The Probability of Dependence on Public Assistance

Abstract

Despite almost incessant discussion of welfare reform (see Levy, 1978, and Department of Health and Social Security, 1978), traditional systems of income support and their associated problems remain largely intact. The most intractable difficulty has been the conflict between the desire to reduce costs by concentrating benefits on the poorest families, and the desire to preserve work incentives by not withdrawing benefits too rapidly if a recipient family starts earning. Hanoch and Honig (1978) show that the incentive effect of income support schemes is theoretically ambiguous. Levy (1979) investigates the labour supply issue empirically. This paper looks at the whole issue of welfare dependency using data on American families receiving benefits under the largest scheme of public assistance, Aid to Families with Dependent Children (AFDC). We concentrate on an important group of poor families-those without fathers-and examine in particular the following issues: (a) the effectiveness of AFDC in selecting the very poorest families for the largest benefits; (b) the extent to which the number of recipients is influenced by the generosity of benefits and the implicit tax on earnings; (c) the extent to which recipients take advantage of the opportunity to supplement benefits with earnings; (d) whether the implicit tax on earnings discourages families from working; (e) whether the disproportionate share of blacks among recipients reflects a difference in attitudes about work and welfare between blacks and whites, or whether it is explained by the adverse circumstances of blacks. Four different levels of dependence are considered: none; partial (i.e. benefits less than earnings); substantial (i.e. benefits more than earnings); and total. The probability of each level of dependence is taken as a function of family characteristics, including earning potential and of the characteristics of the AFDC system. Section I describes a theory of family behaviour based on the assumption that observed differences in the behaviour of apparently identical families reflect an underlying distribution of unobserved differences. Section II discusses the statistical method used to estimate the probabilities, a straightforward generalization of logit analysis. Section III presents and interprets the statistical results, and Section IV summarizes our answers to the five issues posed above. Two comments on the choice and use of the data should be made before proceeding. The choice of data for 1967 was influenced not only by the availability of two surveys, the Survey of Economic Opportunity and a survey of AFDC recipients, but also by the fact that 1967 was the last year in which AFDC was a relatively pure scheme. Subsequent legislation imposed constraints on the extent to which states could withdraw benefits as recipients' incomes rose. More recently, the explosion in the size of the food stamp programme (still very small in

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 45 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.