The number of households receiving public assistance income in the ACS (see Table B19057 and CP03, line 72) showed a statistically significant increase between 2019 and 2021, despite administrative data indicating a decline in caseload.[U.S. Department of Health and Human Services TANF Caseload Data]. The public assistance amount that the Census Bureau intends to capture for this question in the ACS are those received through cash welfare programs, like Temporary Assistance for Needy Families (TANF), Welfare or welfare to work, General Assistance, General Relief, Emergency Assistance, and Diversion Payments.

The question is not intended to capture assistance from non-cash benefits, like Supplemental Nutrition Assistance Program (SNAP), or housing, transportation or childcare assistance. The question is also not intended to capture programs introduced or expanded amidst the coronavirus pandemic, such as stimulus/Economic Impact Payments (EIPs), expanded unemployment insurance, and the advance Child Tax Credit payments.

The ACS has help text across all modes of interviews (mail, Computer Assisted Telephone Interview (CATI), Computer Assisted Personal Interview (CAPI) and Internet) that instruct respondents about which programs they are supposed to include and not include in response to this question. Additional instructions were issued regarding programs introduced or expanded during the coronavirus pandemic. In CAPI, interviewers access the help text and read the instructions to the respondent. CAPI interviewers are trained to understand the definition of income (which includes income received on a regular basis, not one-time disbursements) and to navigate the survey instrument. Therefore, they are more likely to access the help text and guide respondents in excluding these types of payments. In self-response modes, such as mail and internet, the respondent must access the additional help text to read the instructions themselves.

Despite these instructions, there is evidence to suggest that some households may not have read the help text and erroneously reported stimulus payments and child tax credit advance payments in 2021.  Between 2019 and 2021, there was an increase in bunching of public assistance amounts at the same values as stimulus payment amounts, suggesting misreporting of stimulus payments as public assistance. Second, this increase is more evident in self-response modes, where respondents needed to seek out the instructions to learn which programs were meant to be captured for that question.

Due to the small number of households reporting public assistance in general (about 3.4 percent), these erroneously reported amounts did not affect the statistical significance or the direction of the change in either median household income or poverty rate estimates at the national level.  Additionally, we found that zeroing out all public assistance amounts still did not change the statistical significance or the direction of the change in either median household income or poverty rate estimates.  We found only limited differences for smaller sub-groups. As a result, we encourage data users to use caution when looking at reported public assistance receipt and values in the 2021 ACS as they may be inflated due to misreporting of stimulus payments and advance child tax credit payments as public assistance.