CPI Explained: What the Consumer Price Index Actually Measures (and What It Misses)

What the CPI-U actually is

The Consumer Price Index for All Urban Consumers (CPI-U) measures the average change over time in the prices paid by urban consumers for a representative market basket of goods and services. The BLS has published it continuously since 1913, making it the longest-running official inflation series in the United States.

The scale of the data collection operation is large. The BLS surveys approximately 80,000 items every month across 23,000 retail outlets, service establishments, and rental units in 75 urban areas covering about 93% of the U.S. population. Price collectors visit grocery stores, hospitals, auto dealerships, landlords, and utility companies, recording exact prices for specific, pre-defined items.

The result is a single index number. In January 1984 (the standard base period) the index was set to 100. As of early 2026, the CPI-U All Items index stands near 320 — meaning the same basket of goods that cost $100 in 1984 costs approximately $320 today.

The 8 major CPI categories and their weights

The BLS does not weight all prices equally. It uses expenditure weights derived from the Consumer Expenditure Survey (CE Survey), which asks thousands of American households to record exactly what they buy and spend over two weeks. The CE Survey runs continuously, and the BLS updates CPI weights annually using a two-year average of CE Survey data.

The current weights across the 8 major categories are:

Source: U.S. Bureau of Labor Statistics, Handbook of Methods, Chapter 17 (Consumer Price Index). Weights are 2023-2024 CE Survey averages effective January 2025.

Housing alone accounts for one-third of the entire index. This is why shelter inflation — which ran at 5.4% year-over-year in early 2026 — has such an outsized effect on the headline number even when other categories are cooling.

How the BLS handles quality adjustments: hedonic pricing

A laptop that cost $1,000 last year and costs $1,000 this year looks like zero inflation. But if this year’s laptop has twice the RAM and a faster processor, the BLS argues the consumer is getting more value per dollar — so the quality-adjusted price has actually fallen.

This methodology is called hedonic quality adjustment (or hedonic regression pricing). The BLS applies it to categories where technology and product specifications change frequently: computers, televisions, smartphones, vehicles, and increasingly medical equipment. A regression model estimates the value consumers place on each product attribute (screen size, processor speed, memory), then adjusts the listed price downward when those attributes improve.

Why critics say hedonic pricing understates inflation

The criticism is not that the methodology is wrong in theory — it is that it diverges from lived experience. A consumer buying a new laptop writes a check for $1,000 regardless of whether the processor is twice as fast. Their cash outflow has not fallen. Their mortgage payment, grocery bill, and insurance premium are not reduced because their laptop is faster.

Economists like John Williams (Shadow Government Statistics) argue that the cumulative effect of hedonic adjustments, combined with the geometric weighting change introduced in 1999, has reduced the measured CPI by 3-7 percentage points versus the pre-1994 BLS methodology. The BLS disputes this characterization, arguing the adjustments reflect genuine economic reality.

The practical takeaway: for goods-heavy categories like electronics and vehicles, the CPI may look lower than your out-of-pocket experience suggests, because the index is measuring value-per-dollar rather than dollars spent.

What the CPI does not measure

Asset prices

The CPI does not include changes in the price of stocks, bonds, or real estate (capital appreciation). A house doubling in value does not show up in the CPI. What does appear is shelter cost — specifically, rent and Owner’s Equivalent Rent (OER), which tracks what homeowners would theoretically charge if they rented their home to themselves. OER is a large component (roughly 25% of all CPI), but it tracks rents, not home prices, and can diverge sharply from the actual purchase price of housing.

Regional variation

The national CPI-U is an average across 75 urban areas. A household in San Francisco paying $3,500/month rent and a household in Memphis paying $900/month both face the same headline inflation number. The BLS does publish regional indexes (Northeast, South, Midwest, West), but even those mask significant city-level variation. There is no official metropolitan-area CPI published on a monthly basis for most cities.

Substitution bias

The CPI-U uses fixed weights updated annually. If beef prices spike, many consumers switch to chicken — but the CPI still weights beef at its historical share of spending until the weights are updated. This means the CPI can temporarily overstate inflation during price spikes. The BLS’s chained CPI (C-CPI-U) partially addresses this by using contemporaneous weights.

Out-of-pocket vs. total health care costs

The CPI measures what consumers pay directly for medical care — their deductibles, co-pays, and uninsured costs. It does not measure the full cost of employer-sponsored health insurance or Medicare premiums paid on their behalf. This significantly understates the true cost of healthcare inflation for working households.

CPI-U vs. CPI-W vs. C-CPI-U vs. PCE: which measure for what purpose

Sources: BLS Handbook of Methods; BEA PCE methodology; Social Security Administration COLA documentation; IRS Rev. Proc. 2018-18.

One practical implication: your Social Security benefit adjustments use CPI-W, not CPI-U. Retirees who spend heavily on healthcare and housing — categories that consistently outpace CPI-W — have historically received COLAs that do not fully offset their actual cost increases. The BLS publishes an experimental index for Americans 62 and older (CPI-E) that shows this gap empirically.

Why the Fed uses PCE instead of CPI for policy decisions

The Federal Reserve adopted the Personal Consumption Expenditures (PCE) price index as its preferred inflation benchmark in 2000. The Fed’s 2% target — the number that determines whether interest rates rise or fall — refers to PCE, not CPI. There are three structural reasons:

1. Broader scope. PCE includes goods and services purchased on behalf of consumers — most importantly, employer-sponsored health insurance. This makes PCE a more complete picture of total consumption costs.
2. Chain-weighting. PCE uses contemporaneous spending data to update weights continuously, which better captures consumer substitution behavior. CPI-U uses fixed annual weights. This alone accounts for roughly 0.3-0.5 percentage points of the persistent PCE-CPI gap.
3. Historical bias. PCE has consistently run below CPI-U by 0.2-0.5 percentage points over rolling 10-year periods. The Fed believes PCE is a more accurate representation of underlying inflation trends, which matters when setting multi-year rate policy.

In practice, this means the Fed may hold rates steady even when headline CPI appears elevated, because the PCE measure it targets is running lower. When you see a financial commentator say “inflation is above the Fed’s 2% target,” they should specify whether they mean CPI or PCE — the difference can be policy-significant.

How to use CPI data practically

Understanding CPI construction matters for several real-world financial decisions:

• Salary negotiation: If your employer cites “3% CPI” as justification for a 3% raise, check whether the category sub-indices most relevant to your spending (housing, food, transportation) are running higher. They often are.
• Lease escalators: Commercial and some residential leases tie annual rent increases to CPI-U. C-CPI-U runs lower — if you have leverage to specify which index governs escalation, C-CPI-U is typically more favorable to the tenant.
• TIPS and I-bonds: Treasury Inflation-Protected Securities and Series I Savings Bonds both adjust using CPI-U All Items. You are hedging against the official index, which may understate your personal inflation rate if your spending skews toward housing, food, or healthcare.
• Retirement planning: Projecting purchasing power 20-30 years forward using the headline CPI rate may be optimistic. Category-level analysis of your expected retirement spending mix — heavily weighted toward healthcare and housing — may suggest a higher inflation assumption is prudent.

You can explore how CPI has eroded purchasing power over specific time periods using our free Inflation Calculator.

Frequently asked questions about the Consumer Price Index

What does the Consumer Price Index (CPI) measure?

The CPI-U measures the average change over time in prices paid by urban consumers for a representative basket of goods and services. The BLS tracks approximately 80,000 items across 23,000 retail outlets and service establishments in 75 urban areas. It is published monthly and is the most widely cited measure of inflation in the United States.

What are the 8 major categories in the CPI?

The BLS divides CPI into 8 major categories: Food and Beverages (13.5%), Housing (34.0%), Apparel (2.6%), Transportation (16.7%), Medical Care (8.2%), Recreation (5.7%), Education and Communication (5.9%), and Other Goods and Services (3.4%). Weights are derived from the Consumer Expenditure Survey and updated every January.

Why does the Fed use PCE instead of CPI for inflation policy?

The Federal Reserve targets 2% inflation using the Personal Consumption Expenditures (PCE) price index rather than CPI for three reasons: PCE has a broader scope (including employer-paid health insurance), uses chain-weighting to account for consumer substitution, and has historically run 0.2-0.5 percentage points below CPI, making it a more conservative measure of underlying inflation trends.

What does the CPI not include?

The CPI does not include asset prices (stocks, bonds, real estate appreciation), income taxes, Social Security deductions, or investment purchases. It also does not capture regional price variation — a household in San Francisco faces very different prices than one in rural Ohio, but both are measured against the same national CPI. Additionally, the CPI excludes prices paid by rural residents, farm families, and those in institutions.

What is hedonic pricing in the CPI?

Hedonic pricing is a BLS methodology that adjusts CPI for quality improvements in products. If a laptop costs the same as last year but has twice the processing speed, the BLS treats this as a price decrease because the consumer gets more value per dollar. Critics argue this adjustment — applied to electronics, vehicles, and healthcare — understates real-world inflation because consumers still pay the same or higher dollar amounts regardless of quality changes.

Data sources: U.S. Bureau of Labor Statistics, Handbook of Methods, Chapter 17 (Consumer Price Index), 2021 edition; BLS, Frequently Asked Questions about the Consumer Price Index; Bureau of Economic Analysis, PCE methodology documentation; Social Security Administration, COLA methodology. All CPI series IDs verified against accurate.software calculators. Analysis by the staff at accurate.software.