Navigation – Plan du site

AccueilPublicationsFact Sheets2025The impact of skills training on ...

2025
208

The impact of skills training on job integration for job seekers in Brussels

Brussels Studies fact sheet
L’impact de la formation qualifiante sur l’insertion en emploi des demandeurs d’emploi à Bruxelles
De impact van beroepsopleidingen op de arbeidsinschakeling van werkzoekenden in Brussel
Catherine Smith
Traduction de Jane Corrigan
Cet article est une traduction de :
L’impact de la formation qualifiante sur l’insertion en emploi des demandeurs d’emploi à Bruxelles [fr]
Autre(s) traduction(s) de cet article :
De impact van beroepsopleidingen op de arbeidsinschakeling van werkzoekenden in Brussel [nl]

Résumés

L’étude analyse, pour les demandeurs d’emploi bruxellois, l’impact d’une entrée en formation qualifiante sur l’insertion dans un emploi durable. Pour ce faire, la durée de chômage des demandeurs d’emploi entrés en formation qualifiante est comparée à celle des demandeurs d’emploi qui n’en ont pas suivi. Ces deux groupes sont rendus comparables par la méthode de l’appariement. Les résultats mettent en avant un impact positif d’une entrée en formation qualifiante sur l’accès à l’emploi après la formation. De plus, l’impact positif de celle-ci augmente avec la durée de chômage et est plus marqué pour les femmes, les demandeurs d’emploi ayant un diplôme étranger non reconnu et ceux d’une nationalité hors Union européenne.

Haut de page

Notes de la rédaction

In order to see the figures in a better resolution, go to the article online and click on “Original” below them.

Notes de l’auteur

Study carried out by the Service Études et Statistiques at Bruxelles Formation in collaboration with view.brussels, the research department at Actiris.

Texte intégral

Introduction

1Skills training for job seekers is part of the arsenal of public employment policies and is considered an essential tool for occupational integration at European level. In order to meet the objectives of the Europe 2020 Strategy and the European cohesion policy, the Brussels government adopted the Plan Formation 2020 at the end of 2016. The objectives of the Plan Formation 2020 are to develop and strengthen skills training for job seekers, both quantitatively and qualitatively, and to increase the employment rate of the inhabitants of Brussels, by improving their skills and certification levels. Assessing the impact of skills training on job placement is therefore crucial.

  • 1 https://www.bruxellesformation.brussels/wp-content/uploads/2023/03/Analyse-de-limpact-de-la-formati (...)

2The study summarised here analyses the impact of enrolment in skills training on integration into permanent employment in the Brussels Region1. In order to do this, the length of unemployment for job seekers who have enrolled in skills training has been compared with that of job seekers who have not.

3The data are from Actiris, Bruxelles Formation, ONSS and INASTI. The reference population is made up of people aged between 21 and 54 who registered as unemployed job seekers (UJS) with Actiris between January 2013 and December 2015. These people are monitored for four years, and any enrolment in skills training at Bruxelles Formation and its partners within 18 months of registering as unemployed is taken into account. In total, the reference population is made up of 122,373 unemployed job seekers, 2,912 of whom have enrolled in skills training.

4The impact of skills training cannot be measured directly by the difference in the length of unemployment between job seekers who have or have not enrolled in training, due to selection bias. The personal characteristics of job seekers (gender, age, level of education, etc.) influence both enrolment in skills training and exit from unemployment to employment. In order to minimise this bias, job seekers who had undergone skills training (“treatment group”) were matched with job seekers with similar characteristics who had not undergone skills training (“control group”). This matching – using the selected method (Coarsened Exact Matching) – was carried out on the basis of the socioeconomic variables available to us (gender, age, nationality, level of education and place of residence) and the number of months of unemployment during the five years preceding the unemployment period in question.

1. Impact according to the moment of enrolment in training

5We therefore assess the impact of training by comparing the length of unemployment of job seekers in the treatment group (who began training) and the control group (who did not begin training).

  • 2 In our analyses, we use the Kaplan-Meier estimator, which provides a nonparametric estimate of the (...)

6The length of unemployment is analysed using the “unemployment survival function”2, which estimates the proportion of job seekers who are still unemployed at a given point in time, conditional on the fact that they have not left unemployment until then (conditional probability).

7We estimate the impact of training on the unemployment survival function for people who began training in the first month after registering as unemployed, in the second month and so on, up to the 18th month. For each month after registering as unemployed (month 1, month 2,..., month 18), the control and treatment groups are selected from those who were still unemployed at the start of the month in question. The treatment group is made up of UJSs who began training that month, and the control group is made up of UJSs with similar socioeconomic profiles who did not begin training that month but are likely to begin training in the following months. The effect we estimate is therefore that of enrolling in training at a given point in the unemployment period, compared with not enrolling in training at all, at least up to that point [Sianesi, 2004].

  • 3 The length of unemployment may be censored for three reasons: multiple treatment (the person is tak (...)

8After matching, we estimate the unemployment survival function for each group (treatment and control) according to the month of enrolment in training (month 1, month 2,..., month 18). The exit to employment considered is employment of at least one month. The length of unemployment measured begins after enrolment in training and is therefore the difference between the first day of the month of enrolment in training and the date of exit to employment or interruption of unemployment for another reason (referred to as the censure date3). Time spent in training is counted as unemployment. A lock-in effect occurs when the job seeker reduces his or her job search efforts due to a lack of time, or because he or she considers that the training will have a positive impact on his or her job search and therefore prefers to wait until after the training.

9In this study, the treatment effect is calculated as the difference between the unemployment survival curves of UJSs in the treatment and control groups after matching.

10Figure 1 (a to f) shows the unemployment survival curves for the treatment group (in blue) and the control group (in yellow) after matching, according to the time of enrolment in training (first month of unemployment, second month, and so on up to month 18).

11From the start of training and for several months (6-12 months), we observe the lock-in effect of training, as the unemployment survival curve for the treatment group (blue curve) is above that of the control group (yellow curve): as a result of training, the treatment group initially remains more unemployed. It is only after this lock-in effect that we observe a positive effect of training on access to employment, with the unemployment survival curve of the treatment group falling below that of the control group (blue and yellow invert in Figure 1). When training begins within three months of registering as unemployed, the positive effect of training on length of unemployment is insignificant (the blue confidence interval encompasses the yellow confidence interval). For people who enrolled in training between the 4th and 13th month after registering as unemployed, training has a positive and significant effect on exiting unemployment, with the unemployment survival function of the treatment group being lower than the survival function of the control group. Finally, for job seekers who enrol in skills training after 14, 17 and 18 months of unemployment, the impact of skills training is not significant, as the confidence intervals overlap once again. It should be noted that these results can be explained by the small size of the treatment group (between 63 and 112 people).

12The positive effect of training on job integration continues and even increases. In fact, the gap between the unemployment survival curves of the treatment and control groups remains stable and even increases in some cases. The positive effect also appears to be higher for those who were unemployed for longer before enrolling in training. The gap between the unemployment survival curves is greater for people who enrol in training in the 12th,13th and 15th month after registering as unemployed.

Figure 1a. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1a. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1b. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1b. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1c. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1c. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1d. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1d. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1e. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1e. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1f. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Figure 1f. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups

Reading: The first graph for the 1st month compares the unemployment survival curve of people who enrol in training during their 1st month of unemployment (blue) with people with the same characteristics who are not in training (yellow). At the end of the 10th month after enrolment in training (x-axis), the probability that people who have begun training are still unemployed becomes lower than those who have not.

2. Impact according to the personal characteristics of job seekers

13Are there differences according to gender, age, level of education and nationality in the impact of skills training on job integration?

14In order to assess the effect of training on job integration according to individual characteristics, we estimate the “treatment effect” for different sub-populations (i.e. whether the impact of training differs between these populations). This estimator is commonly referred to as the Conditional Average Treatment Effect (CATE - [Gerber and Green, 2012]). For example, for gender, the treatment effect is estimated separately for men and women. These analyses are primarily descriptive and are limited to the difference observed between several groups in the effect of skills training on job integration. They cannot be interpreted as a causal effect (of gender, for example) on the impact of skills training. However, even in the absence of a causal effect, these analyses remain interesting in order to identify the sub-groups in which skills training has a greater impact in terms of job integration.

15In this section, matching data for each of the 18 months following the start of unemployment are pooled in order to obtain the average impact of training on job integration, regardless of the time of enrolment in training [Fredriksson and Johansson, 2008]. Survival curves for the treatment and control groups were estimated based on these aggregated data. In this second analysis, the treatment group is made up of 2,759 individuals, with 153 job seekers from this group having been set aside following matching.

2.a. According to gender

16After the lock-in effect, skills training has a positive impact on job integration for both men and women (Figure 2), but to a greater extent for women. As a reminder, this difference is not a causal effect of gender, but rather the observation of a difference in treatment effect (descriptive statistics) between men and women.

Figure 2. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to gender

Figure 2. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to gender

2.b. According to age

17After the lock-in effect, we observe a positive and significant impact of training on access to employment for the 26-31, 32-39 and 40+ age groups (Figure 3). The impact of training on job integration is highest in the 32-39 age group. For young people in the 21-25 age group, the impact of skills training is low and insignificant. These results could be explained by the fact that when they leave school, young people register with Actiris in order to begin their qualifying period, even if they have short-term employment prospects. Therefore, in the control group, there are young people who will potentially move more quickly from unemployment to employment.

Figure 3. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to age groups (quartiles)

Figure 3. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to age groups (quartiles)

2.c. According to level of education

18After the lock-in effect, training has a positive impact on job integration for all levels of education (Figure 4). However, the positive impact is greater for UJSs with non-recognised foreign qualifications. It cannot be concluded that what is observed is solely the effect of the qualifications; nevertheless, there is a strong link between having foreign qualifications which are not recognised and the positive impact of skills training (Belgian and therefore recognised) on access to employment. Skills training therefore seems to play an important role in access to employment for people whose foreign qualifications are not recognised in Belgium.

Figure 4. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to level of education

Figure 4. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to level of education

2.d. According to nationality

19After the lock-in effect, the impact of training on access to employment is positive for people in all three nationality categories (Belgian, EU and non-EU) – Figure 5. It is for non-EU nationals that the benefits of skills training in terms of access to employment are greatest. These results are consistent with the greater impact for people with non-recognised foreign qualifications, who are often non-EU immigrants.

Figure 5. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to nationality

Figure 5. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to nationality

Conclusion

20The study which this fact sheet is based on highlights both the lock-in effect of skills training and the positive impact on access to employment after the training. It also shows that the positive effect on job integration remains present even 30 months after enrolling in training and is higher for people who were unemployed for longer before beginning the training. The positive impact of training is higher for women, long-term job seekers, job seekers with non-recognised foreign qualifications and non-EU nationals. These results confirm the importance of skills training in public employment policies. Continuity of training pathways is a key factor in allowing a maximum number of people in their qualifying period to access employment.

I would like to thank Isabelle Sirdey from the Service Études et Statistiques at Bruxelles Formation, and Jérôme François from view.brussels, for their advice and support throughout the project. I would also like to thank the teams from the Service Études et Statistiques at Bruxelles Formation and view.brussels for their proofreading and translation, as well as Maritza López Novella and Antoine Dewatripont from the Federal Planning Bureau for their invaluable methodological advice. Finally, I am grateful to the participants in the Belgian Day for Labour Economists 2022 for their constructive comments.

Haut de page

Bibliographie

DELUNA, Xavier and JOHANSSON, Per, 2007. Matching estimators for the effect of a treatment on survival times. Uppsala: Institute for Labour Market Policy Evaluation (IFAU). Working Paper.

FREDRIKSSON, Peter and JOHANSSON, Per, 2008. Dynamic treatment assignment: the consequences for evaluations using observational data. In: Journal of Business & Economic Statistics. 2008. vol. 26, no 4, pp. 435-445.

GERBER, Alan S. and GREEN, Donald P. 2012. Field experiments: Design, analysis, and interpretation. New York: W. W. Norton.

Haut de page

Notes

1 https://www.bruxellesformation.brussels/wp-content/uploads/2023/03/Analyse-de-limpact-de-la-formation-qualifiante-sur-linsertion-professionnelle.pdf

2 In our analyses, we use the Kaplan-Meier estimator, which provides a nonparametric estimate of the unemployment survival function.

3 The length of unemployment may be censored for three reasons: multiple treatment (the person is taking a second skills training course), deregistration from unemployment for a reason other than exit to employment (known only at the end of the month) and the end of the monitoring period (after 4 years the person is no longer monitored).

Haut de page

Table des illustrations

Titre Figure 1a. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-1.jpg
Fichier image/jpeg, 138k
Titre Figure 1b. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-2.jpg
Fichier image/jpeg, 176k
Titre Figure 1c. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-3.jpg
Fichier image/jpeg, 131k
Titre Figure 1d. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-4.jpg
Fichier image/jpeg, 130k
Titre Figure 1e. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-5.jpg
Fichier image/jpeg, 132k
Titre Figure 1f. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) according to the moment of enrolment in training for the treatment (training) and control (no training) groups
Légende Reading: The first graph for the 1st month compares the unemployment survival curve of people who enrol in training during their 1st month of unemployment (blue) with people with the same characteristics who are not in training (yellow). At the end of the 10th month after enrolment in training (x-axis), the probability that people who have begun training are still unemployed becomes lower than those who have not.
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-6.jpg
Fichier image/jpeg, 126k
Titre Figure 2. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to gender
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-7.jpg
Fichier image/jpeg, 114k
Titre Figure 3. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to age groups (quartiles)
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-8.jpg
Fichier image/jpeg, 228k
Titre Figure 4. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to level of education
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-9.jpg
Fichier image/jpeg, 176k
Titre Figure 5. Kaplan Meier unemployment survival curves (time 0 corresponds to enrolment in training) for treatment (training) and control (no training) groups according to nationality
URL http://journals.openedition.org/brussels/docannexe/image/8727/img-10.jpg
Fichier image/jpeg, 185k
Haut de page

Pour citer cet article

Référence électronique

Catherine Smith, « The impact of skills training on job integration for job seekers in Brussels »Brussels Studies [En ligne], Fact Sheets, document 208, mis en ligne le 15 octobre 2025, consulté le 13 novembre 2025. URL : http://journals.openedition.org/brussels/8727 ; DOI : https://doi.org/10.4000/14yca

Haut de page

Auteur

Catherine Smith

Catherine Smith was a researcher in labour economics at IRES, the UCL economics research centre, from 2010 to 2014. She then worked as a researcher and data analyst at Bruxelles Formation, from 2014 to 2025. She is currently a data analyst at Belfius.
ses[at]bruxellesformation.brussels

Haut de page

Droits d’auteur

CC-BY-4.0

Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search