PCE vs CPI

Here is a fact that trips up almost everyone and that you must own: the US has two official inflation gauges, and the one the Fed targets is not the one the news leads with. The news leads with CPI (from the BLS); the Fed targets PCE, the Personal Consumption Expenditures price index, from the BEA (Bureau of Economic Analysis), published inside the monthly Personal Income and Outlays report. Since 2000 the Fed has used PCE as its primary gauge, and its 2% target is defined on headline PCE (while it watches core PCE operationally). The two indices measure "consumer inflation" but differ in four systematic ways, and PCE typically runs a few tenths below CPI, a wedge that averages roughly 0.3–0.4 percentage points over time (check the current gap; it varies). If you quote the CPI number when the Fed is steering by PCE, you will consistently think the Fed is further from target than it is.

What are the four differences between PCE and CPI?

The four differences (worth learning as a set, because the same four ideas explain almost any two price indices that disagree):

1. The formula effect (substitution). CPI uses a fixed-basket Laspeyres-type formula (see consumer price index), which cannot capture consumers substituting away from goods that got relatively dearer. PCE uses a Fisher-ideal chain-weighted formula that updates the basket continuously and does capture substitution. Because substitution always softens the true cost-of-living increase, PCE structurally reads below CPI. This formula difference is the largest single source of the wedge and the deepest reason the Fed prefers PCE: it thinks PCE is simply a better cost-of-living measure.
2. The weight effect. The two indices weight categories very differently because they are built from different source data. The starkest case is housing, which is about a third of CPI but far less of PCE, so when US shelter inflation is high (as in 2022–23), CPI runs hotter than PCE almost mechanically. Conversely, healthcare is far bigger in PCE (next point), so PCE is more sensitive to medical prices.
3. The scope (coverage) effect. This is the subtle, important one. CPI measures only what the urban consumer pays out of pocket. PCE measures all consumption on households' behalf, including the huge slice of healthcare paid for you by employers and by the government (Medicare and Medicaid). So PCE counts spending CPI never sees, and its healthcare weight is correspondingly much larger. PCE is the broader concept: it aims at total household consumption in the national accounts, not just the household's own wallet.
4. Other effects (seasonal-adjustment and residual differences), small in aggregate.

Which should you watch, PCE or CPI?

Two practical points close the loop. First, CPI still matters enormously even though the Fed targets PCE: CPI is what indexes Social Security cost-of-living adjustments, tax brackets, and TIPS (inflation-protected Treasuries); it is released earlier in the month (BLS, ~mid-month) than PCE (BEA, ~month-end), so CPI moves markets first and PCE confirms. Second, professionals read both, plus core, plus a trimmed-mean measure, and treat the ensemble as the signal. But when they want to know "how close is the Fed to its target," they look at core PCE, because that is the number on the FOMC's own dashboard. Learn this and you will never again be confused about why "CPI came in at 3.2%" and "the Fed is near its 2% goal" can be true at the same time.