Restoring trust in economic statistics: Why it matters and how we fix it

Economic statistics—like inflation, unemployment, and income—quietly shape nearly every major decision in society. Policymakers rely on them to design programs, businesses use them to plan investments, and households depend on them to understand their financial reality. But what happens if people stop trusting the numbers?

Has trust declined?

Public confidence in federal economic statistics has been steadily eroding. Trends since 2020 show declining trust in government, and it is not too far of a leap to assume that such low levels of public trust in government may translate into a lack of trust in the data produced by the federal statistical agencies. While trust in federal statistics varies across people, evidence suggests that, on average, we have been moving steadily toward less trust over time.

Public trust in federal economic statistics has declined, and rebuilding it will require more than just better data. Trust in economic statistics doesn’t exist in isolation. It reflects broader confidence in government institutions. A Pew Research Center survey shows that public trust in government has fallen dramatically over time (see Figure 1). But this erosion of trust isn’t uniform. While there may be some people who will always trust government data, the group of people who trust it has been shrinking.

Reflecting these low levels of trust, the AmeriSpeak household survey panels, conducted by the National Opinion Research Center (NORC), show that the percentage of individuals agreeing or strongly agreeing that statistics provided by federal agencies are generally accurate decreased from 40% to 32% between June and September 2025. NORC also shows a slight decline in the percentage of individuals who trust federal statistics, falling from 57% to 50% between June 2025 and February 2026. For people who have used federal statistics, however, about 80% trust the data, and that percentage has remained fairly constant (see Figure 2).

What does trust in federal data require?

Trust in the federal economic statistical system requires trust along several significant dimensions:

• Trust that the agencies producing the data are free from political influence
• That data collection, development of statistical estimates, analysis of the data, and dissemination of data follow the highest standards of methodological and statistical practice
• That statistical agencies are committed to quality and professional standards of practice, maintain an active research program to improve the quality of published data, and are forthright and timely in identifying and communicating any problems in the data such as bias or other sources of mismeasurement
• That statistical agencies are transparent in their communications on all of these dimensions

While statistical agencies have the responsibility to adopt the highest standards of methodological and statistical practice in the construction and interpretation of data estimates, even the most sophisticated data users, potentially well versed in the technical detail of statistical techniques, have to trust that statistical agencies are following best practice when it comes to the agencies’ decisions. Trust, in this context, requires clear communication by the statistical agencies on the methods they adopt and an openness to critical review of these methods in the hopes of improving the quality of the data produced.

Ultimately, rebuilding trust demands commitment to principles that have long underpinned the system’s strength, turning skepticism into reliance for a healthier economy.

Why trust is breaking down

There are several challenges associated with a breakdown in trust in federal data:

1. Political Influence

A critical element tied to trust is the degree to which the statistical system is independent of political influence. The belief that any statistical agency is influenced by the political party in power, whether real or imagined, will erode trust and can move us further down the continuum.

2. Falling budgetary resources

Statistical agencies are losing resources, posing a serious threat to the quality of data they can produce. The American Statistical Association reports that eight of 13 agencies have lost at least 16% of their purchasing power since 2009 and that most agencies have lost 20-30% of their staff. These reductions are causing delays in data production, suspensions and cancellations of datasets, and reductions in scope. If declining budgets are a sign of declining quality, then even among the stalwart defenders of the statistical system, trust will be eroded to some degree.

3. Declining data collection and response rates

Perhaps the single most noticeable metric signaling that trust in the data has fallen is falling data collection and response rates. Even if there were no resource constraints, declining trust has likely contributed to a growing recalcitrance to participate in surveys. In addition, declining real resources will mean that the extra efforts needed to maintain response rates, such as devoting more personnel to contacting reluctant eligible sample members, including the use of relatively expensive but effective personal visits, will be increasingly out of reach.

4. Misunderstanding of data processes

The public may lack a clear understanding of how data processes generally work. The data produced by the federal statistical system are often subject to revision. In many cases, it takes time to collect data from sample members, and the desire to provide timely data results in original estimates being developed based on partial returns and then updated in subsequent periods when more data are received. For example, for a large survey like the Current Employment Statistics (CES), there are a number of reasons for the revisions of data estimates, principally including the time it takes to collect data as compared to the requirements of more frequent publication of data estimates. The CES allows firms three months to report monthly payroll data and publishes versions of estimates during that period. Revisions simply reflect the incorporation of the later reports, which improve accuracy.

More subtly, the choice of which statistics to emphasize and publish can lead to a lack of understanding of the research subject’s complexity, such as only providing an average value instead of a more informative view of the percentile distribution of the statistical data in question. The problem in this context is that with limited and declining real resources, statistical agencies often have to prioritize which of the most informative statistics they can afford to produce.

5. Unplanned releases of data

Yet another possible factor influencing trust in federal data are the self-inflicted wounds owing to unplanned and unfortunate early releases of data by statistical agencies, which suggest the possibility of political influence, managerial lapses, or promoting favored status among data users. There have been a number of instances of “early” releases of data that resulted from IT system failures, an unintentional release of information through emails to a select group of data users that circulated rapidly beyond their intended recipients,  and even accidental mentions of top-line data by administration officials. It is not clear how injurious such instances have been to statistical agency reputations; the infrequency of these issues makes them difficult for study.

6. Use of alternative data

It is not completely clear how the use of alternative data, such as administrative, private sector, website, or text data, affects the level of trust in official statistics. On one hand, replacing data in surveys with declining response rates with higher quality and more reliable alternative data may improve trust. On the other hand, the mixing of various data sources to produce published estimates may produce a “black box” type of opaque understanding, even among technical users, that results in a feeling of a lack of transparency. The statistical agencies have an opportunity to address this shortcoming through more accessible communication of methods.

How can we improve trust in federal economic data?

One solution to maintain trust is to engage diverse data users—policymakers, journalists, the public—via clear communication with investments in cross-disciplinary training, infrastructure, and adaptive policies. Despite the termination of the statistical agencies’ advisory boards, examples of this kind of engagement include reviews by independent bodies like the Committee on National Statistics (CNSTAT), an independent body that commissions studies of work of the federal statistical agencies, and NBER’s Conference on Research in Income and Wealth, which brings together academic researchers and statistical agency staff to discuss measurement issues. 

Other involvement with data stakeholders includes the National Association for Business Economists (NABE) quarterly meetings with the heads of the statistical agencies, the NBER’s Economic Measurement Research Institute (EMRI) that is focused on innovation and modernization of data collection, and other groups like the Council for Community and Economic Research (C2ER), which hosts the agency heads for similar purposes at their annual meetings.

There is also opportunity for non-governmental organizations like the Friends of BLS, the American Statistical Association, the Council of Professional Associations on Federal Statistics (COPFAS), and NABE to expand their work to include collaborating with advocacy groups to lobby on their behalf. This is especially important in circumstances, such as in the current administration, in which cabinet agencies have expressed serious concerns about the quality of the data produced by the statistical agencies.

In many cases, the challenges that statistical agencies face are not directly under their control. For example, what advice would you give to the heads of statistical agencies on responding to charges of data manipulation that have no basis in fact or, more simply, declarations that the mere size in the movements in data are such that they call into question the quality and reliability of the work of the agencies?

The statistical agencies, despite declining real budgets, are doing an incredible amount of high-quality work to preserve and enhance their measurement of the economy and provide data-driven guidance to inform policy. More resources for modernization or exploratory funding within the agencies to set up pilots, manage upfront data costs, and automate processes could help them continue their important efforts.

There is enormous potential to improve the quality and our understanding of the data produced by the statistical system, and thereby the public’s trust. However, significant changes need to be made, especially in terms of resources, to allow the agencies to accomplish these goals.