How the CPI Is Calculated by the BLS

When a news anchor says "inflation came in at 3.2% last month," they are referring to a single number produced by a process that involves tens of thousands of price observations, complex weighting systems, and decades of methodological refinement. The Consumer Price Index, published monthly by the U.S. Bureau of Labor Statistics, is the most widely cited measure of inflation in the United States — and it underpins everything from Social Security adjustments to Federal Reserve monetary policy decisions to the rent increases in many lease agreements. Understanding how it is built demystifies why it sometimes feels like it doesn't match your own experience of prices at the grocery store or the pump.

This article walks through the CPI methodology from the ground up: what it measures, how the basket of goods is constructed, how prices are gathered, and how the final index number is calculated. It also covers the known limitations of the CPI and how it compares to the Federal Reserve's preferred inflation measure, the PCE.

What the CPI Measures

The Bureau of Labor Statistics publishes several variants of the Consumer Price Index, each designed to measure prices for a different segment of the population. The two you will encounter most often are CPI-U and CPI-W.

CPI-U — the Consumer Price Index for All Urban Consumers — is the headline number you see in financial news. It covers approximately 93% of the total U.S. civilian noninstitutionalized population: essentially anyone living in an urban or suburban area, regardless of their employment status. This is the broadest available CPI measure and the one used in most personal finance calculations, including the inflation calculator on this site.

CPI-W — the Consumer Price Index for Urban Wage Earners and Clerical Workers — covers a subset of the CPI-U population. It is restricted to households where at least half of income comes from hourly wages or clerical work. The CPI-W is the measure that Congress uses to calculate annual Social Security cost-of-living adjustments (COLAs), which is why retirees sometimes notice that their Social Security increases feel misaligned with their actual cost of living — the index used for their adjustment is calculated on a somewhat different population's spending patterns.

Both indexes use 1982–1984 as the base period, meaning that the average price level across those three years is set equal to 100. A current CPI-U reading of, say, 315 means that a basket of goods that cost $100 in the 1982–84 base period now costs $315. This base-period anchoring makes it easy to compute cumulative inflation over any span of decades.

The Basket of Goods

The foundation of the CPI is the concept of a fixed "market basket" — a representative collection of goods and services that reflects what American households actually spend money on. The basket is not invented arbitrarily. It is derived from the Consumer Expenditure Survey (CE), an ongoing BLS program that interviews a rotating panel of roughly 14,000 households per quarter about their purchasing habits. These households report in detail what they spend across all categories of consumption, from rent and gasoline to restaurant meals and prescription drugs.

The CE survey data allows the BLS to determine both what categories belong in the basket and how much weight to assign each one. The weights reflect the share of the average household's budget devoted to each spending category. A category that consumes 30% of the average household's budget receives roughly 30% of the weight in the index, meaning that price changes in that category have roughly 30 times the impact on the overall CPI as price changes in a category weighted at 1%.

The BLS organizes the basket into eight major expenditure categories:

• Food and beverages (groceries and food away from home) — approximately 14% of the CPI-U basket
• Housing (rent, owners' equivalent rent, lodging away from home, household furnishings, utilities) — approximately 33%, the largest single category by far
• Apparel (clothing, footwear, jewelry) — approximately 3%
• Transportation (new and used vehicles, gasoline, auto insurance, airfare, public transit) — approximately 17%
• Medical care (prescription drugs, professional services, hospital services) — approximately 9%
• Recreation (televisions, sporting goods, pets, admission fees) — approximately 6%
• Education and communication (tuition, postage, telephone services, computers) — approximately 6%
• Other goods and services (personal care, tobacco, funeral expenses, financial services) — approximately 3%

The housing component deserves special attention. At roughly one-third of the total index weight, housing price changes have an outsized effect on the reported CPI. Within housing, the BLS uses a concept called owners' equivalent rent (OER) — an estimate of what a homeowner would pay to rent their own home in the current market — rather than tracking actual home purchase prices. This is an important methodological choice: it means that a housing market where home prices surge dramatically may not show up in the CPI immediately if rental rates lag behind purchase prices.

How BLS Collects Price Data

Determining what goes in the basket is only the first step. The BLS then must actually measure what those items cost every month. This is a substantial field data operation.

Each month, BLS data collectors — called economic assistants — visit approximately 23,000 retail and service establishments across 75 urban areas selected to represent all sizes of cities and regions of the country. Within each establishment, specific items are selected using a probability sampling method that weights toward products consumers actually purchase. The same item is priced month after month to track its price change over time. If an item disappears from a store — a product discontinued, a business closed — the data collector substitutes a comparable replacement and records any quality adjustment needed.

The BLS collects prices in person, by telephone, and increasingly through online data collection. For housing costs specifically, BLS collects rent data from approximately 50,000 residential rental units. Renters report their actual monthly rent, and the change in that rent from one period to the next feeds directly into the housing component of the CPI.

In total, the BLS processes roughly 80,000 individual price quotes each month to produce a single CPI reading. The sheer scale of this data collection is what gives the CPI its statistical credibility — it is not a sample of a few hundred prices extrapolated; it is a rigorous, large-scale survey operation conducted continuously.

The Index Calculation

With prices in hand for thousands of items across hundreds of locations, the BLS must aggregate everything into a single index number. The CPI is calculated using a Laspeyres index formula — a method that uses a fixed set of quantities from the base period rather than updating quantities continuously.

In practice, the BLS computes the index at many geographic and category levels simultaneously, then aggregates them upward using the expenditure weights from the Consumer Expenditure Survey. A price increase of 5% in medical care — weighted at about 9% of the basket — contributes less to the total CPI change than a 2% increase in housing costs — weighted at about 33% of the basket.

Known Limitations of the CPI

The CPI is an enormously useful tool, but it is not a perfect measure of inflation. Economists have identified several systematic biases and practical limitations that affect how well the index captures the true experience of price changes for any individual household.

Substitution bias is the most widely discussed limitation. The Laspeyres formula uses fixed base-period quantities — it assumes you keep buying the same things regardless of price changes. In reality, when beef prices spike, many consumers substitute chicken. When one brand gets more expensive, shoppers switch to a less expensive alternative. Because the fixed basket doesn't allow for this substitution, the CPI can overstate the true cost of maintaining a given standard of living during periods of relative price changes.

New product bias arises because new goods — smartphones, streaming services, new medications — take time to enter the basket after they appear in the market. During the period between a product's introduction and its inclusion in the CPI basket, consumers may be gaining utility from it that doesn't register in the index at all. By the time the item is added, early-adopter price premiums have often already fallen.

Outlet bias refers to the historical tendency of the CPI to lag in capturing consumers' shift toward lower-price retailers. If shoppers increasingly buy from discount warehouses or online retailers rather than traditional stores, but the BLS price surveys are still weighted toward traditional retail outlets, the index may overstate average prices paid.

Quality adjustment is a challenge running in the opposite direction. When a car model gets a new safety feature or a computer gains a faster processor, the BLS attempts to separate genuine price increases from quality improvements. This is done through a technique called hedonic regression — but these adjustments are estimates, and reasonable economists disagree about how well they capture real-world quality changes.

Perhaps most importantly for individuals: your personal inflation rate may differ significantly from the published CPI. Retirees typically spend a higher share of their budget on medical care and housing, and a lower share on transportation and apparel. Young urban renters may experience housing inflation far above the national average. The CPI is a national average — it describes no individual household exactly.

CPI vs PCE: Why the Fed Uses a Different Measure

The Federal Reserve does not use CPI-U when it states its 2% inflation target. It uses the Personal Consumption Expenditures (PCE) price index, published by the Bureau of Economic Analysis. The Fed prefers PCE for several reasons that have real implications for monetary policy.

The most important difference is in how the weights are updated. The CPI uses fixed basket weights that are revised every two years in bulk updates. The PCE uses chain-weighting — the basket weights update continuously to reflect actual consumer spending patterns. This makes PCE more responsive to the substitution behavior that the CPI's fixed basket misses. As a result, PCE tends to produce lower inflation readings than CPI-U during periods when consumers are actively substituting away from expensive items.

Historically, PCE has run approximately 0.3 to 0.5 percentage points below CPI-U on average. This is not a trivial difference: when the Fed says inflation is at its 2% target using PCE, CPI-U might be showing 2.3–2.5%. Knowing which measure is being discussed matters when interpreting inflation reports and policy announcements.

PCE also uses a broader scope, incorporating spending estimates for goods and services paid on behalf of consumers by third parties — such as employer-paid health insurance premiums and Medicare payments for medical services. CPI measures only what consumers pay directly out of pocket. This broader scope gives PCE a somewhat different profile during periods of rapid healthcare cost changes.

Using the Inflation Calculator

Our Inflation Calculator uses official BLS CPI-U annual average data going back to 1913. The annual average approach — averaging the 12 monthly CPI readings for each year — is the most common method for year-to-year purchasing power comparisons, and is the standard used in academic and government research.

Note that the most recent year's data may be preliminary. The BLS releases monthly CPI figures throughout the year, and the annual average for the current calendar year is not finalized until the final monthly reading (December) is published the following January. Our calculator notes where preliminary data is in use.

If you want to understand how inflation affects your real wage over time, pair the inflation calculator with our Salary Calculator. For a broader discussion of why inflation matters for investment returns, see Real vs Nominal Returns: Why Inflation Matters for Investors. And for the conceptual background on what inflation is and how it affects purchasing power, our companion article What Is Inflation and How Does It Affect Your Savings? covers the full picture.