Jennifer Stewart and Martin D. Dooley
Abstract
This paper analyses the dynamics of social assistance use among lone mothers in Ontario. Specifically, we study the duration of both welfare spells and off-welfare spells during the period 1990 through 1994. This research is a timely subject for at least two reasons. First, social assistance (henceforth SA) policy reform has been high on the agenda in Ontario and in other provinces yet no research of the sort we propose has ever been done with the caseload data for Ontario. Second, lone mother families are more reliant on welfare income than any other group except the disabled and these families contain a large and growing fraction of poor Canadian children whose problems have been the subject of much recent policy concern.
The Canadian literature on this topic is extraordinarily small despite the strong policy interest. There have been several published studies of welfare participation among Canadian lone mothers which employed data from a single cross section and one paper has used a time series of cross sections. These studies have generally found the expected associations of welfare participation with personal characteristics and policy parameters. Survey data, though informative, is still limited in the directions that it can provide to policy makers, in part, because it provides little or no information about the propensity to stay on welfare (extend a spell) or to return to welfare after a spell terminates. Several recent papers have used administrative caseload data from British Columbia and Quebec to analyse some of these dynamic aspects of welfare use in those provinces.
The Ontario Ministry of Community and Social Services (MCSS) has made available to the authors an administrative data set which provides some information concerning welfare use in the province during the period 1983-1994. These data are most complete for the period 1990-1994. In this paper, we analyse the association of three sets of factors with the duration of the spells on-welfare and off-welfare for lone mothers: the personal characteristics of these lone mothers such as age, marital status, number of dependents, age of youngest child, employability and history of previous welfare use; policy parameters such as the level of welfare benefits and the benefit reduction rate; and labour market conditions such as the regional unemployment rate.
The analysis of the data begins by examining the empirical survival and hazard functions. The survival function is defined as the proportion of the original sample with an ongoing spell at time t. The hazard function is defined as the number of spells that end at time t as a proportion of the spells that lasted until time t-1. These functions give us an idea of the distribution of the length of spells and the rate at which spells end, but they do not allow for an examination of the factors discussed above. To examine the impact of these factors a semi-parametric technique suggested by Prentice and Gloeckler(1978) and used in Meyer(1990) is used in this paper. The advantages of this technique include that the baseline hazard is nonparametrically estimated and that a test for one form of unobserved heterogeneity can be conducted. Inappropriate assumptions about either of these issues can lead to incorrect conclusions.