Bureau of Labor Statistics Report to the Appropriations Committees on Modernizing the Current Population Survey, April 2026

The following report provides an overview of the Current Population Survey modernization plan in response to Senate Report 119-55. The referenced report directed the BLS as follows:

Current Population Survey — The Committee remains concerned by declining response rates facing all survey programs and other challenges which necessitate actions by BLS and the Census Bureau to modernize the operations of the Current Population Survey [CPS] and maintain this resource. The Committee believes steps must be taken to modernize the CPS. The Committee understands work on an Internet self-response mode and other improvements are underway. The Committee requests a CPS modernization plan, including an implementation timeline and resource needs, not later than 30 days after enactment of this act and follow-up briefing not later than 30 days after submission of the plan. Such plan and briefing shall include input from and participation of the Census Bureau.

The Current Population Survey (CPS) faces intensifying headwinds as the costs of data collection rise and it becomes more difficult to reach respondents who are increasingly reluctant to participate using existing collection methods. The Bureau of Labor Statistics (BLS) and the Census Bureau have begun an initiative to modernize the CPS, including an initial step to develop an internet self-response (ISR) instrument. The introduction of this collection tool will allow CPS to control data collection costs while reaching respondents who are otherwise unwilling to report in person or by telephone.

This report will provide additional information on the importance of modernization, the work that has been completed by the BLS and the Census Bureau, and the next steps required to ensure the introduction of a high-quality web instrument that protects the reliability and accuracy of CPS estimates while helping to ensure the sustainability of the survey by moderating long-term costs.

Importance of CPS Modernization 

The challenges facing the CPS are growing. Response rates have historically been very high but have declined from the low 90 percent range to the upper 60 percent range over the last 15 years (see chart below). Extensive use of in-person interviews allows the CPS to be extremely timely while still maintaining relatively high response rates.

 

In more recent years, the demand on the CPS has increased demonstrably, not just by stakeholders, but by respondents alike. The demand for additional and more granular data to evaluate components of the labor force continues to increase. Likewise, respondents whose information is vital to the relevance and quality of the survey are requesting to report by web, which is not currently available within the survey’s infrastructure.

The features that make the CPS unique and valuable (high response rates and timely data) are the same features that are proving problematic in the current environment. The extensive use of in-person field interviewers to collect data makes these two features possible. However, data collection costs are the most expensive component of the CPS program and are continuing to rise at an unsustainable pace. Costs have increased for pay and benefits for field representatives and for other collection costs (e.g., travel for in-person data collection), requiring operational changes to data collection.

Over the last several years, the BLS and the Census Bureau have taken numerous measures to maintain the sustainability and viability of the survey and the critical data used to inform federal, state, and local policies. The Census Bureau has received funding to develop the enterprise software that will power the ISR and funding to conduct internal testing for the Annual Social and Economic Supplement to the CPS. Within existing resources, the BLS has undertaken limited activities to coordinate with the Census Bureau on this effort. However, the BLS and the Census Bureau have not yet been able to conduct field testing necessary to ensure the operational readiness.

The CPS sample size in 2026, which consists of 60,000 eligible households, is the same as it was in 1981, even though the population used for the CPS has increased by 61%. Coupled with the decline in response rates, the CPS has seen a decrease in reliability over this period, impacting not just the CPS and its supplements^[1], but also state and local area labor force statistics used for the allocation of federal funds^[2]. In just the last 20 years, a single respondent has gone from representing about 2,100 people to about 3,500 people. At current sample sizes, a net change of fewer than 50 survey responses could move the headline unemployment rate by 0.1 percentage point. This greatly reduces the ability of the CPS to measure granularity and impacts measures of the labor force for subgroups, including demographic groups^[3]. This decline in reliability over time can also be viewed as a timeliness issue. For example, 20 years ago the CPS could detect that an over-the-month change in the unemployment rate as small as 0.18 percentage point was statistically significant, while now it would take two months of data before one could determine that this same change is statistically significant. This means that policymakers and businesses need to wait twice as long to have the same confidence in estimates.

Current Status of CPS Modernization

The first major milestone of CPS modernization will be the deployment of an internet survey response (ISR) instrument. The CPS questionnaire, designed to be completed as an interviewer-administered survey, is being modified for web respondents. This creates substantial risk of introducing adverse mode effects that could impact estimates as respondents might respond to questions they see online differently than questions that are asked by trained interviewers. For example, if the ISR instrument led people to respond to questions about job searches differently it could lead to a change in the unemployment rate. Data users would not be able to identify if the increase was based on actual labor market changes or collection mode effects. As such, extensive development and testing work is required to ensure data quality.

Work on this project ramped up considerably after 2020 as the Census Bureau began expansion of Data Ingest and Collection for the Enterprise (DICE)^[4] enterprise system development to refresh legacy collection systems across the agency and the BLS began work to modify the CPS labor force questionnaire for self-administration. The chart below provides an overview of the work conducted on this project from FY 2023 through the work that is planned for this year.

Recently, the 2025 field test was the first full test of the newly developed ISR instrument. The test included a limited sample with a total number of 8,000 new respondents and 8,000 existing respondents (these existing respondents had recently completed their 8 months of CPS reporting before being included in this test). Testing at this scale was primarily used to test the instrument itself along with the collection logistics involved in managing a monthly multi-mode survey that has a quick 10-day turnaround. The agencies were able to include some experiments to identify the proper contact strategies and piloted the use of incentives to encourage reporting. Within existing resources, the BLS has undertaken limited activities to coordinate with the Census Bureau on the new ISR instrument. The Agencies have not yet conducted field testing at the scale or design needed to measure mode effects and determine the accuracy of the new survey collection design. 

In FY 2026, both agencies are analyzing the results from the 2025 field test to make design updates for a limited follow-up field test in 2026. This includes evaluations of contact strategies, incentives, software performance, collection logistics, and questionnaire design. This test will primarily be used to further refine collection logistics.

Implementation Plan

To ensure the successful transition of the ISR into production, it is critical that the CPS is able to conduct a large-scale field test to identify mode effects that emerge from the introduction of the ISR instrument. The BLS and the Census Bureau propose accomplishing this by conducting a large ongoing parallel survey. This parallel survey would be used to generate experimental estimates and allow survey experts at the agencies to identify and mitigate mode effects that impact the accuracy and reliability of the modernized CPS. Estimates from the experimental sample will also be made available to provide transparency to data users.

Details about the parallel survey:

• Proposed sample size of 45,000 households (with a goal of 40,000 eligible households) a month.
  • At this sample size, the CPS will be able to detect differences in the unemployment rate between the parallel survey and the official CPS of 0.25 percentage point over 2 months of data.
• The Census Bureau will provide dedicated collection staff and supervisory infrastructure during the parallel collection period to allow the survey to be conducted with the same reference week and collection period as the official CPS.
  • This allows for ideal test conditions to compare estimates of the official and parallel surveys, which improves our ability to detect differences.
  • Especially needed to make sure the testing doesn’t limit resources on current production that may affect estimates.
• The parallel survey will be conducted monthly.
  • This allows for ongoing comparisons. The sample size will allow the CPS to detect mode effects and implement questionnaire/procedural changes in real time.
• The CPS will produce experimental estimates from the parallel survey.
  • While the first several months may be used only for internal research, the intent will be to begin providing these experimental estimates to the public to allow for a better understanding of survey differences along with efforts to mitigate these impacts.
• The CPS plans to run this parallel survey for approximately 18 months.
  • This should provide sufficient time to identify and mitigate (to the extent possible) adverse mode effects.
  • Exact timing will be dependent on when data from the parallel survey can be sufficiently evaluated and any changes that are needed to improve accuracy have been implemented.
• After conclusion of the parallel survey, the CPS would be able to move the ISR into the official CPS.

The graphic below displays a timeline of this plan.

Implementing the ISR without the parallel survey would put CPS estimates, including the unemployment rate, labor force participation rate, and employment-population ratio at considerable risk. For example, this could include major impacts to the unemployment rate that the BLS is not able to adequately explain or even the inability to produce estimates for one or multiple months.

The last major re-design of the CPS was in 1994, when the survey moved to electronic collection through Computer Assisted Personal Interviewing (CAPI) and Computer Assisted Telephone Interviewing (CATI) and the questionnaire was re-designed to address changes recommended by the National Commission on Employment and Unemployment Statistics (commonly referred to as the Levitan Commission). The CPS conducted three robust tests during this 8-year project, including field testing of more than 100,000 individuals in total and an 18-month parallel survey of 12,000 households.    

The parallel survey^[5] helped to reassure data users that changes seen in official estimates after the implementation of the re-designed survey were economic changes and not as a result of survey differences. When the CPS deploys the ISR, it will be critical to conduct a parallel survey.

The BLS and the Census Bureau have been in contact during this project with our counterparts in other countries who have gone through comparable modernization efforts. The experience in countries like the United Kingdom^[6], Germany^[7], and Canada^[8] have validated the importance of running a parallel survey to identify accuracy concerns prior to reaching official estimates. 

Resource Requirements

The agencies estimate that executing the plan above would cost approximately $60 million annually, including the cost to support the parallel survey described above.

Modernization beyond ISR

The deployment of the ISR is the first step towards modernizing the CPS and returning the survey to sustainability. This project includes a broader Census Bureau transition to an enhanced technology platform for data collection, which will allow for greater flexibility in how data are collected and processed.

The BLS and the Census Bureau have also explored the use of administrative data and other survey data to be used to support the CPS. However, within existing resources, the Agencies have prioritized the ISR instrument. Additional research has been conducted on adaptive design and other techniques to improve existing survey collection.

The CPS in the coming years will need to consider a larger scale re-design to ensure that it continues to measure the evolving labor force. For example, there are opportunities for the CPS to lead the way in identifying how AI is impacting the labor force.

Both agencies remain in agreement that a continuous process of modernization is necessary to ensure the ongoing accuracy, relevancy, and timeliness of the CPS.