Consumer Price Index (CPI) geographic revision for 2028

To accurately reflect changing U.S. population patterns, the Bureau of Labor Statistics (BLS) continuously updates the geographic sample underlying the Consumer Price Index (CPI). Reflecting population data derived from the decennial census, BLS redraws a geographic sample roughly once every 10 years shortly after the completion of the decennial census. During the inter-decennial period, BLS then transitions CPI survey operations from the older geographic sample to the newly drawn geographic sample in a phased approach.

With the release of January 2028 indexes in February 2028, BLS will introduce the geographic sample reflecting U.S. populations as reported by the 2020 Decennial Census. This sample design will replace the design based on the 2010 Decennial Census, introduced in 2018.

Population

The target population of the CPI remains all urban consumers excluding those living in rural nonmetropolitan areas, on military installations, farm households, religious communities, and in institutions such as prisons and mental hospitals. For more information, see the CPI Handbook of Methods.

Geographic sampling unit

As with the 2010 decennial CPI sample, BLS chose the core-based statistical area (CBSA) concept as the geographical primary sampling unit (PSU) of the 2020 decennial design. The CBSA is a geographic region made up of one or more counties or county equivalents containing a core urban area with a population of at least 10,000, plus adjacent counties that have strong social and economic ties through commuting. Use of the CBSA creates a national standard for collecting and publishing federal statistics. BLS used the CBSAs as delineated under the Office of Management and Budget (OMB) July 2023 Bulletin No. 23-01. There were 925 CBSAs in the U.S., 387 metropolitan areas (with an urban core of at least a population of 50,000), and 538 micropolitan areas (with an urban core of at least a population of 10,000).

Number of PSUs

The total number of PSUs remains unchanged at 75. This represents the required number of PSUs needed to adequately represent the target population with available resources.

Sample design

Each of the 925 CBSAs are grouped into homogeneous geographic strata, and one CBSA is selected to represent each group. To keep sample transition costs low, CBSAs previously selected based on the 2010 decennial sample were retained where possible. Each selected CBSA is weighted to represent the population of its geographic stratum. Strata are divided into self-representing (only one CBSA in the group) and non-self-representing (more than one CBSA in the group).

• Self-representing strata represent the largest metropolitan areas in the country, which are distinct as retail markets based on size. The population cut-off for self-representing strata under the 2020 design is 2.8 million, compared to 2.5 million under the 2010 design. This results in 21 self-representing strata for the 2020 design; the same list of 21 PSUs selected for the 2010 design (see Appendix 1).
  • Urban Alaska and Urban Hawaii are special cases. The four CBSAs in Alaska (Anchorage, Fairbanks, Juneau, and Ketchikan) and the four CBSAs in Hawaii (Honolulu, Hilo-Kailua, Kahului-Wailuku, and Kapaa) are each grouped into urban state strata, based on isolation of their economic markets from the 48 contiguous states. For survey operational reasons, Anchorage and Honolulu are selected with certainty to represent their respective urban state stratum and are weighted accordingly. These selections remain unchanged from the 2018 CPI revision.
• Non-self-representing strata are groupings of the remaining 902 CBSAs. These CBSAs are stratified into 52 homogenous groupings, with groupings constrained by the nine U.S. Census Divisions geographical boundaries. To determine the homogenous groupings within each Census Division, regression models were analyzed to identify socioeconomic and demographic variables explanatory of consumer spending and correlated with price change for select CPI components. The final stratification model for the 2020 design had four explanatory variables:
  • Percent of population age 15+ with income >$75,000 in 2017 dollars
  • Percent of households with a computer
  • Percent of population age 25+ with at least a bachelor’s degree
  • Percent of population classified as urban

For the 2020 design, BLS used a novel sample selection feature for non-self-representing CBSAs aimed at yielding survey efficiencies. Sample selection for non-self-representing strata occurred in two stages. In the first stage, 38 larger CBSAs (population greater than 200,000) were selected. In the second stage, 14 smaller CBSAs (population less than 200,000) were selected, with eligibility conditioned on being conterminous (geographically adjacent) to selected stage 1 CBSAs. This two-stage process helped prevent the selection of small cities located in remote areas and isolated from other cities in the sample. By clustering the sample this way, BLS expects to more efficiently manage in-person survey data collection.

Sample transition

Effective with January 2028 indexes published in February 2028, BLS will calculate and publish CPI data aggregated from basic indexes (the first stage of CPI estimation, for more information see the CPI Handbook of Methods, calculation section) with areas as defined by the 2020 geographic sample. The set of basic areas remains unchanged from the 2010 sample design at 32 areas: 21 self-representing PSUs, Urban Alaska, Urban Hawaii, and 9 non-self-representing Census Division areas. The sample of 52 CBSAs defining each of the 9 non-self-representing Census Division areas, however, will transition over a period of 3 waves: 38 CBSAs continue, 14 existing CBSAs will drop, and 14 new CBSAs will be added. See Appendix 1 for a list of continuing and new CBSAs, and Appendix 2 for a list of dropping CBSAs.

• Timeline for wave 1. Wave 1 contains 6 new CBSAs and 11 dropping CBSAs. The new CBSAs are:
  • Chambersburg, PA
  • Green Bay, WI
  • Indianapolis, IN
  • Pittsfield, MA
  • Punta Gorda, FL
  • Sacramento, CA

Survey operations for the wave 1 new CBSAs commences with initiation into the Housing Survey starting October 2026 and initiation into the Commodities and Services (C&S) Pricing Survey in February 2027. Repricing activities for these samples will occur later in 2027. Data from new wave 1 CBSAs will be used in CPI estimation beginning with the January 2028 index. Data collection will stop in the dropping CBSAs after December 2027.

• Timelines for wave 2 and wave 3. It is expected the wave 2 and wave 3 PSUs will transition in 2029 and 2030, respectively, subject to resource availability. Starting in January 2028, prices for the new wave 2 and wave 3 CBSAs will be represented in index calculations by data collected in wave 2 and wave 3 dropping CBSAs, or by alternative imputation methods, as indicated in Appendix 2.

Last modified date: November 2, 2026