Consumer Price Index (CPI)

CPI is not the inflation number the Federal Reserve actually targets — that is the PCE price index, produced by the Bureau of Economic Analysis. Two differences drive most of the gap. In scope, PCE counts spending made on consumers' behalf — notably employer- and government-paid healthcare — so it covers a broader slice of the economy than CPI's out-of-pocket basket. In weighting, PCE updates more frequently and lets the mix shift as people substitute between goods, while CPI holds its basket fixed for longer. Together these make CPI typically run a few tenths of a point hotter than PCE, so a 3% CPI and a 2.5% PCE can describe the same economy. Because the Fed's 2% target refers to PCE, the two gauges can seem to disagree even when both are right.

CPI is reported as an index, not a percentage: the level (around 310 in recent years) is set against a 1982-84 = 100 base, so 310 means consumer prices are roughly 3.1 times their early-1980s level. What markets trade on is the change in that index. The seasonally adjusted month-over-month (MoM) figure is the real-time pulse — a single 0.4% monthly rise annualizes to nearly 5%. The year-over-year (YoY) figure is the headline and smooths seasonality, but it carries a trap called base effects: because YoY compares with the level twelve months ago, the annual rate can fall simply because a large past increase is dropping out of the 12-month window, even while current prices are still climbing. Reading MoM and YoY together avoids that illusion.

A fixed basket has a built-in upward bias. When beef gets expensive, shoppers buy more chicken, but a fixed-weight index keeps pricing the old beef-heavy basket — overstating the rise in the true cost of living. The BLS counters this in two ways. Hedonic quality adjustment strips out price changes that reflect a better product rather than inflation: if this year's laptop costs the same but is twice as fast, the index records a price decline. And the Chained CPI re-weights the basket as consumers substitute, so it rises more slowly than headline CPI. The 1996 Boskin Commission famously estimated that methodological biases had overstated CPI inflation by roughly 1.1 points a year, prompting many of these refinements.

CPI is not only watched — it is embedded in contracts and securities, which is why a single release moves real money. Subtract CPI from a nominal figure and you get its real value: if wages rise 4% while CPI runs 3%, real wages grew just 1%, so one report can show a raise and a pay cut at once. The principal of Treasury Inflation-Protected Securities (TIPS) is adjusted by CPI-U, and US Series I savings bonds reset their rate to it twice a year. Social Security's annual cost-of-living adjustment, along with many union contracts and commercial leases, also escalates against CPI — so the index doubles as both a thermometer and a lever.