learn$ave Project: Final ReportTable 7.2 Impacts on Participation in Education <strong>and</strong> Training (Percentage Points or Average), during the 54 Months, Education Stream –AdjustedControlGroupIncidence orAverageImpact ofMatchedSavingCreditsImpact ofServiceswhenOffered withCredits ∑CombinedImpact ofCredits +ServicesOverall (Program or Individual Course)Enrolled in any education/training since baseline (%) 81.5 6.6*** 1.7 8.2***Educational ProgramsEnrolled in courses toward a degree, diploma, or certificate (%) 56.0 9.1*** 3.5 12.6***Program Type (first program) (%):▪▪English as a second language (ESL) 4.1 0.2 1.0 1.2▪▪High school 2.8 1.1 -0.9 0.3▪▪Registered apprenticeship 6.1 0.6 -1.4 -0.9▪▪Community college 30.0 3.3 1.9 5.2*▪▪University 18.4 6.7*** 2.5 9.2***Completed program (%) 39.5 5.0* 1.0 6.0**Individual Courses, not Part of a ProgramEnrolled in other (non-program) education courses, seminars, etc. (%) 47.5 4.3 0.3 4.6Number of courses (average) 0.9 0.2** -0.1 0.1Completed one or more courses (%) 43.7 1.5 -1.1 0.4Spouses (among those with a non-student spouse at baseline)Enrolled in any education since baseline (%) 62.3 4.3 1.5 5.8Enrolled in courses toward a degree, diploma, or certificate (%) 41.9 11.0** -0.6 10.4*Enrolled in other (non-program) education courses, seminars, etc. (%) 35.8 -2.8 5.1 2.4Source: Calculations from 18-month, 40-month <strong>and</strong> 54-month survey data.Note: The sample sizes for the control, learn$ave-only <strong>and</strong> -plus groups are 568, 842 <strong>and</strong> 859, respectively for the 54-month survey, while thesample sizes of participants with a spouse not in school at baseline are 135, 203 <strong>and</strong> 205, respectively.Sample sizes vary for individual measures because of missing values.Two-tailed t-tests were applied to differences in characteristics between the program <strong>and</strong> control groups.Statistical significance levels are indicated as * = 10 per cent; ** = 5 per cent; *** = 1 per cent.Rounding may cause slight discrepancies in sums <strong>and</strong> differences.∑The figures in this column show the extra impact of the financial management training <strong>and</strong> enhanced case management services whengiven to those eligible to receive matched credits. It does not represent the impact of those services alone for those not eligible to receive thematched saving credit; it represents the impact of the services when provided with the credits.financial management training services combined withthe matched credits offered to the learn$ave-plus groupled to an non-negligible increase in educational participation,of 8.2 percentage points. In U.S. IDA programs, asnoted in Chapter 1, no education impacts were <strong>report</strong>edin the evaluation of the American Dream <strong>Demonstration</strong>IDA experiment, though impacts on homeownershipwere found (Mills et al, 2008a), whereas large educationimpacts (versus the comparison group) were <strong>report</strong>edin the evaluation of the Assets for Independence IDAprogram (Mills et al., 2008b).More interesting than just looking at the overall rate ofparticipation in adult learning of any kind, the learn$aveIDA program had a positive impact on participationin the type of education that is most likely to lead toreturns in the form of greater employment earnings:certificate or degree programs at the post-secondarylevel. The matched credits alone <strong>and</strong> the credits <strong>and</strong>services combined had a significant impact on enrolmentin programs of any type (by 9.1 <strong>and</strong> 12.6 percentagepoints, respectively), which again represents improvementsover impacts at 40 months. Moreover, the mainprograms impact was on PSE participation: university(by 6.7 <strong>and</strong> 9.2 percentage points for credits alone <strong>and</strong>combined with the services, respectively) <strong>and</strong> college (by5.2 percentage points, credits <strong>and</strong> service combined).No enrolment impacts were observed for other typesof educational programs, i.e., high school equivalency,English as a Second Language (ESL), or apprenticeship.82 | Chapter 7 <strong>Social</strong> <strong>Research</strong> <strong>and</strong> <strong>Demonstration</strong> <strong>Corp</strong>oration
learn$ave Project: Final ReportIn proportional terms, the impacts on post-secondaryeducational enrolment were large. Relative to the controlgroup’s enrolment rate, the impacts were 50 per cent(9.2/18.4 percentage points) <strong>and</strong> about 17 per cent(5.2/30.0 percentage points) for university <strong>and</strong> collegeprograms, respectively. The proportional impact onparticipation in education programs overall was 23 percent (12.6/56.0 percentage points) <strong>and</strong>, on education ortraining of any kind, it was about 10 per cent (8.2/81.5percentage points).The large impact at the university level is interestingfor two reasons. First, the greatest labour marketreturns to education are derived from university degrees(Statistics Canada, 2009) <strong>and</strong> it is therefore promisingthat such impacts on university enrolment are takingplace. Second, as observed in Chapter 3, about half theparticipants already had a university education whenthey entered the project with a substantial overlapbetween the participants with higher education <strong>and</strong> thosewho where recent immigrants to Canada. On the basis ofthis overlap, the impacts on participation in universitylevel education suggests several participants might havebeen using learn$ave to upgrade their university educationor to obtain a Canadian recertification for a foreigncredential.The evidence also indicates that learn$ave had animpact on the completion of programs. The learn$avematched credits, alone <strong>and</strong> in combination with learn$aveservices, increased the completion rate of educationalprograms, by 5 <strong>and</strong> 6 percentage points, respectively. Thisis somewhat surprising given that the data were collectedjust six months after participants could no longer cashout their earned credits for education or training. Since somany programs are of a longer duration than six months,this finding may be linked to early take-up of education<strong>and</strong> training by the sub-group of participants identifiedearlier as “early savers, high investors.”Contrary to expectations, learn$ave did not affectparticipation in individual shorter term courses. Theunderlying hypothesis of learn$ave did not differentiatebetween programs <strong>and</strong> courses, instead suggesting thatlearn$ave would increase participation in both forms ofadult learning. However, one might have expected greatereffects for individual courses which cost less. In fact asshown, statistically significant impacts were observed forjust programs. However, while learn$ave did not have animpact on the proportion taking courses, it did have aninfluence on the number of courses taken: the averagenumber of courses was 0.2 higher for the learn$ave-onlygroup than the control group (a difference that is significantat the 5 per cent level), indicating that the matchedcredits had some impact on that indicator, albeit quitesmall.If participants were unable or unwilling to use someor all the funds accumulated in their learn$ave IDA, theyhad the option of transferring the right to use them toanother adult member of their immediate family whowas not a full-time student <strong>and</strong> who would otherwisehave been eligible for the project themselves at baseline.This is a feature of the program design not dissimilar totransfer rules under existing policy instruments such asRegistered Retirement Savings Plans. To measure whatthis feature had on the results, additional analysis focusedon spouses, as they were the ones who typically met theage <strong>and</strong> education requirements (18 years or older <strong>and</strong>not going to school) at baseline.The results of this analysis indicate that participants’spouses <strong>report</strong>ed higher enrolment rates in educationprograms than control group participants’ spouses — by11 percentage points for the learn$ave-only group <strong>and</strong>10.4 percentage points for the learn$ave-plus group (seethe bottom panel of Table 7.2). When this extra educationtaken by spouses is included, the total impact of matchedcredits on participation in education programs (notshown in table) rises by 1.4 percentage points over theincrease in participation among learn$ave accountholdersthemselves for the learn$ave-only group; however, theimpact does not change for the learn$ave-plus group.Education participation program impacts by subgroupAnalysis was conducted to determine if impactspertaining to participation in education programs variedappreciably for certain subgroups. Table 7.3 showsimpact results by selected subgroups as defined by theparticipants’ baseline characteristics associated withparticipation in education <strong>and</strong> training: enrolment age,labour force status, educational attainment, householdincome in the year prior to entry in the project <strong>and</strong>immigration status. Also shown are results for a variablerelevant in the present context: self-<strong>report</strong>ed savingregularity. The degree to which education program enrolmentimpacts were statistically significant for particularsubgroups is indicated by asterisks. The degree to whichthe impacts varied between subgroups (say between agegroups) is indicated by daggers. More detailed results arepresented in Appendix F, Table F.7.3.The analysis reveals that the impacts were widespread<strong>and</strong> particularly large for certain subgroups.• Labour force status at baseline: learn$ave-plusparticipants who worked for pay or were self-employedsignificantly increased their participation in education by<strong>Social</strong> <strong>Research</strong> <strong>and</strong> <strong>Demonstration</strong> <strong>Corp</strong>oration Chapter 7 | 83