Consumer Price Data

Consumer price index (CPI)

The CPI is used to measure the change in the out-of-pocket expenditures of all urban households for a particular set of goods and services. In terms of its coverage, the CPI measures the cost of spending made directly by households for the items in its basket, with the notable exception that it also includes a measure of the rents that homeowners implicitly pay instead of renting their home. The CPI is constructed by the Bureau of Labor Statistics and is released around the middle of each month, with a one-month publication lag. More detailed information on the CPI is available in this slideshow.

Personal consumption expenditures (PCE) price index

The PCE price index measures the change in the prices of goods and services consumed by all households and nonprofit institutions serving households. Compared with the CPI, the PCE price index has broader coverage because it includes spending made directly by or on behalf of households, and it includes a broader range of nonmarket prices for goods and services for which households receive some benefit. The PCE price index is constructed by the Bureau of Economic Analysis and is released toward the end of each month, with a one-month publication lag. More detailed information on the PCE index is available in this slideshow.

Differences between the CPI and the PCE price index

While the CPI and PCE price index both provide measures of how prices are changing over time, they are not constructed in the same way. One difference is the smaller number of items in the basket of the CPI. The CPI reflects out-of-pocket expenditures of all urban households, while the PCE price index also includes goods and services purchased on behalf of households. In the case of medical care outlays, for example, the PCE price index would not only include the out-of-pocket expenses paid for by households, but also the medical care services paid for by an employer and by the government. Another difference is the expenditure weights assigned to each of the CPI and PCE categories of items. Part of the reason for the different weights reflects the different coverage of the indexes—there are items in the PCE price index that are not in the CPI. In addition, the indexes use different data sources for the weights, with the weights in the PCE price index updated more frequently depending on changes in households’ spending patterns. There are other differences that include the source(s) used for an item’s price and seasonal adjustment procedures. Taken together, the differences in the two indexes result in CPI inflation readings that are generally higher than PCE inflation readings as shown in the chart for the 1995–2016 period.

To see more recent data and plot other inflation series, visit the Inflation Charting section.

The PCE price index has several advantages over the CPI that include its ability to capture the changing composition of spending that is more consistent with consumer behavior (including consumers’ substitution toward relatively cheaper items), as well as weights that are based on a more comprehensive measure of expenditure. One drawback of the PCE price index, however, is that it can be substantially revised, while the (non-seasonally adjusted) CPI is never revised. This could give an edge to the CPI for some purposes (e.g., contract indexation) and also explains its use for Treasury Inflation-Protected Securities (TIPS).

A more detailed discussion of the differences between the CPI and PCE price index can be found in this Cleveland Fed Economic Trends.

Measures of underlying inflation

Inflation rates tend to exhibit both temporary and persistent movements. Because economists often are interested in the persistent movements in inflation, they have proposed a number of measures to capture this particular component, which can be thought of as the inflation trend or underlying inflation. Three of the best-known measures of underlying inflation are median inflation, trimmed mean inflation, and core inflation (all items excluding food and energy). All start with an aggregate price index such as the CPI or PCE price index, and then adjustments are made to remove transitory changes in the index (“noise”), so that a measure of the underlying (persistent) component of inflation can be calculated.

The median and trimmed mean are based on the same notion that the source of the noise in the price data is the lowest and highest price changes in the basket. The difference between the median and trimmed mean is the location of the cutoffs for the lowest and highest price changes to be excluded from the index. The core measure is based on the idea that the source of the noise in the price data is associated with particular items.

• Medians eliminate all price changes in an index except “the one in the middle.” The median price change is found by ranking all the price changes of the goods and services components in the index from smallest to largest, accumulating the weight of each item (the share of total expenditures it accounts for), and then selecting the price change of the item whose expenditure share is the 50th percentile.
• Trimmed means are similar in that they rank the price changes in the index in ascending order and then remove some portion of the highest and lowest price changes using a cut-off percentage based on the expenditure weights. Any percentage cut-off can be specified, and the goal is to find the percentages that eliminate as much noise as possible while retaining enough of the price change data to be informative about the underlying inflation rate. Once the cut-off percentages are determined, the trimmed mean is constructed as a weighted average of the remaining price changes wherein original weights are rescaled to sum to 100.
• Core measures remove specific components from the price index, usually those that appear in the food and energy categories (although in other countries other items are removed, and there are some modest differences in the way that core CPI and core PCE remove food categories in the United States). The core price index is then constructed as a reweighted price index using a similar approach to that for the trimmed mean—the expenditure weights of the remaining items are rescaled to sum to 100.

Underlying measures of CPI inflation

• Median CPI
  The median CPI is constructed monthly by the Federal Reserve Bank of Cleveland. The median ranks all the price changes of goods and services components of the CPI from smallest to largest, and then selects the price change of the item whose expenditure weight is the 50th percentile. Here is more detailed information on the median CPI.
• Trimmed-Mean CPI
  The 16% trimmed-mean CPI is constructed on a monthly basis by the Federal Reserve Bank of Cleveland. Like the median CPI, the 16% trimmed-mean CPI ranks all the price changes of goods and services components of the CPI from smallest to largest, but then removes price changes whose expenditure shares fall below the 8th percentile and above the 92nd percentile. Here is more information on the trimmed mean CPI.
• Core CPI (excluding food and energy)
  The core CPI is constructed on a monthly basis by the Bureau of Labor Statistics. In contrast to the median and trimmed mean measures of underlying inflation, the core CPI associates the noise in an aggregate price index with particular items—food and energy. The core CPI excludes food and energy prices from the CPI.

Underlying measures of PCE inflation

The underlying measures of CPI inflation have analogs for PCE inflation.

• Median PCE Inflation
  Median PCE inflation is constructed monthly by the Federal Reserve Bank of Cleveland. The median ranks all the price changes of goods and services components of the PCE price index from smallest to largest, and then selects the price change of the item whose expenditure weight is the 50th percentile. Here is more detailed information on median PCE inflation.
• Trimmed Mean PCE
  The trimmed mean PCE inflation rate is constructed on a monthly basis by the Federal Reserve Bank of Dallas. The trimmed mean ranks all the price changes of goods and services components of the PCE from smallest to largest, then removes price changes whose expenditure shares fall below the 24th percentile and above the 69th percentile, and then calculates a reweighted average of the remaining price changes. Compared with its CPI counterpart, the trimmed mean PCE inflation rate involves trimming higher proportions of price changes and trimming an unequal (asymmetric) proportion from the highest and lowest price changes. A more detailed discussion of the trimmed mean PCE inflation rate can be found here.
• Core PCE (excluding food and energy)
  The core PCE price index is constructed on a monthly basis by the Bureau of Economic Analysis. Conceptually similar to the core CPI, the core PCE index excludes food and energy prices from the PCE index. However, while core CPI excludes all food, core PCE only excludes food purchased for consumption at home; food services (that is, food purchased for off-premises consumption) are included in core PCE.

Other measures of consumer price inflation

There are other measures of consumer price inflation that either provide alternatives to or a different focus than those previously discussed.

• Federal Reserve Bank of New York Underlying Inflation Gauge (UIG) provides two underlying measures of CPI inflation on a monthly basis. The UIG measures are derived using a dynamic factor model—a statistical technique that can be applied to large information sets. One measure—the “prices-only” UIG—is estimated from a large number of disaggregated CPI series. The other measure—the “full data set” UIG—incorporates additional macroeconomic and financial variables. A more detailed discussion of the UIG measures is provided here.
• Federal Reserve Bank of Atlanta Sticky and Flexible CPI recognizes that the frequency of prices changes of all goods and services is not the same. That is, the prices of some items change more frequently and are “flexible” compared with prices of items that change less often and are “sticky.” The differences in price-setting behavior across these types of items can have implications for inflation. Because sticky prices usually change infrequently, they may be more likely to respond to persistent factors driving inflation, whereas flexible prices may be more likely to respond to transitory factors. A more detailed discussion of these two types of prices and the implications for forecasting inflation is discussed here.