What to Watch for in Next Week’s Release of Census Data on Income, Poverty, and Health Insurance in 2025

Next Tuesday, the Census Bureau will release 2025 data on household income, poverty, and health insurance.

Median income could rise thanks to growth in inflation-adjusted earnings in 2025, but a key measure of poverty may grow because of the rising cost of basic needs and other factors.

The 2025 data will be a key benchmark for tracking the future effects of the harmful 2025 Republican reconciliation law’s cuts to millions of people’s food assistance and health care, the largest such cuts in U.S. history. The data, reflecting circumstances before the bulk of the reconciliation law’s cuts take effect, will show the important role food assistance from the Supplemental Nutrition Assistance Program (SNAP) played in lifting people above the poverty line and that Medicaid and Affordable Care Act (ACA) marketplace coverage played in providing affordable health coverage. Future years’ Census data will show the growing impact of the reconciliation law’s deep benefit cuts if they are not reversed. Other data show the reconciliation law has already taken away food assistance and health coverage from millions of people, and these losses will continue to grow over time as more of the provisions take effect.

And, although the Trump Administration has begun to undermine the data that policymakers need to evaluate policy impacts and to encroach on the Census Bureau’s long-established independence — such as by ending long-standing collection of food insecurity data (traditionally released in early September) and taking steps to interfere with the content of the 2030 Census — we continue to have confidence in the integrity of the Current Population Survey (CPS) figures due out next week.

Here are four main points to be aware of regarding the 2025 data.

1. Poverty and health insurance figures for 2025 won’t yet reflect most of the 2025 reconciliation law’s cuts to food assistance and health coverage.

They will, however, show the considerable impact programs like SNAP, Medicaid, and ACA marketplace coverage had on reducing poverty and uninsurance before the cuts, even while poverty, inequality, and the number of people without health coverage remained high in 2025.

The 2025 data will be a useful benchmark against which to measure the effects of the harmful Republican reconciliation law. Some of the law’s changes to SNAP began in late 2025, including cuts that took away food assistance from certain immigrants with lawful statuses and expanded SNAP’s existing provision that takes away benefits from people not meeting a harsh work requirement; states also started raising participation barriers in an effort to reduce error rates. But the law’s larger impacts on annual poverty, uninsurance, and hardship are expected to hit in 2026 and deepen further in 2027 and beyond, as more provisions — including one taking away Medicaid coverage from people not meeting a work requirement — go into full effect.

Other data already available for 2026 show that millions of people are losing SNAP, as well as health coverage through Medicaid and the ACA marketplace. Between the 2025 reconciliation law’s July 2025 enactment and May 2026, the number of people participating in SNAP has fallen 12 percent, or roughly 5 million people, which we estimate includes well over 1.5 million children. There is no sign that the decline is due primarily to an improving economy (unemployment rates have barely changed).^[1] At the same time, nearly 2.2 million fewer people were enrolled in ACA marketplace coverage in February 2026 than December 2025, after Republicans allowed premium tax credit enhancements to expire and people’s premium costs soared.^[2] The 2025 reconciliation law’s largest Medicaid cuts will start to hit in early 2027, including a policy that takes health coverage away from people who do not meet a work requirement.^[3]

Overall, the reconciliation law, the expiration of the enhanced ACA premium tax credits that made marketplace coverage more affordable, and Trump Administration rule changes will cause about 15 million people to become uninsured by 2034, the Congressional Budget Office (CBO) projects.^[4] For SNAP, CBO estimated that ultimately 4 million will lose all or a substantial amount of their benefits. But as we’ve seen, more people are losing SNAP, and faster, than CBO predicted.^[5]

2. A less commonly cited snapshot of health coverage at the time of the survey could show some of the initial impact of Republican cuts to ACA marketplace coverage.

While the headline CPS coverage estimates measure whether people had coverage at some point during the previous calendar year, a less commonly cited snapshot of coverage at the time of the survey, roughly March 2026, could be useful in providing an early look at the effects of the expiration of enhanced premium tax credits at the end of 2025. Following the expiration of the enhancements, roughly 3 million fewer people were enrolled in ACA marketplace coverage in February 2026 compared to February 2025.^[6] Tuesday’s CPS data could show an increase in the uninsured rate from March 2025 to March 2026.

3. Median household income figures may rise as a result of rising real earnings in 2025, although inequality after taxes is likely to remain high.

Average weekly earnings of non-supervisory workers rose 1.3 percent faster than inflation in 2025, according to the Bureau of Labor Statistics (BLS), suggesting median annual household income could rise modestly as well.

Inequality in after-tax income, however, could remain at or near its highest in recent years. Although the federal income tax system generally narrows income inequality, Congress chose to use the 2025 reconciliation law to instead continue and expand tax breaks that, if anything, are likely to slightly widen gaps in after-tax income.^[7] Last year’s figures show inequality in after-tax income already stood at its highest in at least 17 years, in data starting in 2009, with the household at the 90^th percentile making 9.85 times more than the household at the 10^th percentile.^[8]

4. The Supplemental Poverty Measure (SPM) is likely to rise due to a large rise in the SPM poverty thresholds, in part reflecting higher costs for many necessities.

SPM poverty thresholds rose between 4.4 percent and 6.3 percent from 2024 to 2025, according to BLS data released in August. The rise was due partly to rapid growth in a measure of prices for key necessities (food, clothing, shelter, utilities, and phone and internet access) that BLS uses in calculating the SPM thresholds; these costs rose 3.4 percent in 2025, BLS reported, ^[9] outstripping the overall all-items consumer price index (CPI-U), which rose 2.6 percent.

Not all of the increase in SPM thresholds relates to 2025 prices. Driving the poverty thresholds up further were factors related to rising inflation-adjusted consumer spending on key necessities from 2019 to 2024. The SPM is intended to measure the number of people facing real strain to afford the basics. To keep this level relevant to society’s gradually evolving living standards, and not just prices, it uses spending data from earlier years.

In 2025, the poverty threshold for a two-adult two-child family renting in an average-cost community is $41,701.

We will particularly be watching how poverty trends may differ across racial and ethnic groups. One area of concern is that average monthly employment of Black people age 16 and older fell sharply in 2025, BLS data show, which could result in worse trends for Black households than white households.^[10]

Technical Changes and Revisions Could Complicate Next Week’s Findings

CBPP will also be watching for how the following technical concerns and data revisions affect interpretation of the 2025 estimates or trends:

• Survey response rates. Response patterns – the share and characteristics of people in a survey who finish the survey – may affect the accuracy of the income, poverty, and other data due out next week. In recent years, Census has regularly assessed non-response and its impact on income and poverty estimates, finding that survey response has been falling since 2020 and has been especially low for people with low income. The Census Bureau concluded that in 2024, non-response artificially raised median income estimates by “2 or 3 percent” and lowered the official poverty rate by 0.4 percentage points.^[11] Last year, the Census Bureau released this assessment when the data were released. We will be watching for an updated analysis Tuesday. Census has sound ways to improve response, such as by increasing the number of attempts to reach non-responders and allowing online responses, but these will require increased funding from Congress.
• Delays and loss of detail related to new privacy practices. In June, a Commerce Department political appointee issued a directive telling Census to alter the methods it uses for protecting survey respondents’ privacy.^[12] The new order limits the approved methods to coarsening (e.g. rounding or combining) or simply not publishing certain estimates. Widely criticized by demographers and data users, the order will in effect prevent Census from releasing data with as much detail as in the past, especially for small locations or groups.^[13] (The same directive led Census to delay release of 2025 American Community Survey data, which it would customarily release next Thursday but now plans to release later this year.^[14]) Key questions are what data will be withheld, if any, in Tuesday’s reports, and whether the 2025 published data will be comparable with prior years’.
• Revisions of pre-2025 SPM thresholds. BLS and Census on July 17 released a routine technical correction of SPM thresholds for 2019 through 2024, and Census has released revised SPM poverty figures for those years. We expect Tuesday’s report to include these corrections, which modestly raise or lower SPM poverty rates depending on the year and a family’s home ownership status.^[15]

Next Week’s Data Should Be Reliable, but Political Interference Elsewhere at Census Threatens Public Trust and May Hide Policies’ Harmful Effects

Although we have confidence in the integrity of the survey data due out next week, there have been instances of interference elsewhere at the long-independent Census Bureau.

Since January 2025, the Trump Administration has terminated federal data collection on topics that don’t align with the Administration’s views, such as food insecurity, climate change, or the LGBTQ+ population.^[16] For example, the Administration last year eliminated CPS food insecurity data, collected since 1996, that would have helped evaluate the impact of the Administration’s SNAP and health care cuts on families’ ability to afford food. Until this year, the CPS’s annual food insecurity data were typically released in the first week of September.

This week, the Administration has proposed a regulatory change to exclude some immigrants from the next decennial census (used for apportioning representation in Congress among other purposes),^[17] a move seemingly at odds with long-standing legal understandings of the 14^th Amendment’s requirement that the decennial census count the “whole number of persons.”^[18] The proposed rule would also prohibit the Census Bureau from including questions on race, ethnicity, and sexual orientation in the decennial census questionnaire; the decennial count has captured data on race since the first census in 1790.

Last month, in an apparent case of brazen meddling, the Census Bureau published a highly unusual report with no named authors that the White House said backed its claim of widespread fraud in the 2020 general election, a claim that has been repeatedly debunked. Independent news reports indicated that career Census Bureau staff were not involved in the report and former Census employees noted that the report’s lack of named authors, clear methodology, or discussion of data limitations mark striking departures from Census’s usual rigorous and transparent procedures.^[19] Around the same time, the Commerce Department, the Census Bureau’s parent agency, appears to have quietly altered its Scientific Integrity directive to remove language prohibiting political influence over federal data.^[20]

We will monitor the forthcoming data to be sure that they continue to meet the Census Bureau’s trusted standards.