Food Stamp Program:

Statistical Validity of Agriculture's Payment Error-Rate Estimates

RCED-87-4, Oct 30, 1986

Contact:

John W. Harman
(202) 512-5138
contact@gao.gov

 

Office of Public Affairs
(202) 512-4800
youngc1@gao.gov

In response to a congressional request, GAO reviewed the quality control (QC) system that the Food and Nutrition Service (FNS) used to set fiscal year (FY) 1984 error rates in Food Stamp Program benefit payments and to assess sanctions against states for their errors.

GAO noted that FNS annually estimates each state's food stamp payment error rate in a two-step process, using statistically selected FNS and state case samples. FNS: (1) computes a regressed error rate based on a review of its subsample of the state's sample; (2) adjusts the regressed error rate upward if the state did not complete the number of sample case reviews prescribed in its approved QC review plan; and (3) determines the state's sanction from the resulting number, which is the official error rate. A sanction involves reducing the federally funded share of a state's administrative costs by a predetermined percentage tied to a target error rate. GAO found that, overall, FNS statistical policies for estimating the regressed and official error rates generally conformed with accepted statistical theory. GAO believes that, because small changes in the official error rates could affect sanction amounts, FNS should use the soundest practical statistical methods in developing the payment error-rate estimates.

Status Legend:

More Info
  • Review Pending-GAO has not yet assessed implementation status.
  • Open-Actions to satisfy the intent of the recommendation have not been taken or are being planned, or actions that partially satisfy the intent of the recommendation have been taken.
  • Closed-implemented-Actions that satisfy the intent of the recommendation have been taken.
  • Closed-not implemented-While the intent of the recommendation has not been satisfied, time or circumstances have rendered the recommendation invalid.
    • Review Pending
    • Open
    • Closed - implemented
    • Closed - not implemented

    Recommendations for Executive Action

    Recommendation: The Administrator, FNS, should amend FNS policy and guidance to require that an acceptable expected minimum-stratum sample size be established when approving a state's sampling plan.

    Agency Affected: Department of Agriculture: Office of the Secretary: Food, Nutrition and Consumer Services: Food and Nutrition Service

    Status: Closed - Not Implemented

    Comments: With the passage of the Hunger Prevention Act, FNS will be pursuing the requirements of the law and is not considering the recommendation.

    Recommendation: The Administrator, FNS, should amend FNS policy and guidance to require the routine calculation of the sampling error of payment error-rate estimates.

    Agency Affected: Department of Agriculture: Office of the Secretary: Food, Nutrition and Consumer Services: Food and Nutrition Service

    Status: Closed - Implemented

    Comments: FNS included this calculation capability in its programming at the Washington Computer Center. FNS will use this capability to generate the sampling error of each state's payment error rate annually.

    Recommendation: In conjunction with actions taken in response to studies that the Department of Agriculture (USDA) and the National Academy of Sciences (NAS) are conducting on food stamp error rates and sanctions, the Administrator, FNS, should amend FNS policy and guidance to eliminate the assumption that cases for which reviews have not been completed have the same error rate as completed cases.

    Agency Affected: Department of Agriculture: Office of the Secretary: Food, Nutrition and Consumer Services: Food and Nutrition Service

    Status: Closed - Not Implemented

    Comments: With passage of the Hunger Prevention Act, FNS no longer believes this recommendation to be valid, and intends to implement the requirements of the law.

    Recommendation: In conjunction with actions taken in response to studies that USDA and NAS are conducting on food stamp error rates and sanctions, the Administrator, FNS, should amend FNS policy and guidance to require that the standard deviation of the regressed payment error rate, instead of the standard deviation from the state samples, be used in calculating the adjusted, or official, error rate.

    Agency Affected: Department of Agriculture: Office of the Secretary: Food, Nutrition and Consumer Services: Food and Nutrition Service

    Status: Closed - Not Implemented

    Comments: FNS believes that the standard deviation of the state-reported error rate is a more accurate basis for the calculation. Although FNS agreed that either method may be used, it believes that there is no basis for FNS to change its method, and therefore, do not intend to implement this recommendation.