Canadian lone mother employment rates, policy change and the US welfare reform literature

Abstract

The article examines the rise in Canadian lone mother employment rates during the 1990s using data from the Canadian Labour Force Survey and methods borrowed from the United States welfare reform literature. Patterns of lone mother employment rate increases in Canada are found to be similar to those in the United States. Income support policies in both countries changed in similar directions and in both cases increased the incentive to work. Despite these parallel changes it appears that, unlike the United States, policy reforms account for only a small part of the rise in Canadian lone mother employment rates.

Notes

¹ The employment rate is measured by the employment–population ratio. A lone mother is defined as a woman with children aged 0–24 years and family type designated as lone-parent. The US study by Grogger (2003) focuses on female headed families. The focus in US studies such by Eissa and Liebman (1996), Elwood (2000), O’Neill and Hill (2001) and Meyer and Rosenbaum (2001) is on unmarried women (divorced, widowed or never married) with children. Samples in these studies are typically restricted to women under age 45 years and above, either age 16, 18 or 19 years. Samples are also often restricted on the basis of a child's age. For example, Eissa and Liebman (1996) define a child as under age 19 years or if a full-time student under age 24 years. The sample in the current article will include lone parents with children 18–24 years and will also include lone parents whose marital status is married or separated.

² Similar patterns are found if the lone mother group is refined to focus exclusively on never married lone mothers or divorced/separated single mothers.

³ Unattached women are defined as women in single person households.

⁴ Total 36.9% of lone mothers aged 20–44 years report receiving welfare income during 1984–1992 compared to 12.0% of women aged 20–44 years in single person families. In the United States, roughly 35–40% of lone parents received welfare in 1991–1992, see Elwood (2000), Appendix Table 2.

⁵ Survey of Consumer Finances data shows that 70% of 25–29-year-old lone-parents received social assistance sometime during the year for an average year in the period 1983 to 1989 this compares to only 6% of similarly aged unattached women. For 40–44-year-old women, the difference in welfare receipt rates was only 20% points.

⁶ Data on real social assistance benefits are taken from National Council of Welfare (2002). Gorlick and Brethour provide a review of welfare-to-work programs for Canada by province. Bossenkool (1997) describes the changes to welfare programs in Alberta, where benefit cuts went hand in hand with changes in administration and program objectives. Schafer et al. (2001) provide an overview of changes in each province. Quaid (2002) discusses changes in Ontario, Alberta and New Brunswick with emphasis on workfare policies.

⁷ Under the Canada Assistance Plan (CAP) the Federal government paid up to half of provincial welfare program costs on the condition that these programs provide funding to all in need. Starting in 1996, transfers to the provinces for welfare programs were combined with those for health and post-secondary education resulting in the Canada Health and Social Transfer (CHST). The CHST relaxed the requirements provincial welfare programs had to meet to secure funding opening the door to provincial program changes, see National Council of Welfare (1997).

⁸ When introduced, Ontario, PEI, Nova Scotia, Manitoba and British Columbia treated the NCB Supplement as unearned income when calculating provincial welfare and/or child benefits. Alberta, Quebec and Saskatchewan directly reduced benefits in line with the NCB payments, see National Council of Welfare (1997). Only Newfoundland and New Brunswick did not reduce welfare or other benefit payments when the NCB was first introduced. As a result of these different responses, the policy initially increased work incentives in eight of 10 provinces. Changes in how Manitoba (July 2000), Nova Scotia (August 2001) and Quebec (July 2001) treated the NCB worked to reverse these incentives, see National Council of Welfare (2002).

⁹ Thorough descriptions of the NCB program and associated provincial initiatives can be found in the annual NCB progress reports, Government of Canada (1999–2002).

¹⁰ The size of the effect diminished after about 13 months as employment rates of the control group rose. The maximum effect is used below to provide a rough upper bound for welfare reform effects.

¹¹ The 3% point figure is obtained by making two adjustments to the maximum 12.9 point effect of the SSP. First, the SSP effect is multiplied by 0.6 since average welfare benefit cuts in the 1990s were only roughly 60% of the size of the average SSP incentive. Second, since the 12.9 point effect is for lone mothers on welfare rather than the entire population of lone mothers this figure must be scaled by the share of lone mothers on welfare (assumed to be 40%, see footnote 4). Although the experimental design of the SSP may be cause for more confidence in its estimated effects than other Canadian econometric studies there are considerable difficulties that must be overcome in translating its effects to estimating welfare reform impacts. First, the SSP was a temporary incentive (for up to 3 years) while reforms might more reasonably be regarded as permanent. It is not entirely clear how this might affect behaviour as on the one hand the temporary nature of the incentive means it is less generous than a permanent incentive, while on the other hand its temporary nature may induce intertemporal substitution thus inflating its effects. Also, as noted in the previous footnote, the size of the SSP effects varied substantially over time and the 12.9 point figure is a maximum effect.

¹² This assumes a $1600 supplement.

¹³Elwood (2000) based his initial welfare aggressiveness classifications on the size of real benefit cuts during 1986–1997, whether time limits were introduced pre-1996 and whether sanctions were implemented under earlier reforms.

¹⁴ Another version of the calculation assumed that if working the lone mother would earn the provincial minimum wage and work full time, full year. The rankings obtained with this measure were not much different from those reported in Table 3 although British Columbia was more likely to be classified as aggressive due to large increases in its minimum wage in the mid-1990s.

¹⁵ Elwood (2000) suggests a statistical measure of reform aggressiveness that can be viewed as a more sophisticated way of judging aggressiveness based on welfare recipient trends. The idea is to estimate a model of the incidence of welfare benefit receipt on the pre-reform sample period and then use the model to predict the probability that someone in the post-reform period will receive benefits. Those jurisdictions where actual rates of welfare receipt fall further below the predicted probability of welfare receipt are then classified as aggressive reformers. The author experimented with the construction of such a measure for Canada. The major shortcoming of the exercise is that the dataset that provided annual data on welfare benefit receipt in the pre-reform period was replaced in 1998 creating a break in the data. Consequently, the statistical measure constructed here provides an indication of reform aggressiveness only up to 1997. The results suggested that Nova Scotia, Ontario and Alberta might be classified as most aggressive as of 1997.

¹⁶ Skill categories were aggregated due to concerns about underlying sample sizes in the smaller provinces. The time periods were chosen to be at similar points in the business cycle.

¹⁷ Both log and level versions were considered the results reported in Table 5 are for the level version. Results using the log of welfare benefits were much the same.