This section provides some evidence on levels and trends in participation in the individual learning schemes reviewed in this report. It also describes the type of training undertaken by participants, which tends to be mostly non-formal and of short duration. Finally, the section reviews the limited evaluation evidence on ILS: to what extent is training undertaken aligned with labour market needs, and what impact do they have on participation and training?
Individual Learning Accounts
Panacea or Pandora's Box?
3.1.1. Participation levels
Actual participation in ILS varies across schemes, but is generally modest. Due to complex and sometimes changing rules for eligibility, it is most often not possible to provide accurate coverage rates, i.e. number of participants as a proportion of the eligible population.
In relation to a common denominator, the labour force, participation in the Singapore scheme is by far the highest (Figure 3.1).1Participation in the UK ILA scheme was also relatively high, but in part due to widespread mis-selling and fraud which actually led to close the scheme after only 15 months of activity (National Audit Office, 2002[9]). The French and Upper Austrian schemes are in a middle position as they benefit about 2% of the labour force. The Geneva and Scottish schemes cover slightly less than 1% of the labour force and less than 0.5% of the labour force participated in the Portuguese scheme. Carta ILA in Tuscany concerned only a limited number of individuals, representing about 0.2% of the labour force and 3% of the unemployed per year over the 2009-2014 period (for the two cities for which data is available - Arezzo and Pistoia). The US ITA scheme appears very minor in relation to the total labour force, and covered only a very limited share of the unemployed (1.2% in Michigan and 0.6% in Washington state in 2016), owing to strict eligibility criteria. Finally, participation in the pilot Canadian and US savings schemes was very low even in absolute numbers.2
Note: The labour force covers individuals aged 15 and over for Singapore and the United States, and concerns only resident individuals in Singapore.
Source: Cour des comptes (2015[2]), National Audit Office, Centro de Relações Laborais (2017[10]), Skills Development Scotland, Land OOE Bildungskonto, Caisse des Dépôts et Consignations, Workforce Investment Standard Record Data and WIOA Participant Individual Record Layout, SkillsFuture Singapore, OECD LFS and OECD Regional Labour Indicators.
3.1.2. Trends in participation
The evolution of participation over time varies across the countries for which data is available.
Participation in the French CPF scheme has progressively increased over the four years of its existence (Figure 3.2). The CPF being the only “real” account scheme, its use could initially have been limited by the need for individuals to accumulate credit before they could draw down their rights. In practice, however, employees had their account credited at the start of the CPF scheme with the hours accumulated under the Droit Individuel à la Formation (DIF) scheme created in 2004. Besides, social partners and the state agreed for 2015-2017 to credit jobseekers’ accounts with a minimum of 100 hours each.3 Hence, slow take-up is unlikely to be explained by the fact that individuals needed to accumulate rights first. Participation in the CPF was initially low and largely dominated by jobseekers, with only 0.5% of employees mobilising their account in the first two years, compared to 8% of jobseekers. Strong CPF take-up by jobseekers was largely driven by the “Plan 500,000”, a large-scale training plan for jobseekers launched in the same period. But employees have increasingly used their CPF and in 2018 there were more employees than jobseekers participating in the scheme (Figure 3.2, Panel B).
1. The total number of WIOA participants who received training services in this program of study through the adult or dislocated worker programs and who completed, withdrew or transferred from this program of study in the reporting period.
Source: OECD Secretariat estimates based on Land OOE Bildungskonto for Austria and the Workforce Investment Standard Record Data and WIOA Participant Individual Record Layout for the United States; Caisse des Dépôts et Consignations SI-CPF for France; and Skills Development Scotland.
In part, the rather slow increase in CPF participation results from the complexity of the scheme. The actors involved (funders, firms and training providers) needed some time to understand and learn to navigate it (IGAS, 2017[8]). It was also complicated for individuals to access their training rights because: i) the Droits Individuels de Formation credits were not simple to recuperate and ii) the CPF information system (SI CPF) was not easy to navigate (IGAS, 2017[8]):
Many employees failed to provide the code d’activité principale (APE) and the code de nomenclature d’activité française (NAF) describing the activity of their employer, and even when they could provide it, this code did not always correspond with the branch code that was required to determine eligible training lists. Resulting errors in the lists of training displayed sometimes led to refusals by the training funds (OPCAs).
The language used in the SI CPF was not always accessible (“éditeur de listes”, “OPCA”, “autorité certificatrice”), and not always reflecting the diversity of situations of applicants. For example, people interested in using their CPF were advised to contact the head of human resources or the employee representative body – but this is not relevant for employees in SME’s.
The steps to be followed were not always clear. For example, the proposed alternative to use the CPF “alone” or “without one’s employer” could be confusing for a non-informed user, as well as the requirement to provide information on the dates of training when those are often defined afterwards by the training provider.
The search engine for eligible training programmes was not easy to use and only provided eligible programmes on the different lists, but not the associated training providers or training sessions, which the individual needed to search for independently and with no guidance4.
The modified CPF scheme, which will be fully in place in January 2020, aims to simplify the process, notably by eliminating the lists of eligible training programmes and improving the web application.
In the three other schemes presented in Figure 3.2, trends in participation mostly reflect financial/budget constraints as well as the impact of the 2008 global financial crisis, although the complexity of the system and a heavy administrative burden also played a role in the Upper Austrian and Scottish schemes.
While participation in the Bildungskonto had increased since it was launched in 1994, it has tended to decline again since 2009 from 21 315 to 14 558 participants in 2017 (Figure 3.2, Panel A), which represents about one percentage point of the labour force. As highlighted by the regional audit court, the rise in participation before 2009 was often accompanied by overspending of budgets (Oberösterreichischer Landesrechnungshof, 2009[11]). This happened despite rather complex regulations which implied high administration costs in terms of counselling, problems with applications and managing complaints – all of which could have led to underspending. However, after levelling off in 2009-2012, regional budgets for the scheme have been cut significantly (by almost 40% in nominal terms) between 2012 and 2018. This led to a progressive reduction in the support provided to individuals – from 50% of the training fee to 30% for the basic support package; and from 80% of the training fee to 60% for the higher one5 – which likely reduced the number of applications.
Continuous tightening of public spending since 2011-2012 also largely explains the reduction in participation in the Scottish ILA scheme, which had almost tripled over the period 2007-2010.6 While the scheme had no fixed annual budget and was in principle demand-led, the level of demand was managed through reviewing the eligibility criteria. Course eligibility was tightened in 2011 and holders of a degree lost eligibility in 2012. Another way of reducing participation was to reduce the marketing effort. Recently, budgeted amounts appear to have become binding, as access to the new ITA scheme was restricted for the first time in January 2019 with a message on the Skills Development Scotland website announcing that applications were no longer accepted because all places had been filled.7 In addition to the very low amount of support offered (which provides limited incentives for individuals to apply), the rather cumbersome process in the previous ILA programme may also have limited participation. Applicants had to request a paper form to be sent out to them, and return it after completion. Only 50% of application packs sent out ever resulted in an application for an account and only 20% of application packs resulted in an account being opened.8
In the United States, the number of individuals who can train with ITA support depends on the funding provided by the federal government for the WIOA programmes. Funds are allotted to the states through a formula that accounts for the state’s population and labour force and –in the case of the dislocated workers programme – the state’s share of total US unemployment. Not all WIOA funds are used for training, and not all training funds go to ITA.9 In Michigan and Washington, the number of individuals who received training financed by an ITA increased noticeably during the global financial crisis, thanks to an injection of federal funds through the American Re-investment and Recovery Act (ARRA) enacted in February 2009. Funding for the programme then decreased progressively from 2011/12 to 2016/17 to levels below those reached before the recession.
Recruitment into the pilot individual saving account for training in Canada (learn$ave) initially relied on outreach through local NGOs but achieved very low results, as the method was fairly time-consuming and ineffective (Leckie et al., 2010[4]). The second wave effort focused on targeting clients directly through advertising on local media and turned out to be more effective. As the scheme was set up as a randomised controlled trial, the possibility for participants to be assigned to the control group discouraged some from participating, as did the burden associated with completing the paper application.
In Portugal, the Cheque formação appears to have been taken up mostly by employers rather than by individuals directly, which suggests that the cheque assists with employers’ short-term needs and that the low amount of support involved is not attractive to adults (Araújo, 2017[12])
The type of training that individuals can undertake relates strongly to the amount of support and the associated duration of training that the ILS allows. As indicated in Section 2.2, most schemes provide rather limited support thus by design limiting the possibility for individuals to undertake long-duration training.
3.2.1. Type of qualification
Available data suggest that a minority of ILS participants undertake training programmes to prepare a formal examination or a professional certificate with a qualification level recognised by the state (Figure 3.3). Reflecting the small amount of support, fewer than 5% of the courses undertaken through the Scottish ITA prepared to a national degree or certificate in 2018. In Michigan, about a fifth of the WIOA participants undertaking training with an ITA in 2016/17 were preparing a degree or a license recognised by the state or the federal government. In Upper Austria, one fourth of those getting the Bildungskonto in 2017 were preparing a formal examination, in particular apprenticeships and the Matura to access higher education. In France, this share was slightly higher, with about one third of all CPF users preparing a formal degree or a professional certificate in 2016 slightly more than 40% of jobseekers using their CPF and 17% of employees. Indeed, jobseekers undertook or intended to undertake longer training than employees (451 hours on average compared to 102 hours for employees over the period 2015-2018). This partly reflects the fact that jobseekers received much higher co‑financing than employees did: hours accumulated on the CPF accounted for 4% of the total training duration for jobseekers, against almost half of the training duration for employees; in both cases, the remainder was financed by the additional funding included in the 2014 law creating the CPF.10
In Upper Austria and Michigan, individuals mostly undertook occupational training programmes. Slightly more than half of the applications to the Bildungskonto relate to medium-level, non-formal courses (i.e. courses that lead to qualifications not recognised as formal by the national authorities and which cannot be classified within the International Standard Classification of Education) to improve occupational competences, mainly in the health, technology and business sectors. In Michigan, ITA participants most often chose training programmes leading to an industry-recognised certificate (Figure 3.3); nursing assistant and truck driver programmes were particularly popular.
Source: OECD Secretariat estimates based on Land OOE Bildungskonto for Austria and the Workforce Investment Standard Record Data and WIOA Participant Individual Record Layout for the United States; DARES based on SI-CPF for France; and Skills Development Scotland.
3.2.2. Nature and duration of training
In a number of schemes, training programmes are strongly concentrated on a small number of short certificates often providing transversal skills (rather than occupational ones) or accreditation to work in some sectors/activities:
Language courses are very popular in France and Geneva. English language certificates (BULATS and TOEIC) represented 40% of the CPF training files validated for employees in 2016 ( (IGAS, 2017[8]), and almost two thirds of the training programmes applied for in the Geneva region through the Chèque annuel de formation over 2010-14 concerned languages.11
IT accreditations come second for French employees (TOSA and PCIE made up 8% of the validated training files in 2016), and were also often chosen in Scotland (although less so over time), reflecting the increasing share of youth among ILA/ITA users.12 Basic IT literacy courses are also frequently taken by the SFC users aged 40‑59 in Singapore, and even more so by those aged 60 and over.13
Basic entry-level certificates providing proof that individuals have the required training and qualifications to work in sectors such as construction, transport and logistics, security and the licensed (alcohol) trade (i.e. bars and restaurants) are the most commonly accessed forms of qualification in the Scottish ILA/ITA. For example, the Construction Skills Certification Scheme (CSCS), managed by the Construction Industry Training Board and normally required as a condition of contract for those working on building sites, made up between 18% and 23% of the ILA/ITA training courses over the 2015-2018 period. CSCS cards are available at different levels, from Trainee and Labourer through to Skilled Craftsperson and Supervisor, but the high volume courses funded through ILA/ITA are mostly at lower levels.14 Some employers and contractors pay for their staff to get a CSCS card but it is not a statutory/legal requirement. The high share of self-employed sub-contractors in the sector may partly explain the frequent use of the ILA to finance a CSCS card, as responsibility for such training is passed from the shoulders of employers to those of workers. The Security Industry Authority (SIA) license is also a requirement to get a job as a front-line security guard or door person, and accounted for 5-6% of ILA training places between 2011 and 2018. Jobseekers in France also often use their CPF to get a Certificat d’aptitude à la conduite en sécurité (CACES) a security clearance training which is required to drive forklifts and work in handling and storage and used to be financed by employers before the CPF was created.15 These certificates are of limited validity in time.
In France, jobseekers most often undertake a programme called Cléa, which was developed by the social partners in 2016 to allow individuals who do not have formal qualifications to assess their basic skills in seven areas.16 If the candidate lacks the necessary skills in one or more of these domains, he/she can get training through short-term and practical programmes to obtain the certificate (OECD, 2017[13]). In second position comes a training programme compulsory for those who want to become self-employed and create or take over a craft business, the “stage de préparation à l’installation”.
Other training programmes undertaken through ILS schemes also appear to be of (sometimes very) short duration. This is the case notably in Geneva, where the Court of Audit notes that the amounts of support provided cannot finance programmes that lead to a formal qualification17, and in Scotland where support is even lower. It also seems to be the case in Singapore, where popular programmes such as effective communication, food and beverage safety and hygiene procedures, as well as private hire car driver licenses last less than 20 hours. In Singapore also, while SkillsFuture Credit does not aim in principle to promote the short-term labour market relevance of training courses, in practice, the courses approved appear to be mostly related to immediate occupational demand.
Part of the participants in the Upper Austrian and French schemes access more expensive/longer duration training. Data on training duration is not available for the Austrian scheme, but Figure 3.4 shows that about a quarter of the 2017 participants received between EUR 500 and EUR 1 000 support, and 10% received between EUR 1 000 and EUR 2 400. Participants combining the Bildungskonto with the educational leave programme (3% of all participants in 2017) received the highest average support and are able to participate in longer educational programmes. In France, almost half of the training files validated in 2016 were for programmes of less than 100 hours, but almost a quarter of them were for programmes of more than 500 hours (Balmat, 2018[14]). Most of these long-duration training programmes concerned jobseekers and were permitted by regions and/or Pôle Emploi adding additional funding to the individual’s CPF hours. As mentioned above, such additional funding covered more than 95% of the hours in the training files of jobseekers validated or closed over 2015-201818. Although the average training hours were much lower for employees using their CPF, 7% of them also accessed such long-duration training programmes – allowing real reskilling – thanks to “abondements” (i.e. complementary funding) from the sectoral training funds or because they undertook an individual training leave (CIF, see 2.5)19.
Source: OECD Secretariat based on Land OOE (2017[15]).
Source: DARES based on SI-CPF.
No data is available on the amount actually spent by participants in the US ITA in Michigan and Washington,20 but some elements are available on the duration of training. The two popular programmes mentioned above, nursing assistant and truck driver, are rather short-duration training (between 2 to 10 weeks), but some longer programmes also appear on the list of eligible training on the Michigan Talent Connect website21. In fact, the average duration of training undertaken through the ITA scheme over the budget year 2016‑17 was not short, reaching about 21 weeks for participants in the WIOA Adult Programme in Michigan and Washington, and 29 and 22 weeks for participants in the Dislocated Workers Programme in Michigan and Washington respectively.
Evaluations of the impact of ILS are relatively rare, and counterfactual impact evaluations establishing causal links even more so. Non-alignment with the labour market objectives has led to reforms in a number of schemes. ILS are often found to increase enrolment in training but deadweight costs can be important.
3.3.1. Alignment with labour market objectives
In a number of cases, ILS impact is found to deviate from its stated objectives. In Flanders, for example, an internal VDAB evaluation in 2008 showed that 60% of the Opleidingscheques were used for training where the employee did not acquire any skills relevant for the labour market, but instead for leisure training (such as cooking classes, language courses, etc.) (Euréval, 2012[3]). In Scotland, training was often taken by inactive individuals to pursue hobby courses, which resulted in significant diversion of the funding away from the groups originally targeted22. This lead to the reform of the scheme in October 2017. In the United States, the main use of the LiLA scheme appeared not to be to obtain a promotion in the current job, since about half of the training programmes undertaken were outside the sector of activity of the participant. One might have expected to observe this pattern in sectors with low wages and poor working conditions sector (such as hotels/restaurants), but perhaps less so in other sectors, like the industrial sector for example (Euréval, 2012[3]).
3.3.2. Impact on training participation and employment
Another goal of ILS is to stimulate investment by individuals in their own education and training and improve their employment prospects. Some evaluations suggest that ILS have increased investment in education, but also that deadweight costs may be present. Employment effects are very rarely evaluated.
In Geneva, a survey done in the Winter 2005-2006 showed that 86% of the participants intended to train even before they learnt about the Chèque annuel de formation; while around 68% of those who had no training project said they had been motivated to undertake some course after hearing about the scheme. Overall, 9% of the surveyed participants were estimated to have been directly motivated by the scheme.
In Canada, the learn$ave scheme – the only scheme covered in this report for which a randomised controlled trial was implemented in 3 out of the 10 delivery sites – was found to have a positive impact on education, as enrolment rates over the period of implementation were 6.6 percentage points higher for participants in the basic learn$ave and 8.2 percentage points higher for those in learn$ave +, i.e. those benefiting from intensive counselling (mostly financial but also employment-related) (Leckie et al., 2010[4]). This effect was mainly driven by increased enrolment in higher education, which was most likely to lead to returns in the form of higher earnings. By contrast, enrolment in individual shorter-term courses did not increase. The programme also increased the completion rate of higher education programmes by 5 percentage points for the basic learn$ave programme, and by 6 percentage points for learn$ave +.
In Austria, an audit of labour market policies in Vienna and Upper Austria over the period 2005-2008 concluded that deadweight was relatively high in the Bildungskonto given that 63% of the participants would have undertaken training even in the absence of the scheme (Rechnungshof Österreich, 2011[16]). These relatively high figures are not surprising given that participants in the Bildungskonto can only apply to the programme once their training has been completed.
A randomised controlled trial evaluation was also carried out in the early 2000s for the US ITA. The evaluation does not allow to establish the net impact of the scheme (compared to non-participation) but rather compares the relative efficiency of the three approaches described in Section 2.4 (i.e. the structured choice approach, the guided choice approach, and the maximum choice approach). The experimental design tried to assess whether providing the customer/participant with more autonomy in his/her training choice made a difference to labour market outcomes. The evaluation found no significant difference between the approaches in terms of their impact on employment (employment rate, hours worked, occupation) or earnings (McConnel et al., 2006[17]). Very few of the participants voluntarily requested counselling (if they were not required to receive it). Mandatory counselling discouraged the use of ITA‑funded training for participants in the Workforce Investment Act programmes. Receiving counselling delayed the start of training but broadened the set of training options considered by participants. Participants in the first approach (i.e. the structured customer choice approach) had to contribute fewer of their own resources to the training, which resulted in longer training duration.
← 1. The exact coverage rate calculated as the number of participants divided by the citizen population aged 25 years or above is 6.5% in 2017 and 5.8% in 2018.
← 2. 2 388 individuals participated in learn$ave over the 2001-2008 period; comprehensive data is not available for the LiLA scheme in the United States, but 359 individuals participated to the first experiment in Chicago, Northeast Indiana and San Francisco over 2001-2007, and 100 workers and 25 employers in the state of Maine.
← 3. The Fonds Paritaire de Sécurisation des Parcours Professionnels (FPSPP), a social partner organisation in charge of mutualising and redistributing funds collected through a 0.2% training levy for employees and jobseekers, paid for these hours, at a fixed rate of EUR 9 per hour.
← 4. The absence of information on training supply was a deliberate choice by social partners, in order to avoid direct interference on training markets. However, the choice of a particular certificate can be linked to that of the training provider, as practical modalities of the training programmes (date, place, etc.) are determining in the choice made (IGAS, 2017[8]).
← 5. At the same time, maximum amounts of support were raised from EUR 900-1 300 to EUR 2 000-2 400.
← 6. After having risen from GBP 1.4 million in 2005-06 to GBP 9.2 million in 2010-11, spending on ILA has been constantly decreasing to GBP 3.6 million in 2016-17. The share of the ILA in the overall training expenditure by Skills Development Scotland has also diminished from 8.3% to 3.2% between 2010-11 and 2016-17.
← 7. At the beginning of May 2019, the “My World of Work” website indicated that “SDS and the Scottish Government are exploring options to maximise the impact of the scheme, within the available budget”. SDS has now imposed a cap on numbers (15 000) and a cut-off date for applications (September), both of which had never been used before.
← 8. Source: interview of SDS official.
← 9. Data on expenditure on training funded through ITA is not available. Michigan and Washington spend respectively one third and two thirds of their WIOA funds (USD 8 million in Michigan and USD 21 million in Washington in 2016‑17) on all types of training.
← 10. CPF and DIF hours represented 17 hours and 46 hours of the total for jobseekers and employees respectively over 2015-2018 (source: CDC, système d’information du compte personnel de formation).
← 11. In Singapore, Korean and Japanese courses come in second position of the most often chosen programmes for SkillsFuture Credit users under the age of 40.
← 12. The numbers taking ECDL, an established ICT accreditation which provides a recognised benchmark for digital literacy have declined steadily from 7% of ILA applications in 2011/12 to 4% in 2017/18. In part, the increased take‑up of the ITA/ILA by the younger age group may be explained by the reduction in the number of places in available alternative programmes such as the Employability Fund.
← 13. Four of the top five SFC courses for those aged 60 years and over, and two of the top five SFC courses for those aged 40-59 were basic IT literacy courses such as learning how to use smartphones, tablets and computers.
← 14. In August 2018, there were 87 ITA approved courses providing this type of training across Scotland.
← 15. CACES comes in third position of the top 10 certifications chosen by jobseekers using their CEP over the period 2015-2018 (source: Caisse des Dépôts et Consignations/système d’information du compte personnel de formation).
← 16. Cléa accounted for 11% of the validated training files of jobseekers using the CEP in 2016 (Balmat, 2018[14]) and was the top certification chosen over the 2015-2018 period (Source CDC, SI CEP).
← 17. Professional retraining can be made through other financing schemes, such as Qualification +.
← 18. Source: CDC, SI CPF.
← 19. 3.5% of the validated and closed files for employees over the 2015-2018 period were treated by the CIF financing institutions (OPACIF/FONGECIF); overall, over the 2015-2018 period, additional funding (by employers, training funds, CIF financing institutions, etc.) covered 36% of the total expenditures of the validated training files for employees (Source: CDC, SI CPF).
← 20. Data on total training expenditure on WIOA Adult and Dislocated Workers programmes are available, but not on ITA-training.
← 21. Duration of training programmes is very rarely provided; one example is 16 weeks for a cybersecurity operations training.
← 22. Source: interviews of Scottish officials.