An Economic Analysis of Welfare Administration

在1988财政年度，375万个家庭收到了AFDC付款(对有受扶养子女的家庭的援助)。总共有1090万名受助者，其中三分之二是儿童。现金支付总额约为170亿美元。管理AFDC项目的总成本为22.9亿美元，约占现金支付总额的13%。在每个接受者的基础上，各州在管理AFDC计划的成本方面有很大的差异。例如，1988年，每个受助人的行政费用从密西西比州和西弗吉尼亚州分别为53.64美元和85.92美元的低位，到爱达荷州和马萨诸塞州每年分别为389.16美元和412.08美元的高位不等。尽管各州在福利水平、资格标准等方面缺乏统一，但行政程序对AFDC项目的影响受到经济学家的关注有限。除了AFDC计划的这些限制之外，还有一个重要的欺诈控制问题。这项研究将经历以下四个项目：(1)包含严格错误和宽大错误的福利申请决定的理论模型；(2)关注庭外和解(偿还)与起诉的可能性的欺诈控制过程的博弈论模型；(3)对包括欺诈控制在内的行政程序对AFDC计划的影响的州级时间序列横断面研究；以及(4)欺诈控制过程本身的聚合嵌套Logit模型。这项研究将有助于向福利政策制定者解释哪些因素影响欺诈水平，以及可以采取哪些措施来更好地控制欺诈水平。经验性工作还将估计控制欺诈活动对福利工作量和支出的影响程度。

During fiscal year 1988, 3.75 million families received AFDC payments (Aid to Families with Dependent Children). There were 10.9 million total recipients, two thirds of whom were children. The total cash payments were approximately $17 billion. The total costs of administering the AFDC program were $2.29 billion, or around 13 percent of total cash payments. On a per recipient basis, there was substantial variation across the states in the cost of administering the AFDC program. In 1988, for example, per recipient administrative costs ranged from lows of $53.64 and $85.92 in Mississippi and West Virginia, respectively, to highs of $389.16 and 412.08 per year in Idaho and Massachusetts, respectively. The impact of administrative procedures on AFDC programs has received limited attention from economists, despite the lack of uniformity in benefits level, eligibility standards, and the like among states. Add to these limitations of the AFDC programs the important issue of fraud control. This study will undergo the following four projects: (1) a theoretical model of the decision to apply for welfare which incorporates both errors of stringency and errors of leniency; (2) a game-theoretical model of the fraud control process which focuses on the likelihood of "out-of-court" settlement (repayment) versus prosecution; (3) a state-level time-series cross-section study of the effects of administrative procedures, including fraud control, on the AFDC program; and (4) an aggregate nested logit model of the fraud control process itself. This research will help explain to welfare policy-makers what factors influence the level of fraud and what might be done to better control the level of fraud. The empirical work will also offer an estimate of the extent to which fraud control activities influence welfare caseload and expenditures.