When the news says inflation was 3.2% last month, that number comes from a single government report — the Consumer Price Index, or CPI, published by the Bureau of Labor Statistics. CPI tries to measure how the average price of a representative basket of goods and services has changed over time. It’s used for Social Security cost-of-living adjustments, tax-bracket indexing, union contracts, and every news headline about “inflation.” Understanding how it’s built — and what it leaves out — explains why your personal inflation may feel very different from the official rate.
What’s in the Basket
The BLS tracks the prices of about 80,000 specific items every month, in 75 urban areas across the country. Field staff visit stores, restaurants, doctors’ offices, and rental properties. The basket is grouped into eight major categories with rough weights:
- Housing — about 35% of the basket (rent, owners’ equivalent rent, utilities, household furnishings). The biggest single category by far
- Transportation — about 17% (new and used vehicles, gasoline, auto insurance, public transit)
- Food and beverages — about 14% (groceries plus food away from home)
- Medical care — about 8% (insurance premiums, doctor visits, prescription drugs)
- Recreation — about 5%
- Education and communication — about 6%
- Apparel — about 3%
- Other goods and services — about 3% (personal care, tobacco, etc.)
The weights are updated regularly to reflect how households actually spend money, based on the Consumer Expenditure Survey of about 24,000 families. If Americans collectively start spending more on streaming and less on cable, the weights shift over time.
Headline CPI vs Core CPI
- Headline CPI — everything in the basket, including food and energy. The number most often reported in the news
- Core CPI — the basket excluding food and energy. The Federal Reserve watches this more closely because food and energy prices swing on weather, supply shocks, and global oil markets — noise that obscures the underlying trend
This is why a news headline can say “inflation rose 0.4% last month” while a separate analysis says “core inflation was steady.” Both can be true; they’re different measures.
CPI-U vs CPI-W vs Chained CPI
BLS publishes several CPI variants. They use the same basket and the same prices but weight households differently:
- CPI-U — “all urban consumers.” Covers about 93% of the U.S. population. The most commonly cited measure
- CPI-W — “urban wage earners and clerical workers.” A narrower group. Used by the Social Security Administration to calculate the annual cost-of-living adjustment
- Chained CPI (C-CPI-U) — adjusts for the assumption that consumers substitute cheaper goods when prices rise (buy chicken when beef gets expensive). Used to index federal income tax brackets since the 2017 tax law
Chained CPI tends to rise slightly slower than CPI-U or CPI-W, which is why the choice of index matters: lower indexed thresholds mean bracket creep over time, smaller indexed benefits over time. Whether to switch Social Security from CPI-W to chained CPI has been a long-running political debate for that reason.
Why Your Inflation Differs from the Headline
CPI is a national average across an aggregate basket. Your household’s actual inflation depends on what you spend money on:
- Renters in high-cost cities often see inflation well above headline when rents are rising fast in their metro
- Retirees often see higher inflation than headline, because they spend a larger share of income on medical care — the category that tends to rise faster than overall CPI year after year
- Families with school-age kids may see different inflation than empty-nesters because their food and childcare weights are different
- Drivers with long commutes are exposed to gasoline price swings the headline averages away
BLS publishes a separate index called CPI-E, an experimental measure tracking the basket weighted to households 62 and older. It typically runs slightly higher than CPI-W because of the medical-care weighting. Some advocates want it adopted for Social Security COLAs — this is the case for what’s sometimes called “the CPI-E switch.”
PCE Inflation — The Fed’s Preferred Measure
The Federal Reserve doesn’t target CPI. Its 2% inflation target is based on a different index — Personal Consumption Expenditures (PCE), produced by the Bureau of Economic Analysis. PCE differs from CPI in three ways that matter:
- Broader scope — PCE includes spending by employers and the government on behalf of households (employer-paid health insurance counts; CPI only counts what you pay out of pocket)
- Variable weights — PCE re-weights its basket every month based on actual spending patterns. CPI re-weights its basket every two years. PCE captures consumer substitution faster
- Generally lower — PCE typically runs 0.3 to 0.5 percentage points below CPI because of the weighting differences. When the Fed says it wants 2% inflation, that’s roughly 2.3 to 2.5% on CPI
What CPI Doesn’t Capture Well
- Quality changes — if a $1,000 phone gets twice as good but stays $1,000, CPI may record that as a price decrease through what’s called “hedonic adjustment.” Critics argue this understates real inflation
- Shrinkflation — a smaller package at the same price is supposed to register as a per-unit price increase, but the methodology doesn’t always catch it quickly
- Asset prices — stock prices, home prices, and crypto aren’t in CPI. Home prices show up indirectly through “owners’ equivalent rent,” a rental-equivalent estimate
- Big-ticket once-in-a-decade purchases — the weight given to cars or appliances reflects average annual spending; in the year you actually buy one, your personal inflation may diverge sharply
The Bottom Line
CPI is the best monthly measurement we have of national consumer inflation, and it’s the basis for the inflation news you hear. But it’s an average across a representative basket; your household has its own basket. A retiree spending 25% of income on medical care lives in a different inflation world than a renter in a stable rental market. Knowing what’s in the index — and what isn’t — is the difference between reading the headline and reading the headline accurately.