Impact Evaluation of Training and Wage Subsidies for the Unemployed in Greece

Report

Connecting People with Jobs

8 February 2024

4. Evaluation of training programmes for unemployed workers in Greece

Abstract

Training programmes are important to reskill and upskill jobseekers and help to access good jobs. To understand the impact of Greece’s training programmes for unemployed people on their labour market outcomes, this chapter presents the results of a counterfactual impact evaluation of the main training programmes available for unemployed people. In doing so it considers a broad range of outcomes including the likelihood of finding employment and earnings, but also total days worked, occupational mobility, unemployment benefit receipt, registration with the Greek public employment service, and inactivity.

4.1. Introduction

Training programmes are an important component of active labour market policies (ALMPs). They support unemployed persons to reskill and upskill, which can make them more attractive as employees thereby supporting stronger labour market outcomes. However, training programmes require resources to fund and not all programmes are equally successful. Indeed, even when outcomes of participants are tracked and people in training programmes are known to find employment, it is not immediately clear to what extent these results are causal, after all some participants would have found work without the training. To understand the causal impact of training programmes in Greece this chapter provides counterfactual impact evaluations of the main training programmes available for unemployed people. The methodology used and the programmes covered are presented in Chapter 3.

This chapter finds that Greece’s training programmes are effective. The findings indicate that the training programmes increase the probability of employment, wages, cumulative days in employment and cumulative earnings, while they decrease the probability of registering as unemployed, and inactivity. The programmes support employment outcomes in the short, medium, and long term, except during the initial months when participants are involved in training activities. The estimated effects are higher than those of most training programmes that have been studied internationally. They also support a wide range of different jobseekers into employment, across age, gender, and unemployment duration. These results are consistent with the results of another recent study of Greek training programmes and suggest a strong prima facie business case for increasing the reach of training programmes in Greece.

The rest of this chapter is laid out as follows. Section 4.2 examines the effect of Greek training programmes on the probability of employment, registration as unemployed, inactivity, the probability of being on a benefit, as well as the effect of training programmes on wages, days in employment, and occupational mobility. Section 4.3 looks at whether these programmes are effective for men, women, people of different ages, those who live in different places, and people who have different durations of unemployment. Section 4.3 also breaks down the results by the three different training programmes included in this study. Finally, Section 4.4 compares the findings of this report with international studies and related evidence on the effectiveness of Greek training programmes.

4.2. Training programmes are effective at supporting workers’ employment outcomes

To answer the question of whether or not a programme is effective, it is important to understand what would have happened to training participants had they not taken part in training – the counterfactual. As explained in detail in Chapter 3, the methodology used to estimate this counterfactual matches participants in training programmes to similar non-participants using a technique called propensity score matching. This section discusses the results of this analysis which makes use of linked administrative data from the register of Diofantos on training participants, the Greek public employment service (DYPA) register and the ERGANI register, as described in Chapter 3.

4.2.1. Training programmes support people into employment

Figure 4.1 shows the results of the propensity score matching methodology for the likelihood of being employed. Panel A shows the percentage of people who find employment in both the treatment and the matched control group over time (in months since the start of training). Panel B shows the overall effect of the training programmes, which is calculated as the difference between employment rates of the treatment and control group shown in Panel A.

The analysis shows that training programmes are effective at supporting people into employment. Twelve months following the start of the programme, participants’ probability of employment is about 8 percentage points higher than the matched control group (Figure 4.1, Panel B). This effect is persistent and stable over the medium to long term, 12 to 36 months after the training commenced.

The analysis also shows the importance of comparing the results to a similar group of people when calculating the effect of the programme. Sometimes analysis of the effects of training or other ALMPs considers only the “gross effect” of the programme, or how many people are employed following training. Indeed, at the 12‑month mark, about 18% of participants are employed (Figure 4.1, Panel A). However, without a comparison group this figure would be hard to interpret as some of these persons would have found employment even without training. The use of a matched comparison (or control) group however provides this counterfactual. Panel A shows that about 11% of the control group are employed 12 months after they enter the control group (that is when similar people enter the training programmes). So the difference between the two, 7 percentage points, shows the impact of training on the probability of being in employment 12 months after entering training. This is the counterfactual effect at 12 months. The effect of training on the likelihood of employment remains positive and sizeable at longer time horizons and stands at 8 percentage points three years after starting the training.

Training takes time to have a positive impact on the likelihood of employment. This kind of training for unemployed people in Greece typically lasts five to seven months (see Chapter 3). During this time, participants have much less time to work, or search for work, so employment rates are low and the effects of training during these initial early months are even slightly negative at times. These short- to medium-term effects of trainings are known as “lock in” effects in the literature (Card, Kluve and Weber, 2017[1]).

Figure 4.1. Training participants are more likely to be employed than comparable non-participants

Share of individuals in employment and percentage point effect in employment probability

4.2.2. Training programmes improve a broad range of outcomes

While finding employment is one of the most studied outcomes of training programmes, it is far from the only important one. The rich administrative data provided by the Greek authorities allow for a detailed look at not only whether training participants find a job following training participation, but also the effects of training on unemployment, unemployment benefit receipt, inactivity (defined as not employed, registered as unemployed, or receiving a benefit), earnings, the number of days they worked, and occupational mobility.

The results presented above showed the outcomes for the treated and control groups separately. While this is helpful to have a full account of how the treatment effects are arrived at (and to illustrate the methodology) for brevity the results in this section focus only on treatment effects, while the separate outcomes for the treatment and control group are presented in the technical report (OECD, 2024[2]).

Figure 4.2 presents the estimated effect of training programmes on labour force status (unemployment, benefit receipt and inactivity). These are defined on the basis of administrative data1 rather than the survey-based measure. These results thus merit caution given the large discrepancy between registered and survey-based unemployment in Greece, with survey-based unemployment steadily decreasing over the past 10 years even as registered unemployment has remain relatively constant and leading to a large discrepancy between the two (see discussion in Chapter 2).The impact on employment is copied over from Figure 4.1 (Panel B) for completeness. Overall, training programmes are found to:

Decrease registered unemployment: Registered unemployment is briefly higher during the initial “lock-in” phase when participants are still in the training but falls thereafter. Twelve months following the start of training, participants are about 5 percentage points less likely to be registered as unemployed, an effect which mostly persists over the medium term, but falls slightly over the longer term and amounts to 2 percentage points three years after entering training.
Increase unemployment benefit receipt: Starting from 15 months after training begins, the effect of training on unemployment benefit receipt is positive and rises to nearly 10 percentage points by month 36. This positive effect on unemployment benefit receipt over the medium and longer term is not because participants are working less – in fact, in general they are working more because of the positive effects of programme participation on employment. However, a larger share of them also cycle in and out of employment, often with periods long enough to trigger the payment of unemployment benefits when they become unemployed. This also explains why there is no effect of training on benefit receipt during the first year.
Reduce inactivity: Programme participants are less likely to be inactive than non-participants. During the early months, this reflects a higher likelihood of being registered as unemployed, but over the longer term it is driven by higher rates of employment.
Increase employment (as discussed in Section 4.2.1 above).

These effects are statistically significant, as shown in the technical report (OECD, 2024[2]) where the confidence intervals of these point estimates are reported.

Figure 4.2. Training programmes support better outcomes

Percentage point effect on being in different labour market states

Moving beyond the effects of training programmes on labour force status, it is important to analyse the characteristics of jobs that training participants find following training. Figure 4.3 looks at the sustainability of employment by observing the effect on cumulative days in employment since programme start, whether participants are able to earn more (also expressed cumulatively since the start of the training programme) and whether occupational mobility is improved. Specifically, the results show that participation in training programmes leads to:

Increased cumulative days in employment: Cumulative days in employment for training participants begin to rise compared to the control group from about nine months following the start of the programme and continue to rise until the end of the study period, three years later. At this three‑year mark, training participants have experienced 66 more days in employment than similar jobseekers who did not participate in training.
Increased cumulative earnings: Cumulative earnings are also positively affected, following the lock-in period during training participation.2 Cumulative earnings effects build consistently over time and, three‑years following the start of training, participants have earned EUR 1 500 more than the matched comparison group (in 2015 prices).

The effects on days in employment and cumulative earnings are each linear, showing a consistent trend that shows no signs of abating. This is consistent with the finding of a stable increase in employment each month as shown above.

One outcome that is not discussed in this section is occupational mobility. Although this is an important question, as training could help people to move into different, and potentially higher paying occupations, it turns out that the results vary crucially based on the type of training provided. This aspect is thus discussed in Section 4.3.1.

Figure 4.3. Training programmes improve cumulative earnings and days in employment

Cumulative days worked and cumulative earnings for those who found a job

4.2.3. Certain caveats should be taken into account when interpreting the results

Some caution should be used when interpreting these results. In particular, there are some concerns with omitted variable bias (especially motivation, which would cause the results above to be overstated), unobserved employment in the informal economy (which has an ambiguous effect on the results), and displacement effects (which could cause overall employment effects to be overstated).

One of the most important caveats is that participants largely self-select into training programmes (see Chapter 3). As a result, participants are different to non-participants in a number of ways. This is the motivation for using the propensity score matching methodology which controls for the way the treatment and control group differ on observed characteristics. However for propensity score matching to effectively identify causal effects it is necessary that after matching on these observable characteristics participants are also similar on unobserved characteristics (Rosenbaum and Rubin, 1983[3]). Unfortunately, it is not possible to be sure if all unobserved characteristics are controlled for. A notable unobserved characteristic is motivation: It is plausible that those jobseekers who seek out and enrol in training are more motivated to find work than those who do not participate. If this is the case, training participants could be expected to have been more likely to find work anyway, even without participating in training. This may lead to an overestimation of the effectiveness of training.

There are reasons to believe that omitted variables bias (including from differing levels of motivation, informality) is mitigated. In particular the technical report (OECD, 2024[2]) shows that the matched participants are not only comparable to the comparison group in the characteristics that were matched upon, but also on other variables that were not explicitly accounted for, showing that at least some “unobserved” characteristics are accounted for. The methodology accounts for a plethora of characteristics, and as Chapter 3 shows, the participants appear similar to the control group on these matched characteristics.

A second important caveat is that this study looks at the effect of training on formal employment and cannot account for informal employment. Ideally this study would include informal employment both as an outcome variable and as a control variable. As an outcome it would be beneficial to understand if training has an effect on informal employment. As a control it would be beneficial to account for informal employment which may or may not affect whether participants are more or less likely to enter training and may or may not affect whether participants are more likely to find formal employment in the future. Of course, neither controlling for informality or studying the effects of training on informality is possible with the administrative data used in this study. Likewise, it is unclear how controlling for informal employment would change the results of this study.3 However, the magnitude of the bias could potentially be material given that informal employment is fairly present in Greece (European Labour Authority, 2020[4]), with the informal economy making up perhaps a quarter of GDP (IMF, 2019[5]).

Another possible caveat concerns displacement effects. In the context of training, these occur when those who have completed the training fill a job vacancy at the expense of non-participants (whose job is “displaced”) thereby leading to no overall change in the number of jobs. It is not possible to examine these using this study’s methodology. And indeed, given the technical difficulties of studying this phenomenon, few studies have looked at displacement effects for ALMPs, and those that have find differing results. However, being cognisant of these potential sources of bias means treating the estimates with some uncertainty and viewing them as less precise than they might otherwise appear.

To address concerns over not accounting for all relevant factors in constructing treatment and control group (omitted variable bias due to unobserved heterogeneity), in the future Greece could consider conducting randomised control trials. In such a setting some eligible participants would be randomly offered training. Randomisation ensures that treated and control group individuals are, on average, the same, including for characteristics that are unobserved such as motivation and previous experience in the informal economy. Such randomised trials require actively managing the randomisation process during the roll out of the programmes and so necessitate that researchers are involved during the design and implementation phase.

4.3. Greece’s training programmes are effective across different groups of jobseekers

The previous section showed that training programmes in Greece are effective. This section breaks down these aggregate results by training programme and by group to see if all three programmes studied are effective and if training is effective for different groups of jobseekers.

4.3.1. All three training programmes analysed have positive employment effects, but only ICT training helps occupational mobility

There are several reasons for the results for the three programmes studied to differ by programme. First, the programmes offered different content, with trainings focused on different subjects. The high-demand sectors training covered a wide range of different occupations, while the two other training programmes studied focused on the ICT sector. Second, the programmes started at different times. The training for high-demand sectors occurred in the second half of 2017, while the training for ICT participants occurred during the first part of 2020. This means that participants in the later programmes graduated during the turbulent labour market of the COVID‑19 era in 2020, while the other pre‑COVID participants graduated during a period with high unemployment (even higher than that during the COVID‑19 pandemic in 2020, see Chapter 2). It would not be unusual for training effects to differ over the economic cycle with Card et al. (2017[1]) finding that ALMPs tend to be more effective during recessions. Furthermore, the short-time work scheme which was fairly widely adopted in Greece and provided necessary support during the crisis (OECD, 2022[6]), may further affect the results. Third, the trainings targeted different groups, and different groups may have different needs and respond differently to training. One ICT programme targets persons aged 25‑29 while the other targets those aged 30‑45. The high-demand industries training covers a wide range of age groups. Moreover, only 9% of those participating in the high-demand industries training and the ICT training for persons aged 30‑45 have been registered as unemployed for less than one year, compared to 45% of ICT training participants for those aged 25‑30.

Despite these differences, all three training programmes studied increase employability in Greece (Figure 4.4). Indeed, all programmes achieve similar levels of effectiveness at supporting people into employment, at least one to two years following programme enrolment, where estimates are similar and not statistically different across programmes. As the ICT programmes studied start in the first part of 2020, it is not possible to study the effectiveness of these programmes beyond 27 to 30 months post-programme start due to a lack of follow up data (data are only available until August 2022).

Figure 4.4. All three training programmes studied improve jobseekers’ employment outcomes

Effect of training on employment probability across three training programmes

Quite importantly, both ICT trainings have a positive effect on occupational mobility, whereas the high-demand industries training does not (Annex Figure 4.A.1). Beginning from six months after entry into the programme, a clear pattern emerges: ICT training leads to considerable boosts in occupational mobility for younger individuals and moderate boosts for older ones, with the groups entering occupations that pay on average six and 3 percentage points more, respectively. Participants in the training for high-demand sectors, on the other hand, experience a small negative effect on their occupational mobility (although the effect is statistically significant only during a small majority of the intervals examined). These comparisons are made difficult by the relatively small sample sizes, particularly for the ICT training, leading to effects that are imprecisely estimated (with large confidence intervals). These results nevertheless underscore the potential for the ICT training to facilitate individuals making a career switch in line with the digital transition, enabling the Greek economy to restructure to more high value‑added jobs.

Interestingly, the result that ALMPs provide a greater boost to the occupational mobility of younger jobseekers fits in with a broader set of similar findings on the effects of ALMPs on occupational mobility of different age groups. Specifically, a similar set of findings is found in the evaluation of Greece’s wage subsidy programme (see Section 5.3.2 in the next chapter). Furthermore, qualitatively similar results were also reported in a recent evaluation of Lithuania’s ALMPs (OECD, 2022[7]). This nascent but growing body of empirical evidence suggests that for individuals who experience unemployment, “climbing the career ladder” is a salient feature particularly for individuals earlier on in their working life. ALMPs may help boost career trajectories, but they may also impose a small negative effect on occupational mobility – albeit one that is counteracted by considerable increases in employment rates (as in the case of Greece’s training for high-demand sectors).

How do the different training programmes compare in terms of their effects on participants’ earnings and days worked? As discussed in the technical report (OECD, 2024[2]), all three programmes are equally effective at increasing cumulative days in employment and cumulative earnings over the medium term. It is true that there are short-term “lock-in effects” during the initial 12 months for the ICT training for young workers, reflected in a negative effect on days worked and cumulative earnings. Nevertheless, the estimated effects on both days worked and earnings are relatively similar at the longest time horizon for which reliable estimates are available, approximately 27 months. At this point, the estimated effects on cumulative net earnings amount to roughly EUR 1 000, but with a clear upward trajectory.

Ideally, estimates of the effects of training programmes could be compared with the costs to gauge their cost-effectiveness. As described in Chapter 3, the per-participant costs of ICT training are roughly 50% higher than the training for high-demand sectors (EUR 5 990 versus EUR 3 995 respectively) in principle. However, the present estimates and data available do not permit for cost-effectiveness comparisons without untenable assumptions. Most prominently, this is due to the relatively short time horizon for which outcomes of the ICT training can be examined. Training programmes represent investments in individuals’ human capital that continue to yield payoffs over long time horizons – indeed, the point estimates for the earnings effects of the ICT training for youth are negative for all periods during the first year after entry into the programme but exhibit a sharply positive trajectory. The estimates of the effects of training suggest the direct effects could be durable, requiring a longer observation window.4 A second important caveat relating to comparisons of the cost effectiveness is that the ICT programmes were implemented directly during the onset of the COVID‑19 pandemic. This means that the estimates of the effects over time may be considerably impacted by the effects of the COVID-related shocks, limiting the extent to which the programmes’ effects can be directly compared to one implemented in an entirely different set of economic circumstances. Finally, even if the previous aspects were completely ignored, any comparison of the programmes’ relative effectiveness would need to account for the differing composition of the training participants. This is relevant because the programmes’ observed outcomes reflect not only the effectiveness of the programmes themselves, but also on how the specific groups of participants may benefit from them. As the next section makes clear, these effects vary considerably across different types of participants.

4.3.2. Groups of jobseekers such as the short-term unemployed disproportionally benefit from training

Even if the training programmes are equally effective overall, they may not be effective for all jobseekers. This section examines the effects for groups of jobseekers defined by gender, age, education, region, and unemployment duration. These groups may have different needs and thus might experience different effects from training. Although the estimated results differ slightly for some of these characteristics, it should be noted that positive effects on outcomes such as employment probability are present for all groups.

Younger workers experience a disproportionately large boost to employment (Figure 4.5). Two years after entering training programmes, men under age 30 are 19 percentage points more likely to be employed than their matched control group peers. For women in this age group, the effect is almost as large, 16 percentage points. Older groups of jobseekers experience systematically lower effects, with the employment boost to men over 50 amounting to 6 percentage points (1 percentage point lower than for similar age women). The effects over longer time horizons yield qualitatively similar results and are reported in the accompanying technical report (OECD, 2024[2]), although they refer only to results for the training for high-demand sectors (the time horizon for ICT training permits only results up to 27 months to be reported). In contrast to the results by age, the effects across men and women do not vary systematically.

Individuals with higher levels of education appear to benefit slightly more from training. This is partly driven by the larger effect of individuals in the ICT training, where participants are required to have more than secondary education in order to participate. However, a qualitatively similar – albeit statistically insignificant – result is observed when examining only participants of the training for high-demand sectors.

One possible explanation for the higher effectiveness of training amongst those with higher levels of education may be that they are better poised to seize the opportunities offered by reskilling due to a high degree of qualification mismatch among workers. In Greece, high shares of workers are employed in jobs with a qualification or field of study that does not match the job’s requirements. OECD estimates indicate that 42% of workers have a field of study that does not match their job’s requirements, the highest in the OECD, where it averages 32% (OECD, 2023[8]).5 Furthermore, 25% of workers had a level of qualification that exceeded the requirements for their job – a rate that is also among the highest in the OECD, where the average is 16%. In such an environment, vocational training may play an important role in reskilling workers for jobs in occupations and sectors that make better use of their talents and skills – an opportunity that individuals with higher levels of education may be better placed to exploit (as argued by, for example, Schultz (1975[9])).

Individuals with shorter unemployment spells benefit considerably more from training compared to individuals with longer unemployment spells. The difference between the two groups is sizable, amounting to 5 percentage points. The estimated effect may partly reflect differences in motivation: given that entry into training is largely done at an individual’s own initiative, this difference may reflect the fact that such individuals are more motivated to acquire the skills necessary to get a new job as quickly as possible. They may also thus benefit more from the training offered.6

Finally, examining the effects by location shows that individuals in cities experience a greater boost in employment probability than individuals in other locations. This effect may partly be explained by the broader density of training providers in larger cities, giving individuals the possibility to choose a specific training that better matches their personal career aspirations. It may also be tied to local labour market effects – as discussed in Chapter 3 (Section 3.3.1), Attica and Thessaloniki had lower rates of jobseekers exiting unemployment compared to jobseekers in the other two geographic breakdowns shown. One possible interpretation of these factors together is that the training may have stronger effects in weaker labour markets, which is precisely the finding of a meta‑analysis comparing the results of many impact evaluations – a topic discussed in the next section.

Figure 4.5. Training programmes are effective for many groups of jobseekers

Percentage point effect in employment probability at 24 months after starting training

4.4. Training programmes in Greece appear effective in comparison with trainings in other countries

Many training programmes have been evaluated in the international literature. To examine how the results for Greece compare with these studies, this section compares the results on Greece with those from two meta‑analyses. The first, conducted by Card, Kluve and Weber (2017[1]), covers 49 countries in total, and summarises estimates from over 200 impact evaluations of ALMPs. Of these, 51 impact evaluations include point estimates of the employment effects of training programmes comparable to the ones in Greece. The second meta‑analysis covers programmes funded by the EU’s European Social Fund (ESF) and includes estimates from 20 studies examining vocational training as well as 19 classified as mixed interventions, combining e.g. vocational training with other types of support (European Commission and Ismeri Europa, 2023[10]).

The distributions of the estimates from these two meta‑analyses are shown in Figure 4.6 alongside the results from this study. The meta‑analysis by Card, Kluve and Weber (2017[1]) does not provide estimates of the effects of other outcomes analysed for Greece, such as earnings or days worked or occupational mobility. The meta‑analysis of the ESF programmes contains some studies that reported the effects on earnings, but there are not enough studies of vocational training with these estimates to make meaningful comparisons.

The comparison shows that the effects of the training programmes in Greece are quite large. In most of the comparisons, Greece’s training programmes are more effective at supporting people into employment than 75% of training programmes studied in each of the short, medium, and long run (in years one, two and three after the start of training). The relatively low estimates for the short-term effects in the comparison studies may be partly attributable to differences in estimated lock-in effects across programmes, which lead to negative short-term effects for the duration of the programme participation.7

Figure 4.6. Greek programmes are estimated to mostly be in the top quarter of training programmes evaluated internationally

Percentage point effect in employment probability

In addition to these international studies, there are two recent studies that estimate the effect of trainings on labour market outcomes in Greece. The first is a report by IOBE (2021[12]) which estimates the impact of a different set of training programmes that have a larger target group than that of the trainings analysed in this report, namely not only unemployed persons but also students who have recently finished high school and adults who are working at the time the training commences. IOBE (2021[12]) uses similar data to this report, namely data from the registers of DYPA, Diofantos and Ergani. However, it includes only a random sample of unemployed non-participants and not all persons registered as unemployed over the study period, which is the sample in the analysis conducted in this report. In terms of the methodology, both IOBE (2021[12]) and this report adjust for systematic differences between participants and non-participants, even if the exact method differs – regression analysis in IOBE, compared to propensity score matching in this report. The two reports reach similar conclusions except that the IOBE report finds that not all programmes decreased the likelihood of unemployment.

Another study, by the World Bank, also found positive impacts of a Greek ALMP, in this case a pilot programme launched in 2018 at the Elefsina local office (World Bank, 2021[13]). This programme was comprehensive and the estimated impact captures the effects of all its different interventions, including training. Training was a significant component of the programme: of the 948 persons participating in the programme (who all received at least counselling services alongside profiling) about 80% received training. In addition, 8% of programme participants participated in a wage subsidy programme. The World Bank study found participants in the Elefsina pilot had improved labour market outcomes, with about a 3 percentage point increase in employment and a 5 to 6 percentage point decrease in registered unemployment, though the authors note the results should be interpreted with caution due to the potential for unobserved differences between participants and non-participants.

4.5. Conclusion

This chapter has shown that training programmes for unemployed persons are effective in Greece. The large magnitude of the effects suggests they are even more effective than most training programmes that have been studied internationally and these findings are buttressed by evidence of effectiveness found in other studies of training programmes in Greece. These findings have several possible interpretations. The findings suggest that the training programmes they have been successful at targeting skills needed in growing sectors, as witnessed by the fact that the largest uptake has been in the same sectors and occupations which have also experienced the highest employment growth (CEDEFOP, 2023[14]). It also suggests that the training imparts useful skills that employers find relevant. At the same time, given the extremely low expenditures on training in Greece during the period studied, part of the explanation could also be that the highly positive effects of training reflect a large unmet need for reskilling. However, as the scale of training for workers and the unemployed is set to increase substantially during the 2022‑25 period, the impact of these investments might become more moderate due to diminishing returns to such human-capital investments. Nevertheless, any potential reduction in the impact of training due to increased provision may be offset by enhancements in the quality of training programmes. Specifically, DYPA’s new strategy, which involves integrating outcome‑based payments for training providers, could effectively address this issue by strengthening the incentives for better-quality training programmes.

References

[15] Card, D., J. Kluve and A. Weber (2017), “What Works? A Meta Analysis of Recent Active Labor Market Program Evaluations”, Journal of the European Economic Association, Vol. 16/3, pp. 894-931, https://doi.org/10.1093/jeea/jvx028.

[1] Card, D., J. Kluve and A. Weber (2017), “What Works? A Meta Analysis of Recent Active Labor Market Program Evaluations”, Journal of the European Economic Association, Vol. 16/3, pp. 894–931, https://doi.org/10.1093/jeea/jvx028.

[14] CEDEFOP (2023), 2023 skills forecast: Greece, https://www.cedefop.europa.eu/files/skills_forecast_2023_greece.pdf.

[11] European Commission (2023), , Meta-analysis of the ESF counterfactual impact evaluations – Final report, Publications Office of the European Union, Directorate-General for Employment, Social Affairs and Inclusion, Pompili, M., Kluve, J., Jessen, J. et al.,, https://data.europa.eu/doi/10.2767/580759.

[10] European Commission and Ismeri Europa (2023), Meta-analysis of the ESF counterfactual impact evaluations – Final report, Publications Office of the European Union, https://doi.org/10.2767/580759.

[4] European Labour Authority (2020), Training labour inspectors to use the new IT tools, https://www.ela.europa.eu/sites/default/files/2021-09/EL-TrainingInspectorsToUseNewITtools.pdf.

[5] IMF (2019), “Greece”, IMF Staff Country Reports, Vol. 19/341, https://doi.org/10.5089/9781513520261.002.

[12] IOBE (2021), Απόδοση επένδυσης στο πεδίο της επαγγελματικής εκπαίδευσης και [Return on investment on Vocational Education and Training], Unpublished.

[2] OECD (2024), “Technical report: Impact Evaluation of Training and Wage Subsidies for the Unemployed in Greece”, OECD, Paris, https://www.oecd.org/els/emp/Greece_ALMP_Technical_Report.pdf.

[8] OECD (2023), Skills for jobs: mismatch by country, https://stats.oecd.org/Index.aspx?DataSetCode=S4J2022_MISMATCH.

[7] OECD (2022), Impact Evaluation of Vocational Training and Employment Subsidies for the Unemployed in Lithuania, Connecting People with Jobs, OECD Publishing, Paris, https://doi.org/10.1787/c22d68b3-en.

[6] OECD (2022), “Riding the waves: Adjusting job retention schemes through the COVID-19 crisis”, OECD Policy Responses to Coronavirus (COVID-19), https://doi.org/10.1787/ae8f892f-en.

[3] Rosenbaum, P. and D. Rubin (1983), “The central role of the propensity score in observational studies for causal effects”, Biometrika, Vol. 70/1, pp. 41-55, https://doi.org/10.1093/biomet/70.1.41.

[9] Schultz, T. (1975), The Value of the Ability to Deal with Disequilibria, American Economic Association, pp. 827-846.

[13] World Bank (2021), Greece: Improving the Design and Delivery of ALMPs - Phase II: Monitoring Report #2 – Elefsina Pilot Program, https://documents1.worldbank.org/curated/en/955371622093970259/pdf/Monitoring-Report-No-2-Elefsina-Pilot-Program.pdf.

Annex 4.A. Additional results relating to occupational mobility

Annex Figure 4.A.1. ICT training often leads to positive occupational mobility, training for high demand sectors generally leads to slightly negative job mobility

Change in occupational index for those who found a job

Notes

← 1. Given the way these are defined in this report, inactivity is treated as a residual category (not registered as unemployed with DYPA or registered as employed in Ergani data) each person is in at least one category. However, it is possible to be in more than one category – for example, registered as unemployed and receiving benefits. People are classified as unemployed, employed or receiving benefits in a given month if they have respectively on registered day of unemployment, employment, or any benefit earnings in that month. The measures here are based on administrative data and so are not the same as those measured in survey data from the Labour Force Survey. In particular for Greece, it is known that there are high levels of informality (which may be measured in the Labour Force Survey as employment if respondents report it but by definition does not occur in Ergani) and the downward trend in the unemployment rate has not been marked by a decreased in registered unemployment of the same magnitude (see Chapter 2 for discussion).

← 2. This income does not include unemployment benefits but only direct earnings from the employment registered in ERGANI.

← 3. For example, if (after matching on other variables) past informal employment is associated with people being more likely to participate in training but less likely to be employed, then this study would understate the effects on training given that this study does not control for past informal employment. However, the sign of the bias would be reversed if informal employment makes people less likely to be in training. Indeed, the sign of the bias would be reversed again if informal employment is associated with an increased likelihood of formal employment in the future.

← 4. In addition, the ICT training programme in particular could have positive indirect, general equilibrium effects on the Greek economy, for example by increasing its productive capacity.

← 5. Statistics are calculated based on survey data sources and relate to the last year available (mostly 2019, including for Greece).

← 6. The process of matching individual participants to similar control group individuals explicitly takes into account past employment history in the matching process, matching only individuals who have similar patterns of employment.

← 7. This is especially true given that the time horizons for estimating the short-term effects vary considerably, with some studies measuring effects as early as in the first month after entry.

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Countries

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Publications

Publications

Featured publications

Data

Data

Featured data

News & Events

News & Events

Featured Events

About OECD

About

Engage with us

Work with us

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment