap is the substitution effect made visible.

CPI is published first, roughly two weeks after month end. PCE follows about two weeks later, at the end of the month. Because several PCE components are constructed from CPI inputs, most of the PCE figure is estimable once CPI and the producer price index are known, which is why the later release usually moves markets less.