Designing quality control (QC) systems in public welfare programs exposes competing policy interests. Federal officials see the systems as the principal means for recouping misspent federal dollars. State and local service deliverers see the systems as either a means to improve managerial capabilities or as a major financial and managerial headache. To welfare advocates, the systems raise the specter of benefits denied to clients in need. Those competing interests have created a volatile controversy around QC in the nation's three principal family assistance programs--Aid to Families with Dependent Children (AFDC). Medicaid, and Food Stamps. The controversy has led to administrative appeals of federal financial penalties levied against the states, litigation about those penalties. Congressional interest in system reform, including a new round of studies, and now, reconsideration of the appropriateness of the QC systems as designed.
Some of the controversy concerns how well the QC systems help states improve the management of their large financial assistance programs. But principal concern has been about the application and effect of financial penalties (or so-called sanctions and disallowances), and, specifically, the measurement and estimation procedures that determine who will pay, and how much, for payment errors. In 1985 and 1986, to resolve the growing dispute between the states and the federal government and the increasing backlog of uncollected penalties--then estimated at $3 billion through 1989--Congress mandated that the National Academy of Sciences (NAS) and the Departments of Agriculture and Health and Human Services (HHS) conduct independent studies of the quality control systems in the Food Stamp, AFDC, and Medicaid programs. The findings of those studies and associated analyses can now inform the discussion among proponents of opposing views.
Although the debate that led to the congressional study mandates focused largely on the statistical issues surrounding error rate estimates and financial penalties, the statistical issues are really masks for deeper public policy concerns, since statistical design choices are not exclusively technical choices. Rather, they can balance competing interests in assigning and collecting penalties, and they can influence how the programs themselves are run and how well those programs will fulfill intended objectives.
This article is an overview of the policy context in which the statistical issues have been raised, and of the design, measurement, and estimation questions that expose competing policy interests. The article describes the administration of the family assistance programs, the operation of their QC systems and major analytic issues raised by their design, and selected NAS and Administration study findings and recent legislative reforms that relate to the issues raised.
2. THE POLICY CONTEXT
2.1 Structure and Operation of the Family Assistance Programs
AFDC is the principal federal cash transfer program for needy children and their (mostly single) parents. Medicaid is the principal federal program for providing health care to needy persons. The Food Stamp program provides coupons with which low-income individuals can buy food and related products. The three programs serve overlapping populations and have related objectives.
In all three programs, administrative costs are in most cases split 50-50 between the states and the federal government. The federal government also pays at least 50% (up to 83% depending on the state's ability to pay as measured by its per capita income) of AFDC and Medicaid benefit costs and all of Food Stamp benefits. In fiscal year (FY) 1987 the three programs received about $49 billion, about 40% of federal spending for benefit programs for the poor. In that same year, AFDC served nearly 11 million individuals, Medicaid 22 million, and Food …