This post was updated on May 12, 2022, to correct a minor error.
Inflation refers to changes over time in the overall level of prices of goods and services throughout the economy. The government measures inflation by comparing the current prices of a set of goods and services to previous prices. That turns out to be more complicated than it sounds. Here’s how inflation measures work.
What is the Consumer Price Index?
The Consumer Price Index (CPI), produced by the Bureau of Labor Statistics (BLS), is the most widely used measure of inflation. The primary CPI (CPI-U) is designed to measure price changes faced by urban consumers, who represent 93% of the U.S. population. It’s an average, though, and doesn’t reflect any particular consumer’s experience.
The CPI is constructed each month using 80,000 items in a fixed basket of goods and services representing what Americans buy in their everyday lives—from gasoline at the pump and apples at the grocery store to cable TV fees and doctor visits. The BLS uses a survey of American families called the Consumer Expenditures Survey to determine which items go in the basket and how much weight to assign to each item. Different prices are weighted according to how important they are to the average consumer. For instance, Americans spend more on chicken than tofu, so changes in the price of chicken have a greater impact on the CPI.
The federal government uses a version of the CPI—the CPI for Wage Earners and Clerical Workers—to adjust Social Security benefits for inflation.
How does the government get price data for the CPI?
The BLS collects price data each month by conducting two surveys: one records the prices of most goods and services, the other the price of housing. For most goods and services, BLS representatives visit (online or in person) or call various stores across the country and record what different items cost. During each trip, the data collector records the prices of the same goods and services as last month. Prices in New York, Los Angeles, and Chicago are collected every month, as are food and energy prices across the country. Prices for commodities from all other places (which tend to represent a smaller chunk of the overall basket) are updated every other month.
In the housing survey, the BLS collects the prices of 50,000 residences through personal visits or telephone calls. If a housing unit isn’t rented but is owned by the resident, they use what is called the owners’ equivalent rent: the BLS finds what it would cost the owner to rent a similar place and uses that as the price for housing instead. Since rentals do not change frequently, the rent of each unit is sampled every six months, allowing the BLS to survey more houses overall.
HOW ARE CHANGES IN THE CPI REPORTED?
The BLS reports the change in prices from one month to the next. The CPI rose 1.3% from May to June 2022, adjusted for the usual seasonal fluctuations, but didn’t change at all between June and July 2022. The CPI can be volatile from one month to the next, but the trend in the monthly change over several months can be an important indication of inflation. Another commonly used measure posted by the BLS and often reported in the press compares the CPI in one month to the same month a year earlier – in other words how much prices have risen over the past 12 months. In July 2022, for instance, the CPI stood 8.5% above the year earlier reading. This approach is less influenced by a month with a particularly small or large change but both ways of reporting changes in the CPI are accurate.
How are tax brackets adjusted for inflation?
Another version of the CPI called the Chained Consumer Price Index for All Urban Consumers has been used to adjust tax brackets for inflation instead of the primary CPI since Congress changed the law in 2017.
The primary CPI can overstate inflation because it prices the same basket of goods from one month to the next (although items are updated every two years), and it doesn’t take into account substitutions between similar goods. So, if a good (say, apples) becomes more expensive, and people choose to buy more of its substitutes (like peaches), the CPI calculates the price level as though people are still buying the same amounts of each item, just at a different price.
The chained CPI, however, takes into account the substitutions between similar items. It does this by updating its basket according to what people buy from one period to the next. Basically, the BLS calculates one measure of inflation using the basket from the first period, and another measure from the basket in the second period (which can have fewer apples and more peaches), and reports their average. This “chains” the impact of price changes across months, making the Chained CPI better at capturing consumer spending patterns and measuring the true impact of higher prices.
Because it factors in substitutions away from higher-priced items, inflation measured by the Chained CPI runs slightly lower than primary CPI. The effect can add up over time. Between 2000 and 2020, the CPI went up by 54.5%; the Chained CPI by 45.9%. Using the Chained CPI to inflation-adjust tax brackets means that tax bracket thresholds increase more slowly and Americans pay more in taxes over time than they would if the primary CPI was used. (For more on the Chained CPI, see The Hutchins Center Explains: The Chained CPI.)
What is the price index for Personal Consumption Expenditures (PCE)?
The price index for Personal Consumption Expenditures (the PCE price index) is another measure of inflation, this one produced by the Bureau of Economic Analysis (BEA) using data on prices from BLS. The PCE price index measures the change in prices for all consumption items, not just those paid for out-of-pocket by consumers. For example, the weight on health care in the PCE reflects what consumers pay out-of-pocket for premiums, deductibles, and copayments as well as the costs covered by employer-provided insurance, Medicare, and Medicaid. In the CPI, only the direct costs to consumers are reflected. This difference in scope means that the PCE deflator and the CPI have very different weights. For example, the weight on health care is 22% in the PCE index, but just 9% in the CPI. The weight on housing is 42% in the CPI, but just 23% in the PCE index. That means that a given increase in health care prices will affect the PCE index much more than it will affect the CPI.
The Fed uses the PCE price index as its main measure of inflation. Its long-run target for inflation is for the PCE price index to increase at an annual rate of 2% over time.
The PCE is also a chained index, while the primary CPI is not. This means that, like the chained CPI, the PCE is better at accounting for substitutions between similar items when one of them becomes more expensive. Because its formula uses updated data, the PCE is believed to be a more accurate reflection of price changes over time and across items. Over time, the two measures tend to show a similar pattern, but the PCE tends to increase between 2 and 3 tenths less than the CPI. For example, the CPI-U increased 1.7% per year, on average, from 2010 to 2020; the PCE price index increased 1.5% per year on average over this period.
What is core inflation?
Core inflation, whether the CPI or the PCE price index, is defined as the change in prices excluding food and energy prices, which tend to be volatile. While food and energy are, of course, major parts of any household’s budget, core inflation is often seen as a better indicator of the underlying pace of price changes.
Other approaches to discern the underlying inflation trend include the Cleveland Fed’s trimmed-mean CPI (which excludes the CPI components that show the most extreme monthly price changes) and median CPI (which reflects only the change in price in the center of the distribution of price changes). These measures get rid of “noise” by leaving out whichever prices behave erratically each month, rather than singling out food and energy prices every time.
So, compared to the trimmed-mean of median CPI, core CPI is more susceptible to temporary swings in prices that aren’t food or energy related. For example, when the price of used cars surged during the COVID-19 pandemic, rising 29.7% in May 2021 compared to the previous year, it pushed the core CPI up 3.8%. The trimmed-mean CPI, on the other hand, rose only 2.6% in the same month, showing that most items were only moderately more expensive.
How do price indexes account for quality change?
Accounting for quality change is one of the thorniest issues in price measurement. Tracking the price of the same good over time works well when the exact same good—a dozen large eggs, for instance—is sold from one period to the next. But often that isn’t the case. For example, new versions of the iPhone are introduced to the market on a regular basis. The iPhone someone buys today is of better quality than the one bought five years ago. In addition, entirely new goods are sometimes introduced that might make people much better off.
Economists think that the benefits of those goods should also be captured in price indexes so we can distinguish between price increases that reflect better quality and those that reflect true price inflation. The BLS has made some progress over time in accounting for quality—for example, for some goods they use “hedonics” to try to adjust prices for the value of new attributes—but price indexes still likely reflect a mixture of true inflation and quality improvements. This is particularly true because the sectors of the economy in which quality is particularly hard to measure, such as health care and education, are growing as a share of the overall economy.
For more on these issues, see “The measurement of output, prices, and productivity” by Brent Moulton and “GDP as a measure of economic well-being” by Karen Dynan and Louise Sheiner, both from the Hutchins Center Productivity Measurement Initiative.