CPI vs PPI vs PCE.

Scope: who and what is covered

The biggest single source of divergence is what each index covers. CPI captures direct consumer spending — what households actually pay out of pocket. PCE captures the cost of all consumption, including healthcare paid by employers and Medicare, and other items consumed on behalf of households. For PCE, the weight of healthcare is roughly twice the CPI weight, because so much U.S. healthcare is paid by parties other than the consumer.

This single difference accounts for a meaningful share of the persistent gap between CPI and PCE. CPI's medical-care weight is ~8%; PCE's is ~17%. When medical inflation runs above the basket-wide trend (which it usually does), that shows up disproportionately in PCE.

Formula: how prices aggregate

CPI uses a modified Laspeyres formula at the top level — fixed weights from a base period, allowing prices to move but not allowing the basket composition to shift. PCE uses a Fisher-ideal index — effectively the geometric mean of a Laspeyres and a Paasche, which allows for consumer substitution between goods as relative prices change.

The substitution effect alone has averaged about 0.25 percentage points lower inflation for PCE versus CPI per year over the past two decades. When eggs spike and consumers buy more chicken, Fisher-ideal PCE captures that shift; CPI's fixed weights don't.

Source data

PCE inflation is calculated by the Bureau of Economic Analysis as part of the National Income and Product Accounts. The BEA uses CPI as a primary input but supplements it with PPI for items where producer prices give a better signal (medical care is a major example) and with its own data sources. PCE isn't an independent measurement of prices — it's a re-weighted, methodologically adjusted version of mostly-CPI data.

PPI: the upstream view

PPI measures prices at the producer level, before retail markup, taxes, and consumer-facing distribution. It's the cleanest read on input cost pressure. Many PPI sub-indexes lead the corresponding CPI sub-indexes by 1 to 6 months — for example, finished-goods PPI tends to lead apparel and household goods CPI. But the lead-lag relationship varies by industry and the magnitude of pass-through depends on how much pricing power producers and retailers have.

The CPI-PCE gap is usually a few tenths of a point on a year-over-year basis, with PCE running slightly below. But the gap has occasionally widened to over a full percentage point, with important policy implications.

2021–2022 inflation peak. CPI hit 9.1% in June 2022; headline PCE peaked at 7.2% the same month — a gap of 1.9 percentage points. The widening was driven primarily by the difference in shelter weights (CPI ~33%, PCE ~16%) at a time when shelter inflation was the dominant driver. With shelter so heavy in CPI and so light in PCE, the same housing-cost surge produced a much larger CPI move.

2008 oil spike. CPI ran above PCE during the oil-led inflation peak, but the gap was smaller because the energy weights are closer in the two indexes.

1990s convergence. The CPI-PCE gap narrowed and was sometimes near zero during the disinflationary 1990s, partly because shelter inflation was muted and the methodology differences had less to bite on.

The Fed's preference for PCE became formal in 2000 (later codified at 2% in 2012) partly to avoid the structural overstatement Fed economists believed was embedded in CPI. The post-2021 episode showed that the gap can also reflect substantively different economic stories — shelter-driven inflation moves CPI more than PCE; healthcare-driven inflation moves PCE more.