US CPI explained — why one monthly number moves rates everywhere
The US consumer price index measures how fast prices are rising for urban American households, and because it shapes Federal Reserve expectations, asset prices worldwide move with it
The three lines
- CPI is published monthly by the Bureau of Labor Statistics in headline and core (ex-food-and-energy) forms
- Markets read the core month-over-month change first, and react to the gap from consensus rather than the level
- Shelter is about a third of the index and updates slowly, which is why CPI lags reality in both directions
Key questions
- What is CPI measuring
- The Bureau of Labor Statistics surveys prices for hundreds of goods and services that urban US households actually buy — groceries, gasoline, rent, medical care, airfares, car insurance — and weights each by its share of household spending. The result is an index; the news reports how much it changed over one month or twelve. Crucially it measures the rate of change, not the level: falling inflation means prices rise more slowly, not that anything got cheaper.
- Headline versus core — which one matters
- Headline includes everything. Core strips out food and energy, because those swing on weather and geopolitics and obscure the trend. Central banks and markets weight core for policy judgments; households feel headline. The hardest months to interpret are the ones where the two point in different directions.
- Why do markets outside the US care
- Three channels. US rate expectations move the dollar, and the dollar moves flows into and out of other markets. US Treasury yields are the reference rate for global asset valuation, so equity multiples everywhere reprice. And other central banks cannot ignore rate differentials and currency effects when setting their own policy. For an investor in Seoul or Frankfurt, US CPI is not foreign news — it is their own discount rate.
Once a month, at 8:30 a.m. Eastern time, a handful of numbers are published. Treasury yields move within seconds, and hours later Asian markets open into the result. This is the US consumer price index — CPI.
This reference explains what it measures, how it is built, what to read first when it lands, and why it is not a purely American concern.
1. A speedometer, not an odometer
The Bureau of Labor Statistics surveys prices across the country for hundreds of items urban households buy, weights each by its share of spending, and combines them into one index. The change over one month is month-over-month; the change over twelve is year-over-year. The figure in most headlines is year-over-year headline CPI.
The essential point is that CPI measures how fast prices are rising, not how expensive things are. When inflation falls from 3% to 2%, nothing got cheaper — prices simply climb more slowly. That single distinction accounts for most of the gap between the data and how it feels.
2. Four inflation gauges, distinguished
| Gauge | Publisher | Character | Primary audience |
|---|---|---|---|
| Headline CPI | BLS | All items, volatile | Media, households |
| Core CPI | BLS | Excludes food and energy | Markets, central banks |
| PCE price index | Bureau of Economic Analysis | Adjusts for substitution; the Fed's official target | Fed policy |
| PPI | BLS | Producer prices, read as a leading signal | Analysts |
The third row causes the most confusion. When the Federal Reserve names a 2% target, it means PCE, not CPI. But CPI is released first and feeds much of the PCE calculation, so markets treat it as the preview — which is why a gauge the Fed does not formally target still moves rates.
3. Shelter, and the built-in delay
Shelter is the single largest component, conventionally cited at about a third of the index, and it produces CPI's defining weakness.
Shelter is measured across all existing leases, not just newly signed ones. Since leases typically run a year, changes in market rents take many months to appear. Owner-occupied housing enters through owners' equivalent rent — an estimate of what the home would rent for, not any actual payment.
The result is that CPI's largest component moves late and smoothly. Inflation is understated at the start of an upswing and overstated as it turns down. That is why analysts compute "core services excluding shelter" separately within minutes of the release.
4. The order markets read it in
| Order | What is checked | Why |
|---|---|---|
| 1 | Core CPI, month over month | Fastest read on trend; a 0.1pt surprise moves markets |
| 2 | Headline, year over year | The headline number, sentiment driver |
| 3 | Core services excluding shelter | Trend with the lagging component removed |
| 4 | Distance from consensus | The surprise, not the level, sets the price move |
| 5 | 2-year Treasury yield reaction | The clearest thermometer of rate-cut expectations |
Row four is what newcomers most often miss. A print of 2.8% is not in itself information — that expectation is already in prices. What moves markets is that consensus was 2.7% and the print was 2.9%. The same number can be a rally or a rout depending on what was expected.
5. How it travels abroad
| Channel | Mechanism | Lag |
|---|---|---|
| Currency | Rate expectations → dollar → cross-border flows | Same day to days |
| Discount rate | Treasury yields → global equity valuation | Same day |
| Local policy | Rate differentials → domestic central bank constraints | Weeks to months |
| Real economy | US consumption → export volumes | Months |
The top two arrive immediately, the bottom two slowly. When a market halfway around the world gaps at the open the morning after a CPI release, it is almost always the first two.
6. Three common misreadings
"Lower inflation means lower prices." No — it means slower increases. Prices only fall when inflation is negative, which usually signals something worse.
"My costs rose more than CPI, so CPI is wrong." They measure different things. CPI is the average basket of urban US households; no individual's basket matches it. A renter and an outright homeowner cannot feel the same index.
"One month tells you the trend." Seasonal adjustment, one-off items and methodology changes all move a single month, and revisions change history. A three-month average is the safer read.
7. What remains unverified
The "about a third" shelter weight is an approximation that changes with BLS reweighting; the published weights are the authority. The release time shifts by an hour outside US daylight saving time.
How to read the US jobs report is in "The US jobs report, explained" and "How the unemployment rate is calculated." The path from rate cuts to household deposits and loans is in "When America cuts rates, what happens to your savings." The Fed's late-August signalling event is in "What Jackson Hole is." This reference is updated with each release.