Response Rates for the Consumer Price Indexes : U.S. Bureau of Labor Statistics

Bureau of Labor Statistics > Consumer Price Index > Data > Tables

Response Rates for the Consumer Price Indexes

CPI response rates: overview and methodology

Response rates are the percentage of the sample that responds to or participates in a survey. Any missing responses result in a loss of statistical information, which may increase the risk of nonresponse bias.

For the CPI, BLS calculates response rates for two surveys: the Commodities and Services (C&S) Survey and the Housing Survey. The response rates are unweighted and treat all data points or categories as equally important, regardless of their sample size or frequency. Additional information is available in the Calculation section of the CPI Handbook of Methods chapter.

Key definitions

Targeted unit: A unit belonging to the target population, selected and initiated into the sample, and targeted for regular price collection.
Collected unit: A unit from which an observable price was collected.
Collection rate: The number of collected units divided by the number of targeted units.
Unit used in estimation: A unit whose observable or imputed price is used in the calculation of the CPI.
Estimation rate: The number of units used in estimation divided by the number of targeted units. This includes some units with imputed prices.

Commodities and Services (C&S) Survey

Response rates are calculated at the data collection (or repricing) and data estimation phases. Rates are only calculated for probability samples. Some CPI data are obtained from alternative sources, such as records from purchase transactions, rather than from probability samples. Data collected from nonsurvey sources are excluded from response rate calculations.

The C&S survey provides response rates for two primary units:

Outlets are the establishments, such as websites or physical locations, from which prices are collected.
Quotes are observations of the price and other characteristics of specific consumer goods or services being priced.

**Table 1: C&S Survey response rate definitions**
Data collection stage	Definition	Treatment in estimation
Initiation	The attempt to gain an establishment’s initial agreement to participate.	Focuses on initial contact with the respondent; data are only collected and proceed to estimation if respondent agrees to participate and establishment meets entry-level item eligibility criteria.
Collection (repricing)	The collection of the same price quote in the same outlet each month to estimate price changes over time.	Data collectors contact the respondent to collect an observable price; data proceeds to estimation after review.
Estimation rate vs. collection rate	The denominator for both rates is the same. However, the estimation numerator is typically smaller because it only includes a subset of collected data.	Quotes that are rejected (such as for being non-comparable) are included in collection rates but excluded from estimation rates. Quotes that are estimated using carry forward imputation and substitution are included in estimation rates but excluded from collection rates.

A distinction between the collection rate and the estimation rate in the C&S Survey is how BLS handles missing data. While the collection rate tracks successful contact and response, the estimation rate includes carry forward imputed data and quotes undergoing substitution to ensure the index remains representative. Carry-forward imputation sets the current month’s price of a missing product or service to its previous month’s price. Home cell imputations and different cell imputations are not included in the estimation rate calculations. The denominator for both rates is the same: the total number of targeted outlets or quotes sent out for pricing in that period.

Housing survey (Shelter)

The Housing Survey measures price changes for rental units. Currently, no alternative data are used in these calculations; all figures are derived from probability samples.

Units are categorized below as the three outcomes used in the CPI Housing Survey estimation.

**Table 2: Housing Survey outcomes in estimation**
Data collection outcome	Definition	Treatment in estimation
Data reported	Rent price information was successfully obtained from the respondent.	The actual rent payments reported are used directly in calculations.
Vacant	The unit was located but found to be unoccupied at the time of the survey.	BLS applies a special imputation process to estimate current market rent. More information on this special imputation process is available in the Rent and Owner’s Equivalent Rent factsheet.
Other	The unit is targeted and occupied, but no data could be obtained (for example, no contact made or refusal).	Rent is imputed and used in estimation to ensure the weight of the unit still contributes to the overall housing index even if no price information is collected.

In total, the sum of the data reported, vacant, and other outcomes equals the total targeted sample used in estimation. This minimizes the loss of statistical representation when direct contact is not possible.

A distinction between the collection rate and the estimation rate in the Housing Survey is how BLS handles missing data. While the collection rate tracks successful contact, the estimation rate includes imputed data to ensure the index remains representative. The denominator for both rates is the same: the total number of targeted housing units sent out for pricing in that period.

The number of responding quotes plus the number of imputed quotes will not exactly equal 100 percent of quotes. The most common reason for this discrepancy is quotes that have been initiated but not yet priced.

Available data

Detailed annual XLSX files are released every year with the January news release published in February. Monthly XLSX files will be available on this page within a few business days of release. Monthly and annual average rates are also provided quarterly on the BLS household and establishment survey response rates page.

Methods BLS uses to maximize response rates

BLS uses several techniques to maintain sample sizes for estimating the CPI. Initial outlet sample sizes are larger than necessary to cover initial nonresponses for the following reasons: out-of-business, out-of-scope, refusal, sample items not available, and unable to locate. Different modes of collection are also used, such as telephone, email, video, and web collection at the request of survey participants.

Recent methodology changes to calculate response rates

In June 2025, BLS adopted a revised methodology to calculate response rates. This new methodology better accounts for the way BLS has steadily increased its use of alternatives to traditionally collected survey data over the last 15 years. BLS meticulously studied how alternative data sources would affect its indexes and variance estimates before incorporating alternative data into production.

Response rates using revised method

Note: A time series dating back to January 2019 is available using this revised method. These figures may differ slightly from the legacy data published prior to June 2025, both in the CPI annual response rates files and on the BLS household and establishment survey response rates page.

Response Rates Data

Archived Response Rates Data

Last Modified Date: June 11, 2026