Abstract

It is well documented that a positive correlation exists between receiving welfare as a child and depending on welfare as an adult. However, previous studies have not been able to explore many aspects of this relationship. This paper uses a unique administrative dataset for California, which follows welfare recipients since their teenage years until early adulthood, to study the causal effects of different lengths of welfare exposure as a child (conditional on welfare receipt) on future welfare dependency as a young adult.

The econometric analyses in this paper use a recently developed method from the program evaluation literature, based on the estimation of a generalized propensity score (GPS). As in the binary-treatment case the GPS permits removing the biases associated with differences in the observed characteristics of individuals. In addition, for some analyses, family-level unobserved heterogeneity is controlled for by relying on pairs of siblings exposed to different lengths of exposure.

The results show that there is no causal effect of length of exposure on future welfare dependency, nor on teenage childbearing. Conditional on teenage childbearing, there are no effects of length of exposure on adult welfare dependency either, but this dependency is almost three times larger for teenage mothers than for non-mothers. All results hold when controlling for unobserved heterogeneity. The results indicate that policies like time-limits are not likely to reduce the intergenerational correlation of welfare dependency.