Revisiting the Work Incentives of Welfare Programs

The standard static model of the e¤ect of welfare programs on labor supply is newly estimated with maximum response heterogeneity. Both reduced form and structural models are estimated allowing response heterogeneity whose shape is estimated nonparametrically. Estimated marginal treatment e¤ects representing the e¤ect of program participation on labor supply for the historical AFDC program are found to be quite heterogeneous, varying with the initial level of participation and initial levels of welfare program parameters, all of which a¤ect the composition of those brought into or who leave the program on the margin. Applying the estimated model to major historical changes in bene…ts and tax rates in the AFDC program in 1967 and 1981 shows that those reforms had marginal e¤ects on labor supply which were quite di¤erent because the populations were di¤erent. The author would like to thank Marc Chan, Kai Liu, Shaiza Qayyum, and Lu Xu for research assistance as well as the participants of a large number of conferences and departmental seminars for comments. Joe Altonji, Richard Blundell, Eric French, Michael Keane, Chris Taber, and James Ziliak also provided helpful comments and feedback. Research support from the National Institutes of Health is gratefully acknowledged. There is a long and extensive literature on estimating the e¤ects of welfare programs on labor supply. Studies in the early 1970s estimating labor supply equations with cross-sectional data to forecast the e¤ects of welfare programs (Cain and Watts (1973)) were followed by results from a series of negative income tax experiments (Burtless (1987),Mo¢ tt and Kehrer (1981)). A literature using more sophisticated econometric methods to handle the selectivity of welfare participation developed in the 1980s (Burtless and Hausman (1978),Hausman (1981),Mo¢ tt (1983)) which was followed by a literature of reduced-form estimates of the e¤ects of speci…c welfare reforms in the 1980s and 1990s (Gueron and Pauly (1991),Grogger and Karoly (2005)). Reviews of these literatures have been published by Danziger et al. (1981), Mo¢ tt (1992), and Blundell and Macurdy (1999). For the most part, the parts of this literature which aim to estimate the e¤ects of welfare tax rates and guarantees on labor supply have found them to be in the expected direction, negative in both cases. More recently, progress on dynamics and human capital e¤ects of welfare programs have been studied (Blundell et al. (2016)). Heterogeneity in labor supply response has, however, not been a source of major interest in this literature, yet the importance of heterogeneous response is increasingly recognized in other literatures. The older estimates of welfare e¤ects using the standard static model typically assumed constant coe¢ cients, for example, although this was occasionally relaxed, as in the Burtless and Hausman (1978) study, where the coe¢ cient on nonlabor income was allowed to be heterogeneous but in a restrictive parametric form. The literature on nonlinear budget constraints of which the Burtless-Hausman paper was a part (including, as previously referenced, Hausman (1981) and Mo¢ tt (1983)) did allow heterogeneous preferences in the additive error terms which were assumed to include preference heterogeneity and therefore a¤ected utility on di¤erent segments of the constraint, so heterogeneity did have an impact on which segment to locate on but not on the change in hours of work from welfare participation for a given change in tax rate and