Contents

• Background
• Estimating the CPI
• Medical Care
• Home Ownership
• The "Core" CPI
• Improving the CPI
• Local Area CPIs
• CPI Data Via the Internet
• Historical Data
• CPI Calculations

Summary

The Consumer Price Index (CPI) is perhaps the most widely reported measure of inflation. A number of federal government programs are regularly adjusted to account for changes in the CPI, such as Social Security benefits and the personal income tax rate schedule. Thus, the behavior of the CPI has important consequences for a large number of people. Many, however, may be unfamiliar with how the CPI is estimated.

For Congress, the CPI is of particular interest because of its significant effect on the federal budget. Changes in the CPI can have substantial effects on both revenues and outlays, and those changes may either reflect underlying economic conditions or result from methodological changes in the way the CPI is calculated.

The CPI is based on a number of sample surveys. One of these surveys estimates the purchasing patterns of the "typical" household to determine how that household spends its money. Another survey determines where those households shop, and a third survey collects prices on the goods and services purchased by those households.

The CPI measures the price level relative to a particular period. Currently, the CPI number for each month is a measure of the price level relative to what it was between 1982 and 1984. The CPI is available for a number of metropolitan areas but it does not allow comparisons of the cost of living in different cities.

---

The Consumer Price Index: A Brief Overview

The Consumer Price Index (CPI) is probably the most widely used measure of inflation. A number of federal government programs, such as Social Security benefits and civil service retirement, are tied to increases in the CPI. In addition, the personal income tax rate schedule is indexed to the CPI.¹ Economists use the CPI to calculate constant-dollar estimates of other economic indicators, such as retail sales and hourly earnings, which allow analysis of changes in these variables excluding the effect of changes in the price level. Each year, the CPI is used to update the income levels that determine the poverty rate. Periodic increases in many union wage and other contracts are also tied to increases in the CPI. Thus, the behavior of the CPI has major consequences for a significant portion of the population; but many may be unfamiliar with the details of its calculation.

Background

The CPI is published by the Department of Labor's Bureau of Labor Statistics (BLS).² There is no specific legislation authorizing or requiring BLS to calculate and publish the CPI. Neither has legislation ever been enacted to require BLS to adopt any particular methodology in calculating the CPI. When the Bureau of Labor was first created in 1888, its task was, among other duties, to "acquire and diffuse among the people of the United States useful information on subjects connected with labor, in the most general and comprehensive sense of the word.... "³

In 1913, the Bureau of Labor was transferred to the newly created Department of Labor and renamed the Bureau of Labor Statistics. BLS was given slightly more specific instructions:

The Bureau of Labor Statistics, under the direction of the Secretary of Labor, shall collect, collate, and report at least once each year, or oftener if necessary, full and complete statistics of the conditions of labor and the products and distribution of the products of the same ... and said Secretary of Labor may collate, arrange, and publish such statistical information so obtained in such manner as to him may seem wise.⁴

Two CPIs are published by the BLS, the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W), and the Consumer Price Index for All Urban Consumers (CPI-U). The CPI-W is based on the purchasing patterns of only those in the population who earn more than half of their income from clerical or wage occupations, and were employed at least 37 weeks in the previous year.⁵ The CPI-W population makes up about 32% of the total population. Prior to 1978, the CPI-W was the only CPI published. Beginning in 1978, the CPI-U was introduced so that a broader share of the population would be represented in estimates of changes in the price level. The CPI-U is based on the expenditure patterns of all urban consumers and covers about 87% of the population. The CPI-U is usually the more publicized of the two price indexes.

Although the CPI-U and CPI-W are slightly different indexes, the numerical difference between the two measures is typically small. Between 1989 and 2009, the CPI-U increased, overall, by 73.0% compared with an increase of 71.0% for the CPI-W. Over the same period, that translates into an average annual rate of change of 2.8% for the CPI-U and 2.7% for the CPI-W.

Both the CPI-W and the CPI-U are used for inflation indexing by the federal government.⁶ One advantage to using the CPI in indexing is that the CPI is rarely revised.⁷ Definitions, the index base year, the goods and services accounted for, and the methodology used to calculate the CPI may change from time to time, but, once published, the actual index number is final.⁸ Using other measures of the price level, such as one of the price indexes associated with Gross Domestic Product, for indexing purposes poses the problem of which number to use, the preliminary estimate or one of many subsequent revisions.⁹

Since the release of January 2007 data, the CPI and all of its component indexes have been published rounded to three decimal places rather than one. The change was not meant to imply any increase in the accuracy of the CPI. Instead it is being done to maintain precision in published estimates of percentage changes. Prior to 2007, when BLS published the CPI rounded to only one decimal place, it based published figures of percentage change in the CPI on those rounded numbers so that they could be replicated by users of published CPI data. But doing that meant that some precision in the published percent change data was lost.¹⁰ The effect of the change was expected to be small.¹¹

Estimating the CPI

Both CPIs are based on retail market prices. These prices, for more than 80,000 separate items, are collected in 87 urban areas across the country from thousands of outlets, such as grocery and department stores, gasoline service stations, and hospitals, among others.¹² BLS selects these retail establishments based on a survey showing where people do their shopping. Actual prices (except those for food) are not published because they are collected on a confidential basis. Price indexes are available in considerable detail. Examples of items for which CPI data are available include white bread, men's shirts, automobile tires, haircuts, funerals, automobile repair, and bedroom furniture.

The "all-items" CPI is the index most often referred to and it is a composite index, a weighted average, based on the indexes for all of the goods and services whose prices are collected.

The all-items CPI measures the price change of a fixed market basket of goods and services over time. The mix of goods and services making up the market basket is based on spending patterns established by the Consumer Expenditure Survey (CES).¹³ Based on the CES, weights are assigned to each of the goods and services that make up the market basket. These weights determine how much the price change for a given good will affect the all-items measure. For any given interval, the total price change, as measured by the all-items CPI, is the weighted average of the price changes of all of the components. With the release of data for January 2010, the CPI market basket is based on purchasing patterns described by the CES in 2007 and 2008. BLS updates the expenditure weights every two years.

Table 1 shows the major expenditure categories included in the CPI and their relative importance in the CPI-U as of December 2009. Relative importance reflects both the expenditure weights and changes in relative prices. Either a larger expenditure weight, or an increase in the price of a good relative to prices for other goods may cause the relative importance to increase, and vice versa.

Medical Care

Based on the measures of relative importance shown here, some might be concerned that medical care costs have too small a weight in the all-items index. In particular, the elderly typically spend a relatively larger share of their outlays on medical care. In calculating the CPI, however, the share of the market basket accounted for by medical care is based on "out-of-pocket" costs. This includes direct out-of pocket costs for medical care as well as indirect out-of-pocket costs for health insurance.¹⁴ An increasing share of medical costs are paid for by employers and government, so that out-of-pocket expenses on medical care are not as great as total outlays on medical care.

It is also important to appreciate that there is considerable variation among consumers (and among elderly consumers as well) in the demand for medical care, and the relative importance of medical care in the CPI is based on an average. No single price index can accurately describe the inflation experience of every single person. Different population groups (e.g., the elderly) tend to have different purchasing patterns, and individuals' purchases vary significantly within those groups. Although many elderly may spend more on medical care than is taken into account in the CPI, there are also some elderly who spend less. For those who spend less, if medical care costs rise more rapidly than do the prices of other goods and services, the CPI will tend to overstate increases in the cost of living, other things being equal. BLS is currently investigating the behavior of an experimental CPI for the elderly population.¹⁵ This experimental CPI for Americans aged 62 and older rose by an average of 2.9% per year between December 1989 and December 2009. During the same period, both the CPI-U and CPI-W rose at an annual rate of 2.7%.¹⁶

Considerable effort is made to ensure that the CPI is a meaningful, reliable measure of changes in the price level. But it does not necessarily reflect the inflation experience of each individual consumer. To the extent that individuals spend relatively more on those goods and services whose prices are rising faster than average, they may experience a higher inflation rate than that measured by the CPI. Similarly, those who spend relatively less on goods and services whose prices are rising faster than average, may experience a lower inflation rate than that measured by the CPI.

If purchasing patterns change significantly, then in the short run the CPI may tend to overstate the inflation rate. The CPI is a fixed-weight index and does not immediately take into account changes in spending patterns due to changes in relative prices. There may also be a tendency for the CPI to overstate the inflation rate because some price increases reflect improvements in the quality of goods and services. Taking quality changes into account in a price index is difficult, but BLS does attempt to make some adjustments to the CPI for quality improvements in a number of areas, including automobiles, apparel, and a number of consumer electronic goods, personal computers in particular.¹⁷

Home Ownership

Home ownership costs in the CPI are treated in a special way.¹⁸ Prior to 1983, the home ownership component of the CPI measured changes in the cost of purchasing a new home. Since 1983 for the CPI-U and 1985 for the CPI-W, changes in the cost of home ownership have been based on the concept of "rental equivalence." Rather than measuring changes in the cost of buying a house in each period, which would include finance charges, the CPI attempts to estimate the rental value of owner-occupied housing. Thus, the CPI measures changes in the consumption aspect of housing costs and not changes in the investment value of owner-occupied housing.¹⁹

The "Core" CPI

In addition to the all-items CPIs, BLS also publishes an index which excludes the effects of both food and energy prices. Increases in the CPI less food and energy are also referred to as the "core" inflation rate. The reason for excluding food and energy prices is that, at times, they may be more volatile than other prices and thus mask the underlying trend rate of inflation.²⁰ That can be useful for policymakers to whom inflation is an important variable. The "core" rate is often mentioned prominently in press reports covering the monthly releases of CPI data. This seems to be an occasional source of confusion, leading some to conclude that food and energy prices are not taken into account in calculating Social Security cost-of-living adjustments.²¹ In fact, food and energy prices are taken into account. The all-items CPI is used to make cost-of-living adjustments, not the core index.

Improving the CPI

In December 1996, a special commission chaired by economist Michael Boskin reported to the Senate Finance Committee that the CPI tended to overstate the actual rate of inflation by about 1.1% per year.²² Although a number of specific recommendations were made in the report, Congress took no legislative steps to require any changes in the way BLS calculates the CPI.²³

But the methodology of calculating the CPI has changed much since it was first published and is likely to continue to do so. BLS continues to look at methods that might lead to a more accurate measure of the cost of living. As part of that process, with the release of data for July 2002, BLS introduced an alternative CPI that makes use of "chain-weights." This index is referred to as the C-CPI-U. The expenditure weights for the C-CPI-U are updated more frequently than either the CPI-U or the CPI-W, and the index itself is subject to revision. The C-CPI-U has not replaced either the CPI-U or the CPI-W, and they will continue to be used for indexing.²⁴

Local Area CPIs

The CPI is currently published for 26 metropolitan areas. For most of these cities, however, indexes are not published on a monthly basis. These metropolitan area indexes are only intended to compare inflation rates between cities. The metropolitan area CPIs may not be used to compare the actual cost of living between cities.²⁵ Table 2 shows the metropolitan areas for which the CPI is published as well as the publication frequency.

CPI Data Via the Internet

BLS has set up an Internet home page for the CPI where visitors can get a copy of the most recent CPI press release, as well as up-to-date information regarding the CPI program. The Internet address for this page is http://www.bls.gov/cpi/.

Detailed CPI data are also readily available. BLS has set up a number of ways on their website, http://www.bls.gov/cpi/#data, to obtain CPI data. By making selections from each of a succession of menus, users of BLS's website may specify the particular data they want. There is also a link to an "inflation calculator," allowing users to make their own inflation adjustments to dollar amounts.

Historical Data

BLS has published estimates of the CPI going back as far as 1800, which makes it the longest, continuous price index series available. These data are shown in Table 3. Data for 1800 through 1912 were derived by splicing price indexes collected in three separate, nongovernmental, studies. Prior to 1978, there was only one CPI available. For 1978 and after, the data in Table 3 correspond to the CPI-U. Given that the accuracy of the CPI remains a topic of discussion, the data for the early years of this series should be considered to be fairly crude estimates.

The long series allows a noteworthy observation. Between 1800 and 1945, there was no long-term trend in the price level. During that period prices tended to fall as often as they rose. Since 1945, however, there has been a decided upward trend in the level of prices with the price level having fallen in only a few years. The decline in 2009 was the first since 1955.

The figures in Table 3 show year-over-year changes in the CPI, and the index numbers for each year are the average of the monthly data for that year. Changes in the CPI are sometimes published on a December-over-December basis. That number can vary significantly from the year-over-year number depending on how the monthly changes are distributed over the period. In 2009, for example, the year-over-year change was -0.4% whereas the change from December 2008 to December 2009 was 2.7%.

CPI Calculations

The CPI is an indicator of changes in the price level. At present, those changes are expressed relative to the average level of prices in the years 1982, 1983, and 1984. Thus, the average of all of the monthly CPI numbers for those three years is equal to 100. Determining the change in consumer prices between any two years is a simple percent change calculation using the formula:

percentage change in the CPI = (( CPI₂ ÷ CPI₁ ) - 1 ) x 100.

For example, in 2009, the CPI-U was 214.537, and in 1989 the CPI-U was 124.0. The total percentage change in the CPI-U between 1989 and 2009 was:

(( 214.537 ÷ 124.0 ) - 1 ) x 100 = 73.0 percent.

Calculating the percentage change between any two years at an annual rate is slightly more complicated, and requires the formula:

annual rate of change in the CPI = (( CPI₂ ÷ CPI₁ )^1/n - 1 ) x 100,

where n is the number of years covered in the interval. To calculate the average annual rate of change in the CPI-U between 1989 and 2009, we use an n of 20 and the same CPI-U values as in the previous example. The average annual rate of change in the CPI-U between 1989 and 2009 was:

(( 214.537 ÷ 124.0 )^1/20 - 1 ) x 100 = 2.8 percent.

Another common use of the CPI is to adjust dollar amounts for inflation, so that amounts from different years can be compared in terms of dollars of the same purchasing power. Suppose the question is how much money would have been required in 2009 to buy the same quantity of goods and services as $100 bought in 1989. To get such an estimate, the 1989 dollar value needs to be adjusted to account for the increase in consumer prices between 1989 and 2009. That requires the formula:

equivalent purchasing power in period 2 = ( CPI₂ ÷ CPI₁ ) x dollar amount in period 1.

For example, using the same CPI values as in the above examples, the equivalent purchasing power in 2009 of $100 in 1989 is:

( 214.537 ÷ 124.0 ) x $100 = $173.01.

This same calculation can be reversed to find the purchasing power in an earlier period of a dollar amount of a more recent vintage. To do this use the formula:

equivalent purchasing power in period 1 = ( CPI₁ ÷ CPI₂ ) x dollar amount in period 2.

For example, using the same CPI values as in the above examples, the equivalent purchasing power in 1989 of $100 in 2009 is:

( 124.0 ÷ 214.537 ) x $100 = $57.80.

Using these formulae, dollar values of constant purchasing power can be compared for any two periods for which CPI data are available. Constant dollar values are always compared in terms of the dollar's purchasing power in a particular year, known as the base year. When comparing dollars of constant purchasing power, it is important to specify the base year.

Footnotes

1.	See CRS Report RL34168, Automatic Cost of Living Adjustments: Some Economic and Practical Considerations, by [author name scrubbed].
2.	BLS began periodic publication of a National CPI in 1921. At first, the CPI was only released semi-annually. Quarterly releases were begun in 1935, and monthly releases began in 1940.
3.	29 U.S.C. 1.
4.	29 U.S.C. 2.
5.	Specifically, clerical workers, craft workers, operatives, service workers, or laborers.
6.	The CPI-U is the index used to adjust various provisions of the personal income tax code. The CPI-W is the index used to adjust Social Security benefits.
7.	Historically, BLS has published monthly changes in the CPI on both a seasonally adjusted and not seasonally adjusted basis. The seasonally adjusted changes are subject to revision because seasonal adjustment factors may change over time. BLS also makes available a seasonally adjusted index number that is subject to revision. Seasonal adjustment does not affect yearly changes because the adjustment factors cancel each other out over any given 12-month period.
8.	An exception to this occurred in 2000. A mathematical error was found in the way the CPI was calculated. As a result, numbers that had already been published were revised.
9.	See U.S. Department of Labor, Bureau of Labor Statistics, Using the Consumer Price Index for Escalation, available at http://www.bls.gov/cpi/cpi1998d.htm.
10.	See Elliot Williams, "The Effects of Rounding on the Consumer Price Index," Monthly Labor Review, October 2006.
11.	See additional info at the BLS website, at http://www.bls.gov/cpi/cpithreedec.htm.
12.	See "In the Field With the Price Indexers," New York Times, June 20, 1995, p. D2.
13.	Detailed information on the CES is available on the Internet at http://www.bls.gov/cex/.
14.	See U.S. Department of Labor, Bureau of Labor Statistics, "Measuring Price Change for Medical Care in the CPI," at http://www.bls.gov/cpi/cpifact4.htm.
15.	See U.S. Department of Labor, Bureau of Labor Statistics, "Experimental Consumer Price Index for Americans 62 Years of Age and Older, 1998-2003," http://www.bls.gov/cpi/cpiexpcpie2005.pdf.
16.	See CRS Report RS20060, A Separate Consumer Price Index for the Elderly?, by [author name scrubbed].
17.	See U.S. Department of Labor, Bureau of Labor Statistics, "How BLS Measures Price Change for Personal Computers and Peripheral Equipment in the Consumer Price Index," http://www.bls.gov/cpi/cpifaccomp.htm.
18.	See U.S. Department of Labor, Bureau of Labor Statistics, "Consumer Price Indexes for Rent and Rental Equivalence," at http://www.bls.gov/cpi/cpifact6.htm.
19.	Because of the timing of this change in the way the CPI is calculated, a permanent upward bias in the level of the CPI was introduced. BLS has published estimates of the CPI-U using the rental equivalence approach for the years 1967 to 1982. This series is known as the CPI-U-X1. The Census Bureau now publishes historical estimates of real money income based on the CPI-U-X1 measure. See U.S. Bureau of the Census, Money Income of Households, Families, and Persons in the United States, Current Population Reports, Series P-60, no.174, 1991, pp. 8-9.
20.	See CRS Report RS22705, Inflation: Core vs. Headline, by [author name scrubbed].
21.	John S. Greenlees and Robert McClelland, "Addressing Misconceptions About the Consumer Price Index," Monthly Labor Review, August 2008.
22.	U.S. Congress, Senate Committee on Finance, Final Report of the Advisory Commission to Study the Consumer Price Index, committee print, 104th Cong., 2nd sess., December 1996. For a discussion of methodological changes implemented after the Boskin report, see David S. Johnson, et. al., "Price Measurement in the United States: A Decade After the Boskin Report," Monthly Labor Review, May 2006.
23.	Because of the CPI's budget implications, Congress has, from time to time, taken some interest in how it is calculated. To date, however, no legislation has been enacted that would directly affect BLS' methodology. See David S. Johnson, Stephen B. Reed, and Kenneth J. Stewart, "Price Measurement in the United States: A Decade After the Boskin Report," Monthly Labor Review, May 2006, pp. 10-19.
24.	See CRS Report RL32293, The Chained Consumer Price Index: How Is It Different, and Would It Be Appropriate for Cost-of-Living Adjustments? , by [author name scrubbed].
25.	The CPI expresses the price level relative to a particular period of time. Currently, the CPI reflects the level of prices relative to the 1982-1984 period. Thus, the average index value during that period for each metropolitan area will be 100, no matter how expensive any area may be in which to live. That does not mean that each of these metropolitan areas is equally expensive. See CRS Report RS20942, Adjusting Federal Benefits for Geographic Differences in the Cost of Living, by [author name scrubbed].