Future Schedule for Expenditure Weight Updates in the Consumer Price Index

The Bureau of Labor Statistics (BLS) announced today that it will be updating the consumption expenditure weights in the Consumer Price Index for all Urban Consumers (CPI-U) and the CPI for Urban Wage Earners and Clerical Workers (CPI-W) to the 1999-2000 period, effective with release of data for January 2002. The newer weights will replace the 1993-95 weights, which were first used in the index effective with January 1998 data. Additionally, CPI expenditure weights will be updated at two-year intervals subsequent to the 2002 updating. Thus, for example, CPI expenditure weights will be updated to the 2001-02 period effective with release of CPI data for January 2004. As a result of this change, expenditure weight data will be, on average, "two years old" when introduced into the CPI, and four years old when replaced. By contrast, the 1993-95 weights were, on average, 3½ years old in January 1998, and they replaced weights that were about 15 years old.

Historically, the introduction of a comprehensive new set of expenditure weights attached to the categories of goods and services in the CPI "market basket" has taken place in the context of the periodic major revisions of the index. Such major revisions have taken place approximately once each decade—in 1940, 1953, 1964, 1978, 1987 and, most recently, in 1998. The CPI-U and CPI-W expenditure weights are constructed using household spending patterns during specified base periods, as reported in the Consumer Expenditure Survey (CEX). Effective with data for January 1998, the CPI’s expenditure base period was updated from 1982-84 to 1993-95.

In June 1997, in a paper prepared for the Chairman of the Joint Economic Committee, the BLS said that it was considering a more frequent schedule of updates to follow the planned January 1998 update. In an August 1997 response to a General Accounting Office report, the BLS indicated that more frequent updates would be preferable, that the future schedule was under review, and that the decision would be based on a consideration of what frequency would yield the most accurate CPI and best support the many uses of the index. The review is now complete and is the basis for the policy announced above.

As the BLS has stated previously, the cost-of-living-index (COLI) provides the measurement objective for the CPI. The theory of the COLI, however, does not specify any particular expenditure base period as the appropriate one, nor does it specify the proper interval between updates of the base period expenditure patterns. Furthermore, the BLS does not view the choice of update frequency as a means of addressing the problem sometimes referred to as "upper-level substitution bias" in the CPI. Although it has sometimes been argued that using more current, and more frequently updated, expenditure weights would lower the index’s rate of growth by reflecting consumer response to changes in the relative prices of CPI item categories, there is little evidence of any historical link between the CPI’s growth rate and the age of its underlying expenditure weights. The BLS believes that consumer substitution in response to relative price change is better dealt with through the use of a "superlative" cost-of-living index formula. As previously announced, a superlative CPI index will be published in 2002 as a complement to the CPI-U and CPI-W.

In the BLS view, the goal in employing more current expenditure weights is to make the CPI reflect, as much as possible, the inflation currently experienced by consumers. More specifically, the use of current weights will help to ensure that the relative importance of CPI item categories, such as food away from home, college tuition, or medical care services more accurately reflects how consumers are allocating their spending. The CPI’s current item sample rotation procedures are similarly aimed at ensuring that the individual items priced in the CPI are representative of current purchases within the CPI item categories. The BLS also has initiatives underway that will expand the sample size of the CEX beginning in 1999, and that will enhance the computer systems used to introduce new expenditure weights into the CPI. Both initiatives have the purpose of reducing the average age of those weights.

Based on this overall objective of making the CPI representative of consumer experience, the BLS has decided to update the index’s expenditure weights every two years, beginning with the release of data for January 2002. This schedule will allow for the use of the expanded CEX sample mentioned above, and the consequent implementation of a two-year expenditure base period (1999-2000) rather than the three-year base period used in the 1987 and 1998 revisions. In the planned updating schedule, the 2001-02 expenditure weights will replace the 1999-2000 expenditure weights effective with the CPI for January 2004. As noted above, under the new updating schedule, CPI expenditure weights will be substantially more current, both at the time of their introduction and at the time they are replaced, than under the schedule previously followed.

As noted above, this decision is not intended or expected to have a large, systematic effect on the CPI’s rate of change. Nevertheless, to examine the quantitative impact of moving to a two-year update policy, the BLS has analyzed historical CPI-U data to estimate what the growth in the index would have been, had the new policy taken effect subsequent to the 1987 major revision. Specifically, the simulated policy included the introduction of 1986-87 expenditure patterns in January 1989, of 1988-89 patterns in January 1991, and similarly thereafter through the introduction of 1994-95 expenditure patterns in January 1997. The simulated increase from December 1988 through December 1997 under this policy was 31.9 percent, compared to 33.9 percent for the published CPI-U. On an average annual basis, the policy would have lowered the measured rate of CPI-U growth by 0.17 percentage point. It is very important to recognize, however, that this estimated historical effect may not be indicative of the future effect of the policy. The lower growth associated with the simulated policy is explained entirely by the replacement of 1982-84 weights by 1986-87 weights in 1989. The subsequent biennial updates had virtually no net effect; a simulated policy of maintaining 1986-87 weights from 1989 forward yielded a total increase in the CPI-U between December 1988 and December 1997 of 31.8 percent, 0.1 percentage point less than under the two-year updating policy. It is likely that incorporating new weights more frequently in the future will have a small upward effect on the index in some years, and a small downward effect in other years.

It should also be emphasized that the policy announced today does not mean that the full range of activities involved in a major CPI revision henceforth will occur every two years. In addition to updating expenditure weights, major revisions of the CPI have comprised updating the geographic (area) sample and sample of housing units, revising the item classification and publication structure, and introducing technological and methodological enhancements. (A description of the 1998 and previous revisions can be found in the December 1996 issue of the Monthly Labor Review.) Not all of these activities are feasible or advisable on a biennial basis. For example, fundamental reworkings of the index’s item classification and publication structures are both costly and potentially disruptive for users. Also, so long as the BLS continues to rely on decennial census data for selecting new CPI area and housing samples, it will be possible to update those samples only about once every ten years.

Bureau of Labor Statistics December 18, 1998

¹ Over these years, the expenditure base period for the CPI-U was 1982-84.

² The simulated biennial updates occur in January of 1989, 1991, 1993, 1995, and 1997, using expenditure base periods of 1986-87, 1988-89, 1990-91, 1992-93, and 1994-95, respectively.

Bureau of Labor Statistics December 18, 1998

Last Modified Date: October 16, 2001