JULY 14, 2025 — The U.S. Census Bureau today released new estimates from its National Experimental Well-Being Statistics (NEWS) project.

NEWS is a research-driven project with the goal to improve how the Census Bureau measures income and poverty. NEWS does this by exploring new methods for creating these economic “well-being” indicators by looking back at data collections from previous years, using a wider span of data sources, and applying new advanced analysis techniques.

This new release (version 2.5) includes an expanded set of vintage years (2016, 2017, 2019, 2020 and 2021) where these newly accessed data sources and advanced analysis techniques are applied. Estimates for vintage year 2018 were released as part of NEWS versions 1.0 and 2.0 which served as a proof of concept, or way to experiment with a new methodology to ensure the estimates met our high-quality standards before investing resources into full development.

These newly produced estimates are based on:

• Money income. Income received on a regular basis (exclusive of certain money receipts such as capital gains, profit made after selling assets) before payments for personal income taxes, Social Security, union dues, Medicare deductions, etc. This was the only income measure released in version 1.0. Money income is used to create estimates of official poverty.
• Disposable income. Money income minus taxes plus tax credits (state and federal).
• Disposable income plus noncash benefits (used to create the Supplemental Poverty Measure), including:
  • Housing assistance.
  • Supplemental Nutrition Assistance Program (SNAP).
  • Special Supplemental Nutrition Program for Women, Infants, and Children (WIC).
  • National School Lunch Program (NSLP).
  • Energy assistance.

Version 2.5 findings show that using an expanded set of data sources helps overcome data bias challenges.

Bias in data is a common challenge in the statistical world and can occur for a variety of reasons. Misreporting or underreporting bias occurs if a survey respondent answers with inaccurate information. Nonresponse bias can result when someone declines to answer.

The Census Bureau uses different methods to mitigate these challenges, including linking administrative data to make sure people who decline to respond are still represented in the data. Greater access to 2016-2021 administrative data and improved methodologies help NEWS to address these types of bias in income surveys.

Additional NEWS 2.5 findings:

• Starting in 2019, nonresponse rates increased more for low-income than high-income households. (Nonresponse bias.)
• Starting in 2020, income declined in categories that are generally better reported (such as wage and salary earnings) and increased in categories often underreported (like unemployment insurance compensation, which increased substantially during the pandemic). (Misreporting bias.)

These biases can vary by year and by group:

• For those age 65 and older, NEWS estimates of the official poverty measure were between 3.2 and 4 percentage points lower than the official estimates every year from 2016 to 2021.
• For children under age 18, NEWS estimates were not statistically different from the official poverty measure in 2019 but were 2.5 percentage points lower in 2020.
• The Supplemental Poverty Measure, which accounts for taxes, credits, and in-kind transfers, was lowest in 2020 (5.6 percent), 3.5 percentage points lower than the survey estimate.

By addressing each source of bias simultaneously, NEWS allows the Census Bureau to estimate income and poverty more accurately over time and by group than would be possible with survey or administrative data alone. This work may help inform how we produce statistics in future official reports on income and poverty in the United States.

NEWS version 2.5 data are available on the NEWS webpage. 

No news release associated with this experimental data product. Tip sheet only.

###