Welfare Work Requirements with Paternalistic Government Preferences

Work requirements in means-tested transfer programs have grown in importance in the U.S. and in some other countries. The theoretical literature which considers their possible optimality generally operates within a traditional welfarist framework where some function of the utility of the poor is maximized. Here we consider a case where society is paternalistic and instead has preferences over the actual work allocations of welfare recipients. With this social welfare function, optimality of work requirements is possible but depends on the accuracy of the screening mechanism which assigns work requirements to some benefit recipients and not others. Numerical simulations show that the accuracy must be high for such optimality to occur. The simulations also show that earnings subsidies can be justified with the type of social welfare function used here. Work requirements in means-tested (or “welfare”) programs have seen a resurgence in the last twenty years in the U.S. and in a few European countries. These requirements generally stipulate that benefit recipients must work or engage in some work-like activity for some minimum number of hours each week. Most economists who have studied the welfare system have opposed work requirements. Milton Friedman, for example, believed that welfare should be administered through a negative income tax with benefits based solely on income. Other economists have criticized work requirements for being administratively unworkable, for introducing discretion into the system on the part of caseworkers, and for generating incentives on the part of individuals to avoid the requirements (Barth and Greenberg, 1971; Browning, 1975; Lurie, 1975). In addition, the classic optimal tax framework of Mirrlees (1971), which generates a negative income tax as an optimal solution in a utilitarian maximization model, allows only the manipulation of guarantees and tax rates and permits no other policy instrument to be used. However, there is a body of research papers operating within the optimal tax framework which does consider the possibility that work requirements might be part of a solution. Akerlof (1978) noted that if those with low job skills who have extreme difficulty in becoming employed-that is, those who are deemed, by some definition, to be unable to work--can be partly identified by some measurable, observable characteristic (or “tagged”), then it is possible that social welfare could be improved if those who are untagged, and hence able to work, are denied support because the tagged individuals can then be paid higher benefits. Diamond and Sheshinski 2 (1995), Parsons (1996), and Salanié (2002) have similar models which broaden the set of assumptions and applications of the general idea of using an imperfect observable indicator of need to separate the population. There is also a literature which assumes no tagging (i.e., no observable characteristic to differentiate individuals) but which finds work requirements to be optimal only under restricted conditions in that case (Brett (1998), Beaudry and Blackorby (2004), Cuff (2000), Fortin et al. (1993), Besley and Coate (1992,1995); see Kaplow (2004) for further discussion). Of these, only Besley and Coate are able to provide a strong case for work requirements, and only then by assuming a nonwelfarist social welfare function with an income maintenance criterion which puts little value on leisure. This study provides a new model of optimal work requirements. The primary departure from the previous literature is to use a nonwelfarist social welfare function which represents society as caring directly about the level of work of the poor, not just their utility. Unlike Besley and Coate, the major previous model with a nonwelfarist social welfare function, it is assumed that work, rather than income, enters the social welfare function directly. Society is assumed to value of the leisure time of some families, but only if their true skill (=ability) is low; if their true skill is sufficiently high, society expects them to work. The assumption that work per se is valued is based upon observation of the history of redistributional policy in the U.S. over the last thirty years, which, it is argued here and has been argued elsewhere (Moffitt, 2002a, 2003a), makes clear that voters do not accept the preferences of the poor but instead wish to force onto them specific allocations toward food, medical expenditure, housing, and other goods. It is commonly observed that this attitude is responsible for the rise in in-kind transfer programs for food, medical expenditure, and housing, for example, and for the decline in pure cash programs