Effects of COVID-19 Pandemic and Response on the Consumer Price Index

The Consumer Price Index (CPI) is a measure of the average change over time in the prices paid by urban consumers for a market basket of consumer goods and services. If you have questions, contact the CPI staff.

Consumer Price Index COVID-19 Impact Summaries

Each month from March 2020 to January 2022, BLS published a monthly summary of the impact of the COVID-19 pandemic on Consumer Price Index data. The monthly impact summaries were discontinued effective with publication of February 2022 data in March 2022. The previously published files will remain on the website for customer use.

Beginning in March 2022, the data are available in various formats from our CPI tables page. Note that these data will no longer be directly related to the COVID-19 pandemic. Information on standard errors is available in news release tables 6 and 7 on the CPI supplemental files page.

Questions and Answers

How are prices collected for the CPI? Price data used to calculate the CPI are primarily provided by two different surveys that are administered continuously each month:

Commodities and Services Pricing Survey, an establishment survey of businesses selling goods and services to consumers, used to provide the price data for the CPI.
Housing Survey, a survey of landlords and tenants used to provide rent data for CPI’s shelter indexes.

Survey operations for CPI pricing surveys may be affected by limitations on data-collection staff, the availability of survey respondents, and the availability of items. Note that CPI data are collected throughout the entire month. Specifically, any given price in the CPI sample is collected in one of three defined pricing periods, corresponding roughly to the first 10, second 10, and final 10 days of the month. BLS uses several data-collection modes for CPI surveys that include telephone, internet, and automated electronic data capture. However, the majority of data are collected by personal visit. About 65 percent of CPI price data and 50 percent of CPI rent data are typically collected by personal visit. This type of collection has been suspended since March 16, 2020. (It was suspended on March 5th in the Seattle area.)

What happens if BLS cannot collect CPI data? The percentage of prices in the CPI sample that may be unavailable, either because the outlet is closed or the item is out of stock, is expected to increase. When BLS cannot obtain a price either because of data-collection limitations or the item being unavailable, it will generally be considered “temporarily unavailable.” The CPI program has specific procedures for handling temporarily unavailable prices. Missing prices are generally imputed by the prices that are collected in the same or similar geographic area and item category. Essentially, the price movement of items that are not collected is estimated to be the same as those that are collected for a given item and geographic area. See the "Cell-relative imputation" section on page 20 of the CPI Handbook of Methods chapter for a brief technical discussion of this procedure. Note that this type of imputation is used in the CPI every month, especially in categories where response rates are relatively low.
Were there any changes to CPI data-collection operations? Yes. BLS suspended all in-person data collection in the Seattle, Washington, area on March 5, 2020. On March 16, 2020, BLS suspended all in-person data collection. Upon suspension of in-person data collection, CPI data collectors were instructed to attempt to collect data normally collected by personal visit by telephone, email, or by internet from the website of the establishment, if a website exists. CPI data collectors were specifically instructed not to contact establishments by telephone when it would cause an undue burden on respondents. These types of establishments include hospitals, physician’s offices, grocery stores, department stores, and restaurants. Although states are beginning to reopen, in-person data collection is still mostly suspended.
Do CPI weights change every month in response to changes in consumer expenditures? One characteristic of the CPI-U and CPI-W is that they are fixed-weight indexes. The weights are updated every 2 years based on new consumer expenditure data. In between those weight updates, relative importances change somewhat as relative prices change. See the "How to estimate an updated relative importance" section on the Relative Importance page. However, since we do not get data on consumer expenditures for several months after the calculation of the CPI for a given month, we are not able to estimate indexes based on contemporaneous consumption patterns. So, for instance, decreases in gasoline consumption due to current conditions is not reflected in the CPI, although the relative importance of gasoline has decreased because of the falling gasoline index.
Will data collection for CPI expenditure weights be affected? The Consumer Expenditure Survey (CE), a household survey capturing consumer spending data, is used to calculate relative importances (weights) of goods and services in the CPI market basket. CE in-person data collection ceased on March 19, 2020. CE data are collected by the U.S. Census Bureau through an agreement with BLS. The Census Bureau is transitioning to collecting these data through telephone. Changes to CE survey operations will not have an immediate impact on CPI data, but may have long-term impacts. These weights are used in the chained CPI index (C-CPI-U). The March 2020 weights were incorporated in the final March 2020 chained CPI indexes, which were released in February 2021. BLS also incorporates the CE weights in a biennial weight update to the CPI-U and CPI-W indexes. These weight updates will be effective with the January 2022 indexes, released in February 2022. BLS is working on mitigation strategies to reduce measurement error of CPI weights caused by a potential loss of CE survey data.
How were the C-CPI-U weights affected? The relative importance is the expenditure for an item as a percentage of all expenditures. Changes to relative importance weights can be caused by changes in expenditures on a given component and by changes to total expenditures. Many changes to expenditure weights in the first half of 2020 were likely a result of the COVID-19 pandemic. We have more information about the pandemic effects on C-CPI-U weights.
How will the CPI for airline fares be affected? The CPI airline fares index measures price change of personal commercial air travel. The CPI airline fares sample is selected based on variables such as points of origin and destination, airline, class of service, and fare (First Class, Coach Full Fare, and Coach Discount Fare). A price quote will be collected even if an airline carrier reduces flights, as long as one flight option is available in the time band. The number of price quotes may fall if a route is cancelled entirely and the route is in the CPI sample. Price quotes for the current period are not subject to adjustment once collected; thus, a change or cancellation of a flight in the month that prices are collected is not reflected in the price index. More information on the airline fares consumer price index is available at Measuring price change in the CPI: Airline fares.
Will CPI data be published as scheduled? The CPI monthly estimates of price change will be published as scheduled. See the schedule of CPI release dates. BLS is closely monitoring fallback collection efforts to ensure that data meet quality standards. While it is probable that estimates will be based on a smaller amount of collected prices than usual, CPI procedures are designed to allow for estimates even when data collection is disrupted.
Under what circumstances would some data not be published? An index is not published if it fails a data-quality standard known as an adequacy ratio. Specifically, if BLS fails to collect at least one price in a geographic area that account for more than half the geographic weight of the index, the index is not published. Even in months without disruptions, some minor indexes with small samples occasionally fail this standard and are not published. (One example is Repair of household items.) Data-collection disruptions would have to be extremely severe for major CPI indexes not to be published based on this standard. Data-collection disruptions may be more severe in some area than others, and it is possible that some data for metro areas may fail data quality-standards and not be published. BLS will continue to monitor data-collection disruptions.
Will BLS attempt to quantify the overall impact of the COVID-19 pandemic on CPI estimates? The primary goal for the CPI is to provide accurate estimates of price change. While it is not possible to precisely quantify the impact of COVID-19 pandemic and response efforts on price change estimates because its effects cannot be separated from other influences on the economy, we can show some impacts on data collection for each affected month.
How does the CPI handle forgiveness, reduction, or nonpayment of rent in its shelter indexes? To collect rent values, CPI data collectors identify respondents for each sampled housing unit when they are initiated into the sample. Respondents could include its occupant (the renter), its owner (the landlord), a property manager, or an authorized representative of the occupant. However, in a given month when the unit is priced, the data collector could attempt to contact more than one of these respondents to get data on a sampled housing unit. For example, if the occupant is unavailable, the property manager might be contacted.

Specifically, the respondent is asked:
1. How much rent (are you/is the tenant) paying for this (house/apartment) now?
2. What period of time does that rent cover?

Most often the answer to question 2 is 1 month, but occasionally rent might be for a different length of time, or rent for a part of a month may be prorated.

When an unusually large price change is reported by a respondent, the data collector typically attempts to investigate the situation and confirm the change. For instance, if a tenant reports a large decline, the data collector will attempt to confirm this with the landlord or property manager. Relevant to the COVID-19 pandemic, when a tenant reports being unable to pay rent, the data collector is instructed as follows (from the manual used by CPI data collectors):

Be sure to probe these situations to determine if any rent obligation will be forgiven.
- If the landlord expects payment in full, regardless of when, enter the full rent amount that is due.
- If some or all of the rent is being forgiven, enter the amount the landlord/manager has agreed to accept.
- If the rent is not paid or not expected to be paid AND the landlord/manager is unsure about the future, enter $0.00.
These are all longstanding procedures, with the exception of the final bullet addressing situations of uncertainty, which have generally not arisen in the past.

Although some units in the rent sample, notably those under rent control, are excluded from owners’ equivalent rent calculation, a unit would not be excluded from the computation of owners' equivalent rent simply on the basis of a large decline in rent paid.

Note that a free or $0 price in the CPI is adjusted to a small positive value, typically a value equal to a 95-percent reduction from the previous price. (Prices of zero do not work well in the formulas used to compute the CPI.)

Promotional free rents. When rent is free for a defined period of time as a promotion, it is handled differently than nonpayments. The sample of rental units used to calculate the indexes for rent and for owners' equivalent rent in the CPI is separated into six panels. Each panel, which contains around 6,000 units, is priced every 6 months. When rent is free for one month as a promotion (such as “first month free”), the data collector enters the rent as 5/6 of the contract rent. Implicitly, the tenant’s rent is being reduced by 1/6 for the 6-month period. Less frequent situations of multiple months free are handled similarly, with the entered rent reduced by a fraction equal to the number of free months divided by six. This is also a longstanding procedure.

For additional information on the calculation of the index, see “Measuring Price Change in the CPI: Rent and Rental Equivalence.”
How does the CPI handle discounts and rebates on motor vehicle insurance? Discounts and price reductions to motor vehicle insurance policies are captured in the CPI; however, rebates are not. So, if a provider in our sample discounted a policy by 25 percent in April, that decline would be captured. However, if a provider in the sample paid customers a cash refund equal to 25 percent of the premium, that refund would not be captured because it is not visible to us. Note that the one-time reduction is captured in the single month that it occurs; it is not spread out over the course of the policy.
How does the CPI capture food at home prices? While our personal-visit data collection is suspended, the CPI program is collecting primarily online prices for food at home with a small amount of collection by telephone. Some stores in the sample do not have any prices online and many stores do not have all products available online, which results in CPI food at home indexes being based on smaller sample sizes than usual. For some stores in the sample, the online price might be different from the in-store price for the same day and time. This difference between in-store and online prices contributed to the increase in the CPI for Food at Home from March to April, but we are unable to quantify the impact. For items that were previously priced in store and are now being priced online, all explicit delivery and shipping charges or fees were excluded.
How does the more widespread use of online classes affect the CPI tuition indexes? Does the CPI quality adjust for this? The CPI does not do direct quality adjustment of the prices used in the index for college tuition and fees. BLS staff researched creating quality adjustments for online versus in-person classes. However, the lack of historical data made this difficult; there were few institutions that offered both in-person and online learning before the COVID-19 pandemic. Given this, we were unable to create a reasonable hedonic model to estimate a quality adjustment. Similarly, we do not directly quality adjust the prices used in the index for elementary and high school tuition and fees. The tuition and fee CPI indexes are based on changes in prices charged by the colleges and schools in our sample without regard to whether classes are online or in person. Detailed information about how these indexes are calculated is available at Measuring price change in the CPI: College tuition and at Measuring price change in the CPI: Elementary and high school.
Will CPI seasonal adjustment be affected? For the vast majority of CPI indexes, seasonal adjustment is accomplished through direct adjustment using X-13ARIMA-SEATS software. Direct seasonal adjustment involves applying seasonal factors to unadjusted data to remove within-year seasonal patterns. Seasonal factors represent regularly occurring, within-year price patterns and are estimated from historical price index data. For a number of series, the COVID-19 pandemic caused extreme price movements, which were unrelated to seasonal influences. To mitigate the effects of these extreme price movements on seasonal adjustment, CPI conducted intervention analysis on 2020 data for indexes with seasonal factors that were adversely affected by the COVID-19 pandemic. Intervention analysis involves estimating and removing the effects of certain important, nonseasonal events from the movement of an index before testing for seasonality and, if the presence of seasonality is detected, developing seasonal factors. The resulting seasonal adjustment is based on a representation of the series with the seasonal pattern emphasized. A list of indexes for which CPI conducted intervention analysis to specifically mitigate the effect of the COVID-19 pandemic is on the seasonal adjustment page.

Last Modified Date: January 30, 2023