institut c.d. howe communiqué

C.D. Howe InstituteInstitut C.D. Howe Communiqué

Embargo: For release Wednesday, April 9, 1997

Alberta welfare reformsa model for other provinces,

says C.D. Howe Institute study

The welfare reform program that Alberta embarked on in 1993 has reduced the province’swelfare caseload, as a percentage of its population, to levels not seen since before the early1980s’ recession, a success that has important lessons for other provinces, concludes aC.D. Howe Institute Commentary released today.

The study, Back to Work: Learning from the Alberta Welfare Experiment, was written byKenneth J. Boessenkool, a Policy Analyst at the C.D. Howe Institute.

Boessenkool points out that, until 1993, shifts in welfare use in Alberta followed nationaltrends: during recessions, the number of recipients shot skyward, but in subsequent periodsof economic growth, the number failed to fall to prerecession levels. Then, beginning in 1993,Alberta decided to make welfare a program of last resort by implementing a two-stage reformprocess.

The first stage involved a change in the administrative culture of Alberta Family and SocialServices, as a result of which welfare applicants are now routinely turned away unless theyhave exhausted all other sources of support. In the second stage, the province brought benefitlevels in line with wages earned by Albertans with low incomes. Together, these reforms appearto have contributed to a nearly 50 percent decline in the number of Albertans on welfare,Boessenkool says.

The decline was accomplished mainly by reducing the number of new welfare cases,rather than by pushing off the rolls individuals who were already receiving welfare. “In late1992, over 14,000 individuals with no previous history of welfare use came onto the caseloadeach quarter,” Boessenkool reports. “By late 1995, that number had dropped to 4,000.” Most ofthose who were denied access to welfare, he says, were young, single, employable individuals.

As to what happened to Albertans who were turned away from welfare, Boessenkool findslittle evidence to support the assertion that failed applicants migrated to British Columbia totake advantage of more generous provisions in that province. Data on the province of originof welfare recipients in British Columbia, he says, fail to reveal a change in the proportion fromAlberta, despite the massive swings in Alberta’s welfare rolls. Boessenkool also finds no

evidence that failed applicants may have turned to crime, and little evidence of increased useof federal unemployment insurance (UI) benefits.

There is evidence, however, that a portion of those clients who were refused benefits foundalternative sources of support in the form of provincial education and training programs aswell as federal training programs under UI — programs that cost far more than regular benefits.The annualized cost of the voluntary workfare program, for example, was more than threetimes the single, employable benefit level.

Most of these programs were of relatively short duration, however, and therefore cannotaccount for the continued drop in Alberta caseloads over the 1993–96 period he surveyed,Boessenkool says. Nonetheless, he argues, these programs, including grants to allow individu-als to pursue postsecondary education, did play a role in reducing the number of new welfarecases and in increasing the human capital of those who, under the former system, might haveended up on passive welfare assistance.

Boessenkool examines evidence that individuals who were turned away from welfarereturned to work. He points out that the Alberta economy grew more slowly during the 1993–96period than it did during the 1980s, when welfare caseloads in Alberta and other provinces fellvery little, if at all, and he argues: “Those who point to economic growth as the reason whyAlberta welfare use was cut in half must also explain why previous stronger and longer periodsof growth both in Alberta and in other provinces failed to have anywhere near that effect.”

Boessenkool notes that total employment growth in Alberta in 1994 and 1995 was thestrongest in more than ten years and that, for the first time in a decade, employment growthfor those most affected by the welfare reforms kept pace with economic growth: “During the1980s and early 1990s, a barrier prevented Albertans under age 25 from enjoying the rewardsof economic growth in the form of new jobs. The administrative and benefit changes made tothe Alberta welfare program appear to have played a role in lowering that barrier,” Boessenkoolconcludes. “Alberta may have discovered a key link to the youth employment mystery.”

Boessenkool cautions that his analysis is suggestive rather than conclusive: “Had Albertakept track of those who were turned away from welfare, one could make a more confidentappraisal. Unfortunately, these data appear not to exist, and merely tracking where formerrecipients went would not provide an accurate sense of what happened to the many individualswho were turned away from welfare and therefore never became ‘former’ recipients.”

“Policy experimentation is a great asset of Canadian federalism,” Boessenkool concludes.“The 1993 Alberta welfare reform is one experiment from which all Canadians and theirgovernments can benefit.”

* * * * *

The C.D. Howe Institute is Canada’s leading independent, nonpartisan, nonprofit economic policy researchinstitution. Its individual and corporate members are drawn from business, labor, agriculture, universities,and the professions.

- 30 -

C.D. Howe Institute / Institut C.D. Howe Communiqué / 2

For further information, contact: Ken Boessenkool; Susan Knapp (media relations),C.D. Howe Institute

phone: (416) 865-1904; fax: (416) 865-1866;e-mail: [email protected];

Internet: www.cdhowe.org/eng/pr/new.html

Back to Work: Learning from the Alberta Welfare Experiment, C.D. Howe Institute Commentary 90, by KennethJ. Boessenkool (C.D. Howe Institute, Toronto, April 1997). 32 pp.; $6.00 (prepaid, plus postage & handlingand GST — please contact the Institute for details). ISBN 0-88806-409-8.

Copies are available from: Renouf Publishing Company Limited, 5369 Canotek Road, Ottawa, OntarioK1J 9J3 (stores: 71 1/2 Sparks Street, Ottawa, Ontario, phone 613-238-8985; 12 Adelaide Street West, Toronto,Ontario, phone 416-363-3171); or directly from the C.D. Howe Institute, 125 Adelaide Street East, Toronto,Ontario M5C 1L7.


C.D. Howe InstituteInstitut C.D. Howe Communiqué

Embargo : à diffuser le mercredi 9 avril 1997

Les réformes de l’assistance sociale en Albertasont un modèle pour les autres provinces,affirme une étude de l’Institut C.D. Howe

Le programme de réforme de l’assistance sociale qu’a entrepris l’Alberta en 1993 a diminué lenombre d’assistés sociaux dans cette province, exprimé en pourcentage de la population, à unniveau qui n’avait pas été observé depuis la période précédant la récession du début des années80; il s’agit là d’une réussite qui comporte d’importantes leçons pour d’autres provinces,conclut un Commentaire de l’Institut C.D. Howe publié aujourd’hui.

L’étude, intitulée Back to Work: Learning from the Alberta Welfare Experiment (Retour autravail : les leçons de l’expérience albertaine en matière de bien-être social), est rédigée par KennethJ. Boessenkool, un analyste de politique à l’Institut C.D. Howe.

Boessenkool souligne que jusqu’en 1993, les changements dans l’utilisation du bien-êtresocial en Alberta suivaient les tendances nationales : durant les récessions, le nombre d’assistéssociaux grimpait en flèche, mais durant les périodes de croissance économique qui ont suivi,ces nombres diminuaient sans retomber aux niveaux de prérécession. Puis, à partir de 1993,l’Alberta a décidé de faire du bien-être social un programme de dernier recours par la mise enœuvre d’un processus de réforme en deux étapes.

La première étape portait sur un changement dans la culture administrative des servicessociaux et à la famille Alberta, qui a fait que les requérants sont maintenant automatiquementrefoulés à moins d’avoir déjà épuisé toutes les autres sources d’aide. Dans une deuxième étape,la province a aligné les niveaux de prestations sur les salaires des Albertains à faible revenu.Ces réformes semblent avoir réussi à diminuer de près de 50 p. 100 le nombre d’Albertains quireçoivent de l’assistance sociale, explique Boessenkool.

Cette diminution découle principalement d’une réduction du nombre de nouveauxprestataires, plutôt que d’une élimination des individus qui l’étaient déjà. « Vers la fin de 1992,plus de 14 000 personnes n’ayant encore jamais eu recours au bien-être social s’ajoutaient aunombre de prestataires chaque trimestre, indique Boessenkool. À la fin de 1995, ce chiffre étaittombé à 4 000 ». La plupart de ceux qui se sont vus refuser l’aide sociale, ajoute-t-il, étaient desindividus jeunes, célibataires et aptes au travail.

Pour ce qui est du sort des Albertains à qui l’on a refusé l’aide sociale, Boessenkool indiquequ’il ne semble guère établi que les requérants rejetés ont émigré vers la Colombie-Britannique

pour profiter des dispositions plus généreuses de cette province. Les données fournies par lesassistés sociaux de la Colombie-Britannique sur leur province d’origine n’indiquent pas dechangements dans le pourcentage d’individus provenant de l’Alberta, et ce en dépit desénormes fluctuations dans le nombre de prestataires en Alberta. Boessenkool ne trouve aucunepreuve que les requérants rejetés se sont tournés vers des activités criminelles, et peu depreuves d’une utilisation accrue des prestations fédérales d’assurance-chômage (a.-c.).

Par contre, il existe des preuves qu’un pourcentage des clients à qui l’on a refusé desprestations ont eu recours à d’autres formes de soutien, sous la forme de programmesprovinciaux d’éducation et de formation, ainsi que des programmes fédéraux de formationsous les auspices de l’a.-c. — des programmes qui sont bien plus coûteux que les prestationsordinaires. Le coût annualisé du programme de travail volontaire, par exemple, était trois foisplus élevé que la prestation à une personne célibataire et apte au travail.

Toutefois, la plupart de ces programmes sont relativement de courte durée et ne peuventdonc expliquer la diminution constante du nombre de prestataires en Alberta durant la périodequ’il a étudiée, soit de 1993 à 1996. Boessenkool soutient néanmoins que ces programmes, ycompris les bourses permettant aux individus de poursuivre des études postsecondaires, ontcontribué à la réduction de nouveaux cas d’assistés sociaux et à accroître le capital humain deceux qui, sous l’ancien système, auraient pu se retrouver à l’assistance sociale passive.

Boessenkool recherche l’hypothèse selon laquelle les individus à qui l’on a refusé l’aidesociale sont retournés au travail. Il souligne que l’économie albertaine a connu un essor pluslent durant les années 1993 à 1996 que durant les années 80, lorsque le nombre d’assistés sociauxtant en Alberta que dans les autres provinces n’a baissé que sensiblement, et il émet lecommentaire suivant : « Ceux qui invoquent la croissance économique pour justifier la dimi-nution de moitié des dépenses d’assistance sociale en Alberta, doivent alors expliquer pourquoidurant des périodes d’essor plus fort et plus soutenu en Alberta et dans d’autres provinces, onn’a pas observé le même phénomène ».

Boessenkool indique qu’en 1994 et en 1995, la croissance de la population employée enAlberta fut la plus importante depuis plus de 10 ans et que pour la première fois en unedécennie, la croissance de l’emploi pour ceux qui ont été le plus touchés par la réforme del’assistance sociale a suivi le rythme de la croissance économique. « Au cours des années 80 etau début des années 90, des obstacles empêchaient les Albertains de moins de 25 ans de profiterdes avantages de la croissance économique sous la forme de nouveaux emplois. Or, lesmodifications administratives et de prestations que l’on a apportées au programme de bien-être social semblent avoir contribué à réduire ces obstacles, conclut Boessenkool. L’Alberta apeut-être découvert la clé du mystère du chômage chez les jeunes ».

Boessenkool prévient qu’il s’agit là d’une analyse suggestive et non concluante : « Sil’Alberta avait suivi ceux à qui elle a refusé l’assistance sociale, on serait en mesure d’établirune évaluation plus solide. Malheureusement, ces données ne semblent pas exister; en nesuivant que les anciens assistés sociaux, on ne peut véritablement avoir une idée de ce qui estadvenu des nombreux individus à qui l’on a refusé l’aide sociale et qui ne sont donc jamaisdevenus “d’anciens” prestataires ».

« Les essais de politique constituent un important avantage du fédéralisme canadien,conclut Boessenkool. La réforme de l’assistance sociale de 1993 en Alberta en est un exempledont tous les Canadiens et leurs gouvernements peuvent tirer profit ».

* * * * *


L’Institut C.D. Howe est un organisme indépendant, non-partisan et à but non lucratif, qui joue un rôleprépondérant au Canada en matière de recherche sur la politique économique. Ses membres, individuels etsociétaires, proviennent du milieu des affaires, syndical, agricole, universitaire et professionnel.

– 30 –

Renseignements : Ken Boessenkool;Susan Knapp (relations avec les médias)

Institut C.D. Howetéléphone : 416 865-1904;

télécopieur : 416 865-1866c. électr. : [email protected]

Internet : http://www.cdhowe.org/fr/index.html

Back to Work: Learning from the Alberta Welfare Experiment, Commentaire no 90 de l’Institut C.D. Howe, parKenneth J Boessenkool, Toronto, Institut C.D. Howe, avril 1997, 32 p., 6,00 $ (les commandes sont payablesd’avance, et doivent comprendre les frais d’envoi, ainsi que la TPS — prière de communiquer avec l’Institutà cet effet). ISBN 0-88806-409-8.

On peut se procurer des exemplaires de cet ouvrage auprès des : Éditions Renouf ltée, 5369, chemin Canotek,Ottawa ON K1J 9J3 (librairies : 71 1/2, rue Sparks, Ottawa ON, tél. 613 238-8985 et 12, rue Adelaide ouest,Toronto ON, tél. 416 363-3171), ou encore en s’adressant directement à l’Institut C. D. Howe, 125, rueAdelaide est, Toronto ON M5C 1L7.


Welfare Policy

Back to Work:

Kenneth J. Boessenkool

before

Main Findings of the Commentary

• Over the past 15 years, welfare use has increased dramatically in Canada —ratcheting up during recessions but not falling to prerecession levels duringsubsequent periods of growth.

• Welfare use in Alberta had a similar history up to 1993, when the provinceintroduced important changes to its welfare program.

• The first change was in the administrative culture at the Alberta Department ofFamily and Social Services. First-time welfare applicants were routinely turnedaway (unless they could prove exceptional circumstances, such as family abuse)until they had exhausted all other sources of support.

• The second change was a reduction in welfare benefits to bring them into line withthe wages earned by lower-income Albertans.

• Following these reforms, Alberta welfare caseload levels began to decline, eventu-ally falling by nearly 50 percent.

• Closer inspection reveals that Alberta accomplished its reductions primarily bylessening new inflows into welfare, rather than pushing individuals off the rolls.

• Young, single, employable individuals were the people most affected by thechanges.

• What happened to the individuals who were turned away from welfare? Althoughthe evidence is tentative,

• there is little support for the proposition that they went to British Columbia,went on federal unemployment insurance benefits, or turned to a life of crime;

• there is some support for the proposition that individuals turned to federal andprovincial training programs, including postsecondary education, financed bygrants from the Alberta Students Finance Board.

• There is also support for the proposition that many went back to work. Adjustedfor economic growth, job growth was particularly strong in the year following thereforms. In particular, the long-standing failure of youth employment to keep upwith the economy appears to have been reversed.

• The Alberta experiment holds lessons for other Canadian provinces: administra-tive culture matters, as do benefit levels, particularly those levels in relation to theearnings of low-income individuals.

• Data problems are pervasive in looking at Canada’s welfare system. An improvedsystem of recordkeeping, perhaps run by the provinces, would do much to increaseunderstanding of this important part of Canada’s safety net.

During the 1980s and early 1990s, thenumber of Canadians on provincialwelfare increased nearly every year. Inbad times, welfare enrollments shot

skyward. In good times, they fell slightly insome jurisdictions and rose in others, butacross Canada after each recession, welfarecaseloads stayed well above the level that pre-ceded the downturn.

In 1993, Alberta began a determined effortto reduce its welfare caseload. The objectivesincluded integrating income support and em-ployment-related services, providing new em-ployment and training opportunities,improving access to mainstream student as-sistance, and removing disincentives to workand train.1

Fiscal savings also motivated the changes.The Alberta government began a period ofsubstantial fiscal retrenchment in 1993, andthe welfare package was expected to — andeventually did — contribute to the overall re-duction in spending.

In the three years following the changes,Alberta’s welfare caseload fell by nearly half(from just under 95,000 to just under 50,000).The consequent reduction in costs exceededthe fiscal targets set in 1993. Although theshift affected individuals across the entire demo-graphic spectrum, much of the decline was theresult of preventing young, single, employableindividuals from gaining access to welfare.

Alberta’s own review of the changes2 fo-cuses on the caseload and fiscal impacts,largely ignoring the question of what happenedto individuals who might formerly have beenon welfare. Did they relocate to jurisdictionswith more generous welfare provisions? Didthey go into the underground economy, or,worse, were they forced into a life of crime?Were they redirected to other government pro-grams? Or did they return to work? The lackof clear answers to these questions hasprompted headlines such as “After all the cut-backs, many welfare cases have gone, butwhere?”3

The analysis here, based on a macro evalu-ation of the changes, suggests that a large

number of former and potential welfare recipi-ents were directed toward training programs.There is also tentative evidence that manyfound their way back to the workforce.

Other provinces are now remodeling theirwelfare programs, or thinking about doing so.British Columbia and Ontario recently beganchanges, while Quebec, which has shelved thechanges it initiated in the late 1980s, is in theearly stages of its own efforts. It is important,therefore, to learn from the Alberta experience.What is known about the effects of the changes?How might other provinces adjust their welfaresystems in that light? What sort of monitoringsystems would be desirable?

This Commentary is a preliminary attemptto answer these questions. In the openingsection, I provide some background, includingan explanation of the way social assistance isorganized in Alberta. The second section de-scribes the changes instituted in 1993, andthe third evaluates their impact. In the sub-sequent two sections, I examine several expla-nations of where individuals and families wentonce they were off or refused welfare; somescenarios seem much more likely than others.Implications for welfare reform in other Cana-dian jurisdictions conclude the study. (For acomment on the data used, see Box 1.)

The Context

During the past 15 years, the total number ofwelfare cases in Canada has ratcheted up-ward. They have increased during recessionsbut have never declined to prerecession levelsduring subsequent periods of economicgrowth (see Figure 1). The early 1980s’ reces-sion increased caseloads and, despite strongemployment and economic growth, the cross-Canada total did not fall over the 1985–90period (a modest decline was erased by a risein 1989). With the recession in the early 1990s,caseloads rose sharply, and did not fall in 1993and 1994 despite modest economic growthand fairly strong employment growth.

Individual provinces, of course, had variedexperiences. Of the four largest, Ontario and

C.D. Howe Institute Commentary / 3

Alberta more or less tracked the cross-Canadarecord throughout the 1980s, but British Co-lumbia and Quebec managed to translate thestrong employment and economic growth ratesof the late decade into a reduction in welfarecaseloads. The recession at the beginning ofthe 1990s sent recipient levels upward in allfour provinces, with the increase most pro-nounced in Ontario.

Following that recession, the number ofrecipients in British Columbia, Ontario, andQuebec declined modestly but stayed well aboveprerecession levels. Alberta’s welfare program,however, took a distinctly new path in 1993.Its recipient levels began falling quickly anddecisively. By 1996, Alberta recipients as apercent of population had fallen to levels notexperienced since 1982.

Benefit levels also diverged provincially. In1986, benefits for single, employable individu-als in Alberta were 53 percent higher than theaverage in British Columbia, Ontario, andQuebec, and the rate for married couples withchildren was 22 percent higher. But by theearly 1990s, Ontario had become the high-paying large province, with benefit levels about40 percent higher than those of the otherthree.4

The Alberta Welfare Program

Before I review the 1995 changes in Albertawelfare policy, a quick synopsis is in order.

Alberta’s Department of Family and SocialServices (AFSS) delivers two programs: a wel-fare program called Supports for Independence(SFI), and Assured Income for the SeverelyHandicapped (AISH).

AISH provides support for many individu-als with severe physical and mental disabili-ties. But individuals whose need exceeds AISHbenefit levels (because they have dependents,for example) are transferred to SFI, from whichthey receive benefits from a subprogram, As-sured Support, that is designed for people whoare assessed as permanently unemployable(“unable to work”) or who face multiple severeemployment barriers, such as age, low educa-tion, and/or inadequate skills (“unavailablefor work”).

From 1993 through 1996, the period un-der study, the number of cases in AISH andthe related SFI subcategory of Assured Sup-port grew at a rate similar to that of previousexperience. Aside from a slight increase inbenefits, the changes during and following1993 were not directed at those with severe

Box 1: The Data Problem

One of the important reasons for analysts’ relativeinattention to provincial welfare programs — es-pecially compared to the attention lavished on thefederal unemployment insurance program — isthe difficulty in finding consistent data acrossprovinces.

Each March, the federal government collectsdata on total caseloads and the number of indi-viduals on welfare (a family of four is counted asone case, but obviously represents four individu-als). These data are available and fairly easy touse for interprovincial comparisons, but the in-formation they provide is limited. More specificdata, by demographic or employment charac-teristics, can often be found only at the provinciallevel. This information is based primarily oncaseloads, and the provinces differ considerablyin records they keep, as well as in the way they

categorize cases. Each may also alter its proce-dures over time, making the construction of timeseries difficult, if not impossible.

For the interprovincial comparisons in thisCommentary, I report the federal data, usingindividuals (as a percentage of population), ratherthan caseloads, to make the likenesses and con-trasts meaningful. All the monthly or quarterlyfigures come from data provided by Alberta’sDepartment of Family and Social Services. Al-berta substantially changed its classification anddata collection system midway through 1993.Portions of the data for December 1992 andMarch 1993 were recalculated by the departmentto provide a continuous data series. Because thelatter recalculation was much more extensivethan the former, I use March 1993 as the initialdata point for my analysis.

4 / C.D. Howe Institute Commentary

disabilities. Therefore, the remainder of thisCommentary focuses on the SFI program, andall the Alberta statistics given below are basedonly on SFI clients (including those on AssuredSupport).

The Changes

In 1993, Alberta made two kinds of changes toits welfare program. Early in the year cameadministrative changes, such as tightened eli-gibility and a renewed focus on misuse. Laterthe same year, it adjusted its welfare benefitlevels.

A New Administrative Culture

Alberta began its welfare reform with changesin the administrative approach of AFSS, mak-ing admission to the welfare rolls more difficultto achieve and emphasizing subsequent checksto reduce misuse of the system.

A Shift in Emphasis

The Alberta minister who implemented thereforms was Mike Cardinal, whose experiences,

both as an aboriginal and a former socialworker on native reserves in northernAlberta, had led him to believe in thedamaging effect of long-term welfare de-pendency. As he says in a representativestatement:

[t]he system was a corrupting influence,and was devastating to the native popu-lation in particular. Prior to 1950, com-munities in northern Alberta wereindependent from government and com-pletely self-sufficient. Everybody worked,there was no welfare system, we had ourown health care system, alcoholism wasvery limited, family breakdowns were verylimited, people practiced their culture,and lived off the land in a traditional way.

We changed that with good intentions,but within 15 or 20 years (by 1970) 80 to90 percent of the members in those com-munities had moved onto the welfare sys-tem completely....I know, and we all know

in Canada, that welfare is not the best wayof dealing with poverty and unemploy-ment.5

Cardinal concluded that the emphasis in wel-fare programs should shift from providing fi-nancial assistance to providing assistance inreturning to the workforce. During the late1980s and early 1990s, under the auspices ofthe previous minister, he piloted many of hisideas in poor, mostly native, welfare-depend-ent northern Alberta communities. These pilotprojects formed the backbone of the reformsimplemented when he became minister in late1992 and presided over a major reorientationof the culture of AFSS.

Departmental statements insisted that SFIwas a program of “last resort.”6 (Official docu-ments frequently say as much, but in Albertathe reality seems to have corresponded to therhetoric.) After 1993, applying for welfare be-came a two-step process for individuals deemedemployable. Officials routinely denied the firstapplication (making exceptions in cases of hard-ship, such as family abuse7). The second stagewould commence only if the applicant couldshow the welfare office that he or she hadtaken steps to exhaust other avenues, such astraining programs and employment searches.

Figure 1: Canadians on Welfare, Canadaand Selected Provinces, 1981–96

0

2

4

6

8

10

12

14

1981 1983 1985 1987 1989 1991 1993 1995 1

ABCDE

%of

pop

ula

tion

QuebecBritish Columbia

Ontario

Alberta

Canada

Sources: Canada, Department of Human Resources Develop-ment; and various provincial departments.


Cardinal’s views were reinforced by grow-ing evidence that the propensity to receiveincome support increases with an initial stinton that support. Program participants “learn”how to use the system and alter their economicbehavior to take advantage of it.

How important is this effect? A recent studyof unemployment insurance (UI) finds that oneepisode of collecting it permanently increases,by 4.2 to 12.3 percent, the probability of anindividual’s returning to the program.8 Theauthors explain:

As workers became exposed to UI for thefirst time for a variety of reasons, theylearned about the functioning of the sys-tem and adjusted their behavior accord-ingly....[f]or lower skilled workers, beingemployed on a part-year basis may be anattractive long runoption.For theseworkers,a first experience may have a significanteffect, therefore, on the future probabilityof receiving UI benefits.9

This effect is likely to be more pronounced forwelfare than for UI because welfare recipientsare likely to come from the bottom of theincome distribution and often have low skills.10

By making an initial stint on welfare moredifficult, Alberta could permanently reduce itslevels of welfare over the longer run.

Fraud and Eligibility Reviews

The other administrative change has been arenewed focus on clients’ eligibility for thebenefits they receive, thus reinforcing the goalof preventing individuals from getting on thewelfare rolls.11 Although the department com-mitted no additional funds to these activities,the declining caseload meant an increase inemphasis on them.

Three programs in particular stand out.First, department review officers perform fol-low-up checks of new clients to verify eligibilityfor benefits. Such reviews resulted in closureof 22 percent of the cases checked from Aprilto November 1995.

Second, 30 investigators and 12 adminis-trative support staff review complaints of fraud

and misuse. From April to December 1995, forexample, the department received 6,204 com-plaints, of which it investigated 1,468. Theresults were 288 charges and 188 convictions,which netted $1.3 million in court-orderedrestitution and $3.5 million in out-of-courtsettlements.

Third, department staff also randomlycheck files and conduct home visits to confirmboth eligibility and the appropriate level ofbenefits. From November 1994 to November1995, officers reviewed 21,000 files and con-ducted 7,000 home visits. This work resultedin the closure of nearly 1,500 files and therevision of many others for overpayment ofbenefits; total savings amounted to nearly$1 million.

Unfortunately, the sparse nature of thedata available makes it difficult to determinewhether the success of these initiatives in-creased after 1993. Nevertheless, the depart-ment’s own reviews suggest that its admin-istrative focus on tighter eligibility require-ments applied not only to potential clients butto existing ones as well.

Benefit Changes

Following the administrative changes thattook hold in early 1993, AFSS moved to thesecond stage of its reform — a reduction insome benefits to ensure that “those on socialassistance received a level of support not ex-ceeding that earned by working Albertans.”12

Alberta did not cut its benefits uniformlyfor all recipients. Rather, it focused its mostsevere reductions on single, employable indi-viduals. They faced a 19 percent decline inbenefits, which brought the amount just belowthe minimum wage. A single parent with asmall child faced a 13 percent decline, and atwo-parent family with two children a 12 per-cent decline.

Did the Alberta government cut benefitstoo far? Had its benefits before the cuts beentoo high? Answering these questions requiresa foray into the economics of designing welfareprograms.


The design of any social assistance pro-gram faces a fundamental tradeoff. The bene-fits floor level must be high enough to providean acceptable level of income to a householdwith no outside earnings. To target only thosehouseholds, the marginal tax rate on earnedincome must be high enough to exclude mostpeople who work at low wages. But — and hereis the tradeoff — this high marginal tax rate isa disincentive to earn income. If the ultimateaim of welfare policy is to return employableindividuals to work, program designers mustaddress this tradeoff.

Factors Difficult to Quantify

The raw level of welfare benefits and the tax-back rate clearly affect this tradeoff. So do anumber of other factors. Some of them arequantifiable, and I discuss them below. Othersare more difficult to quantify but are importantto keep in mind when evaluating the level ofwelfare benefits in any jurisdiction.

One of these factors is the relative cost ofworking versus being on welfare. In setting awelfare benchmark, outlays for items such astransportation, appropriate clothing, and toolsor equipment play an important role in deter-mining whether a particular recipient is betteroff working than on welfare. These costs aredifficult to quantify because they differ accord-ing to the skill-set, location, and many otherpersonal characteristics of recipients. For ex-ample, unattached singles do not have to worryabout childcare, but its cost may provide sin-gle parents with a strong incentive to acceptwelfare benefits rather than a slightly higherwage.13

A second difficult-to-quantify factor is thein-kind benefits available to welfare recipients,such as low-cost housing, free prescriptiondrugs, and dental care for themselves or theirchildren. The employment income at whichthese benefits cease, coupled with the avail-ability of comparable employment benefits,may strongly influence the work-welfare deci-sion. Thus, any comparison of benefit levelsacross jurisdictions or over time must consider

how individuals’ incentives are affected by in-kind benefits and the cutoff levels for theirreceipt (which may differ from the cutoffs forcash benefits).

The Welfare Wall

A common image in thinking about benefitdesign is the welfare wall, which is created bythe interaction between levels of benefits andtaxes, including taxbacks. The wall appearswhen the interaction results in a prohibitivetax rate, sometimes more than 100 percent,which can affect the economic incentives ofwelfare recipients who are contemplating earn-ing some income.14

Before the 1993 benefit reductions, theAlberta welfare taxback rate for single indi-viduals was 90 percent (after a small earningsexemption of about $200). On top of that,Canada Pension Plan premiums had to be paidon earnings of more than $3,500, and UIpremiums and provincial income tax kicked infor earnings of more than $7,000.

Thus, a single recipient who earned $5,000in a year ended up only $308 better off than asimilar recipient who earned $1,000 (see Ta-ble 1). And welfare benefits ceased entirely ifearned income reached $5,571.

When Alberta lowered benefits in 1993, italso modestly decreased the welfare taxbackrate to 75 percent. The nature of the funda-mental tradeoff changed a bit, but the totalmarginal rate facing individuals who earnedbetween $4,000 and $6,000 is still a prohibi-tive 80 percent.

Under this new regime, the difference be-tween the two welfare recipients has increasedapproximately threefold. But the absolute dif-ference is only $908, total takehome is lessthan it was for the same effort before thereform, and welfare payments will cease whenearned income reaches $5,793 (a slightlyhigher cutoff than before the changes).

In brief, people on welfare still have littleincentive to find low-paying or part-time work,especially if doing so means an increase inwork-related costs or the loss of in-kind benefits.


Primarily because the taxback rate remainshigh, the changes have done little to dismantlethe welfare wall for those who remain on wel-fare. The lower benefit levels, however, shouldhave increased incentives to forgo welfare forwork. The new wall is not much lower, but itsnew placement means fewer people are forcedto scale it.

Relative Benefit Levels

The welfare wall is an important considerationin setting taxback rates. However, it offers littleguidance for setting the appropriate absolutelevel of benefits, a task for which some sort ofbenchmark is well suited. One such guide isthe level of wages actually earned by workersnear the bottom of the wage scale (see Boxes 2and 3 for a discussion of alternate bench-marks).

The relationship between actual wages andwelfare benefits is important because havingthe latter exceed the earnings of a sizableportion of the workforce creates incentives forlow-paid workers to turn to welfare. The like-

lihood of this disincentive’s having aneffect is probably partially related to indi-viduals’ previous experience with incomesupport programs. Recall the previousdiscussion of the learning curve that seemsto affect individuals’ propensity to usesuch programs. Low-income workers, whoare vulnerable to economic fluctuations,may find themselves looking for help dur-ing an economic shock. If they go on awelfare program whose benefits are rela-tively high, they quickly learn that theireconomic position has improved. Oncethe economic shock has passed, this im-provement may tempt them to continueto receive higher welfare benefits ratherthan take an income cut by returning tothe workforce.15

Welfare benefits above actual earn-ings levels may also undercut the effec-tiveness of training and placementprograms. Suppose an individual com-

pletes such a program and finds that the jobsimmediately available provide remunerationnear the bottom of the wage scale while return-ing to welfare would provide larger benefits. Hewill clearly be tempted to forgo the job and itslong-term advantages for the higher incomelevel on welfare.

Overall, if welfare benefits appear too highrelative to the lower portion of the wage scale,reducing them should mitigate the temptationto turn to welfare. Many employables will endup with lower takehome incomes, but that paywill come from the workplace, rather thanwelfare, a situation society has two reasons toprefer. First, employment maintains or in-creases human capital, which erodes for peopleon welfare. Indeed, human capital deprecia-tion is a severe problem, especially for youngand likely low-skilled employables. A reduc-tion in human capital at a young age can bringabout a dramatic reduction in lifetime earn-ings. The downside of lower benefits should,therefore, be offset over time by the gains tothe stock of human capital.

Second, welfare use can have long-termintergenerational effects. Longitudinal studies


in the United States suggest that the childrenof parents who work have much better educa-tional and labor market attainments than chil-dren whose parents have had previous stintson welfare.16

Movement over Time

What has been the historical relationship be-tween earnings levels and welfare benefits inAlberta? Between 1982 and 1993, the ratio ofwelfare benefits to after-tax income for singleindividuals stayed remarkably constant. Wel-fare benefits as a percentage of after-tax second-decile earnings straddled 40 percent (Table 2).

The same ratio for couples with two childrenstayed closer to 100 percent. In terms of theless-reliable first decile (see Box 2), the ratiobounced around a bit more, but showed adownward trend over this period.

The October 1993 benefit changes (reflectedin the 1994 figures in Table 2) moved the ratioof benefits to after-tax income at the seconddecile for single employables to 34 percent, a17 percent decline from its previous ten-yearaverage. For couples with children, the ratiofell to 91 percent, a more modest 10 percentdecline. In terms of first-decile earnings, themagnitude and direction of the changes weresimilar.

Box 2: Alternative Benchmarks for Setting Welfare Benefits

Finding an appropriate benchmark for welfarebenefits is fraught with methodological as well asideological minefields. Not surprisingly, there areas many possibilities as there are commentatorswho use them.

Roughly speaking, benchmarks are either rela-tive or absolute. Those who prefer relative meas-ures prefer a benchmark designed to highlightdisparities in income, while those who preferabsolute measures prefer a benchmark designedto highlight deprivation.

The National Council on Welfare, for example,compares welfare income to the low-income cut-off (LICO) and to the average income in eachprovince. Both measures are, more or less, rela-tive ones* and may be important to show therelative gap between various levels of income. Yetvarious conceptual difficulties suggest that theseindicators are insufficient by themselves. TheLICO, for example, is extremely income sensitive.The 1995 LICO for single people living in largecities was a little less than $17,000. Welfarebenefits set at 50 percent of LICO may soundunappealingly low — yet Canada’s per capitadisposable income (adjusted for inflation) was atthis level only 30 years ago.** Comparing welfarerates with the average wage also gives little morethan an idea of the relative position of welfarebenefits to earned income and thus is of little helpin setting absolute levels of benefits.

The relative nature of these measures has ledsome to prefer a more absolute benchmark. Onesuch possibility is a needs-based benchmark —for example, the cost of a basket of goods deemed

essential.† It seems inarguable that benefit levelsshould not fall below what is required for basicsurvival, and a needs-based measure can providea useful floor.

Yet needs-based measures also have their dif-ficulties. What is a necessity in the eyes of oneperson may well be a luxury in the eyes of an-other. Furthermore, a wealthy society such asCanada may wish to provide welfare recipientswith more than the bare necessities. Thus, adefensible value for a ceiling is needed. One can-didate is income levels at the bottom end of theincome distribution. The economic rational isthat welfare benefits should not exceed what anindividual could reasonably earn outside the pro-gram. This measure is a relative one, but it fitswell with the goal of ensuring that incentives forthose on welfare are biased toward working, ratherthan staying on welfare.

* Technically, the LICO is needs based in that thecalculation involves individuals’ or families’ spending20 percent more than the average on food, shelter, andclothing. The current LICOs, however, still use a basecalculated in 1978, which is updated using the con-sumer price index but has not been adjusted forchanges in income levels or in the relative costs of food,shelter, and clothing. The LICO has thus become, forall intents and purposes, a relative measure.

**These numbers appear in William M. Scarth andWilliam B.P. Robson, Equality and Prosperity, PolicyStudy (Toronto: C.D. Howe Institute, forthcoming).

† For an example, see Christopher Sarlo, Poverty inCanada, 2d ed. (Vancouver: Fraser Institute, 1995).


These changed ratios should have changedthe incentives for individuals facing the labormarket. For single employables, welfare in-come fell well below average first-decile earn-ings, which would have improved theirincentives for taking these low-paying jobs.Similarly, the incentives should have improvedfor individuals in households with children,although the effect would have been smaller,given the smaller decline in the benefits-to-earnings ratio.

The Impact

How did the reforms affect welfare in Alberta?The short answer is that, between early 1993and mid-1996, the number of welfare benefi-ciaries in the province declined by nearly half,from about 95,000 to just under 50,000, andthe cost of benefits fell by nearly 40 percent.(The reduction in benefit levels and recipientswas somewhat offset by increases in benefitsto some categories of people, primarily theseverely disabled.)

Because the administrative and benefitchanges took place at separate times (the for-mer early in 1993, and the latter near the end

of that year), it is possible to determine theimpact of the two independently. Fortunately,sufficient data also exist to determine the kindsof people who were most affected by thesechanges.

New and Reopened Cases

Alberta, like other provinces, has a sizableturnover of welfare cases each month. In 1993,for example, both the number of opened cases(112,497) and closed cases (128,032) exceededthe highest caseload level in that year. Thisturnover suggests the possibility of two quitedifferent approaches to reducing welfare roles:increasing the number of cases closed, anddecreasing the number opened.

The result of Alberta’s administrative focuson reducing entrants to welfare shows up in areduction in the number of “opened” (new andrepeat) caseloads between 1993 and 1996.During the second half of 1992, that numberhad been about 37,000 per quarter. As earlyas mid-1993, it had fallen to 25,000, and byearly 1996 to about 20,000.

Opened cases can be divided into newcases (clients with no previous contact with

Box 3: Using Actual Wage Levels as a Benchmark for Welfare Benefits

In considering the level of welfare benefits overtime, every program designer and analyst facesthe problem of what to measure them against. Asthe text suggests, a good theoretical benchmarkis an amount slightly less than a low marketwage. The target should be based on wages forindividuals who are employed full time and for afull year. A suitable target would be within thefirst decile of such earnings.* But obtaining suchinformation on market wages presents somepractical difficulties.

Statistics Canada publishes long historical se-ries of quintiles of all market earnings, but specialruns are necessary to generate deciles for employ-ment earnings. The data used in most of thisCommentary are average earnings in each decilebased on full-year, full-time wage and self-employ-ment earnings for males. Since these data includeself- employment income (which can be negative),

the average amounts tend to fluctuate sharplyover time, and the average first decile earningstend to be unrepresentative of actual earnings atthis level. The second decile is more stable and,therefore, more convenient for analysis acrosstime.

The data in Table 4 use a more recent set ofunpublished data that include the top earningsin each decile, so the first decile is less prone tofluctuations. (The cost of getting the entire his-torical series of data on this basis would havebeen prohibitive.)

* Although few Canadian families earn this level ofincome (unattached individuals made up 78 percentof the lowest quintile), those that include children stillface a labor market in which potential earnings beginat this level. Statistics Canada, Income after Tax:Distributions by Size in Canada, 1994, cat. 13-210.


the welfare system over a specified period) andrepeat cases (Figure 2). In Alberta, the reduc-tion in the number of new cases was the mostimportant cause of the falloff. In late 1992,they had been about 18 percent of thecaseload; by 1996, they had stabilized at

7 percent. The number of repeat cases,however, increased as a percentage of thetotal caseload.

Indeed, the shifting numbers revealan interest phenomenon. Effective July1993, Alberta redefined “reopened” casesfrom persons who had received SFI bene-fits in the previous six years to those whohad received benefits in the previous twoyears. This change should have de-creased the proportion of repeat casesand increased the proportion of newones. Yet repeat cases failed to fall (sug-gesting that most had access to welfarewithin the previous two years), and newcases continued to decrease.

Also interesting is the fact that theproportion of closed cases decreased verylittle during the period reflected in Fig-ure 2.

In brief, only a small portion of the de-crease in the Alberta welfare caseload can becredited to moving existing cases off the rolls.The significant reduction came from a sharp

Figure 2: Welfare Caseloads, Alberta,mid-1992 to mid-1996

0

5

10

15

20

25

30

35

40

45

50

1992 1993 1994 1995 1996

ABCD

%of

tota

lca

selo

ad

closed cases

a Opened cases in which the clients had received benefitssometime in the previous two years, had subsequently leftthe welfare system, and then returned.

b Opened cases in which the clients had no record of benefitsin the previous two years.

Source: Alberta, Department of Family and Social Services.

repeat casesa

new casesb


decrease in individuals who were applying forwelfare for the first time.

The most important falloff in new clientsoccurred early 1993, and thus much of thecredit must go to administrative, rather thanto benefit, changes (although there was cer-tainly some acceleration of moving people offwelfare once the benefit changes took place).This conclusion could be made firmer hadAlberta tracked application and acceptance/rejection rates. Unfortunately, the provincedoes not publish this data.

Demographics

What were the demographic characteristics ofthe Albertans who were kept from the welfarerolls? Since much of the impact of welfarereform was in reducing inflows, it would havebeen extremely helpful if the Alberta govern-ment had kept more detailed track of thecharacteristics of its applicants and closures.No such records exist, however, so a demo-graphic profile must rely on changes to the netnumber of welfare clients, rather than tochanges in flows.

Comparing the demographics of the wel-fare caseload in early 1993 with that of early1996 suggests that young, single people madeup a large portion of the total reduction. InMarch 1993, single individuals made up 47 per-cent of the welfare caseload. Over the followingthree years, singles made up almost half of thetotal caseload decline (Table 3). Yet despitetheir massive falloff, single individuals stillmade up 47 percent of the welfare case-load inMarch 1996. In other words, although singlesmade up a substantial portion of the decline,they continue to be a significant part of thetotal caseload.

Single parents made up 24 percent of thetotal decline. As a proportion of the total case-load, however, they increased from 30 percentin March 1993 to 36 percent in March 1996.

In terms of age, welfare rolls were heavilyweighted toward the young, with 26 percent ofthe total in March 1993 consisting of peopleunder age 25 (and 60 percent under 35). While

the total caseload fell by 48 percent from March1993 to March 1996, clients under age 25 fellby 64 percent; teenaged clients alone fell by 75percent. And clients under age 35 made up 70percent of the total caseload reduction.

The unambiguous conclusion is that thechanges in Alberta prevented the young fromgaining access to welfare.

Client Categories

The classification of welfare recipients pro-vides another piece of the puzzle. Alberta di-vides recipients into five categories. Panel C ofTable 3 reveals that changes across the catego-ries varied considerably. Some had increasesfrom 1993 to 1996; in particular, the numberof clients in training programs nearly doubled.But overwhelming the rises were decreases,especially in the “available for work” category.

These numbers suggest two things. First,Alberta tightened eligibility requirements for“available” and ”unavailable for work” clients.Second, following the changes, a number ofindividuals received training rather than wel-fare benefits — that is, they moved from pas-sive to active assistance.

Another notable observation is that thecaseload for the Assured Support programgrew by 22 percent over the three years. Thisincrease was in line with the increases in theAISH program described previously.

Profile

The kind of generalizations I have just pro-vided inevitably exclude important segmentsof individuals who were affected by the Albertawelfare reforms. Nevertheless, it is possible toconclude that, from early 1993 to early 1996,Alberta managed to keep from its welfare rollsa substantial number of employable, young,single individuals who were applying for wel-fare for the first time.


Where PotentialClients Did Not Go

The sizable falloff in Alberta’s welfare recipi-ents, especially single, young employables,leads to a natural and potentially importantquestion: What happened to them?

Unfortunately, no specific information ex-ists (see Box 4). I believe it possible, however,to use the profiles developed above and someother macro information to rule out some pos-sibilities and suggest the likelihood of others.

Such an exercise seems worthwhile in theface of the plethora of theories that have beendeveloped, often backed up with little evidence.Sorting through this speculation is important.If 50,000 potential and former welfare recipi-ents merely migrated from Alberta to British

Columbia, turned to crime, or transferred tofederal income support programs (possibilitiesconsidered in this section), the Alberta modelis not necessarily one that should be copied. Amore favorable view would result from evi-dence that the changes prompted individualsto return to work or to use training programsto enhance human capital and eventually theirjob potential (possibilities considered in thenext section).

British Columbia

One much-repeated theory is that the reduc-tion in Alberta welfare clients was a result ofmigration to British Columbia. But evidence ofa massive shift of cases is difficult to find.


British Columbia caseloads did increasesteadily in the early 1990s. However, the trendstarted in 1991 and continued through 1996,despite changes in that province’s level of eco-nomic growth.

An independent and confidential survey ofsocial assistance clients written for the BritishColumbia Ministry of Social Services for theperiod August 1993 to May 1994 concludedthat former Alberta welfare cases made up onlya very small portion of the surging BC load.17

This study may be a less-than-adequate guideas to what happened, however, since much ofthe reduction in Alberta came from individualswho may not have been on assistance in thatprovince. Fortunately, the BC government be-gan tracking new welfare clients from Albertasoon after the latter revised its benefit levels inlate 1993. These data show that Albertansmade up a fairly stable 28 to 31 percent ofBritish Columbia’s out-of-province recipientsfrom late 1993 to mid-1996 (Figure 3). Duringperiods of large reductions in Alberta’s case-load, British Columbia experienced only smallchanges in flow from that province (in particu-lar, a spike in early 1994).

In fact, the most noticeable changes in BCcases from Alberta was after November 1995,when British Columbia imposed a residencyrequirement. And although that impact de-creased the overall flow, it did not affect theproportion of clients coming from Alberta. In-deed, the proportion of BC welfare clients fromAlberta seems to have been remarkably resil-ient against both changes in Alberta welfare

policy and the imposition of British Columbia’sresidency requirement. (For an alternative ex-planation of the trends in British Columbia,see the Appendix.)

A Life of Crime

Some people have suggested that Alberta’sformer welfare clients may have been forced to“sleep in the streets.”18 In the extreme, a largeshift of individuals onto the streets might re-sult in their turning to criminal activity as away of providing for their basic needs and thusin an increase in property-related crime. Noevidence supports the idea that this hypotheti-cal scenario was played out in Alberta, how-ever. Indeed, the province’s property crime ratereached a high in 1991 and has been droppingsteadily ever since — by 24 percent between1992 and 1994, the years in which the drop inwelfare caseloads was greatest. (Although alogical connection to other types of crimes isever more difficult to make, Alberta has alsoseen a steady decline in its violent crime ratesince peaking in 1991.)

Without a long list of additional variables,this evidence does not conclusively suggestthat former or potential welfare recipients arenot “sleeping in the streets.” It does, however,rule out a massive movement of welfare recipi-ents toward a life of crime. (Whether individu-als turned to the underground economy is thefocus of a subsequent subsection.)

Box 4: Looking for Potential Recipients

Short of conducting extensive follow-up inter-views, there is no way of being certain whereformer, or more important, potential recipients ofwelfare in Alberta have gone. Furthermore, sinceAlberta primarily reduced its welfare caseload bypreventing potential recipients from gaining ac-cess to welfare, it is unlikely that interviewstracking former welfare recipients would shedmuch light on what happened. Only if Alberta hadkept track of and interviewed applicants who

were turned away or referred elsewhere wouldsuch an exercise be fruitful.

A caveat is, therefore, in order. Because mostof the evidence I use is not based on trackingindividuals who were turned away or left welfare,it is suggestive, rather than conclusive. Yet thesheer magnitude of the reduction in caseloads inAlberta suggests that the macro methods usedhere should be able to shed some light on theplight of potential and former recipients.


Onto UI Benefits

Yet another theory is that many former andpotential Alberta welfare recipients turned tofederal unemployment insurance. The logichere is that the system gives provinces incen-tives to provide temporary jobs in order toqualify individuals on provincial welfare pro-grams for UI benefits. In the first half of the1990s, as few as 20 weeks of work entitled anindividual to unemployment benefits for theremainder of a year. Thus, a provincial govern-ment could provide make-work employmentfor the short period, while the federal govern-ment picked up the tab for the long one. Thisapproach could be a profitable venture forprovincial governments if the cost of providingjobs for 20 weeks was less than the cost ofpaying benefits for 52.

There is little evidence, however, that Al-berta employed this make-work strategy (al-though it did provide some jobs that welfareclients had the option of accepting — see thefollowing section). The number of initial andrenewal UI claims in Alberta reached 282,000in 1992, fell over the next two years, and rose

again in 1995 in line with the nationalreduction in job growth.

The number of claims did increase bymore than the national average in 1995(by 8 percent in Alberta versus 3 percentnationally), suggesting that some indi-viduals coming off provincial work pro-grams may have applied for UI. However,the time spent by Albertans on UI fellthroughout the 1993–95 period. The av-erage number of weeks paid in regularbenefits per month in Alberta peaked in1992 at 347,000 weeks. That numberdropped in 1993 and fell quickly in thefollowing years to reach 223,000 weeksby 1995.19 All of these trends were, more-over, exactly mirrored in British Colum-bia, a province that had similar job andeconomic prospects through this period.

Again, this evidence is not conclusive,but if a significant portion of the formerwelfare caseload shifted to UI benefits,

Alberta’s experience should not have followedBritish Columbia’s as closely as it did.

Where PotentialClients Might Have Gone

As part of its shift from passive to active assis-tance, Alberta worked to divert potential wel-fare clients to various training, job-hunting,and educational programs with the ultimategoal of returning these individuals to work.This section reviews enrollment in these pro-grams and examines evidence that suggeststhat some potential clients did return to work.

Provincial Work andTraining Programs

Alberta’s shift to a focus on active assistanceentailed programs that provided temporaryemployment, job placement, search assistance,and training. Some of the programs were newbut many were existing ones expanded to in-clude the new flow of recipients.

To pay for these programs, Alberta divertedto them some of the savings from its loweredwelfare costs. Between fiscal years 1992/93


and 1995/96, the number of welfare cases wasreduced by more than 50 percent, but the totalAFSS budget fell by only 20 percent (from$1.72 billion to $1.37 billion). Clearly, the costper recipient rose over this period — in partbecause of increased benefits for disabled in-dividuals and in part because of costs fortraining and education programs.

The reduction in expenditure came pri-marily from grants to individuals and deliverycosts, which fell by 46 percent over this periodfor a savings of $404 million. “Other” depart-mental expenditures declined by $32.4 mil-lion. Of the remaining departmental spending,employment and training expenditures rosefrom $10 to $32 million, AISH and spendingfor the disabled increased by $26 million, andWidows Pension and Child Welfare combinedwent up $18 million. Thus, the savings inwelfare benefits were partially shifted to otherdepartmental priorities, with the largest in-crease for the disabled and smaller increasesfor employment and training expenditures.

The bulk of new spending came from pro-grams administered through or jointly withAlberta Education and Career Development(AECD). In particular, AFSS transferred fundsto the Students Finance Board to finance edu-cation grants for potential and existing welfareclients.

Overall, what role did work and trainingprograms play in reducing welfare caseloadsin Alberta? The short answer is: a big, al-though often temporary, role. Between March1993 and March 1996, more than 35,000former and potential welfare clients wentthrough the various work, training, and edu-cation programs made available by AFSS and/or AECD.20

Work Projects

Although workfare was not formally part of theAlberta changes, AFSS did initiate several workprojects aimed at long-term welfare recipients.There were three voluntarily programs, whichdiffered primarily by the type of employer.

The Alberta Job Corps (AJC) included workprojects sponsored by local communities, non-

profit groups, and businesses in northern Al-berta, Edmonton, and Calgary. The AlbertaCommunity Employment Program (ACE) in-cluded work projects created by nonprofit so-cieties, municipalities, hospitals, health units,and schools. Individuals in these two pro-grams received $5 per hour plus benefits. Thethird program, the Employment Skills Pro-gram (ESP), provided $6 per hour for jobs invarious provincial government departments.Participants’ maximum stay in these programswas six months, with the average in 1994ranging from 2.9 months for the AJC to 3.4months for the ESP.

What impact did these programs have onthe welfare caseload while individuals wereinvolved in them? Singles who participatedwere removed from the caseload since theirearnings now exceeded welfare rates, makingthem ineligible for supplementary welfare bene-fits. However, families whose primary earner’stotal wages under the program were less thanwelfare received a top-up. On balance, justfewer than one-half of individuals who partici-pated in these programs continued to receivesupplementary welfare benefits and were thusrecorded on the welfare caseload.21

These job creation programs did not comecheap. From mid-1993 to mid-1996, 12,071individuals took part in them. The total expen-diture for the three over this period was$55.7 million, which works out to an averagecost of $4,614 per participant for three months(the average employment duration) for an an-nualized cost of $18,456 per participant. Thatamount is more than triple the benefit rate forsingle employables. It is also more than therate for married couples — but recall thatmany of them receive a welfare top-up in ad-dition to program involvement.

During this period, provinces received fed-eral assistance for welfare under the CanadaAssistance Plan (CAP), whose rules preventedthem from giving recipients payments condi-tional on accepting work or training. Albertawas able to skirt this restriction, however,since Ottawa’s cuts to the CAP released aportion of provincial spending from the federal


requirements.22 (With the CAP’s demise in 1996,the restriction disappeared for all provinces.)

A number of factors muted the role theseprograms played in reducing caseloads from1993 to 1996. First, half the participants re-ceived supplementary welfarebenefitsand there-fore remained on the welfare caseload. Second,the number of individuals taking part wasquite modest during 1993 and 1994, the pe-riod of largest decline in welfare recipients.Finally, these programs were aimed at longer-term users of welfare, who collectively madeup a relatively small portion of the total decline.

It is difficult to determine long-term suc-cess rates for the job creation programs. Theevaluations that are public (and that reportvery high success rates) are non-arm’s-lengthones made by the department itself and involvelittle long-term tracking of recipients. A longer-term, more accurate arm’s-length accountingfor these individuals would provide a bettersense of their success.

Training and Education

The primary success of the Alberta experimentcame, as already explained, from divertingpotential clients away from welfare. Many ofthem were directed to alternate programs, inparticular, those run in partnership betweenAFSS and AECD.

To forward these referrals, welfare officesand provincial employment centers in manycommunities merged to offer all services underone roof. (Selected pilot projects are under wayto offer all related services — UI benefits,federal training programs, provincial trainingand work programs, and provincial welfareprograms — in one place.) In addition, theStudents Finance Board took on many clientsreferred by welfare administrators.

Most of these programs were not new, butthe demand for them increased as administra-tors referred clients following a first applica-tion for welfare. To facilitate this shift, AFSStransferred money to AECD to expand existingprograms to handle the additional demand.

Some of this money went into job-trainingand placement programs, which enrolled 3,371

individuals in 1993/94 and 4,807 in fiscalyear 1994/95.23 Many programs, such as ca-reer workshops and employment-planning ses-sions by AFSS and/or AECD staff, were ofshort duration. Increased funding also went tocareer hotlines, job referral services, and labormarket information centers. The administra-tive changes of early 1993 required all newwelfare applicants to exhaust these services.

A significant number of individuals wentinto education or skills-training programs. In1993/94, AFSS transferred 11,000 clients tothe Students Finance Board. These individu-als moved into the mainstream student sys-tem: those with low education levels receivedtraining in foundation skills, such as academicupgrading, literacy training, and English as asecond language training; others were directedto skills-training programs at technical insti-tutes, colleges, and universities.24 The StudentsFinance Board provided grants to individualswho were upgrading their education, whilefurther funding was available through Alberta’ssystem of student loans.

This training was much more expensivethan passive welfare costs. The provincial gov-ernment used grants and loans to cover thecost of tuition and living expenses, paying forthe remainder through its regular funding ofeducational institutions. (The average annualinstitutional cost per learner in universitiesand technical Institutes was more than$10,000. Vocational colleges were the cheap-est at $4,541, and other colleges were about$8,620 in 1995.25)

Many of these programs take two years ormore to complete, so it is still difficult to tellwhether the additional education will preventformer clients from returning to welfare. Butthe early signs give reason for optimism. Al-though more than three years have passedsince the reforms began, there is little evidencethat individuals directed to education are nowturning to welfare. Indeed, the welfare rollshave continued to fall (though at a much re-duced pace).

Furthermore, it is no secret that increasedemployment and earnings follow increased edu-


cation.26 Indeed, the role of the additionaltraining was to prepare former or potentialwelfare recipients for entry or reentry into thelabor force.

Federal Training Programs

An important part of the welfare changes inAlberta was the shift in emphasis from passiveto active benefits, partly by diverting welfareapplicants to active programs that were al-ready available. Not surprisingly, therefore,there is some evidence of an increase in Alber-tans’ use of federal training programs duringthe period being examined.27

Specifically, the number of people receiv-ing Developmental Uses benefits offered underUI grew during the early 1990s.28 The largestrise occurred during 1991 — the wrong yearto provide any proof that the province’s welfarebeneficiaries were shifted, en mass to the fed-eral program. The increase in Alberta was alsoproportionately smaller than the national av-erage increase, and use declined during 1993and 1994.

Yet comparison of Alberta’s post-1993 datawith those of British Columbia, a province withsimilar economic prospects over this period,suggests that Alberta did benefit from Devel-opmental Uses programs somewhat more heav-ily than did British Columbia through 1994 and1995. While the number of developmental cli-ents per thousand population in British Co-lumbia fell by 15 percent in 1994 and a smalleramount in 1995, use in Alberta declined only8 percent in 1994 and increased in 1995.When the data are adjusted for populationsize, the difference is about 1,500 additionalindividuals in Alberta by 1995, most of themreceiving training benefits.

Despite the difference in trends, both Al-berta and British Columbia had about 2.5developmental clients per thousand popula-tion in 1995. So, while use in Alberta may havegrown slightly faster during 1994 and 1995,by the end of these two years the two provinceshad equal proportions of residents on federalUI training programs.

Back to Work

In the end, the most important question sur-rounding the Alberta welfare reforms iswhether former welfare clients found their wayback into the workforce. Since many formerand potential clients undertook training pro-grams of various lengths, it may yet be tooearly for a conclusive answer to this question.There are promising indications, however, thatindividuals are choosing work over welfare.

Economic Growth in Alberta

Alberta’s growing economy doubtless contrib-uted to the reduction in the welfare caseload.The early 1990s’ recession was much lesssevere in Alberta than elsewhere in Canadaand the recovery was much more notable. Inthis environment, the welfare caseload shouldbe declining.

Canada’s and Alberta’s experience throughthe 1980s and 1990s shows, however, thatstrong growth is not necessarily enough toreduce welfare rolls. During the booming mid-and late 1980s, for example, total Canadiancaseloads failed to fall to prerecession levels.That this phenomenon was heavily influencedby the experience in Ontario, where benefitsrose consistently through the late 1980s, rein-forces the point that growth alone is notenough to account for changes in recipientlevels (see the Appendix). Indeed, during thepast 15 years, no province has experienced adecline in its welfare caseload of anywherenear the magnitude of Alberta’s, although manyhave had higher growth rates for longer peri-ods of time than Alberta experienced from1993 to 1996.

In Alberta, too, robust growth rates be-tween 1984 and 1988 (excluding 1986) failedto have anywhere near the effect on welfarecaseloads that the much more modest growthrates of the early 1990s had. Those who pointto the recovery as the prime reason for reduc-ing welfare caseloads in Alberta must alsoexplain why earlier, stronger periods of growthfailed to have such an effect.


Job Growth

Job growth in Alberta was very robust from1993 to 1995 (see Figure 4). There was morejob growth in 1994 and 1995 than in any ofthe previous ten years. In fact, the number ofnew jobs created in Alberta in 1993 through1995 exceeded the total reduction in welfarerecipients. That is not to say that there

was a direct transfer of individuals fromwelfare to work, but it does suggest that jobgrowth in Alberta was strong enough totake up a substantial number of former orpotential welfare recipients.

Was the job growth in these yearsprimarily demand driven — that is, dideconomic growth provide the stimulus forthe strong job creation — or did supply-side factors played a role as well — thatis, were barriers to hiring, such as incen-tives to stay on welfare as opposed towork, reduced? A comprehensive answeris beyond the scope of this Commentary,but job growth values can easily be ad-justed to take account of the changes ingross domestic product (GDP). Multiply-ing last year’s employment by the current

year’s growth in GDP gives an estimate ofjob change due to changes in GDP. Sub-tracting that value from actual job growthgives a residual — that is, job growth thatwas due to factors other than economicgrowth.

I did these calculations for two agecohorts — under-25-year-olds and 25- to65-year-olds, hoping that the breakdownwould give some idea of whether manyyoung Albertans affected by the changesfound their way into the workforce.

Figure 5 shows the results. The dataup to 1993 reflect what has become adisturbing trend in recent decades —namely, that youth employment has notkept pace with economic growth. In yearswhen the economy grew (1985, 1987–90,and 1992–95), job growth for those belowage 25 was below the level of real GDPgrowth. (That job growth kept pace with

the economy in 1986 and 1991 is small con-solation since the economy contracted in bothof those years.) Something was preventingthose under age 25 from enjoying Alberta’seconomic prosperity during those years.

In 1994 and 1995, however, job growth forthose young people kept closer pace with posi-tive growth in the economy. For the first time

Figure 4: Changes in Welfare Recipientsand Employment, Alberta, 1981–95

-50

-25

0

25

50

T

H

O

U

S

A

N

D

S

82 83 84 85 86 87 88 89 90 91 92 93 94

Change in number of welfare recipients

B

per

son

s

Sources: Alberta, Department of Family and Social Services;and Statistics Canada, CANSIM database on CD-ROM.

50,000

25,000

0

�25,000

�50,000

Change in employment level

Change in no. of welfare recipients

Figure 5: Job Growth Adjusted forGDP Growth, Alberta, 1982–95

-40

-20

0

20

40

60

T

H

O

U

S

A

N

D

S

82 83 84 85 86 87 88 89 90 91 92 93 94

25-64-year-olds (residual job grow

B

Sources: Statistics Canada, CANSIM database on CD-ROM;and author’s calculations.

per

son

s

60,000

40,000

20,000

0

�20,000

�40,000

Age <25 (residual job growth)

Ages 25–64 (residual job growth)


in over ten years, job growth for those underage 25 kept pace with (positive) growth in theeconomy. A barrier to employment appears tohave been removed, and the welfare changesare among the most obvious candidates.

A look at the types of jobs that increasedin those two years does not illuminate theanalysis much. The only notable shifts werean increase in employment in constructionand in primary industries that appears strongerthan in the previous decade.

Underground Activity

Another set of data are suggestive of changesin underground economic activity. The datareported above come from the federal LabourForce Survey (LFS), a phone survey in whichindividuals are asked about their employmentsituation. Another source of information aboutemployment is payroll data, which report em-ployment according to social insurance pre-mium records.

The two data sets are not completely com-parable (for example, self-employment is lesslikely to be reflected in the payroll data), butwhen noticeable numbers of individuals report

to the LFS that they have work but this em-ployment does not turn up in the payroll data,this is at least circumstantial evidence ofchanges in underground activity. Figure 6shows that, twice in the past few years, thepayroll data moved in a different direction fromthe LFS. The first was 1991, the year in whichthe goods and services tax, the first sales taxin Alberta, was introduced. This shift suggeststhat individuals said they were working inresponse to a telephone survey (LFS), but theirincome was not reported as payroll. Althoughthis difference may indicate a move to self-em-ployment, it may also suggest an increase inunderground activity.

Another shift in the payroll employmentdata that did not appear in the LFS dataoccurred shortly after the benefit reductions inAlberta. One explanation is that individualswere moving out of the underground economyback into the “counted” economy.

This evidence is highly circumstantial and,in the absence of a full accounting of otherrelevant variables (such as changes in self-employment), it is indicative only of a possiblemove out of the underground economy. How-ever, if a substantial number of welfare clientswere able to work in the underground econ-


omy, a reduction in welfare benefits wouldlower individuals’ incentive to keep their ac-tivities underground.

Summary

Clearly, job growth in Alberta was particularlystrong during the years immediately followingwelfare reform. It also appears that demandfactors alone cannot explain this growth, par-ticularly when compared with experience overthe previous decade or so. Further, the changesmay have occasioned a movement out of theunderground economy as fewer individualshad incentives to hide their sources of incomein the absence of the receipt of welfare benefits.

The figures suggest, therefore, that jobgrowth during this period had more than astrictly demand-side explanation and that asignificant portion of individuals who mighthave formerly ended up on welfare did findemployment.29 And the fact that employmentgrowth for those under age 25 was strongerthan in the previous decade suggests thatwelfare reforms may have played a role inimproving the supply side of the labor market.Changes in incentives for young employablesmay have been a key factor drawing theseindividuals back into the workforce, a possi-bility that reinforces the importance of benefitchanges in the Alberta welfare experiment.

National Implications

The Alberta welfare experiment holds someimportant insights into welfare reform thatother provinces may wish to emulate. Particu-larly encouraging is evidence suggesting thatpotential recipients found work. Before review-ing these insights, however, a few cautionarywords are in order.

First, as I have emphasized throughoutthis Commentary, my study, which is based onmacro variables, cannot be entirely conclu-sive. Some former and potential welfare recipi-ents, especially the very young (under age 19,at a guess), may have turned or returned tofamily, friends, or community groups for sup-

port. (At least one study30 offers some evidenceof individuals’ taking this course.31)

Welfare reform was bound to put pressureon these other support groups, and it may beimportant — for government and for society atlarge — to ensure that individuals have foundhelp there and are not turning or returning toabusive situations. To the extent that thesesupport groups can bear the burden, however,there may be important social benefits in mak-ing welfare programs a last, rather than a first,resort.

Second, many individuals are still in train-ing or work programs funded by the Albertagovernment.But workfareand “learnfare” shouldnot be replacements for long-term dependenceon welfare: government cannot provide themforever.32 Eventually, these individuals mustuse their human capital (which, one hopes,has been increased) and take their place inmainstream society. It will be important tofollow these individuals in coming years tolearn what works.

Finally, the Alberta experience took placeduring a period of positive economic growth,an environment in which efforts to reducewelfare dependency are most likely to be suc-cessful. The next economic downturn in theprovince will provide an important test of theexperiment. Will the welfare caseload increaseduring a period in which job losses normallyoccur? If it does, will Alberta avoid the recentCanadian experience in which welfare fails tofall to prerecession levels following the down-turn? Unfortunately, only time can answerthese questions.

Lessons to Learn

With these caveats registered, what lessonscan be drawn from the Alberta experience?They seem to be threefold: administration mat-ters; so do benefit levels; and program design-ers and researchers alike need better data.


Administration Matters

Perhaps the strongest implication of the Al-berta welfare reform is the crucial role of ad-ministrative culture (see the Appendix). Themost important administrative changes Al-berta made were a shift to focus on new appli-cants and the integration of welfare with otherhuman capital programs.

The shift in focus was to prevent peoplefrom coming onto welfare, rather than to moveindividuals off the rolls. The province routinelyturned down applicants or diverted them toother programs (although caseworkers wereallowed exceptions for hardship reasons).

These changes suggest that the Albertagovernment understood the long-term behav-ioral shifts that result from time spent onincome support programs. Welfare programs,which usually include many lower-skilledworkers, are particularly susceptible to thisphenomenon. The important lesson is thatpreventing the first exposure to welfare playsan important role in preventing future depend-ence on welfare, particularly for the groupmost affected by the Alberta changes: young,single, and employable individuals.

Some individuals who apply for welfaremay, of course, have a legitimate need for sometype of assistance. By directing individuals totraining and postsecondary education, Al-berta’s active labor market programs played arole in reducing caseloads. This role was madeeasier by integrating its human capital pro-grams. Early in the reform process, AFSS tookover selected activities from AECD, and thetwo departments arranged to jointly adminis-ter a variety of programs for potential welfarerecipients. Many of these services were placedunder one roof, which made transferring orreferring clients much easier in practice.

In Alberta, this practice is expanding toinclude federal programs of a similar nature.Other provinces would do well to emulate thisprocess of consolidation (many have or aredoing so). Current federal commitments totransfer training to the provinces give themanother opportunity to integrate human capi-

tal services more closely with their welfareagencies.33

The Alberta experience also highlights theimportance of allowing provinces to experi-ment with different policy approaches withoutfederal interference. The removal of most of thefederal standards under the CAP is a positivedevelopment for provincial autonomy. Prov-inces can integrate programs in related areas,and all Canadians will benefit from experimen-tation in this important policy area.

Benefit Levels Matter

In addition to the success of the welfare ad-ministrative reforms in their own right, thereduction in benefit levels also played a keyrole in Alberta. The principle that people onsocial assistance should receive a level of sup-port not exceeding that earned by workingAlbertans had important political and eco-nomic consequences.

Welfare is a program of last resort. Beforeentering it, individuals should have exhaustedtheir UI benefits and any available trainingprograms. They should also expect to work atjobs made available to them. Keeping welfarebenefits below actual wage levels ensures thatthe economic incentives point in the directionof taking a job. Furthermore, welfare benefitsthat remain above the level of a significantportion of the population are not likely to enjoybroad-based public support.

No one can know precisely what level ofbenefits will tip the marginal benefits of workover welfare. Individuals differ in their per-sonal circumstances and in their preferences.Single parents face larger barriers to returningto work than single, employable individuals(particularly those who are young). Welfarerecipients may also feel they cannot afford toforgo the present benefits of welfare programsfor work, even given the long-run benefits ofdoing so.

Setting targets for welfare benefits relativeto some existing level of wages makes goodsense. The difficulty is in determining theappropriate target — should welfare be set


equal to the earnings of the tenth percentile?the fifth? the twentieth? Clearly, welfare pro-grams must be generous enough to providenecessities, so some basic needs-based indexcould provide a floor value.34 A reasonableceiling might be the midpoint of the first decileof wage earnings (although the data limita-tions pointed out in Box 2 might cause practi-cal difficulties).

Setting a range for welfare benefits wouldallow jurisdictions to make ample considera-tions for other, less quantifiable factors, suchas the additional costs of working and in-kindbenefits.

Better Data

I have referred several times to the unavailabil-ity of data and the changing nature of whatdoes exist. Most analytical studies plead forbetter data. My plea is particularly urgent,given the nature of Canada’s welfare programs,with provision by ten provinces, two territo-ries, and municipalities. Moreover, since thefederal government is increasingly withdraw-ing from this field, the divergence of provincialpractice reporting and collecting of data hasincreased.

A number of specific sets of data wouldhave aided my and others’ research. First,everyone needs more and better data on bene-fit levels and/or wage levels at low-incomelevels. The National Council of Welfare hasdone a heroic job with information from recentyears, but comparable historical data are ex-tremely hard to find. In addition, the use ofsupplementary and in-kind benefits make thisa difficult exercise.

Second, since the flows in and out of wel-fare are so large, a better account of who isleaving and who is coming would be revealing.Alberta’s average caseload during 1992 wasabout 90,000 but nearly 130,000 (or more)people cycled through the program. It could beimportant to know who those 130,000 were,the acceptance and rejection rates, and possi-bly the referral rates. (Alberta’s placing thesedifferent services at a single access point shouldaid this kind of data collection.)

Third, analysts need comparable inter-provincial data. This problem plagues manyprovince-specific programs, and it is difficultto argue that the federal government is neces-sarily the solution (given its history of move-ment into provincial areas of jurisdiction).Official interprovincial discussions on socialpolicy35 might, however, cover the possibilityof a joint data-collection agency. Indeed, suchan agency need not be an arm of government.Citizens faced with different provincial policypackages may find it in their interests to fundtheir own groups to collect such data, which,in the longer term, might result in a race to thetop, as provinces sought to place themselvesat the leading edge of such comparisons.36

Postscript: RecentChanges in Other Provinces

Recent changes in British Columbia’s and On-tario’s welfare programs offer a final point of

C.D. Howe Institute Commentary© is a periodicanalysis of, and commentary on, current publicpolicy issues.

Kenneth J. Boessenkool, the author of thisissue, is a Policy Analyst at the C.D. HoweInstitute. The text was copy edited by Lenored’Anjou and prepared for publication by WendyLongsworth and Barry A. Norris.

As with all Institute publications, the viewsexpressed here are those of the authors, anddo not necessarily reflect the opinions of theInstitute’s members or Board of Directors.

To order this publication, please contact:Renouf Publishing Co. Ltd, 5369 CanotekRoad, Unit 1, Ottawa, Ontario K1J 9J3 (tel.:613-745-2665; fax: 613-745-7660), Renouf’sstores at 711/2 Sparks Street, Ottawa (tel.:613-238-8985) and 12 Adelaide Street West,Toronto (tel.: 416-363-3171), or the C.D. HoweInstitute, 125 Adelaide Street East, Toronto,Ontario M5C 1L7 (tel.: 416-865-1904; fax:416-865-1866; e-mail: [email protected]).

We also invite you to visit the Institute’sInternet web site at:

www.cdhowe.org

Quotation with proper credit is permissible.

$6.00; ISBN 0-88806-411-X


discussion. Many of the reforms in theseprovinces — in particular, the focus onyoung employables, and adjustments tobenefit levels — mirror changes in Alberta.Yet there are also some important differ-ences in approach. Following are somehighlights of the reforms (I do not claimthey are exhaustive).

British Columbia. The BC government re-cently embarked, with some success, onchanges to its welfare program.37 It fol-lowed Alberta in some respects — in par-ticular, by lowering benefits for single,young employables (see Table 4) and tight-ening eligibility. It requires, however, thatwelfare recipients wait seven months be-fore becoming eligible for active programs,including career planning, skills plan de-velopment, and job-finding clubs.38 TheAlberta experience suggests diverting ap-plicants to these programs before admit-ting them to welfare programs, not sevenmonths later.

Ontario. The much-heralded changes inOntario’s welfare program have been re-stricted primarily to changes in benefit levels,although selected pilot projects for workfarehave been initiated.

Table 4 indicates that Ontario had goodreason to reduce its benefit levels. Relative toearnings at the top of the first decile, Ontariobenefits for single employables before reformsurpassed those of all the larger provinces.Even after the reductions, Ontario was stillmore generous than most jurisdictions.

Ontario’s approach to workfare is moredifficult to understand. Clearly, it plans tomove a substantial number of those presently

on welfare into work programs; indeed, theemphasis appears to be getting existing, long-term clients off welfare. Although Alberta’sreform had a small, voluntary componentalong these lines, most of its success camefrom dealing with potential, rather than exist-ing clients. In its development of workfare,Ontario would do well to keep this focus inmind. It might also remember that the Albertawork programs, on an annualized basis, costmore than three times as much as benefits fora single individual; although the aims of work-fare may be noble, they should hardly beviewed as a cost-saving mechanism.


Appendix:

A Statistical Evaluation ofAdministration and Relative Benefits

Changes in the proportion of the populationon welfare result partly from macroeconomiccircumstances, such as the recessions thatkicked off both the 1980s and 1990s (seeFigure 1). But macroeconomic factors are notthe sole cause. As the previous pages indicatedboth the level of benefits — or perhaps moreexactly, the level of benefits relative to the levelof low-wage earnings — and changes in ad-ministrative culture played an important rolein the Alberta experiment.

The question is: To what extent can onegeneralize from my conclusions about Alberta?It turns out that a simple statistical analysisof only three or four variables can explain asubstantial amount of the change in the pro-portion of individuals receiving welfare bene-fits in three of Canada’s largest provinces.39

Table A-1 shows the explanatory power of theunemployment rate (a proxy for economic con-ditions), benefit levels, low-wage earnings lev-els, and a variable for administrative culturein determining the level of welfare use over thepast decade and a half.

All these variables matter, but the relativeimportance of each has changed over bothtime and place.

Ontario

Ontario has seen large fluctuations in unem-ployment in the past ten years. Nineteen eighty-eight, the year in which the unemploymentrate reached a trough, is an effective point atwhich to divide the province’s experience.

The simple regression results for Ontariosuggest that the unemployment rate and theratio of benefits to average second-decile earn-ings combine to explain a significant amountof the changes in welfare use. Administrativechanges, proxied by changes in government

from Conservative to Liberal and from Liberalto New Democratic do not appear to haveinfluenced welfare use beyond their effect onbenefit changes. Therefore, the table does notreport changes in administrative culture.

The Ontario experience highlights the im-portance of benefit levels in welfare use. Inparticular, for the 1989–93 period, they ac-count for nearly half of the explained increasein welfare use. Increased unemployment ac-counts for another two-fifths, and a decline insecond-decile earnings for only one-seventh.

Alberta

In the early 1980s, base benefit levels in Al-berta were 53 percent higher than in the threelargest provinces. Over the following ten years,inflation ate away at these levels, bringingthem nearer to the Canadian average levels.But a number of things about Alberta’s benefitlevels make their use in this simple regressionproblematic.

First, a large number of supplementarybenefits were given out in various proportionsover the decade; in some periods, they madeup a substantial portion of all benefits, and inothers, they were relatively insignificant. Un-fortunately, no time series data on averagebenefits are available from Alberta, and thereare no suitable proxies that can account forthese shifts.

Second, the province provided very differ-ent benefit levels to its various family classes.Although benefit levels for singles fell throughthe 1980s, those for families remained highrelative to the levels of other provinces.

Third, the regressions could not take ac-count of the large benefit changes that oc-curred before 1982, due to the even greaterscarcity of data for that period.


Fourth, the administrative changes andbenefit reductions took place in the same yearand are therefore impossible to disaggregatefrom the annual data used here.

For all these reasons, the regressions re-ported for Alberta do not include benefit levels(although an accurate account of them wouldsurely have added to the explanatory power).

The remaining variables still manage tocapture many of the shifts. Alberta saw nodramatic change in unemployment, but a sharpbreak in welfare use occurred between 1993and 1994, following the introduction of re-forms by Mike Cardinal. Accordingly, I dividedthe observations at that point.

In the first period, the most importantfactor in the increase in welfare use was adecline in reference earnings levels. After 1993,the cultural change in administration accountsfor more than nine-tenths of the explaineddecline in welfare use. The improved employ-ment picture and changes in reference earn-ings explain only about one-tenth.

British Columbia

British Columbia data extend back far enoughto allow me to set up three periods: the early1980s’ recession, the late 1980s’ boom, andthe years of government by the New Demo-cratic Party. In the first period, welfare usedoubled; and most of the explained changecame from changes in the unemployment rate.In the second period, welfare use fell, andagain, much of the explained change is due tochanges in the unemployment rate. Over bothof these periods, a decline in reference earn-ings caused welfare use to increase (1980–85)

or not fall as far as it otherwise might have(1985–91).

In 1991, the New Democrats came to power,and Joan Smallwood took over responsibilityfor social assistance. Smallwood, the antithesisof Alberta’s Cardinal, made clear to her staffthat the ministry’s task was to serve clients,not to police welfare use. In explaining welfareuse in the 1991–95 period, a small increase inunemployment accounts for about one-fifth ofthe explained change. Changes in benefits andreference earnings combine to roughly offseteach other. The overwhelming portion of theincrease, nearly five-sixths, is explained by thechange in administrative culture under thenew minister.

Conclusion

My regressions are based on limited data se-ries for benefit levels that, in reality, leaveadministrators much latitude. Nevertheless,the results reported here are consistent withthe main theme of this Commentary: that ad-ministration and benefit levels relative to lowwage levels are an important part of explainingchanges in welfare use. The Alberta and Brit-ish Columbia experiences especially highlightthe importance of administrative culture. Irre-spective of the levels of the unemployment rate,benefits and low wages, the changes in admin-istrative cultures in both provinces appear toexplain a significant component of changes inwelfare use.

Overall, my simple statistical methods leavelittle room for the popular position that changesin the economic environment alone are enoughto predict welfare use.


Notes

I thank the federal, British Columbia, Alberta, andOntario governments, which were forthcoming withdata used in this project. Thanks also to Doug Allen,Christopher Bruce, Angela Ferrante, Robert Howse,John Richards, William Robson, William Scarth, andDaniel Schwanen, who provided helpful comments onearlier drafts. All responsibility for errors and interpre-tations remain, of course, with me.

1 Alberta, “Welfare Reform Strategy,” News Release andBackgrounder, Edmonton, April 15, 1993.

2 Alberta, Department of Family and Social Services(hereafter AFSS), Alberta Welfare Reform: Three YearReview (Edmonton, May, 1996).

3 See Jeffrey Simpson, “After all the cutbacks, manywelfare cases have gone, but where?” Globe and Mail(Toronto), February 9, 1996.

4 National Council of Welfare, Welfare Incomes, variousissues.

5 Mike Cardinal, “Welfare Reform In Alberta,” FraserForum, June 1995, p. 6.

6 Alberta, “Welfare Reform Strategy,” p. 10.

7 Ibid., p. 7. Processes were already in place to deal withhardship cases, either directly by the caseworker, byadministrative reviews (casework supervisor or sitemanager), or by regular appeal to the nondepartmentalcitizens’ appeal committees.

8 Thomas Lemieux and W. Bentley MacLeod, UI ImpactsonWorkerBehaviour: State Dependance and Unemploy-ment Insurance (Ottawa: Department of Human Re-sources Development, 1995), p. 26.

9 Ibid., p. 30. US evidence of the same phenomenon inthe welfare system can be found in T. David Ellwood,Targeting “Would Be” Long Term Recipients of AFDC,report prepared for the US Department of Health andHuman Services (Washington, DC: Mathematica PolicyResearch, 1986).

10 See National Council of Welfare, Poverty Profile, 1994(Ottawa, 1996).

11 Information in this section comes from AFSS, AlbertaWelfare Reforms: Progress Report, March 1993–Decem-ber 1995 (Edmonton, 1996), p. 7.

12 Ibid., p. 4.

13 See Robert Howse, “Workfare: Theory, Evidence andPolicy Design” (paper presented to a roundtable at theCentre for the Study of State and Market, University ofToronto, May 17, 1996).

14 The methodology of this section is based on Ken Battleand Sherri Torjman, “The Welfare Wall: The Interactionof the Welfare and Tax Systems” (Ottawa: CaledonInstitute of Social Policy, Summer 1993). That studyexamined the situation in Ontario in 1992.

15 Assar Lindbeck, Hazardous Welfare State Dynamics,reprint series 538 (Stockholm: Institute for Interna-tional Economic Studies, 1995).

16 Robert Haveman and Barbara Wolfe, “The Determi-nants of Children’s Attainments: A Review of Methodsand Findings,” Journal of Economic Literature 33 (De-cember 1995): 1829–1878; and idem, Succeeding Gen-erations: On the Effect of Investments in Children (NewYork: Russell Sage Foundation: 1994).

17 British Columbia, Ministry of Social Services, A Profileof New Income Assistance Clients: August 1993 to May1994, Research, Evaluation and Statistical Branch,Policy, Planning and Research Division, RES 94-04(Victoria, August 1994); and a companion report,idem., A Profile of Income Assistance Clients who Re-turned to the Caseload, RES 95-01 (Victoria, March1995).

18 Simpson, “After all the cutbacks.”

19 Statistics Canada, Annual Unemployment InsuranceStatistics, various years, 73F003XDE on CD-ROM.

20 Summaries of these programs are found in AFSS,Alberta Welfare Reforms: Progress Report.

21 The average of the ACE, AJC, and ESP clients still activeon SFI in April 1994, 1995, and 1996. Data are fromAFSS.

22 The CAP was a 50 percent shared-cost program. Thefederal “cap on CAP” released from restriction about$100 million of its 50 percent contribution. Albertanotionally used these funds to initiate its conditionalprogram.

23 AFSS, 1994/95 Annual Report (Edmonton, 1995).

24 AFSS, EmploymentPrograms and Services, informationpamphlet (Edmonton, April 1996).

25 Alberta, Department of Education and Career Develop-ment, Annual Report (Edmonton, 1995).

26 See, for example, Statistics Canada, Labour Force An-nual Averages 1995, cat. 71-529, table 5 (Ottawa); andidem, Historical Labour Force Statistics, cat. 71-001,table 5 (Ottawa).

27 This point is also made in a study for the EdmontonSocial Planning Council and Edmonton area food banks:Alberta Facts: Hunger in Edmonton, Understanding theNeed for Food Banks (Edmonton: Edmonton SocialPlanning Council, 1996).

28 The statistics throughout this subsection are fromStatistic Canada, AnnualUnemployment Insurance Sta-tistics, various years, 73F003XDE on CD-ROM.

29 Despite the obvious selection bias against finding indi-viduals who are working, a recent survey of individualsusing Edmonton food banks indicates that, ignoringthose waiting for welfare benefits, 40 percent of formerrecipients of welfare had full- or part-time employment.(The report does not distinguish between those who


were refused welfare and those who formerly receivedwelfare for these statistics.) Ibid.

30 Ibid.

31 Research under way to track former recipients mayshed more light on this and other possibilities that thisstudy, due to its methodology, can say little about. SeeCanada West Foundation, Social Services Project (Cal-gary, 1996).

32 Deadweight losses (hiring that would have occurredanyway) and substitution effects (hiring that displacesexisting market jobs) remain serious concerns for gov-ernment-created work projects. Calmfors finds “com-bined deadweight and substitutions effects of the orderof magnitude of 70–90 percent of the gross number ofjobs created” in a review of studies of these effects. SeeLars Calmfors, Active Labor Market Policy and Unem-ployment: A Framework for the Analysis of CrucialDesign Features, OECD Economic Studies 22 (Paris:Organisation for Economic Co-operation and Develop-ment, 1994).

33 Although it appears that the provinces may have muchless than total autonomy under the new transfer. See

Kenneth J. Boessenkool and William B.P. Robson, End-ing the Training Tangle: The Case against Federal-Provincial Programs underEI, C.D. Howe Institute Com-mentary86 (Toronto:C.D.Howe Institute,February1997).

34 Perhaps along the lines of those suggested in Christo-pher Sarlo, Poverty in Canada, 2d ed. (Vancouver:Fraser Institute, 1995).

35 Such as the one that resulted in Ministerial Council onSocial Policy Reform and Renewal, “Report to the Pre-miers,” December 1995, mimeographed.

36 For a proposal along these lines, see William Robson,“Getting Radical: Federalism’s Only Hope,” Gravitas,Spring 1996, pp. 9–12.

37 Ross Howard, “BC welfare rolls slimmer since reform,”Globe and Mail (Toronto), December 23, 1996, p. A4.

38 British Columbia, http://www.gov.bc.ca/bcben/initia-tives.html, at June 26, 1996, p. 5.

39 This appendix draws heavily from work done with JohnRichards, which is also reported in What’s Left to Do(Toronto: C.D. Howe Institute, forthcoming).


Recent C.D. Howe Institute Publications

Schwanen, Daniel. Trading Up: The Impact of In-creased Continental Integration on Trade, In-vestment, and Jobs in Canada. C.D. Howe Insti-tute Commentary 89 (March 1997). 32 pp.; $6.00.

Laidler, David, ed. Where We Go from Here: InflationTargets in Canada’s Monetary Policy Regime.Policy Study 29 (March 1997).

Gibbins, Roger. Time Out: Assessing IncrementalStrategies for Enhancing the Canadian PoliticalUnion. C.D. Howe Institute Commentary 88 (Feb-ruary 1997). 28 pp.; $6.00. With KatherineHarmsworth.

Laidler, David E.W., and William B.P. Robson. “TheBank of Canada and the Economy: Has theReferee Put Away the Whistle?” Backgrounder(C.D. Howe Institute), February 13, 1997. Freeof charge.

Slater, David W. The Pension Squeeze: The Impactof the March 1996 Federal Budget. C.D. HoweInstitute Commentary 87 (February 1997). 32 pp.;$6.00.

Boessenkool, Kenneth J., and William B.P. Robson.Ending the Training Tangle: The Case againstFederal- Provincial Programs under EI. C.D.Howe Institute Commentary 86 (February 1997).16 pp.; $6.00.

Nakamura, Alice. Employment Insurance: AFramework for Real Reform. C.D. Howe InstituteCommentary 85 (October 1996). 12 pp.; $6.00.

Richards, John. Language Matters: Ensuring Thatthe Sugar Not Dissolve in the Coffee. C.D. HoweInstitute Commentary 84 (October 1996). 44 pp.;$6.00.

Sauvé, Pierre, and Daniel Schwanen, eds. Invest-ment Rules for the Global Economy: EnhancingAccess to Markets. Policy Study 28 (September1996). 332 pp.; $19.95.

Mintz, Jack M., and James E. Pesando, eds. PuttingConsumers First: Reforming the Canadian Fi-nancial Services Industry. Policy Study 27 (Sep-tember 1996). 104 pp.; $14.95.

Iacobucci, Edward. Value for Money: ExecutiveCompensation in the 1990s. Observation 41(September 1996). 178 pp.; $14.95. With Mi-chael J. Trebilcock.

Monahan, Patrick J., and Michael C. Bryant. Com-ing to Terms with Plan B: Ten Principles Govern-

ing Secession. C.D. Howe Institute Commen-tary 83 (June 1996). 56 pp.; $9.00. With NancyC. Côté.

Schwanen, Daniel. Drawing on Our Inner Strength:Canada’s Economic Citizenship in an Era of Evolv-ing Federalism. C.D. Howe Institute Commen-tary 82 (June 1996). 20 pp.; $6.00.

Howse, Robert. Securing the Canadian EconomicUnion: Legal and Constitutional Options for theFederal Government. C.D. Howe Institute Com-mentary 81 (June 1996). 20 pp.; $6.00.

Boessenkool, Kenneth J. The Illusion of Equality:Provincial Distribution of the Canada Health andSocial Transfer. C.D. Howe Institute Commen-tary 80 (June 1996). 24 pp.; $6.00.

Horstmann, Ignatius J., C. Frank Mathewson, andNeil C. Quigley. Ensuring Competition: BankDistribution of Insurance Products. Observation40 (May 1996). 110 pp.; $12.95.

Boessenkool, Kenneth J., David E.W. Laidler, andWilliam B.P. Robson. Devils in the Details: Im-proving the Tactics of Recent Canadian Mone-tary Policy. C.D. Howe Institute Commentary 79(April 1996). 20 pp.; $6.00.

Neave, Edwin H. Canadian Financial Regulation:A System in Transition. C.D. Howe InstituteCommentary 78 (March 1996). 28 pp.; $6.00.

Beach, Charles M., and George A. Slotsve. Are WeBecoming Two Societies? Income Polarizationand the Myth of the Declining Middle Class inCanada. The Social Policy Challenge 12 (March1996). 190 pp.; $14.95.

Scarth, William. Beyond the Deficit: Generation Xand Sustainable Debt. C.D. Howe Institute Com-mentary 77 (February 1996). 20 pp.; $6.00.

Cadsby, Charles Bram, and Kenneth Woodside.Canada and the New Subsidies Code. C.D. HoweInstitute Commentary 75 (February 1996). 12 pp.;$6.00.

Barley, Stephen R. The New World of Work. British-North American Committee 40 (January 1996).59 pp.; $18.95.

Robson,WilliamB.P. PuttingSomeGold in theGoldenYears: Fixing the CanadaPension Plan. C.D. HoweInstitute Commentary 76 (January 1996). 28 pp.;$6.00.


institut c.d. howe communiqué

Documents