7 min read · Last updated October 14, 2026
- The Consumer Price Index basket is organized into 8 major categories, and Housing alone carries a 44.469% weight – more than Food, Transportation, and Medical Care combined.
- The Bureau of Labor Statistics runs two separate operations to build the index: a Consumer Expenditure Survey sets each category’s weight, and a different monthly survey collects about 100,000 actual prices.
- The Consumer Price Index for All Urban Consumers and the Consumer Price Index for Urban Wage Earners and Clerical Workers price the same goods but weight them differently – and Social Security’s annual benefit increase runs on the second one, not the more commonly quoted first one.
- Starting with the January 2023 index, the Bureau of Labor Statistics updates the basket’s weights every year instead of every two years, so the index tracks real spending shifts faster than it used to.
The Consumer Price Index is built from an 8-category spending basket weighted by a household expenditure survey, combined with a separate monthly survey that prices about 100,000 items nationwide – which is why one household’s own cost of living can rise or fall faster than the single published number covering the whole country. Housing alone accounts for 44.469% of that basket, more than Food, Transportation, and Medical Care combined.
Say you and a neighbor earn the same income. You own your home outright and rarely see a doctor. Your neighbor rents and manages a chronic condition. The published Consumer Price Index treats you both the same, because it averages across every household in the country at once.
In this article
- The 8 categories, and why Housing dominates
- Two different surveys, not one
- Two versions of the index, and why Social Security uses the other one
- Why your own inflation rate can differ from the headline
- The part most explainers skip: imputed rent
- Frequently asked questions
The 8 categories, and why Housing dominates
The Bureau of Labor Statistics (BLS), the federal agency that builds and publishes the Consumer Price Index, organizes the entire basket into 8 major expenditure groups. Based on 2024 household spending data, the relative weight of each group as of December 2025 was:
| Category | Share of the index |
|---|---|
| Housing | 44.469% |
| Transportation | 16.316% |
| Food and beverages | 14.539% |
| Medical care | 8.423% |
| Education and communication | 5.846% |
| Recreation | 5.137% |
| Other goods and services | 2.902% |
| Apparel | 2.368% |
Housing carrying nearly 45% of the index on its own is the single most important fact for reading any Consumer Price Index headline correctly: a housing-cost swing is doing almost half the work of whatever the all-items number does that month.
Two different surveys, not one
The weights in that table and the monthly price changes that move the index come from two entirely separate Bureau of Labor Statistics operations, and conflating them is the most common misunderstanding about how this works.
The first is the Consumer Expenditure Survey: thousands of households report what they actually bought over a period of months, and that spending pattern sets each category’s weight in the table above. The second is a separate monthly price-collection operation that gathers roughly 100,000 individual prices every month, about two-thirds of it from in-person store visits and the rest by phone or a retailer’s own website. One survey tells the Bureau of Labor Statistics what to weight; the other tells it what things actually cost that month.
Two versions of the index, and why Social Security uses the other one
Two versions of the index get published every month, and the difference between them has a real consequence for anyone receiving a Social Security benefit.
The Consumer Price Index for All Urban Consumers, abbreviated CPI-U, represents over 90% of the U.S. population and is the version that makes most news headlines. The Consumer Price Index for Urban Wage Earners and Clerical Workers, abbreviated CPI-W, covers a narrower slice – about 28% of the population, limited to households where more than half of income comes from clerical, sales, service, or laborer wages. Since 1985, the two indexes have used the exact same underlying prices. The only difference is the spending weights each one applies.
That difference matters because the Social Security Administration calculates its annual cost-of-living adjustment using CPI-W, not the more commonly quoted CPI-U. A retiree reading about “the” inflation rate in the news is usually reading the CPI-U number, while the benefit increase that actually lands in their bank account was calculated from a different index entirely.
Why your own inflation rate can differ from the headline
The Bureau of Labor Statistics itself publishes a worked example of this gap. In a hypothetical case, medical care prices rise 10% and food prices rise 5% over a period, while every other category holds steady. For a household spending close to the national average weight on medical care, the published all-items Consumer Price Index rises just 1.4% – but for a hypothetical household that puts 25% of its budget toward medical care instead of the average share, the same price changes work out to a 3.5% increase in that household’s own effective cost of living.
The direction runs both ways. A household that heats with solar power, for instance, can experience less overall inflation than the general population when energy prices spike, simply because that category carries far less weight in their actual budget than it does in the national average.
The part most explainers skip: imputed rent
Within the 44.469% Housing weight, the single largest piece is shelter costs, and the way the Bureau of Labor Statistics prices shelter for homeowners is the least intuitive part of the entire index. Homeowners don’t pay rent, so there’s no direct transaction to observe. Instead, the agency asks a sample of homeowners a specific question: if someone were to rent your home today, how much do you think it would rent for monthly, unfurnished and without utilities? That estimate, called owners’ equivalent rent, has been used since 1987, and its price changes are then tracked using a separate sample of units that are actually rented out – not the homeowner’s own mortgage payment, property tax bill, or maintenance costs.
This is also a basket that moves faster than it used to. Before 2002, the Bureau of Labor Statistics updated spending weights once every 10 years. From 2002 through 2022, it updated them every 2 years. Starting with the index for January 2023, the agency began updating weights every single year, which means the basket now tracks real shifts in household spending considerably faster than the version most people grew up reading about.
Frequently asked questions
What are the 8 major categories in the Consumer Price Index basket? Housing, Transportation, Food and beverages, Medical care, Education and communication, Recreation, Other goods and services, and Apparel. Housing alone carries a 44.469% weight as of the December 2025 relative-importance table, more than any other category by a wide margin.
What’s the difference between CPI-U and CPI-W? Both use identical underlying prices but apply different spending weights. CPI-U covers over 90% of the population and drives most news headlines. CPI-W covers about 28% of the population, limited to clerical, sales, service, and laborer households, and is the version Social Security uses to calculate its annual benefit increase.
Why does my own cost of living feel different from the official inflation rate? Because the published rate is a national average. If your household spends a much larger share of its budget than average on a category that’s rising sharply – medical care is the Bureau of Labor Statistics’s own example – your personal inflation rate runs higher than the headline, even though the math behind both numbers is correct.
How often does the Bureau of Labor Statistics update the basket weights? Annually, as of the January 2023 index. Before 2002 it was once a decade; from 2002 to 2022 it was every two years. The yearly schedule means the basket reflects real spending shifts faster than in past decades.
Does the Consumer Price Index measure what I actually pay if I own my home? No. For homeowners, the index uses an estimate called owners’ equivalent rent – what the home would rent for today – rather than your mortgage payment, property taxes, or repair costs. That estimate has been part of the methodology since 1987.
Leave a Reply