Sept 28 -- Per email to AEAStat, FNS provides draft materials for public review and extends comment period through October 30, 2023.
July 19 -- The Food and Nutrition Service (FNS), USDA invites comment by September 18, 2023 regarding the proposed Understanding Risk Assessment in SNAP Payment Accuracy Study.
This collection is a new collection. The primary purpose of this study is to provide FNS with information about SNAP State agencies' use of risk assessment (RA) tools to reduce payment errors, the effects of these tools, and best practices for FNS and the SNAP State agencies to consider in the development and use of RA tools. RA tools may apply statistical models using SNAP household characteristics to estimate the relative risk of improper payment.
The Supplemental Nutrition Assistance Program (SNAP) is the largest hunger safety net program in the United States, providing food assistance benefits to roughly one in eight Americans. SNAP is administered by the U.S. Department of Agriculture's (USDA) Food and Nutrition Service (FNS). Although the benefits are federally funded and must be issued in accordance with Federal statutes and regulations, SNAP State agencies are responsible for determining eligibility and calculating appropriate benefit amounts for eligible participants. SNAP State agencies have flexibility in administering the program through a range of policy options, waivers of regulations, and demonstration projects. The challenges associated with accurately documenting households' circumstances and calculating benefits within the context of complex regulations, options, and waivers lead to a degree of improper payments.
The Payment Integrity Information Act of 2019 continues the work of previous related legislation in requiring Federal agencies to track and mitigate improper payments, which are defined as payments that either should not have been made or were made in an incorrect amount. FNS and the SNAP State agencies use SNAP Quality Control (QC) to closely monitor the program for improper payments. SNAP State agencies must conduct a QC review of a random sample of current cases each month (referred to as active cases) to identify underpayments and overpayments and calculate total payment error. At the end of the review period for each month's cases, the SNAP State agencies share the case files and results with Federal SNAP staff, who review a subsample of the cases for accuracy and use the results to calculate an annual official payment error rate for each State agency's official payment error rate.
Some social welfare agencies and criminal justice organizations have begun using risk assessment (RA) tools. These tools apply a statistical model to case characteristics to estimate the relative risk of a particular outcome. Agencies that use RA tools may use the output to allocate staff resources such that the riskiest cases receive the most time and attention. This is intended to improve program outcomes but may have unintended consequences. As RA tool use becomes more common across social sectors, it is critical to address the risk of bias in these tools. Bias can enter RA tools through the data used to build them and the way the tool uses those data to predict risk and may impinge on civil rights by leading to disparate treatments and/or disparate impacts.
FNS is conducting a study, Understanding Risk Assessment in Supplemental Nutrition Assistance Program (SNAP) Payment Accuracy, to develop a comprehensive picture of whether and how SNAP State agencies use RA tools and determine if these tools create disparate impacts on protected classes. The key research objectives follow: (1) determine which States use RA tools to reduce error rates; (2) determine what factors and variables are being used in RA tools; (3) identify how SNAP State agencies act on the results of their RA tools; (4) determine whether SNAP State agencies' RA tools are successful in reducing error rates; (5) determine if the RA tools create (or relieve) racial or other disparities by which individuals are flagged for further review; and (6) determine best practices in the development and use of RA tools.
The study approach includes a survey, case studies, and a request for administrative data from SNAP State agencies. Data will be collected via a web-based census survey of the 53 SNAP State agencies. Case studies will be completed with six SNAP State agencies; these case studies will include telephone interviews with up to five types of State-level staff and up to two types of local SNAP agency staff (as applicable). The types of State-level staff will include RA tool development leads, SNAP Quality Control Directors, SNAP Quality Assurance Directors, IT systems staff, and data analysis staff. The types of local SNAP agency staff will include local agency supervisors and local agency eligibility staff. The study team will also request administrative data from the SNAP State agencies that use an RA tool.
Respondent groups identified include the following: (1) individuals/households (pretest participants); and (2) State, local, and Tribal government (SNAP State agencies and SNAP local offices).
The total estimated sample size and the number of respondents is 100. The team expects all sample units to respond to all relevant data collection activities. The study includes 53 SNAP State agency directors and up to 5 other staff in 6 selected SNAP State agencies that use RA tools. The study also includes a SNAP local office supervisor and a local office eligibility staff member for local offices of relevant selected SNAP State agencies that use RA tools. The study also has 5 pretest participants from a pool of 5 possible pretest participants. Some pretest participants will pretest more than one instrument. The study team expects all SNAP State agencies to respond to the survey. The exact number of SNAP State agencies that use RA tools is currently unknown, but estimates suggest the number is 15 or fewer. Under the assumption that 15 SNAP State agencies have RA tools, the study team expects all 15 of these SNAP State agencies to respond to the survey and provide data on their RA tool. The team expects 6 of these 15 SNAP State agencies to also participate in case study interviews. These estimates assume all potential respondents will eventually respond to their relevant information collections.
SNAP Quality Control:
https://www.fns.usda.gov/snap/quality-control
Draft information collection instrument and technical documentation
https://www.dropbox.com/scl/fi/ks1msezn6r07p4uk1r0q1/Draft-Risk-Assessment-OMB-Package-for-FRN-1.zip?rlkey=akjc6zexcs1x5en0toevt444p&dl=0
FRN:
https://www.federalregister.gov/d/2023-15209