The Self-Sufficiency Project (SSP) offered a generous but time-limited earnings supplement to a randomly assigned group of lone parents—who were also long-term social assistance recipients—if they found full-time work and left social assistance. Employment data was collected for this group over a three-year period following the offer, and for a randomly-assigned control group. This article analyzes the characteristics of the first job that SSP participants found after they left social assistance. The occupations and industries of the first job held are analyzed as is SSP’s impact on hourly wages, weekly hours and job stability. The article finds that SSP increased employment in jobs that were no worse (and no better) than the jobs that participants might have taken in the absence of the program.
Le projet d’autosuffisance a mis à l’épreuve une stratégie de promotion du travail à plein temps chez des bénéficiaires de l’aide sociale de longue période et monoparentaux en Colombie-Britannique et au Nouveau-Brunswick. Le projet offrait un supplément de gains généreux, mais limité dans le temps, à la condition d’abandonner l’aide sociale pour travailler à temps plein. L’effet du programme pouvait s’apprécier à l’aide d’une méthodologie expérimentale à assignation au hasard.
Quoique les suppléments de gains peuvent favoriser l’emploi de façon efficace, il est possible qu’ils encouragent les bénéficiaires de l’aide sociale à accepter un emploi de qualité moindre que celui qu’ils auraient éventuellement accepté en l’absence d’un tel programme. Notre article s’intéresse à ce sujet en analysant l’impact du programme sur la qualité et les caractéristiques du premier emploi que les participants ont déniché après avoir quitté l’aide sociale.
Après avoir décrit les industries et les occupations où les participants ont trouvé un emploi, cet essai cherche à évaluer l’effet qu’a pu avoir le projet d’autosuffisance dans ces mêmes industries et occupations. Il cherche également à vérifier si le projet a eu ou non un impact sur les caractéristiques du premier emploi obtenu après avoir quitté l’aide sociale. Finalement, en se basant sur quatre caractéristiques reliées à l’emploi qui peuvent être qualifiées d’indicateurs positifs de la qualité d’un travail, il évalue l’impact du projet sur la nature des emplois trouvés par les participants au projet.
Au moment où les participants au projet ont quitté le programme d’aide sociale et accepté du travail, ils ont trouvé des emplois dans une sphère limitée d’occupations et d’industries. Plus de la moitié des participants ont trouvé des emplois dans les mêmes dix industries. De façon plus spécifique, ils se sont retrouvés surtout dans l’industrie des services alimentaires (15 %); dans les services sociaux non institutionnels (7 %); dans des magasins d’alimentation (5 %) et dans des résidences privées (5 %).
La proportion la plus élevée des participants (8,5 %) détenait des emplois dans des occupations reliées aux services d’alimentation. Puis, les occupations les plus populaires étaient celles de commis-vendeurs ou de représentants des ventes. Un autre 5,7 % travaillaient comme caissiers et 5,2 % trouvèrent du travail dans des emplois reliés aux soins des enfants. Quoique le projet ait augmenté de façon significative la proportion des bénéficiaires de l’aide sociale qui quittèrent le programme d’aide sociale pour travailler, cela a eu très peu d’effet sur la répartition de l’emploi dans les différentes occupations et industries.
Cet essai cherche aussi à analyser l’impact du projet sur les caractéristiques de l’emploi, incluant les salaires, les heures de travail, l’affectation à plusieurs emplois pour équivaloir au temps plein, la durée, les avantages octroyés par l’employeur et le statut syndical. Nous avons observé que le projet d’autosuffisance augmentait l’emploi dans les postes où les salaires sont relativement faibles, c’est-à-dire dans les emplois rémunérés moins d’un dollar de plus que le taux du salaire minimum. Cependant, il n’existe aucune preuve à l’effet que le projet aurait incité des participants à accepter des salaires plus bas que ceux qu’ils auraient acceptés en l’absence du programme. Puisque le projet comportait une exigence de travail à plein temps, il n’est pas surprenant de constater que l’emploi additionnel généré par celui-ci s’est retrouvé dans des postes offrant plus de trente heures par semaine. Les règles permettaient aux participants d’occuper plus d’un emploi de façon à rencontrer l’exigence d’un travail à temps plein, mais l’effet du projet a été deux fois plus grand sur la tenue d’un seul emploi que sur la tenue de plusieurs emplois. En ce qui concerne la durée, l’impact le plus prononcé s’est répercuté sur des occupations de durées plus longues que douze mois. Par ailleurs, les participants ont accédé à des emplois en grande majorité non syndiqués et sans avantages sociaux.
Pour évaluer l’impact du projet sur la qualité des emplois, nous nous sommes basés sur les travaux déjà publiés sur le sujet et sur les données disponibles pour identifier quatre caractéristiques positives d’une occupation : (1) la présence d’au moins un avantage payé par l’employeur ; (2) un salaire horaire suffisamment élevé pour qu’un travailleur à temps plein pendant un an gagne l’équivalent du seuil de faible revenu ; (3) un travail à plein temps, et (4) un travail d’une durée d’au moins six mois.
L’emploi additionnel créé par le projet n’a présenté que quelques unes de ces caractéristiques positives. Il a avant tout accru l’embauche dans des occupations sans avantages payés par l’employeur. Ensuite, presque tous les individus ayant quitté l’aide sociale et qui étaient retournés au travail à cause du projet ont eu accès à des occupations qui ne leur auraient pas permis de gagner autant que le seuil de faible revenu. Toutefois, le projet a eu un effet appréciable sur le travail à temps plein et de longue durée. En somme, le projet a augmenté l’embauche dans des occupations qui n’étaient pas pires que celles que les bénéficiaires auraient acceptées en l’absence du programme.
El Proyecto de Auto-Suficiencia (PAS) ofreció un suplemento de ingresos, generoso pero limitado en su duración, a un grupo de individuos monoparentales designados al azar – y que eran beneficiarios de la asistencia social desde mucho tiempo – a condición de que ellos encuentren un trabajo a tiempo completo y abandonen la asistencia social. La información sobre el empleo por este grupo fue compilada por el periodo de 3 años después de la oferta, y por otro grupo de control escogido al azar. Este artículo revela las características del primer empleo obtenido por los participantes al PAS después de dejar la ayuda social. Son analizadas las ocupaciones e industrias del primer empleo obtenido así como el impacto del PAS en el salario por hora, las horas por semana y la estabilidad en el empleo. El documento establece que el PAS incrementa el empleo en trabajos que no son peores (ni mejores) que los trabajos que los participantes pueden obtener en ausencia del programa.
When welfare-to-work programs encourage participants to find work, these programs can also bring about dramatic changes in participants’ lives. It is possible that these changes may not improve welfare recipients’ well-being. For example, social assistance recipients who leave welfare for work could lose income and experience increased stress, while their children may receive less care and supervision. Alternatively, such programs might set in motion a series of events leading participants to positive outcomes such as economic self-sufficiency. This article considers one such welfare-to-work program, the Self-Sufficiency Project (SSP), a random assignment demonstration of an earnings supplement implemented from 1992 to 2001 in British Columbia and New Brunswick.
During the period that SSP was being developed and implemented, social assistance programs across Canada were undergoing extensive reform. Major reforms implemented in some provinces emphasized the reduction of welfare caseloads by restricting eligibility and by moving social assistance recipients into employment. Since 1995 every province and territory has introduced some welfare-to-work element into its social assistance system (Gorlick and Brethour 1998). While many reform efforts were motivated, to some extent, by fiscal prudence, a philosophy of work as socially preferable to welfare has also influenced the tenor of reform strategies.
One of the concerns most frequently voiced about the contemporary emphasis on encouraging lone-parent welfare recipients, most of whom are women, to move from welfare to work is that such women might simply be trading poverty-and-welfare for poverty-and-work. If a welfare recipient wishes to be a full-time mother, carefully supervising the many aspects of her children’s development, no amount of money, prestige or job satisfaction will substitute for being home.
Others argue that there are no “bad” jobs, that any job—no matter how poorly paid, no matter how difficult—is preferable to long-term welfare dependence. And even if the first post-welfare job is a “bad” job, in terms of earnings or working conditions or both, some former welfare recipients might eventually be able to move into a “good” job.
Within the context of the Self-Sufficiency Project, there is a concern that even though SSP was successful in achieving significant increases in employment over the first 36 months, the project could conceivably have led participants to take “bad” jobs in order to qualify for the supplement. That is, encouraging self-sufficiency with an earnings supplement could reduce the chances of finding a “good” job.
The SSP earnings supplement encouraged many participants to leave social assistance and take up paid employment. This article uses administrative data as well as data from follow-up surveys to examine the first job that participants held after they left Income Assistance (IA). First, the occupations and industries in which participants worked are described. The impact that SSP had on employment in different occupations and industries is then estimated.
This article also analyzes whether SSP had an impact on characteristics of the first job held after leaving welfare including the following: wages, hours, job duration, the receipt of employer-sponsored benefits and union membership. Finally, the article identifies four job characteristics that may be considered positive indicators of job quality, and estimates the impact SSP had on employment in jobs with these characteristics.
Recent Research on the Job Quality of Former Welfare Recipients
Defining Job Quality
One aim of an analysis of jobs held by former welfare recipients might be to determine if those jobs enable the workers to be self-sufficient. One important factor contributing to self-sufficiency is an income greater than that available through welfare. Such an analysis would focus on earnings (including tax benefits like the National Child Tax Benefit) and benefits. Job duration is also important here since it may indicate that the ability or willingness of the worker to hold a job might be linked to wage progression or other types of advancement.
A different, and perhaps complementary, aim of such an analysis might be normative. Are these the kinds of jobs that the analyst believes will eventually lead to a more satisfying life than that available on welfare? Here, the type of job matters because some jobs may lead to higher paying, more rewarding jobs as the worker gains experience. The type of job also determines the type of activities in which a worker spends a substantial proportion of their time. Those activities determine, in part, whether or not workers find their lives satisfying.
Assessing the quality of a job might seem straightforward: some jobs are seen as being better than others are. Everyone recognizes this fact, both when they discuss jobs in daily conversation and when they must actually choose among jobs. Yet social scientists have no comprehensive measures of a job’s desirability. Sociologists have devised many schemes for ranking occupations, but none for ranking the diverse jobs that fall into the same occupational category. Economists rank jobs according to their pay but have no global measure of jobs’ non-monetary benefits (or costs). Psychologists measure workers’ subjective satisfaction with their jobs but have not, for the most part, tried to rank jobs on the basis of objective characteristics (Jencks, Perman and Rainwater 1998: 1323).
There are no widely accepted job quality scales that translate various job characteristics such as wages, benefits, autonomy, or social setting into a single numerical measure of the quality of the job. One reason is that the quality of a job is largely subjective—the same job might be considered to be “ideal” by one individual, and a “nightmare” for a second individual.
Studies of welfare “leavers” in the U.S. suggest that those who leave welfare for work move into jobs similar to those held by other low-income workers. These jobs pay low wages and offer few benefits. For example, Loprest (1999) studied a group of “leavers” in the period immediately after the passage of the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA), the landmark U.S. welfare reform of 1996. Using the National Study of America’s Families (NSAF), Loprest was able to analyze, for each job, variables such as hourly wage, hours of work, occupation and industry, the provision of health benefits, and whether another job was held simultaneously. Over all Loprest (1999: 9) found that “the types and quality of jobs held by former welfare recipients are similar to those held by other low-income mothers.”
The U.S. government commissioned several state-specific studies of welfare “leavers” in the wake of PRWORA. Most of these studies used state-level administrative data to characterize the employment and earnings of the “leavers.” Several studies involved surveys of “leavers” that provided somewhat more extensive information than could be obtained from the administrative data alone. According to a synthesis by Acs and Loprest (2001), these studies found, in general, that “leavers” had relatively low earnings and few benefits. The studies did not provide much further detail on the jobs held by “leavers.”
Bartik (1997) analyzed a large sample of women who had both been on welfare and worked in the calendar year prior to being interviewed as part of the March Current Population Survey (CPS). The focus of the analysis was on estimating the effects of job characteristics on the probability that the women were employed at the time of March interview. The job studied was the one held in the year prior to the March interview.
Bartik’s major finding was that, holding wages constant, the occupation and industry of the job were important determinants of the probability of working at the time of the March survey. For example, those who worked in hospitals or educational institutions were more likely to be employed at the time of the March survey than others, and cashiers and labourers were less likely to be employed.
A “good” job was one that increased the probability of being employed at the time of the March CPS interview. For example, working as a cook in an eating or drinking establishment lowered the probability of working, whereas working as a waitress in an eating or drinking establishment did not. Wages and hours were also quite important but the magnitude of the occupation and industry “effects” was often larger than the effects of wages or hours.
The implication was “...that the characteristics of jobs matter. Policymakers should consider efforts to target higher-wage jobs, jobs in the hospitals or educational services industry, and jobs with less customer contact and less intense supervisory pressure” Bartik (1997: 41).
In summary, the literature suggests that job quality has at least three important dimensions. First, job quality might be assessed by the nature of the work and the work environment. Work that is interesting, physically comfortable, or which provides access to a social network, for example, might be considered of high quality. Second, job quality can also be related to future job prospects. High quality jobs might be described as jobs that either provide wage growth or lead to other higher paying jobs. Third and finally, job quality can arise from the compensation, whether cash or in-kind, that workers receive for their labour.
For this article, the limitations of the available quantitative data prevent the estimation of the effect of SSP on any measures of the first, and most subjective, source of job quality. Based on lessons from the literature, this article does, however, attempt to identify measures of the second and third dimensions of job quality.
The Self-Sufficiency Project
The Self-Sufficiency Project (SSP) was a random assignment demonstration that tested the effect of a generous financial incentive on the behaviour of long-term social assistance recipients in New Brunswick and British Columbia. SSP was a voluntary program that offered lone parents, who had received Income Assistance (IA) for at least twelve months, an earnings supplement if they found full-time work within one year and left IA. If SSP participants had taken up the supplement within the one-year window, they were then eligible to receive it for the next three years. SSP was designed to “make work pay” more than social assistance. To this end, the SSP supplements could potentially double earnings from minimum wage work. The key features of SSP are described in further detail in Box 1.
In order to assess the impact that SSP had on important outcomes such as employment, income and earnings, potential participants were randomly assigned to either a control group or a program group.
The evaluation of SSP has already shown that the program has had a substantial impact on the lives of participants. Three years after random assignment, SSP had significantly increased full-time employment, reduced the rate of Income Assistance receipt, increased earnings and reduced the proportion of participants experiencing very low income (Michalopoulos et al. 2000). After nearly five years, the program no longer had a significant impact on these outcomes (Michalopoulos et al. 2002).
The evaluations of SSP have focused on overall employment, which averages outcomes across all of the jobs held by participants. As a result, very little is known about the specific characteristics of the jobs that SSP participants held.
Sample and Data Sources
For the evaluation of SSP, data was collected from four sources: a baseline survey, follow-up surveys, administrative data and data from the SSP Program Management Information System (PMIS). The baseline survey was administered at random assignment. It collected demographic information and asked questions about a variety of topics that could inform the evaluation. In particular, the baseline survey collected detailed information about the respondents’ employment history. Follow-up surveys, at 18 and 36-months after random assignment, posed questions similar to those asked in the baseline survey. This article employs data from all four sources.
The sample of long-term welfare recipients used here includes all respondents to the SSP 36-month follow-up survey. Of the 5,729 individuals that were randomly assigned, 4,961 completed the 36-month follow-up survey—2,503 in the program group and 2,458 in the control group. At random assignment, all sample members were lone parents and the vast majority (95.6 per cent) were female. Although 95 per cent of the sample had worked for pay in the past, most were neither employed nor looking for work at random assignment. More than half of the sample had less than a high school education and about a quarter reported a physical health problem. Michalopoulos et al. (2000) provide a more detailed description of the sample members’ characteristics.
The First Post-IA Job
SSP encouraged lone parents, who had been dependent on welfare, to work full time and leave social assistance. The parameters of SSP were such that the program may have differentially affected the types of jobs that individuals held at various points in time. Because SSP offered a financial incentive to participants if they found full-time work within 12 months, some participants may have lowered their expectations (in terms of wages or working conditions) in order to find a job within the 12-month supplement take-up window. Later, once they had secured their eligibility for the supplement, participants may have looked for and found jobs which they preferred.
Given that participants may have held many different jobs, there is a choice to be made about the most relevant job to analyze when considering how SSP affected job characteristics. This article is interested in the initial transition from long-term social assistance receipt to work. Consequently, the article examines the characteristics of the first job that SSP participants held after or during the first post-random assignment month in which they did not receive IA. Some participants may have left IA because they began working in the job that this article describes. Other participants may have left IA for reasons unconnected with the job analyzed here. People often leave welfare because changes in their family composition or income sources have made them ineligible.
Because some people combined work with social assistance after random assignment, about 30 per cent of the SSP sample members were working in the same month they left IA. In these cases, the article analyses the job that was held when the recipients left IA. Some sample members held more than one job simultaneously. In these cases, the job selected was the one in which the recipient usually worked the most hours.
For most of the sample, there was no record of a first post-IA job. Nearly 44 per cent of the sample received income assistance in every month of the follow-up. A further 16 per cent of the sample experienced at least one month without income assistance, but did not work in any of the months of follow-up.
Occupation and Industry of the First Post-IA Job
The SSP follow-up surveys collected detailed information about all of the jobs that participants held during the follow-up period. Specifically, participants were asked about the type of business, industry or service in which they were employed. The survey also asked respondents to describe their most important activities or duties. This information was then used to classify the participants’ occupations and the industries in which they worked. Occupations were classified according to Statistics Canada’s 1980 Standard Occupational Classification (SOC). The industrial classification followed the 1980 Standard Industrial Classification (SIC).
Table 1 indicates the kinds of jobs long-term welfare recipients held when they first left IA; the Table combines SSP participants in both the program and control group but includes only those group members who reported their occupation or industry. Over half of this subset of respondents (all of whom worked after they left IA and provided enough data to classify their jobs’ industries) worked in the same 10 industries.
Moreover, a relatively large proportion of those who held a job after they left IA worked in a single industry—almost 15 per cent worked in the food services industry. In contrast, only 7 per cent of all Canadian women (aged 15 years and older) worked in food services industries (Statistics Canada 1996a). The food services industry includes establishments that are primarily engaged in operating restaurants, take-out food and catering services. The next largest proportion of SSP participants (7 per cent) worked in non-institutional social services, which includes child day-care and nursery school services. Roughly 5 per cent of the SSP participants worked in food stores and another 5 per cent in private households.
The second panel of Table 1 presents the 10 occupations in which the largest proportions of SSP participants worked. The largest proportion of SSP participants (8.5 per cent) found their first post-IA jobs in food and beverage serving occupations. Among all Canadian women, only 3 per cent worked in food and beverage serving occupations (Statistics Canada 1996b). The next most common occupation for SSP participants’ first post-IA jobs was “sales clerk.” Another 5.7 per cent worked as cashiers and tellers and 5.2 per cent found work in childcare occupations.
In all of these occupations, the typical worker earns well below the average for Canadian workers. For example, the average worker in a food and beverage serving occupation earned $28,000 in 1997 compared to $37,400 earned by the average Canadian (HRDC 2000). Moreover, a labour market information tool developed by the Canadian government considers each of these occupations to have fair or limited future prospects in terms of earnings and unemployment rates (HRDC 2000).
Program Impacts on Industry and Occupation of the First Post-IA Job
The offer of a financial incentive led many members of the SSP program group to leave welfare before the 36-month follow-up interview. SSP reduced the number of people who remained on IA throughout the entire 36-month follow-up period by 9.6 percentage points. Most of the people who left welfare also went to work. SSP reduced, by 4.2 percentage points, the proportion of program group members who left IA but then did not subsequently work. The first two rows of Table 2 imply that 47.4 per cent of the program group left IA and worked, compared to 33.6 per cent of the control group, a difference of 13.8 percentage points.
These 13.8 percentage points represent the program group members who would not have worked in the absence of the SSP intervention and much of the remainder of this article is taken up with asking how this 13.8 percentage point impact is distributed across various job characteristics.
Although Table 1 shows that social assistance recipients tend to find jobs in a limited range of occupations and industries, it is possible that by altering their work preferences, SSP encouraged program group members to take jobs in industries and occupations that they might not have otherwise accepted. On the other hand, welfare recipients’ preferences may not be the key influencing factor because they are unable to obtain jobs in other occupations and industries. If the latter were the case, since SSP does not directly change participants’ qualifications, there would seem to be little scope for SSP to affect the occupations and industries in which program group members find jobs.
Table 2 shows the industries and occupations of the first jobs that program and control group members held after they left IA. The occupational and industrial categories presented in Table 2 are at a much higher level of aggregation than in Table 1.
SSP increased work in only four industries. Although most members of both groups worked in accommodation, food and beverage services and other service industries, members of the program group were 6 percentage points more likely to be working in these industries than were control group members. SSP also increased work in wholesale and retail trade industries by 4.2 percentage points. Although statistically significant, SSP had substantively small (about one percentage point) impacts on work in manufacturing and primary industries and work in finance and insurance, real estate and business services industries.
Because SSP had such a large impact on employment, increases in some industrial categories result simply because so many more people were working. That is, the impacts observed in Table 2 do not necessarily reflect a change in the industrial distribution of participants’ post-IA jobs. In order to assess the extent to which SSP might have led to a distributional change in the industries, Table 3 presents the industries for only those who left IA and subsequently worked.
This is a non-experimental comparison, which means that there are differences, both observable and unobservable, between the program and control group members represented in the table. Members of the control group who left IA and worked did so without the offer of a financial incentive. It is, therefore, likely that working control group members possessed characteristics associated with work more often than members of the program group did. For example, control group members who worked after leaving IA may have faced fewer barriers than their counterparts in the program group. The control group members represented in Table 3 might also have been more likely to possess unmeasured characteristics, such as motivation or a preference for work outside the home. Because the experimental and control group members being compared in Table 3 are not the same, on average, any differences between the program and control group cannot be attributed entirely to SSP.
The non-experimental results in Table 3 show that among those who worked after leaving IA, program group members found jobs in different industries than did control group members. Program group members were 6.1 percentage points more likely to work in accommodation, food and beverage services and other services industries and 5.3 percentage points more likely to be working in wholesale and retail trade industries. In contrast, program group members worked less often in government and educational services industries and health and social services industries than did their counterparts in the control group. If it was more difficult to find jobs in these industries, program group members, concerned about finding a first full-time job in order to qualify for supplements within the 12-month take-up window, may have avoided these industries.
Within an industry, occupations can vary substantially. While one worker in the food services industry might be a manager earning $40,000 per year, another might be a cashier earning $20,000. In some ways, therefore, occupation is a more important indicator of the characteristics of a job. In the second panel of Table 2, the occupations of SSP participants are reported. Because all members of the sample are represented in Table 2, the differences between the program and control groups are experimental impacts.
The largest impact was on the proportion working in service occupations; while 13.5 per cent of the program group worked in these occupations, only 8.6 per cent of the control group worked in these occupations. SSP also increased work in clerical and related occupations by 2.8 percentage points and increased work in sales occupations by 2.4 percentage points.
The second panel in Table 3 shows a non-experimental comparison of the occupations of those program and control group members who worked after leaving IA. Among those who left IA and worked, participants who were offered the financial incentive were almost 3 percentage points more likely to be working in a sales occupation. Program group members were 2.2 percentage points less likely to be working in teaching and related occupations and 1.3 percentage points less likely to be working in occupations in the primary sector.
The non-experimental results, when combined with the experimental results, suggest that SSP had relatively little influence on the different occupations in which participants found their first post-IA job. Welfare recipients appear to work in only a very limited subset of occupations. Although SSP encourages participants to choose work over welfare, when SSP program group members work, they find jobs, for the most part, in the same occupations that other former welfare recipients do—primarily in service, sales and clerical occupations.
Impacts on Wages, Hours, Job Duration and Job Stability
In this section, the impact of SSP on characteristics of participants’ first post-IA job other than occupation and industry are analyzed.
As the first panel of Table 4 demonstrates, the largest proportion of SSP participants who left IA and went to work did so in a job that paid within $0.99 of the provincial minimum wage. SSP had the largest impact on work in this category; SSP increased the proportion working in jobs that paid less than or equal to $0.99 above minimum wage by 9.2 percentage points. SSP also encouraged people to work in jobs that paid between $1.00 and $1.99 above minimum wage and jobs that paid between $2.00 and $2.99 above minimum wage. Although a relatively large proportion of the program group, nearly 10 per cent, earned wages that were $3.00 or more above the minimum wage, they were no more likely than control group members to be working in these jobs.
A concern with any program that supplements wages and earnings is that the program will encourage participants to accept lower wages than they might otherwise have accepted. Table 4 provides some evidence that SSP did not encourage participants to accept lower wages than they otherwise would have accepted. The assumption required here is that any differences in the percentages of the treatment and control group in each wage category are created by the individuals who would not have worked in the absence of SSP. If there had been a negative impact in, say, the highest wage category, this assumption would imply that SSP caused some individuals to switch between the highest wage category and a lower wage category. But the impacts in Table 4 are almost all positive, albeit with larger impacts in the lower wage categories. This means that the “extra” workers took relatively low paying jobs—though not necessarily lower than they otherwise would have accepted. Thus, it would appear that the supplement offer did not encourage participants to accept wages lower than they might have in the absence of the program. Analysis of experimental impacts on average wages across all jobs held in month 33 reached similar conclusions (Michalopoulos et al. 2000).
Although SSP did not appear to encourage people who would have worked in the absence of the program to work in jobs with lower pay, there is evidence that the work SSP encouraged was in low-wage jobs. SSP did not have an equal impact on jobs offering various wages. The impact on the proportion working in jobs that paid less than $1.00 above minimum was almost twice as large as the combined impact on all jobs with wages that exceeded minimum wage by $1.00 or more.
The designers of SSP were concerned that the additional income provided by the supplement would cause some people to reduce their work effort. To mitigate against this possibility, a full-time work requirement was attached to the supplement (see Box 1).
The results in the second panel of Table 4 suggest that the full-time hours requirement was effective. Although many program and control group members left IA and worked in a job with fewer than 30 weekly hours, SSP did not increase employment in these kinds of jobs. Almost all of the additional post-IA employment generated by SSP was in jobs that offered at least 30 hours per week.
For this article, job duration is defined as a period of continuous employment with a single employer that is uninterrupted by an unpaid break of more than two weeks in length. Job duration is difficult to measure accurately because the period over which participants are observed is finite. Some respondents left IA and found work earlier than others. For respondents who found work earlier, the study had a longer time horizon in which to observe their job duration. Thus, shorter job duration may be observed for some individuals, not because they ultimately would have worked for their employer for less time, but because they found work later in the follow-up period.
Since we are using data for only three years after random assignment, the job durations are censored at the last interview date for which the participant was a survey respondent. This means that if the respondent was currently working when they were interviewed, then their job was assigned an end date equal to the interview date. The last panel of Table 4 shows that 13 per cent of the control group and 15 per cent of the program group worked in post-IA jobs that were censored. Of all the post-IA jobs held, 35 per cent were censored.
The fourth panel in Table 4 shows the impacts on post-IA jobs of various durations. Because the data are censored, these results must be interpreted with caution. SSP increased employment in jobs in all duration categories. The largest proportions of both groups worked for more than 12 months. Among the program group, 21.4 per cent worked for more than 12 months, compared to 16.6 per cent in the control group. This impact of 4.7 percentage points on jobs with durations longer than 12 months was the largest impact that SSP had on any duration category. This does not mean that SSP necessarily increased job durations. SSP did accelerate the process of leaving IA and finding work. Because of this, researchers are more likely to observe job durations lasting more than 12 months.
While some participants may have lost their jobs, others may have left their jobs voluntarily. Program group members might have accepted the first job they were offered in order to initiate the supplement within the 12-month take-up window. They might then have subsequently searched for better employment. An analysis of jobs that followed the first post-IA job might determine whether job switching was responsible for the lower proportion of post-IA employed program group members with job durations longer than 12 months. This is a question for future research to consider.
Impacts on Employer-Sponsored Benefits and Union Membership
Workers can receive compensation for their labour in the form of wages and also employer-sponsored benefits. Such benefits may include pension plans, health or dental plans, and childcare benefits. An important example is that, while on social assistance, parents are usually eligible for drug and dental benefits and other supplemental health services. These benefits might represent an important source of income. Upon leaving social assistance, some SSP participants could lose such benefits. Finding a job that offered these benefits could have made an important difference to whether welfare recipients decided to leave IA.
Table 5 shows that virtually all of the additional post-IA employment generated by SSP occurred in jobs that did not offer any benefits. SSP increased employment in jobs without any benefits by over 11 percentage points, compared to an impact of 3 percentage points on jobs with at least one employer-sponsored benefit.
Membership in a union can generate many benefits for workers. The collective bargaining activities in which unions engage are associated with wage premiums that can be particularly high for low-skilled workers (Simpson 1985; Card 1996). Unions are also associated with non-wage work benefits, including sick leave, vacation and some of the employer-sponsored benefits discussed earlier (Freeman 1981). In the past, unionized work environments tended to have better safety regulations and worker grievance procedures than non-union counterparts. The advantages of a unionized environment may have diminished over time, however, as provincial legislation has been extended to protect all workers from unsafe environments and from harassment and discrimination. Yet, within the narrow range of occupations and industries in which social assistance recipients find jobs, it is possible that union jobs are better than non-union jobs.
The second panel of Table 5 shows that union work is relatively rare among welfare leavers. Only about 2 per cent of both the program and control groups left IA and worked in a unionized job; SSP had no impact on those proportions. All of the additional employment that SSP created was in non-unionized jobs. Nearly 36 per cent of the program group worked in non-unionized jobs, compared to about 22 per cent in the control group, leading to an impact of 14 percentage points.
Impacts on Job Quality
As established from the literature, making objective observations about the quality of a job is a difficult task. Yet, only when dimensions of quality are identified does it become possible to draw conclusions about which jobs might be better than others. Thus, developing measures of job quality is a useful endeavour. This final section of the article attempts to do so.
As was noted previously, three different dimensions of job quality can be identified from the literature: the nature of the work, future job prospects and compensation. Although all three of these dimensions entail some form of subjective assessment, this is particularly true of the first aspect of quality. Because of the subjectivity required to describe the nature of one’s work, this analysis does not attempt to identify any such measures of quality. The authors recognize that subjective notions of quality, while not easily measured, are nonetheless important.
This article, drawing from the second and third dimensions of quality, suggests some job characteristics that can be considered positive: job duration, wages, hours and the availability of employer-sponsored benefits.
With respect to future job prospects, the article considers measures of job duration. When individuals remain with employers, they are able to develop skills which improve their productivity and might therefore lead to wage progression or promotion. In so far as it can be linked to wage growth, job duration can be used to gauge quality.
If jobs that last longer are better, then how long is long enough for a job to be considered of reasonable quality? When workers have accumulated enough hours of paid work, the Employment Insurance (EI) program insures their employment. A worker would have to work full-time for approximately 6 months, or 700 hours, before their employment was insurable. For this reason, the following analysis adopts durations of at least 6 months as a positive job characteristic.
In terms of compensation, the article examines wages, hours and employer-sponsored benefits. Holding other aspects of the job constant, a higher wage could certainly be considered better than a lower wage. Because many social assistance recipients experience very low incomes, a positive job attribute might be a wage that pays enough so that a full-time full-year worker would earn at least as much as Statistics Canada’s Low-Income Cut-Off (LICO).
Earnings are determined not only by wages but also by the hours worked. Full-time hours are considered a positive job characteristic for this reason. The SSP study adopts the Statistics Canada definition of full-time work, which is 30 or more hours per week.
Employer-sponsored benefits are another form of compensation. Some individuals might value some benefits more highly than others. It would not be appropriate to suggest that some benefits are better than others. Instead, for the following analysis, the availability of any benefits is considered to be a positive job characteristic.
Table 6 shows the impact that SSP had on jobs with the four positive job characteristics, which are:
at least one employer-sponsored benefit;
hourly wage high enough for a full-time full-year worker to earn the equivalent of the LICO;
full-time work; and,
job duration at least 6 months.
While SSP increased employment in post-IA jobs with one positive characteristic by 4 percentage points, the impact on jobs with two positive characteristics was more than double that. Nearly 19 per cent of the program group went to work in jobs with two positive characteristics after leaving IA, compared to only 10 per cent in the control group. SSP also had an impact on employment in jobs with 3 or 4 positive characteristics, but it was much smaller, at less than 2 percentage points.
In summary, when SSP encouraged people to leave IA and take up work, the jobs they found had at least one positive characteristic. SSP had no impact on work in jobs with no positive characteristics. This result is probably driven by the full-time work requirement. Most of the employment generated by SSP was full-time work, which is a positive job characteristic. There is evidence, however, that SSP also encouraged employment in jobs with other positive characteristics.
This article has demonstrated that when social assistance recipients leave welfare and take up work, they tend to find jobs in a relatively narrow range of occupations and industries. Their jobs are concentrated in sales, service and clerical occupations. Evidence reported in the article showed that the generous SSP earnings supplement increased overall employment but did little to increase the range of occupations and industries in which welfare “leavers” first worked.
The earnings supplement offered to SSP program group members seems to have encouraged participants to choose work as an alternative to welfare. However, SSP had comparatively little effect on the kind of work that participants found. There is little evidence that SSP encouraged individuals who would have worked in the absence of the program to select lower quality jobs. Instead, SSP encouraged employment in the jobs that were most common among welfare leavers. SSP generated employment in jobs that offered low wages and few, if any, benefits. On the other hand, participants who left IA and found work because of SSP did tend to work in full-time jobs and experienced longer observed job durations.
The finding that SSP increased employment in jobs that were no worse than the jobs that participants might have taken in the absence of the program could be a result of the unique features of the program. For example, SSP was voluntary. Participants were not required to work and if they did choose to work, they could return to IA at any time. Other programs with different features such as mandating employment might not produce the same results.
Generally, the findings in this article suggest that there are some aspects of job quality that can be improved by interventions that directly target employment in jobs with particular characteristics. SSP increased employment in full-time jobs because the financial incentive was structured to reward only that kind of work. If other job characteristics were identified as positive, interventions may find success by targeting those types of jobs. For example, policy makers could choose to target union work or jobs in specific industries and occupations. Such a strategy would clearly require co-operation between social service agencies and groupings of employers or unions.
That strategy, however, has its limitations. Although some jobs might be better than others within the narrow range of jobs that welfare recipients can typically access, this strategy does nothing to broaden the array of opportunities for welfare recipients. On the whole, the jobs that are realistic alternatives to welfare are worse than the jobs that most Canadians hold.
In focus groups that were part of the SSP evaluation, some members of the program group cited the lack of good job opportunities as an important reason that they did not take up the supplement. One participant commented, “I don’t have education or skills where I’m able to get a nice job. You know, I’m just, like [a] minimum-wage type, and I feel guilty, but I don’t want to do that, I cannot see myself working down at the mall for $5.50 an hour.” (Bancroft and Currie Vernon 1995: 33). Other non-takers in the program group felt that any job they found would be a dead end job that would ultimately lead back to IA, as one participant explained, “...if you have grade 7 education, there’s no way you’re bettering yourself in your job. So three years down the road, after having all this money, you’re going to go back to welfare and say, ‘I can’t make it. Give me my welfare back.’” (p. 33).
To combat the discouraging job prospects available to most welfare recipients, a policy strategy might seek to improve the opportunities that are available to social assistance recipients when they leave welfare. Offering incentives to undertake training or education that improves individuals’ qualifications may broaden the array of occupations and industries in which welfare recipients obtain jobs. Further experimental research might also reveal how other employment services might assist welfare recipients compete for better jobs or seek advancement within their jobs.
The article focuses throughout on pooled results from British Columbia and New Brunswick. Michalopoulos et al. (2000) suggest that there were no important provincial differences in the impacts on economic outcomes.
In Canada, many welfare recipients are male. Nonetheless, because SSP participants were primarily female lone-parents, we focus on issues facing female welfare recipients.
Jencks, Perman and Rainwater (1998) created a job quality scale based on a special U.S. survey that collected detailed information on a wide variety of job characteristics. The survey also asked workers to rate how “good” their job was compared to the average job. By regressing job characteristics on the workers’ ratings, Jencks et al. developed an index of job quality. In the index, they tried to limit themselves to “objective” characteristics (e.g., reported wages and hours) as opposed to characteristics that seemed more subjective (e.g., reports that the work was interesting). The job characteristics about which survey information was collected were drawn from an extensive review of the literature on the variables that affect job satisfaction. Positive characteristics included educational requirements, wages, the existence of on-the-job training, weeks of vacation, control over own hours, and the existence of a union contract.
SSP operated in the lower mainland of British Columbia and the lower third of New Brunswick. Funded by the Human Resources and Development Canada (HRDC), SSP was managed by the Social Research and Demonstration Corporation (SRDC) and jointly evaluated by SRDC and the Manpower Demonstration Research Corporation (MDRC).
Early in the evaluation of SSP, however, Mijanovich and Long (1995) provided a preliminary description of supplement takers’ occupations. Using job titles to categorize the occupations of the supplement takers held in first 26 weeks after take-up, Mijanovich and Long (1995) reported that the largest proportion of supplement takers were working in service occupations. The next most common were clerical and sales occupations.
For a discussion of non-response bias, see Michalopoulos et al. (2000: 87–93).
Because such a large proportion of the sample was female, feminine pronouns are used throughout the article.
Participants were asked about how many paid hours they usually worked during each of their continuous employment spells with one employer. A continuous spell is uninterrupted work without a unpaid break longer than two weeks.
The structure of Table 1 is modeled after similar tables in Bartik (1997).
This includes only women who were working and reported an industry that could be coded according to the SIC.
This includes only women who were working and reported an occupation that could be coded according to the SOC.
While it was possible that the financial incentive encouraged some people to seek training and education, there is little evidence that SSP increased education and training. Indeed, 18 months after random assignment, program group members who did not have a high school diploma at baseline were statistically significantly less likely than their counterparts in the control group to have taken courses toward a high school diploma or a trade/vocational certificate (Lin et al. 1998).
The assumption described is necessary because it is impossible to know for certain whether or not program group members accepted lower wages as a result of the supplement offer. Some program group members would have worked even if they had not been offered the supplement. In order to determine whether SSP caused program group members to accept lower wages, it would be necessary to distinguish between program group members who would have worked in the absence of SSP from those who would not have worked in the absence of the program.
The SSP study followed respondents for 54 months in total, but only 36-month data was available at the time this article was written.
Effective April 1996, British Columbia extended some dental and vision care benefits to children under the age 12 living in low income working families that were not covered by federal or employer sponsored programs. In 1997, these benefits were further extended to children 18 years of age and younger. In New Brunswick, some welfare recipients retain their benefits for a limited period of time after beginning work. However, IA case managers report that these extended benefits are not well advertised. Qualitative evidence suggests that having extended health and dental benefits is important for participants in both British Columbia and New Brunswick (Bancroft and Currie Vernon 1995).
The number of hours required to qualify for EI depends on the local unemployment rate. This measure assumes an unemployment rate up to 6 per cent. In areas with high unemployment rates, fewer hours would qualify for EI.
Low-Income Cut-Off is a measure of low income created by Statistics Canada and varies depending on family size and the population of the region in which the person lives.
The particular economic and political environment in which SSP was implemented may also have affected the results. Michalopoulos et al. (2000) provide a discussion of the effect of such possible effects.
A great deal of research has been undertaken on the efficacy of further education and training for social assistance recipients. For an accessible discussion, see http://www.mdrc.org/inpractice/qff5_answer.htm.
- Acs, G., and P. Loprest. 2001. Initial Synthesis Report of the Findings from ASPE’s “Leavers” Grants. Washington: U.S. Department of Health and Human Services.
- Bancroft, W., and S. CurrieVernon. 1995. The Struggle for Self-Sufficiency: Participants in the Self-Sufficiency Project Talk about Work, Welfare, and their Futures. Vancouver, B.C.: Social Research and Demonstration Corporation.
- Bartik, T. 1997. Short-Term Employment Persistence for Welfare Recipients: The “Effects” of Wages, Industry, Occupation, and Firm Size. Kalamazoo, Mich.: W. E. Upjohn Institute for Employment Research.
- Card, D. 1996. “The Effect of Unions on the Structure of Wages: A Longitudinal Analysis.” Ecta, Vol. 64, 957–979.
- Freeman, R. 1981. “The Effect of Unionism on Fringe Benefits.” Industrial and Labour Relations Review, Vol. 34, 489–509.
- Gorlick, C., and G. Brethour. 1998. Welfare to Work Programs: A National Inventory. Ottawa: Canadian Council on Social Development.
- HRDC. 2000. Job Futures. Ottawa: Author. Retrieved October 9, 2001, from the World Wide Web: http://jobfutures.ca/doc/jf/index.shtml.
- Jencks, C., L. Perman, and L. Rainwater. 1998. “What is a Good Job: A New Measure of Labor-Market Success.” American Journal of Sociology, Vol. 93, 1322–1357.
- Lin, W., P. K. Robins, D. Card, K. Harknett, and S. Lui-Gurr. 1998. When Financial Incentives Encourage Work: Complete 18-Month Findings from the Self-Sufficiency Project. Ottawa: Social Research and Demonstration Corporation.
- Loprest, P. 1999. Families Who Left Welfare Who Are They and How Are They Doing? Assessing the New Federalism Discussion Paper 9902. Washington, D.C.: The Urban Institute.
- Michalopoulos, C., D. Card, L. A. Gennetian, K. Harknett, and P. K. Robins. 2000. The Self-Sufficiency Project at 36 Months: Effects of a Financial Work Incentive on Employment and Income. Ottawa: Social Research and Demonstration Corporation.
- Michalopoulos, C., D. Tattrie, C. Miller, P. K. Robins, P. Morris, D. Gyarmati, C. Redcross, K. Foley, and R. Ford. 2002. Making Work Pay: Final Report on the Self-Sufficiency Project for Long-Term Welfare Recipients. Ottawa: Social Research and Demonstration Corporation.
- Mijanovich, T., and D. Long. 1995. Creating an Alternative to Welfare: First-Year Findings on the Implementation, Welfare Impacts, and Costs of the Self-Sufficiency Project. Vancouver, B.C.: Social Research and Demonstration Corporation.
- Simpson, W. 1985. “The Impact of Unions on the Structure of Canadian Wages: An Empirical Analysis with Microdata.” Canadian Journal of Economics, Vol. 18, 164–181.
- StatisticsCanada. 1996a. Labour Force 15 Years and Over by Detailed Industry (Based on the 1980 Standard Industrial Classification) and Sex, for Canada, 1996 Census (20% Sample Data) Retrieved on November 26, 2001 from the World Wide Web: http://www.statcan.ca/english/census96/mar17/occupa/table3/t3p00t.htm
- StatisticsCanada. 1996b. Labour Force 15 Years and Over by Detailed Occupation (Based on the 1991 Standard Occupational Classification) and Sex, for Canada, 1996 Census (20% Sample Data ) Retrieved on November 26, 2001 from the World Wide Web: http://www.statcan.ca/english/census96/mar17/occupa/table1/t1p00t.htm