美國「貧窮家庭暫時性救助金」（TANF)方案，係依據1996年 「個人責任與工作機會調節法」（PRWORA)第一條之規定而制訂， 旨在賦予各州較大行政彈性，俾有利於貧窮家庭及兒童福利方案的執 行。由於各州有其不同的行政裁量權，似乎代表著全美儼然有五十一 種不同版本的「貧窮家庭暫時性救助金」方案，因此各州救助金方案 的内涵顯然有所不同。 本研究旨在探討各州政府政策工具的選擇與其對「貧窮家庭暫時 性救助金」方案成效的影響。本文將美國五十州以及華盛頓特區，依 據各州獎懲政策工具的不同，劃分爲九大組別。本文應用「變異數分 析法」，並透過所建構之假設，來檢測政策工具與九大組別之間的整體 影響。研究發現，就福利金受助案件數而言，「高獎勵與高懲罰」的州 與「低獎勵與低懲罰J的州之間，有顯著的差異。根據實證資料，本研 究提出不同獎懲程度的政策工具組合，以供決策者決策之參考。

The Temporary Assistance for Needy Families (TANF) program, which was included in the first title of the ‘Personal Responsibility and Work Opportunity Reconciliation Act 1996’，was designed to empower states with a degree of flexibility in the operation of their welfare programs for poor families and their children. Such individual state discretion, however, basically implies that there are 51 different TANF programs within the United States (Washington DC also participates in the program) and that as such, there is significant variance in the substance of the TANF programs across all states. The present study is concerned with the selection of policy instruments by state governments and the effects that these instruments have on the success, or otherwise, of the TANF program. This paper compresses all fifty states, along with Washington DC, into nine groups based on the incentive and penalty instruments used. Using an ANOVA framework, a number of hypotheses are constructed as a means of testing for relationships, in terms of the overall effects, between the nine groups. The study finds that a significant variation exists, with regard to welfare caseloads, between groups of 4high-intensity incentive, high-intensity penalty7 and those of low-intensity incentive, low-intensity penalty5. The policy implications arising from the optimal mix of incentives and penalties proposed by this study, based on the empirical data, are presented as guidance for policymakers.